The Second Column Secunda of Origens Hex

COPYRIGHT
BY
BENJAMIN PAUL KANTOR

2017
THE DISSERTATION COMMITTEE FOR BENJAMIN PAUL KANTOR CERTIFIES THAT THIS IS THE
APPROVED VERSION OF THE FOLLOWING DISSERTATION:
THE SECOND COLUMN (SECUNDA) OF ORIGEN'S HEXAPLA

IN LIGHT OF GREEK PRONUNCIATION
COMMITTEE:
NA'AMA PAT-EL, SUPERVISOR
JOHN HUEHNERGARD, CO-SUPERVISOR
JO ANN HACKETT
GEOFFREY KHAN
BY
BENJAMIN PAUL KANTOR
DISSERTATION
PRESENTED TO THE FACULTY OF THE GRADUATE SCHOOL OF
THE UNIVERSITY OF TEXAS AT AUSTIN
IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS
FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
THE UNIVERSITY OF TEXAS AT AUSTIN

DECEMBER 2017
ACKNOWLEDGEMENTS
First of all, I would like to thank my advisor John Huehnergard. Much of who I am
academically has been shaped by the many hours of time Prof. Huehnergard has invested in
me, whether that be through teaching me various Semitic languages, meeting with me to
discuss the many issues of historical Hebrew linguistics, or the continuous help he has been
to me on this dissertation. He has been a model of both scholarly excellence and humility and
I truly regard it as a privilege to have been one of his final students.
I would also like to thank Jo Ann Hackett and Na'ama Pat-El, who, in addition to their
helpful comments on this dissertation, have also invested countless hours in my education.
Prof. Hackett has always encouraged me to think thoroughly through issues I am researching.
I have been challenged by the high bar she sets for biblical scholarship and I have fond
memories of discussions she led on points of Hebrew grammar in the "Turbo Hebrew"
seminar at UT. I have learned a great deal from Prof. Pat-El with respect to syntax in the
Semitic languages, but I think what I will remember most that she imparted to me is a sense
for careful linguistic argumentation. I also owe much gratitude to Jonathan Kaplan, who has
served as an exemplary model of biblical scholarship and has been a help in many other
practical ways as well. Throughout all my years at UT, these four professors have been a
constant support. They have all demonstrated care for me both academically and personally
and it has been a pleasure to know them. Finally, I would like to thank Geoffrey Khan, my
external reader, who generously agreed to be on my committee and has provided me with
helpful comments and discussion on various points of my work. He has had a great influence
on my approach to the Secunda as a Hebrew reading tradition and I am indebted to him both
for his excellent scholarship and for his specific help with the dissertation.
I also owe a special thanks to the Biblioteca Ambrosiana in Milan, Italy and its
director Federico Gallo for their generosity in allowing me to come and examine the Hexapla
palimpsest in September of 2016. I would also like to thank Stefano Serventi, Trifone
Cellamaro, and Ferdinando Righetto for their warm welcome, utmost kindness, and readiness
to help daily as I examined the manuscript in the reading room. The generosity of everyone at
the Biblioteca to give me such an opportunity has contributed greatly to my dissertation
- iv -
research. Finally, I thank Todd Hanneken and his team for the outstanding work they have
done spectral imaging the palimpsest and giving me access to and discussing the results.
I am also grateful to the UT Austin professors in Arabic and Modern Hebrew, with
whom I have studied and worked. I would especially like to thank Kristen Brustad, who
introduced me to the world of the Arabic dialects, sociolinguistics, and language ideology.
The lessons I have learned from her regarding how to think about and analyze language have
been a perennial help. Likewise, I thank the Middle Eastern Studies department as a whole
for giving me the opportunity to be involved in building and teaching an intensive Biblical
Hebrew course for three years, which served to deeply increase my knowledge of Hebrew.
Discussing linguistic and other matters with my fellow students in the Hebrew Bible
program, for whom I am grateful, has also been a favorite and much appreciated pastime of
mine. I especially thank Philip Zhakevich, a great friend and predecessor in the program, who
has continued to be an encouragement and help in many ways. David Justh, the graduate
program coordinator for most of my time at UT, was a continuous help in all matters relating
to the university. I also owe a great deal of thanks to my professors at the Hebrew University
of Jerusalem, especially in the Classics Department. They invested many hours to train me in
Greek and Latin and gave me a top-notch education. Finally, I would like to thank Randall
Buth (and all those) from the Biblical Language Center, from whom I have learned so much
and with whom it has been a joy to work, teaching Hebrew or Greek together in various and
diverse parts of the world.
I would like to thank my community in Austin and my family near and far. From my
very first weeks in Austin, my church has been a continuous source of support and help in
any possible difficulty I have faced. Most of all, I thank my family. I thank my parents, David
and Sharon, for all the care and support they have given me throughout my life and especially
for supporting my moving to Israel just weeks before my eighteenth birthday for studies. I
thank my brother Jeremy and my sister Andrea for their support. Finally, I would like to
thank my wife Olivia, who has been a loving, supportive, and helpful wife through many
long days of dissertation writing all while expecting our first child. Moreover, no matter how
long I worked, unlike the wife of the seventeenth-century Chrysostom scholar Henry Savile,
she never threatened to burn my manuscripts (see note 332).
-v-
BEǊAMIN PAUL KANTOR, PH.D.
THE UNIVERSITY OF TEXAS AT AUSTIN, 2017
SUPERVISOR: NA'AMA PAT-EL
CO-SUPERVISOR: JOHN HUEHNERGARD
This dissertation addresses the phonology and orthography of the second column
(Secunda) of Origen's (185–254 CE) Hexapla, which constitutes a Greek transcription of
Biblical Hebrew. The transcription text is analyzed in light of its Hellenistic/Roman Near
Eastern background, the phonology and orthography of Roman Palestinian Koine Greek, and
roughly contemporary Greek transcription conventions for other languages.
Aside from the brief introduction (chapter 1) and conclusion (chapter 7), this
dissertation is comprised of five substantial chapters. Chapters 2 and 3 address the historical
and social background of the text of the Secunda. In chapter 2, I argue that Origen did not
have enough Hebrew knowledge to compose the text himself. In chapter 3, on the basis of
comparative evidence from the Hellenistic, Roman, and Byzantine Near East, I argue that the
Secunda originated among the Jewish scholarly community of Caesarea as a didactic aid in
the second or third century CE. Chapters 4 and 5 address the linguistic background of the text
of the Secunda. Chapter 4, based on a thorough analysis of the epigraphic evidence from
ancient Palestine, provides a reconstruction of contemporary Greek pronunciation. Chapter 5,
based on a linguistic analysis of comparative transcription material, surveys typical Greek
transcription conventions from roughly the same period. Chapter 6 applies the data from the
previous sections to the Hebrew vocalization tradition reflected in the text of the Secunda,
addressing the phonemic and phonetic value of the consonants, vowels, and shewa as well as
the syllable structure. Methodologically, the phonology and orthography of Secunda Hebrew
are approached from the perspective of historical (Hebrew) linguistics, Greek pronunciation
and orthography, linguistic studies on cross-language perception, and moraic phonology.
- vi -
TABLE OF CONTENTS
List of Charts and Tables .......................................................................................................ix
List of Figures...........................................................................................................................x
Citation of Primary Texts and Citation Conventions ..........................................................xi
Abbreviations and Symbols .................................................................................................xiii
1. Introduction ........................................................................................................................1
1.1. Introduction .....................................................................................................................1
1.2. Previous Research ...........................................................................................................2
1.3. Methodology ...................................................................................................................4
1.3.1. Historical (Hebrew) Linguistics .......................................................................................................5
1.3.2. Greek Pronunciation and Orthography ............................................................................................5
1.3.3. Cross-Language Perception in Modern Linguistics .........................................................................6
1.3.4. Moraic Phonology ............................................................................................................................9
2. Origen and the Secunda...................................................................................................10
2.1. Introduction ...................................................................................................................10
2.2. Origen's Knowledge of Hebrew ....................................................................................11
2.2.1. The Limits of Origen's Hebrew Knowledge ..................................................................................12
2.2.2. Nature and Extent of Origen's Hebrew Knowledge .......................................................................23
2.2.3. Origen's Use of the Secunda and the Purpose of Including It in the Hexapla................................32
2.2.4. Conclusions: A Portrait of Origen the Hebrew Scholar .................................................................36
2.3. The Secunda and the Compositional History of the Hexapla .......................................38
2.3.1. Origen and the Secunda in Caesarea ..............................................................................................38
2.3.2. Did Origen Commission the Second Column? ..............................................................................38
2.3.3. Early Church Fathers on the Hexapla ............................................................................................40
2.3.4. The Second Column as a Parallel Text...........................................................................................45
2.3.5. Conclusions ....................................................................................................................................46
3. The Secunda in the Hellenistic/Roman Near East.........................................................48
3.1. Introduction ...................................................................................................................48
3.2. Survey of Scholarship ...................................................................................................48
3.2.1. Previous Suggestions......................................................................................................................48
3.2.2. Evaluation.......................................................................................................................................50
3.3. Date of the Secunda: terminus post quem .....................................................................53
3.3.1. Introduction ....................................................................................................................................53
3.3.2. Merger of /ḫ /,/h ̣/ > /h ̣/ and /ġ /,/ˁ/ > /ˁ/ .............................................................................................53
3.4. Sitz im Leben of the Secunda........................................................................................57
3.4.1. Comparative Material.....................................................................................................................57
3.4.2. The Linguistic Situation in Roman Palestine .................................................................................75
3.4.3. Learning Hebrew in Tannaitic/Amoraic Palestine .........................................................................78
3.4.4. Conclusions: The Original Context of the Secunda .......................................................................84
4. Pronunciation of Greek in Roman Palestine .................................................................97
4.1. Introduction ...................................................................................................................97
4.2. Methodology .................................................................................................................97
4.3. Preliminary Phonemic System ......................................................................................99
4.4. Egyptian Koine Greek.................................................................................................100
4.4.1. Vowels ..........................................................................................................................................100
4.4.2. Consonants ...................................................................................................................................108
4.5. Palestinian Koine Greek..............................................................................................110
4.5.1. Introductory Remarks...................................................................................................................110
4.5.2. Orthographic Data ........................................................................................................................110
4.5.3. Analysis ........................................................................................................................................110
4.5.4. Summary ......................................................................................................................................126
4.6. Conclusion ..................................................................................................................131
5. Greek Transcription Conventions in the Hellenistic/Roman/Byzantine Near East .132
5.1. Introduction .................................................................................................................132
5.2. Methodology ...............................................................................................................132
5.3. Linguistic and Orthographic Analysis: Greek Transcription ......................................133
- vii -
5.3.1. Latin .............................................................................................................................................133
5.3.2. Akkadian.......................................................................................................................................146
5.3.3. Arabic ...........................................................................................................................................158
5.3.4. Aramaic ........................................................................................................................................166
5.3.5. Phoenician-Punic..........................................................................................................................168
5.3.6. Summary ......................................................................................................................................172
5.4. Linguistic and Orthographic Analysis: Greek in Hebrew ...........................................175
5.4.1. Greek Loanwords in the Mishnah ................................................................................................175
5.5. Concluding Remarks...................................................................................................181
6. The Phonology and Orthography of the Secunda .......................................................182
6.1. Introduction .................................................................................................................182
6.2. Methodology ...............................................................................................................182
6.3. Consonants ..................................................................................................................183
6.3.1. Stops (/b/, /p/, /d/, /t/, /g/, /k/).......................................................................................................183
6.3.2. Sibilants (/s/, /š /, /ś /, /z/)...............................................................................................................200
6.3.3. Emphatic Consonants (/q/, /t/,̣ /s ̣/)................................................................................................205
6.3.4. Nasals (/m/, /n/) ............................................................................................................................215
6.3.5. Liquids (/l/, /r/) .............................................................................................................................220
6.3.6. Gutturals (/ʕ/, /h ̣/, /ʔ/, /h/) .............................................................................................................225
6.3.7. Semi-Vowels (/w/, /y/)..................................................................................................................228
6.3.8. Consonant Gemination (Cː or CC)...............................................................................................237
6.3.9. Summary ......................................................................................................................................248
6.4. Vowels .........................................................................................................................250
6.4.1. Length and Quality: /ē / = η and /ō / = ω .......................................................................................251
6.4.2. Lax /e/ (< */i/) and /o/ (< */u/), Tense /ı̄ / and /ū /..........................................................................264
6.4.3. Long /ɔ̄/ [ɔː] and Short /a/ [a]/[æ].................................................................................................274
6.4.4. Potential Phonemes: /ɛ̄/ and /ε/.....................................................................................................294
6.4.5. Etymological Long /ı̄ / = Greek ι, ει, and η...................................................................................301
6.4.6. Summary ......................................................................................................................................310
6.5. Shewa and Syllable Structure .......................................................................................310
6.5.1. Shewa............................................................................................................................................311
6.5.1.1. The Concept of Schwa/Shewa: Linguistics and the Hebrew Reading Traditions.................312
6.5.1.2. Review of Scholarship ..........................................................................................................313
6.5.1.3. The Phonetic Realization of Vocalic Shewa .........................................................................315
6.5.1.4. Complex Onsets....................................................................................................................326
6.5.1.5. Vowel Syncope and Consonant Clusters: CvCvCv(C) > CvCCv(C) ...................................339
6.5.1.6. The Conjunction waw and the Inseparable Prepositions ......................................................346
6.5.2. Final Consonant Clusters .............................................................................................................354
6.5.2.1. Segholates (*qVtl).................................................................................................................354
6.5.2.2. Verbal and Participial Forms.................................................................................................357
6.5.2.3. Concluding Remarks: Dialectal Variation ............................................................................358
7. Conclusion.......................................................................................................................360
8. Appendix A: Palestinan Greek Spelling Interchanges................................................363
9. Bibliography ...................................................................................................................386
- viii -
LIST OF CHARTS AND TABLES
Chart 1: Preliminary Vocalic Phonemes: Attic Greek Vowels ...............................................99
Chart 2: Preliminary Consonantal Phonemes: Attic Greek Consonants................................100
Chart 3: Egyptian Koine Greek Vowels: Grapheme-Phoneme Correspondences by Period .107
Chart 4: Egyptian Koine Greek Vowels: Orthographic/Spelling Interchanges by Period .....107
Chart 5: Egyptian Koine Greek Consonants (Classical to Byzantine Period) .......................110
Chart 6: Palestinian Koine Greek Vowels (Roman Period) ...................................................127
Chart 7: Palestinian Koine Greek Vocalic Orthography in (Roman Period) .........................127
Chart 8: Palestinian Koine Greek Diphthongs (Roman and Byzantine Periods)...................128
Chart 9: Palestinian Koine Greek Diphthongal Orthography (Roman/Byzantine Periods) ..128
Chart 10: Palestinian Koine Greek Consonants (Roman Period) ..........................................130
Chart 11: Palestinian Koine Greek Consonantal Orthography (Roman Period) ...................131
Chart 12: Summary of Latin in Greek Transcription: Correspondences ...............................146
Chart 13: Summary of Akkadian in Greek Transcription: Correspondences ........................158
Chart 14: Summary of Arabic in Greek Transcription: Correspondences .............................165
Chart 15: Summary of Aramaic in Greek Transcription: Correspondences ..........................167
Chart 16: Summary of Phoenician-Punic in Greek Transcription: Correspondences............172
Chart 17: Summary of Latin and Semitic in Greek Transcription: Correspondences ...........175
Chart 18: Summary of Greek Loanwords in Mishnaic Hebrew: Correspondences...............181
Chart 19: Consonantal Phonology, Phonetics, and Orthography in the Secunda ..................250
Chart 20: Vocalic Phonemic Inventory of the Secunda .........................................................250
Chart 21: Vocalic Phonemic Oppositions in the Secunda......................................................251
Chart 22: Representation of Long /ı̄ / as ει in the Secunda.....................................................305
Chart 23: Phonetic Realization of the Vocalic Phonemes in the Secunda .............................310
Chart 24: Vocalic Phonology, Phonetics, and Orthography in the Secunda ..........................310
Chart 25: Sonorants and Sibilants in Complex Onsets in the Secunda..................................327
Chart 26: Greek Representation of the Piˁel and Qal 3ms Verbal Prefix in the Secunda ......335
Chart 27: Epenthesis in Segholate (*qVtl) Forms in the Secunda.........................................356
Chart 28: Various Greek Transcriptions of the Segholate Letter Names ...............................358
- ix -
LIST OF FIGURES
Figure 1: Examples of Heavy and Light Syllables in Moraic Phonology .............................9
Figure 2: Example of Word in Moraic Phonology.................................................................9
Figure 3: 'male and female' in the Hexapla............................................................................22
Figure 4: 'the virgin' in the Hexapla.......................................................................................23
Figure 5: Greco-Latin Columnar Glossary (P. Oxy. LXXVIII.5162) ....................................58
Figure 6: Greco-Latin Columnar Glossary of Vergil's Aeneid (PSI VII 756)........................59
Figure 7: Greek-Greek Glossary for Homer's Iliad (P.Oxy. XXIV.2405) ..............................59
Figure 8: Greco-Latin Columnar Translation of Vergil's Aeneid (P.Ryl. III.478)..................60
Figure 9: Greco-Latin Same Column Translation of Vergil's Aeneid (Palimps. Ambros.)....61
Figure 10: Latin Abecedary in Greek Transcription (O.Max. inv. 356) ................................62
Figure 11: Greek-Demotic Columnar Glossary (P. Heid. Inv.-Nr. G 414) ............................63
Figure 12: Latin-Greek-Coptic Trilingual Colloquium (P.Berol. inv. 10582) .......................64
Figure 13: Supralinear Greek Transcription in the Demotic Magical Papyrus of London....65
Figure 14: Sumerian-Akkadian Lexical Text with Greek Transcription (BM 34797)...........66
Figure 15: Latin Receipt of Slave Trader in Greek Transcription (SB III.I.6304).................68
Figure 16: Egyptian Dating Formulary of Pharaoh Hor-Wenefer in Greek Transcription ....69
Figure 17: Old-Coptic in Greek Transcription (Schmidt Papyrus)........................................69
Figure 18: Beth She'arim Funerary Inscription (no. 91) with Transcribed Tag .....................70
Figure 19: Aramaic in Greek Transcription (Dura-Europos Inscription)...............................71
Figure 20: Moraic Representation of µερεσθ ........................................................................223
Figure 21: Moraic Representation of ηλ, δερχ, and λεβ ........................................................246
Figure 22: Moraic Representation of εθνηου and φελλετηνι.................................................247
Figure 23: Moraic Representation of ιερχθη..........................................................................329
Figure 24: Moraic Representation of βνη ..............................................................................331
Figure 25: Moraic Representation of ϊκερσου and ιεσεµου...................................................332
Figure 26: Moraic Representation of αρσαειµ.......................................................................332
Figure 27: Moraic Representation of ηζχορ...........................................................................333
Figure 28: Moraic Representation of ιδαββερ and ιµιν .........................................................336
Figure 29: Moraic Representation of ασσωµριµ....................................................................344
-x-
CITATION OF PRIMARY TEXTS AND CITATION CONVENTIONS
CITATIONS OF THE SECUNDA
Citations of the Secunda are based on my personal examination of the Ambrosiana palimpsest
(O 39 sup.) in consultation with the readings of MERCATI (1958; 1965) and YUDITSKY (2017).
Accent and breathing marks have generally been omitted, but trema (¨) on iota (i.e., ϊ) has
been retained in the representation of the Secunda transcriptions. A transcribed word is
followed by one asterisk (*) when it indicates a reasonably hypothesized emendation, in
many instances following YUDITSKY (2017). For example, we may reasonably suppose that
paleographically similar δ (Δ) was mistaken for λ (Λ) in transmission in the transcription
φαλιθ ‫' פָּדִ יתָ ה‬you redeemed' (Ps. 31:6). Accordingly, it is emended to φαδιθ and represented
with an asterisk: φαδιθ*. Forms with two asterisks ** indicate unattested or impossible
forms. Forms with three asterisks *** indicate reasonably hypothesized but unattested forms
(see 6.4).
I occasionally cite quotations of the Secunda not found in the Ambrosiana palimpsest.
These quotations are often found in the early church fathers' writings or in marginal notes on
manuscripts of the Septuagint (LXX). Like YUDITSKY, I refer to these as being found in
"external sources" ("‫ "מקורות חיצוניים‬in Hebrew) (2017, 1–2, 108). Most of these Secunda
quotations from external sources are found in FIELD (1875), HATCH and REDPATH (1897, vol. 3,
199–216), BRÖNNO (1943), MURTONEN (1988, vol. I/Ba), or YUDITSKY (2017), but I have added
a number of Secunda quotations found in the early church fathers as a result of my own
searches in the TLG database (see note below on TLG). I also once cite an attestation of the
Secunda from the highly fragmentary Cairo Genizah palimpsest (TAYLOR 1900; see 6.4.5.6).
CITATIONS OF ANCIENT GREEK AND LATIN AUTHORS
Most citations of ancient Greek authors, such as the early church fathers, are from the
Thesaurus Linguae Graecae® Digital Library (http://www.tlg.uci.edu) (TLG). Usually, TLG
follows the text of MIGNE's Patrologia Graeca (MPG). When an ancient Greek work is cited
with two or three numbers separated with a period (e.g., Selecta in Genesim [12.100.23]), the
first indicates the volume, the second the page, and the third the line in MPG. Occasionally,
other sources have been used such as SAVILE (1611) for Chrysostom, Die griechischen
- xi -
christlichen Schriftsteller (GCS) for Eusebius's church history (MOMMSEN 1908),
MOUTSOULAS (1973) for Epiphanius, and HOFFMAN (2007) for Nikolaos of Otranto. Ancient
Latin authors, such as Jerome, are typically cited from MIGNE's Patrologia Latina (MPL),
whose system of citation follows that of MPG. Finally, other sources are occasionally used,
such as or Corpus Scriptorum Ecclesiasticorum Latinorum (CSEL) for Jerome's letters
(HILBERG 1910).
CITATIONS OF ANCIENT PAPYRI
The full bibliographical information for each of the papyri, typically expressed in abbreviated
form (e.g., P.Berol.21246, P.Oxy. XLVI.3315, P.Lund I.5), may be found at Papyri.info (http:/
/www.papyri.info), Trismegistos (http://www.trismegistos.org), or POxy: Oxyrhynchus
Online (http://www.papyrology.ox.ac.uk/POxy). Note that names of papyri in italics signify
that they belong to a multi-volume collection.
OTHER ONLINE/ELECTRONIC TOOLS UTILIZED
A number of other online and electronic tools (not mentioned above) have been utilized for
this dissertation. For ancient rabbinic texts, I have made use of the Sefaria online database
(https://www.sefaria.org), which includes the William Davidson Talmud. For Hebrew and
Aramaic lexica, I have made use of the electronic edition of Brown-Driver-Briggs Hebrew
and English Lexicon (BDB), JASTROW's Dictionary of the Targumim, the Talmud Babli and
Yerushalmi, and the Midrashic Literature (1926) (http://www.tyndalearchive.com/tabs/
jastrow), and the Comprehensive Aramaic Lexicon (CAL) (http://cal1.cn.huc.edu). Finally, a
number of Hebrew, Aramaic, and Greek texts have been accessed through Accordance Bible
Software, including the Göttingen LXX critical edition of Psalmi cum Odis (RAHLFS 2008).
EMPHASIS (BOLD OR ITALICS) IN QUOTATIONS
Note that any bold or italic font in a quotation is my own emphasis and not present in the
original, except perhaps in those cases (in modern scholarship) in which it is used in a
conventional manner (italic for Latin words, italic for transliteration of foreign words, etc.).
- xii -
ABBREVIATIONS AND SYMBOLS
BIBLIOGRAPHICAL
A Aleppo Codex
b Babylonian Talmud
BDB BROWN, FRANCIS, S. R. DRIVER, AND CHARLES A. BRIGGS. 2008. The Brown-
Driver-Briggs Hebrew and English Lexicon. Peabody, MA.: Hendrickson
Publishers.
BM British Museum tablet number
CAL Comprehensive Aramaic Lexicon Project (http://cal1.cn.huc.edu)
CIIP AMELING, WALTER, HANNAH M. COTTON, LEAH DI SEGNI, WERNER ECK, BEN-
JAMIN ISAAC, ALLA KUSHNIR-STEIN, HAGGAI MISGAV, JONATHAN J. PRICE, IS-
RAEL ROLL, ADA YARDENI, MARFA HEIMBACH, and NAOMI SCHNEIDER, eds.
2010–. Corpus Inscriptionum Iudaeae/Palaestinae: A Multi-Lingual Cor-
pus of the Inscriptions from Alexander to Muhammad. Berlin/Boston: De
Gruyter.
CSEL Corpus Scriptorum Ecclesiasticorum Latinorum (http://csel.sbg.ac.at/en/)
EHLL KHAN, GEOFFREY, ed. 2013. Encyclopedia of Hebrew Language and Lin-
guistics. Leiden/Boston: Brill.
G-JUDEAN-T The Greek Judaean Desert Manuscripts electronic database accessed
through Accordance Bible Software corresponding to the Greek section of
ABEGG, MARTIN, JAMES BOWLEY, and EDWARD COOK. 2015. The Dead Sea
Scrolls Concordance. Leiden/Boston: Brill.
GCS 1897–. Die griechischen christlichen Schriftsteller der ersten [drei]
Jahrhunderte. Leipzig.
HSM Harvard Semitic Museum
KAI DONNER, HERBERT, and WOLFGANG RÖLLIG, eds. 1971. Kanaanäische und
aramäische Inschriften. Wiesbaden: Harrassowitz.
L Leningrad Codex
LXX Septuagint
m Mishnah
MPG MIGNE, J. P., ed. 1857–. Patrologiae Cursus Completus: Series Graeca.
Paris: Migne.
MPL MIGNE, J. P., ed. 1844–. Patrologiae Cursus Completus: Series Latina.
Paris: Migne.
MT Masoretic Text
t Tosefta
- xiii -
TLG Thesaurus Linguae Graecae® (http://www.tlg.uci.edu)
VAT signature of tablets located in the museums of Berlin
y Jerusalem Talmud
OTHER
AE American English
ASCII American Standard Code for Information Interchange (i.e., the numerical value
of a character/text)
Bab. Babylonian Hebrew Tradition
CG Category-Goodness Difference
cl. classical
cod. codex
F1 first formant
F2 second formant
fem. feminine
fs feminine singular
IPA International Phonetic Alphabet
L1 first/native language
L2 second language
masc. masculine
mp masculine plural
ms masculine singular
MS manuscript
MSS manuscripts
NENA Northeastern Neo-Aramaic
OCP Obligatory Contour Principle
Pal. Palestinian Hebrew Tradition
PAM Perceptual Assimilation Model
PAM-L2 Perceptual Assimilation Model for learners of a second language
PNWS Proto-Northwest Semitic
r recto
SSBE Southern Standard British English
- xiv -
SBH Standard Biblical Hebrew
SC Single-Category Assimilation
SCL Syllable Contact Law
SLM Speech Learning Model
SSP Sonority Sequencing Principle
TC Two-Category Assimilation
Tib. Tiberian Hebrew Tradition
UU Both Uncategorizable
v verso
* precedes etymological or reconstructed form (but note citation of Secunda above)
** precedes unattested or impossible form (but note citation of Secunda above)
*** follows reconstructed Greek transcription form (note citation of Secunda above)
// encloses phonemic transcription of words or phonemes
[] encloses phonetic transcription of words or phones
(*)X > Y X developed into Y
X < (*)Y X developed from Y
µ represents a mora in phonological trees
σ represents the syllable node in phonological trees
ω represents the word node in phonological trees
- xv -
1. INTRODUCTION
1.1. INTRODUCTION
The second column (Secunda) of Origen's (185–254 CE) Hexapla, which contains Biblical
Hebrew transcribed into Greek letters, constitutes the best direct evidence we have for an-
cient Hebrew pronunciation, inasmuch as it is the earliest vocalized Hebrew of any signifi-
cant length. Since the discovery of the Ambrosiana palimpsest (O 39 sup.) at the end of the
nineteenth century, the Secunda has been incorporated into much research on historical He-
brew phonology and has been the focus of several larger works (SPERBER 1925–1934; BRØNNO
1943; JANSSENS 1982; and YUDITSKY 2017).
However, despite the attention the Secunda has received, there are a few areas in need
of further research. First, a firm consensus is lacking regarding the original date and social
setting of the Secunda. Second, while cursory references to ancient Greek pronunciation are
found in various treatments of the Secunda, none of the previous works have engaged in an
in-depth study of contemporary Greek phonology and orthography. Third, a phonetic
transcription of the pronunciation of the Secunda has yet to be produced.
The present dissertation seeks to address these issues by giving greater attention to the
social context of comparable texts in the Hellenistic/Roman Near East, the phonology and or-
thography of Palestinian Koine Greek, the conventions of Greek transcription of other lan-
-1-
guages in the Hellenistic/Roman Near East, and the relevance of modern linguistic studies on
cross-language perception. Primarily, this dissertation addresses the phonology and orthogra-
phy of the Secunda in light of the pronunciation and orthography of Palestinian Koine Greek.
1.2. PREVIOUS RESEARCH
For a long time, the Hexapla only survived as references in other sources. These citations
were collected and published by FIELD in a two volume work, Origenis Hexaplorum quae
supersunt, in 1875. However, within this massive two-volume work, citations of the Secunda
are few and far between. It was not until 1894 that MERCATI discovered a palimpsest
containing eleven mostly-fragmentary Psalms (18, 28, 29, 30, 31, 32, 35, 36, 46, 49, 89). Not
long after, a number of scholars worked on the material.
MARGOLIS used the Greek transcriptions of Origen to treat the pronunciation of shewa
(1905). SPEISER wrote a series of articles in The Jewish Quarterly Review describing and
analyzing various phonological issues in the Secunda (1925–1934). A few more contributions
to research on the Secunda followed. PRETZL and STAPLES published short articles on the
Secunda in 1932 and 1939, respectively. Sperber's 1938 work, "Hebrew Based upon Greek
and Latin Transliterations," which is a collection of the Greek and Latin transcription
material, contains the Greek transcriptions found in Origen's Hexapla. A summary of this
early period of research is presented nicely in YUDITSKY (2013, 803–804; 2017, 6–7).
All this led up to the publication of what is still regarded as the most comprehensive
work published on the Secunda to date. In 1943, BRØNNO published Studien über hebräische
Morphologie und Vokalismus: auf Grundlage der Mercatischen Fragmente der zweiten
Kolumne der Hexapla des Origenes. To describe the work as "utterly comprehensive" is by
no means an exaggeration, in that he covers every form present in MERCATI's manuscript. His
treatment is well-organized and very systematic. He classifies all the forms into appropriate
grammatical, morphological, and phonological categories, while discussing and explaining
unusual forms. He essentially analyzes the data with respect to two reference points, the
-2-
historical etymological form of a word and the Tiberian form of a word, the former playing
only a minor part in the work and the latter being far more prominent. The historical
linguistics in the book is relatively sound, though it cannot help but be a product of its time.
Moreover, treating the Hexaplaric material from the perspective of how it parallels Tiberian
Hebrew can be a problematic methodology. Although it may make the Hexaplaric material
more approachable, it encourages a tendency to see Secunda Hebrew through the lens of
Tiberian Hebrew and thus creates confusion even in historical-linguistic matters—a
shortcoming pointed out by both JANSSENS and YUDITSKY.
The next monograph to be published on the Secunda was JANSSENS's Studies in
Hebrew Historical Linguistics Based on Origen's Secunda (1982). JANSSENS specifically
contrasts his methodology with that of BRØNNO. Though both attempt to provide sound rules
to explain the Secunda vocalization, JANSSENS criticizes the fact that BRØNNO approaches the
material statistically, constantly comparing the Secunda to Tiberian Hebrew. JANSSENS, on the
other hand, treats the material from a historical-linguistic perspective, attempting to delineate
consistent sound rules to trace the development from the etymological (proto-Semitic) form
to the Secunda form. While JANSSENS's work marked an attempt at analyzing the Hexaplaric
material with a more historically-grounded linguistic approach, the historical linguistics in
the book is not sound (for a more comprehensive history of research, see BRØNNO 1943, 1–14;
JANSSENS 1982, 25–36; YUDITSKY 2013, 803–804; 2017, 5–14).
The most recent scholar to work on the material is YUDITSKY, who published a series
of articles from 2005–2016 and has recently followed them up by publishing the culmination
of his work in a monograph in 2017. In this work, he covers the phonology and morphology
of the Secunda comprehensively, dealing with all the forms in the palimpsest and numerous
quotations form external sources. Methodologically, YUDITSKY emphasizes the importance of
treating the Secunda as a Hebrew tradition in its own right and not relying on conformity to
other traditions such as Tiberian. Accordingly, he first analyzes the Secunda by itself and only
-3-
subsequently compares it to the base of Hebrew shared across the various other traditions
such as Tiberian, Babylonian, Palestinian, and Samaritan (2017, 13–14). His work should be
commended and is currently the best treatment of the Secunda material available.
There are, however, three points to be made. First, while his methodology is sound,
the emphasis on treating the Secunda as a tradition in its own right sometimes leads to too
sharp a distinction between the Hebrew of the Secunda and the other traditions. We must
remember that the main Hebrew reading traditions have ancient roots and certain features,
like shewa, are probably quite old (see 6.5). Second, while YUDITSKY does address the
phonology and orthography of Greek in his book (2017, 46), there are two weaknesses in his
approach. The first weakness is that, on the basis of the conservatism of writing, he assumes
that the orthography of the Secunda reflects a pronunciation hundreds of years older than its
composition. However, orthography is only conservative when there is an established
spelling tradition. Transcription, by nature, is far more indicative of current pronunciation.
The second weakness is that the comments YUDITSKY makes about Greek are based on a
general overview of Greek pronunciation and not specific to the Greek spoken in Palestine.
Both of these weaknesses lead to inaccuracies. Third, his assumption that one letter can only
represent one sound in the Secunda, while convenient, is not necessarily the most nuanced
approach for analyzing the transcriptions. Moreover, a comparison with transcription
conventions of other languages into Greek indicates that such an assumption is unfounded.
The work of all of these scholars, especially that of YUDITSKY, will be addressed in
greater detail in the body of the dissertation.
1.3. METHODOLOGY
My analysis of the phonology and orthography of the Secunda transcriptions essentially in-
volves four strands of methodology. While the first methodological approach, namely, histori-
cal (Hebrew) linguistics, is not novel, my dissertation implements three methodological ap-
proaches that have gone either unimplemented or lightly implemented in studies of the
-4-
Secunda: an emphasis on the pronunciation of Roman Palestinian Koine Greek, theoretical
models of cross-language perception, and moraic phonological theory.

1.3.1. Historical (Hebrew) Linguistics
First, an analysis of the phonology of the Secunda is undergirded by principles of historical
linguistics generally and historical Hebrew linguistics specifically. Each form in the Secunda
is analyzed with respect to its etymological form both in Proto-Northwest Semitic (PNWS)
and Proto-Hebrew. It is generally assumed that the realization of a particular word in the Se-
cunda falls somewhere on the spectrum between the Proto-Hebrew form and its realization in
the various reading traditions and dialects of Hebrew attested throughout history if no other
innovative development is apparent (for a more detailed discussion, see 6.2).

1.3.2. Greek Pronunciation and Orthography
Second, an analysis of the phonology of the Secunda is based on an in-depth analysis of the
pronunciation of Roman Palestinian Koine Greek. This is necessary because the Greek text of
the second column is regarded as a transcription and not a transliteration. It is necessary,
therefore, to understand the correspondences between the Greek graphemes and phonemes in
the contemporary local Greek pronunciation. Because there is no evidence that the Secunda
transcriptions emerged out of a centuries-old tradition of transcribing continuous Hebrew
texts into Greek, there is no reason to suggest a conservative spelling system. For a transcrip-
tion to be functional, if it does not reflect an established convention, it must reflect contempo-
rary pronunciation. Accordingly, we will assume that the transcriber chose each particular
Greek grapheme because the Greek phoneme (or phone) that it represented best "approximat-
ed" a given Hebrew sound (see below). An analysis of the conventions for transcribing other
languages into Greek will serve to provide comparative evidence for how such approxima-
tions tend to come to fruition. The methodology for analyzing Greek pronunciation in Roman
Palestinian Koine and Greek transcription conventions in the Hellenistic/Roman Near East is
described in more detail in chapters 4 and 5 (4.2; 5.2).
-5-
1.3.3. Cross-Language Perception in Modern Linguistics
Third, an analysis of the specific correspondences between the Greek graphemes and the He-
brew phones will be aided by modern linguistic studies on cross-language perception. Al-
though it seems obvious that mapping the sounds of one language onto the graphemic-phone-
mic system of another would be inextricably linked to perception, previous scholars writing
on the Secunda have not availed themselves of the advancements of modern linguistics on the
topic. To address this lack in scholarship, I turn to modern linguistic studies on cross-lan-
guage perception to support and complement my interpretations of Secunda phonology.
In order to understand cross-, or non-native-, language perception, we must first begin
by understanding native-language perception. When speakers conceive of their own native
language, unless they are phonologists or phoneticians, they typically process it phonemical-
ly. Thus, two non-contrastive allophonic realizations of a particular phoneme are unlikely to
be intuitively perceived any differently by a native speaker. For example, English speakers do
not typically distinguish the /p/ in happy, realized as an unaspirated [p], from the /p/ in pie,
realized as [ph] (DIRVEN and VERSPOOR 2004, 115; MARTIN and PEPERKAMP 2011, 2334–36).
Because individual speakers' perceptual systems are built to process the phonemes of
their own native languages (MARTIN and PEPERKAMP 2011, 2337), a number of different phe-
nomena occur when processing non-native speech sounds. The modern linguistic discipline
of cross-language perception has yielded primarily two theoretical models for predicting and
describing how non-native sounds are perceived, namely, the Perceptual Assimilation Model
(PAM) and the Speech Learning Model (SLM).
The core principle of the Perceptual Assimilation Model is that non-native speech
sounds are perceived with reference to the phonemes and the phonological space of one's na-
tive language. In the PAM, a non-native speech sound is described as being perceived in one
of three different ways. First, non-native sounds that are similar but not identical to native
phonemes tend to perceptually assimilate to the native phonemic category. In this case, the
-6-
non-native sound's approximation of the native category can be either good, acceptable but
not ideal, or markedly different. For example, speakers of English, in which no ejective con-
sonants exist, were found to assimilate the Ethiopic ejectives /pˀ/ and /tˀ/ to the English non-
ejective plosives /p/ and /t/. Second, a non-native sound may perceptually assimilate as be-
longing to the native phonological space but not to any particular native phoneme. In other
words, it is perceived as a speech sound in between the existing phonemes of the native lan-
guage. Third, a non-native phone may not assimilate at all to the native phonetic space and
thus be regarded as a nonspeech sound. For example, some foreign sounds, such as clicks,
may not even be regarded as part of speech for an English speaker due to their stark dissimi-
larity to any native sounds (BEST 1995, 193–96).
The perception of contrasts between non-native phonemes follows from these various
patterns of assimilation and may be realized in various ways. According to BEST and TYLER's
extension of PAM to learners of a second language (L2), henceforward referred to as PAM-L2,
there are four different ways that L2 contrasts might assimilate to L1 phonological categories.
First, according to the Two-Category Assimilation (TC Type), the non-native sounds assimi-
late to two different native phonemic categories and thus are perceived as contrasting. Sec-
ond, according to the Category-Goodness Difference (CG Type), both non-native sounds as-
similate to one native phonemic category, but one is a significantly better exemplar of the
native category and thus the sounds are perceived as distinct. Third, according to the Single-
Category Assimilation (SC Type), both non-native sounds assimilate to one native phonemic
category, yet both are poor exemplars of the native category and thus are not distinguished
well. Fourth, and finally, according to the Both Uncategorizable (UU Type), both non-native
sounds do not assimilate to any category, yet fall within the native phonetic space, and are
distinguished in perception according to their proximity to one another (BEST and TYLER
2007; FABRA and ROMERO 2012, 493).
-7-
The other theoretical model, namely, the Speech Learning Model, addresses how both
L1 and L2 sounds affect one another. The model is grounded in two primary assumptions.
First, the learning of L2 speech is not relegated to an early developmental stage of life. Sec-
ond, the faculties monolinguals utilize to learn their L1 are accessible for L2 learning
throughout their lifetime. The main claim of the SLM is that the phonemic categories of the
L1 and L2 systems of a bilingual coexist in a "common phonological space" and thus affect
one another. The phonetic categories of each subsystem (i.e., L1 and L2) may either assimi-
late or dissimilate. The SLM argues that learners of an L2 can, but do not necessarily, form
new categories for the sounds of L2. To put it simply, the more similar an L2 sound is to an
L1 sound, the more likely it is to assimilate and the more dissimilar an L2 sound is to an L1
sound, the more likely it is to dissimilate and have a new category formed. Moreover, FLEGE
has shown that the phonetic categories of an L2 can actually influence the production of an
L1 phonetic category so that it comes to differ slightly from that of native monolinguals. This
sort of assimilation and dissimilation depends, to a large degree, on the linguistic experience
of the speaker and the age at which they learned their L2 (FLEGE 2007, 366–376; FABRA and
ROMERO 2012, 493).
The applicability of these theoretical models to the Secunda transcriptions should be
apparent. By necessity, transcribing Hebrew phones into Greek script requires the assimila-
tion of a Hebrew speech sound to a Greek phonetic category. Moreover, it is probably the
case that the transcriber was (at least) bilingual, since there are instances in which he seems
to transcribe Hebrew phonemically rather than phonetically (see chapter 6). Accordingly, we
may assume that the same sorts of principles outlined in the PAM, PAM-L2, and SLM were at
work in the process of transcription. In light of this assumption, modern linguistic studies on
cross-language perception are cited throughout chapter 6 in support of various interpretations
of the phonology of the Secunda. In these cases, studies are sought that best replicate the spe-
cific situation in question. For example, I argue that the Hebrew phoneme /e/ (< */i/), which
-8-
is represented by Greek ε, was phonetically realized as [ɪ]. Because the vocalic system of
Roman Palestinian Koine had [ɛ] (or [e̞ ]), [e], and [i], but not [ɪ], I cite cross-language per-
ception studies of how a non-native [ɪ] is assimilated to the phonetic categories of languages
with [ɛ], [e], and [i] on the front axis but not [ɪ] (e.g., Italian, Catalan, Korean) (see 6.4.2.4).
1.3.4. Moraic Phonology
Fourth, and finally, our analysis of Secunda phonology, specifically with respect to syllable
structure, is based on the moraic model outlined by HYMAN (1985), HAYES (1989), and VAN
OOSTENDORP (2005). Moraic theory essentially regards syllables as consisting of an onset and
one or two morae. Heavy syllables contain two morae and light syllables contain one (VAN
OOSTENDORP 2005). In the following trees, σ signifies a syllable node and µ signifies a mora:
σ σ σ
µ µ µ µ µ
C V C V C C V
Light Heavy Heavy

Figure 1: Examples of Heavy and Light Syllables in Moraic Phonology
If an entire word is represented, the word node is signified by ω:

ω
σ σ
µ µ µ µ
C V C C V V C
(CVC).(CVV).C
Figure 2: Example of Word in Moraic Phonology
The specific application of moraic phonology in the analysis of Secunda Hebrew syllable
structure follows the work of KIPARSKY on Arabic (2003) and KHAN on Biblical Hebrew
(1987; 2013b). A primary distinction of their approach involves interpreting certain conso-
nantal morae in complex onsets and complex codas as extra-syllabic or "semisyllables" (6.5).
-9-
2. ORIGEN AND THE SECUNDA
2.1. INTRODUCTION
If the Greek transcriptions of the second column of the Hexapla were originally composed by
Origen, then the date and provenance of the text are straightforward. It was composed in Cae-
sarea Maritima in Palestine sometime after the year 233 CE (see 2.3.1). If, on the other hand,
Origen acquired the text of the second column as he did the other five columns, then the issue
of its original date and provenance remains an open question. The first of these two alterna-
tives may be dismissed if it can be determined that Origen lacked sufficient skill in Hebrew to
compose the second column himself.
Accordingly, the first part of this chapter will be a comprehensive treatment of Ori-
gen's knowledge of Hebrew based on his own writings. It will be demonstrated that Origen
did not have the requisite Hebrew knowledge to compose the second column of the Hexapla
himself. Also flowing out of the analysis of Origen's Hebrew knowledge is a better under-
standing of how Origen used the second column and why he included it in the Hexapla. Fi-
nally, having pulled the original composition of the Secunda out from under Origen's pen,
this chapter will conclude with a discussion regarding the nature of the second column as it
relates to the compositional history of the Hexapla.
- 10 -
2.2. ORIGEN'S KNOWLEDGE OF HEBREW
The history of scholarship regarding Origen's knowledge of Hebrew begins within a couple
centuries after his death in the writings of Eusebius (260/65–339/40 CE) and Jerome (347–420
CE). Concerning Origen's impetus for compiling the Hexapla, Eusebius writes (Historia Ec-
clesiastica 6.16.1):
So great an exacting study of the divine words was introduced to Origen, that
he thoroughly learned the Hebrew language, and acquired as his own posses-
sion the original scriptures held by the Jews in the very letters of the
Hebrews.1
In the beginning of his account of Origen's work on the Hexapla, Jerome states (De viris il-
lustribus 54):
Who is ignorant also how greatly he was invested in the study of the divine
scriptures, such that even the Hebrew language, contrary to the nature of his
time and his people, he learned thoroughly?2
The verbs used to describe Origen's study of Hebrew both in Eusebius's Greek account
(ἐκµανθάνω) and in Jerome's Latin account (edisco) signify a thorough knowledge gained
through study. It should be noted, however, that Jerome is likely working from Eusebius's
text in this passage (GRAFTON and WILLIAMS 2006, 317). The parallel language (e.g., τοσαύτη
|| quod tantum, τῶν θείων λόγων || Scripturis divinis, ὡς...ἐκµαθεῖν || ut...edisceret) strongly
supports this claim. Nevertheless, the fact that Jerome follows Eusebius in this regard shows
that he found no reason to contradict Eusebius's claim that Origen knew Hebrew well.
Despite the statements of Eusebius and Jerome, most modern scholars who have in-
vestigated Origen's Hebrew knowledge have come to the conclusion that his skill in Hebrew
was by no means expert. The most extensive treatments regarding Origen's Hebrew knowl-
edge have been carried out by ELLIOTT (1877–87, 855–59), HANSON (1959, 167–75), and DE
1. Τοσαύτη δὲ εἰσήγετο τῷ Ὠριγένει τῶν θείων λόγων ἀπηκριβωµένη ἐξέτασις ὡς καὶ τὴν Ἑβραΐδα γλῶτταν
ἐκµαθεῖν τάς τε παρὰ τοῖς Ἰουδαίοις ἐµφεροµένας πρωτοτύπους αὐτοῖς Ἑβραίων στοιχείοις Γραφάς κτῆµα ἴδιον
ποιήσασθαι.
2. Quis ignorat et quod tantum in Scripturis divinis habuerit studii, ut etiam Hebraeam linguam, contra
aetatis gentisque suae naturam edisceret?
- 11 -
LANGE (1976, 21–23, 152–54), all of whom conclude that Origen had a limited knowledge of
Hebrew.3 At the same time, some scholars (e.g., HANSON 1959, 167) acknowledge evidence in
favor of Origen having some knowledge of Hebrew, even if it was not expert.
The purpose of this section is to summarize and build upon the work of previous
scholars to provide the most comprehensive treatment of Origen's Hebrew knowledge to date.
The first part of this section will discuss the limits of Origen's Hebrew knowledge and its
ramifications for the authorship of the Secunda. The second part of this section will address
the nature and extent of Origen's Hebrew knowledge in terms of what he did know. Follow-
ing this, the third and final part of this section will describe how Origen used the second col-
umn and how the way he utilized it relates to the reason that he included it in the Hexapla.
2.2.1. The Limits of Origen's Hebrew Knowledge
Arguments for Origen's limited Hebrew knowledge are based primarily on his deferral to He-
brew experts, his mistaken etymologies, and his reliance on Greek (translation or transcrip-
tion) instead of Hebrew.

2.2.1.1. Origen's Deferral to Experts
One of the most common evidences cited to prove the deficiency of Origen's Hebrew knowl-
edge is the fact that when discussing a matter of Hebrew philology, he often defers to those
who are experts in Hebrew (for the examples below, see ELLIOTT 1877–1887, 856–57; BARDY
1925, 217–19; HANSON 1959, 171–72; DE LANGE 1976, 152; MARCOS 2000, 205).
In Contra Celsum (1.34), when discussing the meaning of the Hebrew word ‫ עלמה‬in
Isaiah 7:14, he states that "it is found, as they say, also in Deuteronomy referring to a vir-
gin."4 The verse in Deuteronomy (22:23) has the same Greek word in the LXX as in Isaiah
7:14 (παρθένος), but a different Hebrew word in the MT (‫( )בתולה‬HANSON 1959, 167; see also
2.2.1.3.3). In Homilies on Genesis (XII), when discussing the etymology of Esau, he prefaces
3. For various opinions on the nature of Origen's Hebrew knowledge, see also WUTZ 1914, 37–38; BARDY
1925, 217–19; KAHLE 1947, 87; LIETZMANN 1950, 302; SPARKS 1959, 276–77.
4. κεῖται, ὥς φασι, καὶ ἐν τῷ Δευτερονοµίῳ ἐπὶ παρθένου.
- 12 -
his information with the the phrases, "as those who interpret Hebrew names say" and "as it
seems to others."5 In Epistula ad Africanum (11.61), when attempting to discern if the play οn
words in the Greek History of Susanna would be present in a Hebrew original, he writes, "I
referred the matter to not a few Hebrews in my attempt to learn the answer."6 In Homiliae in
Canticum Canticorum (I.6), he writes that "the Hebrews say that Cedar is interpreted as dark-
ness."7 In his comments on Psalm 24:10 (Selecta in Psalmos, 12.1269), when discussing the
Hebrew word ‫ ְצבָאוֹת‬, he cites "those who have come to an exact understanding of the matters
of the Hebrews."8 Finally, in Homiliae in Librum Numerorum (16.1), when discussing
whether or not the word referring to God in the story of Balaam refers necessarily to the God
of Israel, he cites "those who read Hebrew literature" as an authority and concludes the dis-
cussion with, "let it be asked of one who is able."9 This homiletical passage10 has been taken
5. ut aiunt qui Hebraea nomina interpretantur ... ut aliis visum est.
6. οὐκ ὀλίγοις Ἑβραίοις ἀνεθέµην πυνθανόµενος.
7. Aiunt Hebraei Cedar interpretari tenebras (13.44).
8. οἱ τὰ Ἑβραίων ἠκριβωκότες. The phrase, referring to those expert in Hebrew, may be a technical term. The
word ἠκριβωκότες 'having investigated exactly' bears a striking similarity to those words in the Hebrew
grammatical tradition formed from the root ‫דקדק‬. For example, when discussing the reading of the shema, the
Mishnah discusses one who "recited but was not exact (‫א דִ יקְדֵּ ק‬Ä) in its letters" (mBer. 2.3). In the Jerusalem
Talmud, it is said regarding R. Hoshaˁyah that he "read and translated all the exact details of the parashah" (‫היה‬
‫( )קורא ומתרגם כל דקדוקי הפרשה‬jYoma 3.8).
9. qui Hebraicas litteras legunt ... de quo qui potest requirat.
10. The full passage is:
In Hebrew literature the name of God, that is, God (Deus), or Lord (Dominus), is said to be written in various
ways. For in one way god is written, whatever is a god. In another way God himself, of whom it is written,
"Hear, O Israel, the Lord (Dominus) your God (Deus), is one God (Deus)." Therefore, that God of Israel, one
God and creator of all, is written with a certain definite sign of letters, which is called the tetragrammaton by
them. Therefore, when God is written under this sign in the scriptures, there is no doubt at all that it is said about
the true God and creator of the world. However, when it is written with other letters, that is, common letters, it is
considered uncertain whether it is said regarding the true God, or regarding another ... Now those who read
Hebrew literature/letters say that in this place, God is not referenced under the sign of the tetragrammaton. Let
it be asked of one who is able.
In Hebraeorum litteris nomen Dei, hoc est Deus, vel Dominus, diverse scribi dicitur. Aliter enim scribitur Deus,
quicunque Deus: aliter Deus ipse, de quo dicitur: Audi, Israel, Dominus Deus tuus, Deus unus est." Iste ergo
Deus Israel, Deus unus et creator omnium, certo quodam litterarum signo scribitur, quod apud illos
tetragrammaton dicitur. Si quando ergo sub hoc signo in Scripturis scribitur Deus, nulla est dubitatio quin de
Deo vero et mundi creatore dicatur. Si quando vero aliis, id est communibus litteris scribitur, incertum habetur
utrum de Deo vero, an de aliquo...Aiunt ergo qui Hebraicas litteras legunt in hoc loco Deus non sub signo
tetragrammati esse positum de quo qui potest requirat.
- 13 -
to imply that Origen was not able to consult the Hebrew manuscripts himself to see whether
‫ יהוה‬or ‫ אלהים‬was written there. However, based on the full context of the passage, it is con-
ceivable that Origen is not simply referring to the difference between the words ‫ יהוה‬and
‫אלהים‬, but rather to some sort of accompanying symbol11 or the practice of writing the divine
name in the Paleo-Hebrew script.12 Origen was not only aware of this practice, but regarded
the writing of the divine name in ancient Hebrew letters as indicative of a most accurate man-
uscript.13 Therefore, Origen's need to defer to experts may involve a more complex distinc-
tion than merely the presence of ‫ יהוה‬or ‫ אלהים‬in the Hebrew text. His comments in this pas-
sage may be—not without confusion—referring to multiple elements at the same time.
In sum, two main facts about Origen's Hebrew knowledge are made clear from the
quoted passages. First, he was in contact with and depended upon Jewish scholars for much
of his knowledge of Hebrew. Second, Origen did not regard himself among those who were
expert in Hebrew and certain questions regarding the Hebrew language were beyond Origen's
linguistic skill.
11. The phrase sub hoc signo 'under this sign' used in the passage (see previous note) may indicate that there
was a particular sign written above the word ‫ אלהים‬to signify that it was referring to the one true God. In the
Babylonian pointing tradition, albeit much later, the dagesh and rafeh signs, written above the word ‫אלהים‬,
served this purpose (YEIVIN 1985, 918).
12. The fact that Origen states that the word for God can be written in communibus litteris 'in common letters'
may indicate that the word for the one true God was sometimes written in a different script in some manuscripts.
It is well-known that the tetragrammaton was written in Paleo-Hebrew script in the scrolls from Qumran.
13. Selecta in Psalmos, 12.1104:
There is a four-letter word unpronounced by them, which is even written on a leaf of gold of the high priest, and
it is said with the appellation Adonai ... Among the Greeks it is pronounced with Kurios. And in the most
accurate manuscripts, the name is found in Hebrew characters, not contemporary Hebrew characters, but rather
the most ancient. For they say that in the captivity Ezra passed down to them characters different from the first
characters.
ἔστι δέ τι τετραγράµµατον ἀνεκφώνητον παρ’ αὐτοῖς, ὅπερ καὶ ἐπὶ τοῦ πετάλου τοῦ χρυσοῦ τοῦ ἀρχιερέως
ἀναγέγραπται, καὶ λέγεται µὲν τῇ Ἀδωναῒ προσηγορίᾳ ... παρὰ δὲ Ἕλλησι τῇ Κύριος ἐκφωνεῖται. Καὶ ἐν τοῖς
ἀκριβεστέροις δὲ τῶν ἀντιγράφων Ἑβραίοις χαρακτῆρσι κεῖται τὸ ὄνοµα, Ἑβραϊκοῖς δὲ οὐ τοῖς νῦν, ἀλλὰ τοῖς
ἀρχαιοτάτοις. Φασὶ γὰρ τὸν Ἔσδραν ἐν τῇ αἰχµαλωσίᾳ ἑτέρους αὐτοῖς χαρακτῆρας παρὰ τοὺς προτέρους
παραδεδωκέναι.
- 14 -
2.2.1.2. Etymologies
The etymologies of Hebrew names in Origen's writings are derived primarily from Jewish
sources (HANSON 1956, 120–22). These Jewish sources may be further subdivided into a list
of biblical names arranged "consecutively" and a more traditional onomastic list of biblical
names accompanied by interpretations (1956, 119).14 It is supposed that only on a few occa-
sions does Origen attempt his own etymology (1956, 103–105). Scholars cite the faulty ety-
mologies belonging to this group as evidence of Origen's ignorance of Hebrew (for the exam-
ples below, see ELLIOTT 1877–1887, 856–58; WUTZ 1914, 37–38; HANSON 1956, 104; HANSON
1959, 170–71).15
By way of example, in Selecta in Genesim (12.100.23), Origen interprets the word ‫כּוּשׁ‬
as 'darkness' (σκότωσις), but at the same time expounds the text as if it refers to dust, presum-
ably on the basis of the meaning of the Greek word χοῦς 'dust' (HANSON 1959, 170). Else-
where in the same book, he interprets ‫ קטורה‬as 'smaller' (cf. ‫( )קטנה‬12.120.39), Συρια as 'lofty'
(cf. ‫( )רום ~ א ֲָרם‬12.117.34), and ‫ אבידע‬as 'the height of my father' (cf. ‫ִירם‬
ָ ‫( ) ֲאב‬12.121.8).
There are also a few other examples of supposedly faulty etymologies cited in the lit-
erature which, in my view, are not inconsistent with a knowledge of Hebrew. In Selecta in
Genesim (12.136.8), Origen interprets Symmachus's rendering of ‫( ָצ ְפנַת ַפּ ְענֵ ַח‬Σαφθφανὴ) as 'he
has revealed hidden things' (κεκερυµµένα ἀπεκάλυψε). HANSON makes the implausible argu-
ment that Origen might have read the word as a combination of ‫' יצא‬to go out' and ἐφάνη 'ap-
peared' (1956, 104). He seems not to realize that Origen's interpretation is perfectly consistent
with the Hebrew root letters. The root ‫ פענח‬is attested in later Hebrew with the meaning of 're-
vealing hidden things' and the root ‫ צפן‬is attested elsewhere in the Bible with the meaning of
14. Origen follows the etymology of Philo in a few instances (HANSON 1956, 103–104). For example, he
interprets ‫ חנוך‬Enoch as 'your favor' (cf. üְ‫( ) ִחנּ‬ELLIOTT 1877–1887, 857).
15. The two most significant works on the etymologies of names in Origen are WUTZ (1914) and HANSON
(1956), both of which contain many more etymologies than are cited here. For their discussions of Origen's
faulty etymologies, see WUTZ (1914, 37–38) and HANSON (1956, 104).
- 15 -
'hiding'. Moreover, the explanation of the name in the various Targumim typically includes
the idea of 'revealing' and 'hidden things'.16
In his Commentarium in evangelium Matthaei (12.16.21) and In Ezechielem Homiliae
(13.4), Origen interprets the meaning of the place name ‫ צִידוֹן‬as 'hunters' (θηρῶντες). HANSON
suggests that this is due to a mistaken reading of Psalms 124:7 based on a divergent text
(1956, 104), but there is no reason that a root-based interpretation (‫ ציד‬associated with the
meaning of hunting) could not explain Origen's etymology. Finally, ELLIOTT cites Origen's in-
terpretation of ‫ שְׁמוּאֵל‬as 'there is God himself' (‫ )שָׁם הוּא אֵל‬as evidence of his "defective" He-
brew (ELLIOTT 1877–1887, 857). However, such an interpretation based on breaking up the
word in a non-etymological manner would not be out of place even among the rabbis. For
example, in Selecta in Genesim (12.133.47), Origen interprets the word which the Egyptian
herald declares before Joseph in Genesis (41:43), °‫( אַב ְֵר‬Αβρηχ), as 'gentle father' (πατὴρ
ἁπαλός) (‫)אב רך‬. Origen goes on to explain that "it reasonably calls Joseph gentle father, since
although he was gentle according to his age, as a father he demonstrated himself to be a ruler
bringing salvation to the Egyptians" (Selecta in Genesim, 12.133.48–50).17 A similar interpre-
tation is found in Rashi's commentary on the Torah, who quotes R. Yehudah saying that
"Joseph is ‫ אברך‬because he is a father in wisdom and gentle in years" (‫אברך זה יוסף שהוא אב‬
‫)בחכמה ורך בשנים‬.18
In my view, criticizing Origen's Hebrew knowledge on the basis of etymologies is an
unfruitful endeavor. The fact that he is even criticized for a method of etymological interpre-
tation found among the rabbis should serve as a humbling reminder that modern scholars are
16. Targum Onkelos interprets the name as 'a man to whom hidden things are revealed' (‫ַמרן גַלי ָן לֵיה‬
ָ ‫)גוּברא דְ ִמט‬,
ָ
Targum Neofiti as 'a man to whom hidden things have been revealed' (‫)גברא דטמירתא גליו ליה‬, Targum Pseudo-
Jonathan as 'a man who reveals hidden things' (‫)גברא דטמירן מפרסם‬, and the Cairo Genizah Targum as 'the hidden
things are revealed to him' (‫)טמירתה מתגליין ליה‬.
17. Εἰκότως πατέρα ἁπαλὸν ἐκάλεσε τὸν Ἰωσὴφ, ἐπειδήπερ ἁπαλὸς ὢν κατὰ τὴν ἡλικίαν, ὡς πατὴρ σωτήριον
ἀρχὴν Αἰγυπτίοις ἐνεδείξατο.
18. As far as I am aware, the connection between Origen's interpretation of °‫ אַב ְֵר‬and the rabbinic explanation
has not been acknowledged in the literature (cf. WUTZ 1914, 347, 522).
- 16 -
too quick to dismiss or misinterpret ancient ways of thinking as ignorance. As ELLIOTT him-
self admits, "it is unsafe ... to attach too much importance to etymological errors" (1877–
1887, 856–57). This is especially true when we take into account the fact that most of his ety-
mological interpretations are derived from other sources and that it is not always clear which,
if any, etymological interpretations are original to Origen. However, it should be noted that in
those instances where Origen reproduces a clear mistake of one of his sources, it would be
evidence of a less-than-expert knowledge of Hebrew.

2.2.1.3. Reliance on Greek Translation/Transcription over Hebrew
Perhaps that which betrays Origen's lack of Hebrew expertise more than anything else is his
utter reliance on Greek translation and transcription. Such a reliance is evident in his writing
not only when he makes mistakes because of it, but also when it leads him to ignore the He-
brew entirely (for the examples below, see ELLIOTT 1877–1887, 856–58; HANSON 1959, 167–
68, 170–71; DE LANGE 1976, 152–53).

2.2.1.3.1. No Acknowledgement of Hebrew Variants
Numerous scholars have pointed out that Origen frequently relies on the Greek translations to
such an extent that he seems oblivious to significant divergences in the Hebrew (ELLIOTT
1877–1887, 856–58; DE LANGE 1976, 152).
For a number of passages, in which there is a significant difference between the LXX
and the Hebrew, Origen expounds the LXX reading without referring to the reading in He-
brew. For Genesis 2:2, he follows the Greek reading of 'on the sixth day' (ἐν τῇ ἡµέρᾳ τῇ
ἕκτῃ) with no comment on Hebrew 'on the seventh day' (‫שּׁבִיעִי‬

ְ ‫( )בַּיּוֹם ַה‬Selecta in Genesim,
12.97.24). For Numbers 24:17, he follows the Greek reading of 'a man will arise'
(ἀναστήσεται ἄνθρωπος) with no discussion of Hebrew 'a scepter will arise' (‫שׁבֶט‬
ֵ ‫( ) ְוקַם‬Adno-
tationes in Numeros, 17.21.38). For Isaiah 53:8, he accepts the Greek reading of 'to death' (εἰς
θάνατον) without a comment on Hebrew 'to him' (‫( )לָמוֹ‬Contra Celsum, 1.54.36). For Jeremi-
ah 11:19, he quotes the Greek reading of 'let us throw wood into his bread' (ἐµβάλωµεν ξύλον
εἰς τὸν ἄρτον αὐτοῦ) with no mention of Hebrew 'let us destroy the tree with its fruit [lit.
- 17 -
"bread"]' (‫שׁחִיתָ ה עֵץ ְבּ ַלחְמוֹ‬
ְ ַ‫( )נ‬In Jeremiam, 10.1.21). However, some of these examples, such
as Deuteronomy 32:8 and Psalms 81:7, may represent text-critical issues.19
There are also a number of passages in which Origen refrains from commenting on
the Hebrew when it would be expected. When discussing the various readings of the begin-
ning words of 1 Samuel, Origen compares the readings of various Greek translations without
reference to the Hebrew (Homiliae in Librum Regnum, 12.998–999.4).20 When discussing the
mark placed on the foreheads of those grieved by the sins of Jerusalem in Ezekiel 9:2–7, Ori-
gen cites the translations of Aquila and Theodotion ('the mark of the taw' [Σηµείωσις τοῦ
Θαῦ]), who merely transcribe the Hebrew (‫ )תָּ ו‬into Greek (Θαῦ), rather than referring to the
Hebrew itself (Selecta in Ezechielem, 13.800.50). Origen's comments that "God wiped out the
name ... of Sarah (Σάρας), calling her Sarrah (Σάῤῥαν)" may convey that he believed her re-
naming consisted of Sarah acquiring an additional r, rather than Sarai (‫ )שׂ ַָרי‬becoming Sarah
(‫ )שׂ ָָרה‬as in the Hebrew (Selecta in Psalmos, 12.1188.40–42).21 Origen's contention that a
grammatical mistake is present in the phrase σπεῖρον σπέρµα κατὰ γένος 'bearing seed ac-
cording to kind' and that it ought to be amended to κατὰ γένος σπεῖρον σπέρµα 'according to
kind bearing seed' seems unreasonable if he was familiar with the Hebrew (Selecta in Genes-
im, 12.92.22–12.93.3). Finally, when discussing the text of Daniel, he bases his claim about
the order of the verses "in the Hebrew versions" (ἐν τοῖς Ἑβραϊκοῖς) on the fact that "so
Aquila, serving the Hebrew idiom, has rendered it in his version" (Οὕτω γὰρ Ἀκύλας
δουλεύων τῇ Ἑβραϊκῇ λέξει ἐκδέδωκεν εἰπών) (Epistula ad Africanum, 11.52.15–21).
19. For Deuteronomy 32:8, Origen follows the Greek 'according to the number of the angels of God' (κατὰ
ἀριθµὸν ἀγγέλων θεοῦ) with no discussion of Hebrew 'according to the number of the sons of Israel' (‫ְל ִמ ְספַּר ְבּנֵי‬
‫( )יִשׂ ְָראֵל‬Commentarii in evangelium Joannis, 13.50.332.6). However, 4Q37 12:14 has ‫ בני אלוהים‬in this verse.
For Psalms 81:7, Origen accepts the Greek 'they served' (ἐδούλευσαν) with no comment on Hebrew 'they pass
over' (‫( )תַּ עֲב ְֹרנָה‬Commentarium in evangelium Matthaei, 11.2.49). It is easy to conceive of how ‫' תעברנה‬they pass
over' could be mistaken as ‫' תעבדנה‬they serve', but Symmachus and Jerome both support a Hebrew text of ‫תעברנה‬
(FIELD 1875, II 233; ELLIOTT 1877–1887, 858).
20. See also ELLIOTT's comments on Origen's treatment of Genesis 45:27 and Exodus 4:10/6:30 (1877–1887,
858).
21. Ἐξήλειψε δὲ ὁ Θεὸς...τὸ ὄνοµα...τῆς Σάρας καλῶν αὐτὴν Σάῤῥαν.
- 18 -
Although these examples seem to paint the picture of a scholar totally unaware of the
Hebrew text, the passages require a more nuanced approach. HANSON has shown that even
though Origen was well aware of the differences between the Hebrew and Greek text, he re-
garded the LXX as divinely inspired. In fact, according to HANSON, while the Hebrew text
was regarded as the original authoritative text for Origen, nevertheless, because of his view
of inspiration, even in instances where he acknowledges that the LXX has changed the origi-
nal Hebrew, he views such alterations as inspired changes (1959, 162–67). Such a view may
seem unusual, but it is essential for understanding how someone like Origen, who knew very
well the differences between the Hebrew and Greek versions, would feel perfectly comfort-
able expounding the text of the LXX in its own right without any reference to the differences
in the original Hebrew text. Nevertheless, even in light of Origen's view of the inspiration of
the LXX, the examples above do not support the idea that Origen knew Hebrew well.
2.2.1.3.2. Mistakes with Hebrew Letters
There are a couple instances in which Origen's errors regarding Hebrew letters are best ex-
plained as the result of approaching the Hebrew through the Greek. In Selecta in Psalmos
(12.1068), when discussing the mention of Abimelech in the superscription of Psalm 34, he
writes:
It seems that the one named Ahimelech, of the first of the kingdoms, is called
Abimelech. The letters among the Hebrews, I refer to kaph and beth, have
such great similarity, so that one may not differentiate between them by any-
thing, except a small tittle only.22
Before proceeding to Origen's error, we ought to note that this passage clearly demonstrates
Origen's familiarity with the Hebrew alphabet. In the Hebrew-Aramaic script of Origen's day,
beth (‫ )ב‬and kaph (‫ )כ‬were indeed distinguished by very little. However, the name Ahimelech
(‫ )אחימלך‬is spelled with a heth—not a kaph—in Hebrew and thus would not have been con-
fused scribally with the ‫ ב‬in Abimelech (‫)אבימלך‬. That Origen suggests scribal confusion be-
22. ἔοικε τὸν τῇ πρώτῃ τῶν Βασιλειῶν Ἀχιµέλεχ ὠνοµασµένον Ἀβιµέλεχ ἀποκαλεῖν. τῶν στοιχείων παρ᾽
Ἑβραίοις, λέγω δὲ τοῦ χὰφ καὶ τοῦ βὴθ, πολλὴν ὁµοιότητα σωζόντων, ὡς κατὰ µηδὲν ἀλλήλων διαλλάττειν, ἢ
βραχείᾳ κεραίᾳ µόνῃ.
- 19 -
tween beth and kaph seems to suggest that he was first thinking of the name in Greek
(Ἀχιµέλεχ) and then envisioning how that name might be spelled in Hebrew (‫)**אכימלך‬.
In Origen's Epistula ad Africanum (11.77), it seems likely that he has conflated He-
brew shin (‫ )שׁ‬and ś in (‫ )שׂ‬based on their identical Greek transcription of σ. When attempting
to demonstrate how a Greek translator might preserve a play on words present in Hebrew also
in the Greek translation, he writes:

The Hebrews say that the woman was called essa and that 'I took' is made
clear from the word, as is evident from the [verse], 'Chos isouot essa', which is
translated as, 'I will take up the cup of salvation'. And the man is is, as is ap-
parent from the [verse], 'Esre ais', which is, 'Blessed is the man'. According to
the Hebrews, then, is [is a man] and essa [is] of a man, for from is, her hus-
band, she was taken.23
In Hebrew, the play on words lies in the phonological similarity between the name for 'man'
(‫ )אִישׁ‬and 'woman' (‫) ִאשָּׁה‬. An equivalent play on words in Greek would have been to attach a
feminine ending to the Greek word for 'man' (ἀνήρ). In fact, Symmachus does just this in his
translation, substituting an invented word ἀνδρίς (ἀνήρ 'man' + fem. ending -ις) for 'woman'
instead of the more common γυνή.24 This translation would have served Origen's argument
far better and would have helped him to realize what precisely the play on words was. In-
stead, the fact that Origen connects Genesis 2:23 (‫' כִּי ֵמאִישׁ ֻל ֳקחָה־זּ ֹאת‬for from man she was tak-
en') with Psalms 116:13 (‫' כּוֹס־י ְשׁוּעוֹת ֶאשָּׂא‬I will take up the cup of salvation') seems to indicate
that he thought that the play on words lay in the phonological similarity between the words
'woman' and 'take'. This only makes sense if he was working from Greek translation and
transcription and not the original Hebrew. Even though different Hebrew verbs are used in
each of these verses—Genesis 2:23 has ‫ לקח‬and Psalms 116:3 has ‫—נשׂא‬they are both trans-
lated into Greek with forms of the verb λαµβάνω 'to take'. Moreover, even though the words
23. Φασὶ δὲ οἱ Ἑβραῖοι ἐσσὰ µὲν καλεῖσθαι τὴν γυναῖκα· δηλοῦσθαι δὲ ἀπὸ τῆς λέξεως τὸ ἔλαβον, ὡς δῆλον ἐκ
τοῦ· Χῶς ἰσουὼθ ἐσσά, ὅπερ ἑρµηνεύεται· ἴς δὲ τὸν ἄνδρα, ὡς φανερὸν ἐκ τοῦ· Ἐσρὴ ἀΐς, ὅπερ ἐστί· Μακάριος
ἀνήρ. Κατὰ µὲν οὖν Ἑβραίους ἲς καὶ ἐσσὰ ἀνδρὸς, ὅτι ἀπὸ ἲς ἀνδρὸς αὐτῆς ἐλήφθη αὕτη.
24. Symmachus (Gen. 2:23): She will be called woman (ἀνδρίς), for from man (ἀνδρὸς, gen. of ἀνήρ) she was
taken. αὕτη κληθήσεται ἀνδρὶς, ὅτι ἀπὸ ἀνδρὸς ἐλήφθη αὕτη (FIELD 1875, 15).
- 20 -
‫' ִאשָּׁה‬woman' and ‫' ֶאשָּׂא‬I will take up' were pronounced with different sibilants in Hebrew dur-
ing Origen's time (/ˀeš š ɔ̄/ [ʔɪʃːɔː] 'woman' and /ˀeś ś ɔ̄/ [ʔɪsːɔː] 'I will take up'), they were both
transcribed into Greek as εσσα.25 The fact that Origen skips over the obvious play on words
in Hebrew brought out in Symmachus's translation for a far less intuitive pun seems to show
that Origen was working primarily from Greek translation and transcription when accessing
the Hebrew (cf. ELLIOTT 1877–1888, 858).
Two other instances of apparent conflation of Hebrew /š / and /ś / (or /s/) due to Greek
transcription are found in Origen. First, as cited earlier, Origen connects the etymology of the
Hebrew word ‫ כּוּשׁ‬/kū š / to Greek χοῦς 'dust' (HANSON 1959, 170). Second, in Selecta in Ju-
dices (12.949.12–20), when commenting on the shibboleth incident in Judges 12:6, Origen
claims that the distinction was between those who could say σεβηλα and those who pro-
nounced it as σεβηλω. He seems to ignore the fact that the distinction in the Hebrew text was
between those who could pronounce /š / and those who pronounced it as [s].26 According to
DE LANGE, Origen heard this illustration or found it in a Greek translation (1976, 152).
2.2.1.3.3. Indiscriminate Copying
In a couple of instances Origen seems to copy a line from the second column indiscriminately
without separating its parts. In Commentarium in evangelium Matthaei (14.16), when discus-
sing the Hebrew words for 'male' and 'female' in the creation story (Gen. 1:27), he writes:
At the same time, notice that regarding those made in the image it is not said
'man and woman', but 'male and female'. This we also have observed in the
Hebrew. For man is signified by the word IS, and male by the word ZACHAR.
And again, woman by the word ESSA, and female by the word OUNKEBA.27
25. For etymological */i/ being realized as /e/ [ɪ] in the Secunda, see 6.4.2.
26. Some scholars suggest that the Gileadites pronounced the ‫ ש‬in the word ‫' שבלת‬stream' as an interdental
fricative /ṯ / [θ], while the Ephraimites, in whose dialect there was no such phoneme, articulated [s] when trying
to pronounce [θ] (see RENDSBURG 2013b). FABER suggests that /š / was present in Gileadite but not Ephraimite
Hebrew and thus the Ephraimites pronounced it as [s] or [ɬ] (1992). However the original context is to be
explained, we must remember that in Origen's day, the only apparent distinction for those familiar with the
biblical text would have been that between ‫[ ש‬š ] and ‫[ ס‬s].
27. ἅµα δὲ πρόσχες ὅτι ἐπὶ µὲν τῶν κατ᾽ εἰκόνα οὐκ ἀνὴρ καὶ γυνὴ εἴρηται, ἀλλὰ ἄρρεν καὶ θῆλυ. τοῦτο δὲ καὶ
ἐν τῷ Ἑβραϊκῷ τετηρήκαµεν. ἀνὴρ µὲν γὰρ δηλοῦται τῇ ΙΣ φωνῇ, ἄρρεν δὲ τῇ ΖΑΧΑΡ. καὶ πάλιν γυνὴ µὲν τῇ
ΕΣΣΑ φωνῇ, θῆλυ δὲ τῇ ΟΥΝΚΗΒΑ.
- 21 -
In the Hebrew of this passage from Genesis, the words 'male' and 'female' come in a pair (‫זָכָר‬
‫ )וּנְ ֵקבָה‬with the conjunctive waw attached to the word 'female' (see also Gen. 5:2; 6:19; 7:3, 9,
16). It seems that Origen has copied the entire line, treating the whole phrase 'and female'
(‫ )וּנְ ֵקבָה‬as one word (ΟΥΝΚΗΒΑ) (HANSON 1959, 168). Such a mistake is unlikely if Origen
knew the Hebrew word, but is far more likely if he was working from the Greek translations
of the Hexapla and simply copied whole-cloth the word in the second column that fell on the
same line, not being careful, or able, to analyze its Hebrew components (see reconstruction
below; based on FIELD [1875, 10]):28
Hebrew Secunda Aquila Symmachus LXX Theodotion
‫זכר‬ ζαχαρ <ἄρσεν> ἄρσεν ἄρσεν ἄρσεν

‫ונקבה‬ ουνκηβα <καὶ θῆλυ> καὶ θῆλυ καὶ θῆλυ καὶ θῆλυ
Figure 3: 'male and female' in the Hexapla
A similar example, cited earlier for a different purpose, occurs in Contra Celsum
(1.34) when Origen is discussing the proper translation of the Hebrew word ‫( ַע ְלמָה‬Isa. 7:14):
And if a Jew, coming up with ingenious arguments, should say that 'Behold,
the virgin' has not been written but instead of it, 'Behold, the young woman',
we will say to him that the word aalma, which the seventy have translated as
'the virgin' but others as 'the young woman', is found, as they say, also in
Deuteronomy referring to a virgin.29
The word ‫ ַע ְלמָה‬occurs in the Isaiah passage with the definite article (‫) ָה ַע ְלמָה‬. Like the previ-
ous example, the fact that Origen cites the word with the article (ἀαλµά) may demonstrate
once again that Origen copied the corresponding line of the second column in its totality30
and was not working from his own Hebrew knowledge (see reconstruction below; based on
FIELD [1875, 443]) (figure 4):
28. Aquila's translation is not attested in Field, but it is not likely that it would have differed.
29. Ἐὰν δὲ Ἰουδαῖος εὑρεσιλογῶν τὸ Ἰδοὺ ἡ παρθένος µὴ γεγράφθαι λέγῃ ἀλλ᾽ ἀντ᾽ αὐτοῦ Ἰδοὺ ἡ νεᾶνις,
φήσοµεν πρὸς αὐτὸν ὅτι ἡ µὲν λέξις ἡ ἀαλµά, ἣν οἱ µὲν ἑβδοµήκοντα µετειλήφασιν εἰς τὴν παρθένον ἄλλοι δ᾽
εἰς τὴν νεᾶνιν, κεῖται, ὥς φασι, καὶ ἐν τῷ Δευτερονοµίῳ ἐπι παρθένου.
30. It is unlikely that the double αα at the beginning of the word signifies the guttural ˁayin. Such a
representation would be unusual for the Secunda; further, the Ambrosiana palimpsest has αλµωθ (46:1) in the
plural construct without a double αα at the beginning of the word.
- 22 -
Hebrew Secunda Aquila Symmachus LXX Theodotion
‫העלמה‬ ἀαλµά ἡ νεᾶνις ἡ νεᾶνις ἡ παρθένος ἡ νεᾶνις

Figure 4: 'the virgin' in the Hexapla
On the other hand, because Origen is discussing how the word is translated in this context,
one could argue that citing the word with the definite article is acceptable.31
2.2.1.4. Conclusions
Based on Origen's own testimony regarding his lack of expertise in Hebrew, his dependence
on secondary sources for etymological meanings, and his heavy reliance on Greek translation
and transcription instead of Hebrew, it is necessary to conclude that Origen lacked the requi-
site skill in Hebrew to compose the second column of the Hexapla himself. It is inconceiv-
able that the same man who was able to vocalize the entire Hebrew Bible and devise a system
of transcription for it could have made the sort of mistakes cited above. Even the idea that
Origen merely transcribed what was dictated to him from an expert in Hebrew may be dis-
missed; the results of such a collaboration would be a much superior Hebrew knowledge than
what is exemplified in Origen's writings and a much inferior system of transcription than
what is exhibited in the Secunda.32 Therefore, barring new evidence that might come to light,
it should be considered a fact that Origen did not compose the second column himself.33
2.2.2. Nature and Extent of Origen's Hebrew Knowledge
The preceding section is sufficient to demonstrate that Origen was by no means an expert in
Hebrew and could not have composed the second column himself. Nevertheless, it would not
be correct to claim that he had no Hebrew knowledge at all. The present section will attempt
31. However, when Origen uses the same formula (ἡ λέξις ἡ + word ) to discuss the Greek word ἐπιούσιον (in
the phrase τὸν ἄρτον ἡµῶν τὸν ἐπιούσιον) in the Lord's prayer, he quotes the word without the article: ἡ λέξις ἡ
ἐπιούσιον (De oratione 27.7.2).
32. Note that Jerome, who had a personal Hebrew teacher, exhibits a deep knowledge of Hebrew in numerous
cases (GRAVES 2007). Also, the idea that someone unfamiliar with the language would transcribe it relatively
consistently and sometimes even phonemically is unlikely on the basis of the principles of cross-language
perception (see chapter 6).
33. This is the view of ELLIOTT (1877–1887, 855–59), KAHLE (1947), LIETZMANN (1950, 302), HANSON (1959,
167–75), and DE LANGE (1976, 21–23, 152–54).
- 23 -
to paint a picture of the nature and extent of Origen's Hebrew knowledge in terms of what he
did know. A survey of the evidence demonstrates that Origen knew the Hebrew alphabet,
knew many Hebrew words, had some grammatical knowledge of Hebrew, and used Hebrew
text-critically.
2.2.2.1. Knowledge of the Alphabet
There are a number of pieces of evidence which suggest that Origen had learned the Hebrew
alphabet (for the examples below, see ELLIOTT 1877–1887, 858; HANSON 1959, 170–71; DE
LANGE 1976, 152–53). Although cited earlier to demonstrate that Origen relied on Greek
translation and transcription, his comment that "kaph and beth maintain a great deal of simi-
larity, such that they differ from each other in nothing but merely a small tittle"34 (Selecta in
Psalmos, 12.1068) points to familiarity with the alphabet. Also, in Selecta in Psalmos
(12.1276.44–47), when discussing the phrase 'my savior' (σωτήρ µου) in the LXX (cf. 'my
salvation' ‫שׁעִי‬
ְ ִ ‫ י‬in the Hebrew), he states that "in this passage, in the Hebrew 'my savior', the
name of our Savior Jesus Christ is written the way in which someone would write Jesus in
Hebrew characters."35 Origen seems to be connecting the root letters of ‫ ישעי‬with the same
letters in the name of Jesus, spelled either ‫ ישוע‬or ‫ ישע‬during the Roman period.36 Last, in
Fragmenta in Lucam (221.1–4) and Scholia in Lucam (17.365.15–20), when discussing Je-
sus's famous saying about "one jot or tittle" (Matt. 5:18), he writes:
Not only among the Greeks is iota one tittle, but also among the Hebrews that
which is called among them ioth [is one tittle]. And 'one iota or one tittle' may
symbolically represent Jesus, since the beginning of his name, not only among
the Greeks, but also among the Hebrews is written starting with ioth.37
34. See footnote 22 for Greek text.
35. ἐν γὰρ τούτῳ τῷ Ἑβραϊκῷ τὸ σωτήρ µου, ὄνοµα γέγραπται τοῦ Σωτῆρος ἡµῶν Ἰησοῦ Χριστοῦ. δι᾽ ὧν ἄν
τις τοῖς Ἑβραϊκοῖς χαρακτῆρσι γράψῃ τὸν Ἰησοῦν.
36. HANSON (1959, 171) wrongly argues that this passage reflects a mistake of Origen, assuming that the name
of Jesus should be spelled ‫ יהשע‬in Hebrew. However, HANSON seem to confuse the longer name ‫שׁ ַע‬ ֻ ‫ י ְהוֹ‬with its
shorter variant ‫ י ֵשׁוּ ַע‬common in the Second Temple Period. Moreover, in the Judaean Desert Texts, the short
spelling of ‫ ישע‬for ‫ ישוע‬is common (MOR 2015, 79).
37. Μία κεραία οὐ παρ᾽ Ἕλλησι µόνον ἐστὶ τὸ ἰῶτα, ἀλλὰ καὶ παρ᾽ Ἑβραίοις τὸ παρ᾽ αὐτοῖς καλούµενον ἰώθ.
δύναται δὲ τὸ ἰῶτα ἓν ἢ µία κεραία συµβολικῶς λέγεσθαι ὁ Ἰησοῦς, ἐπείπερ ἡ ἀρχὴ τοῦ ὀνόµατος αὐτοῦ οὐ
παρ᾽ Ἕλλησι µόνον, ἀλλὰ καὶ παρ᾽ Ἑβραίοις ἀπὸ τοῦ ἰὼθ γράφεται.
- 24 -
In Selecta in Ezechielem (13.800–01), when discussing why the Hebrew letter taw is
to be put on the foreheads of those who are grieved by the sins of Jerusalem, he explains
three Jewish interpretations, all of which are contingent on knowledge of the alphabet. First,
because the taw is the last letter of the twenty-two letter Hebrew alphabet, it signifies the per-
fection of those who are grieved over the sins in the city. Second, it signifies those who have
kept the law, because 'the law' (ὁ νόµος) is called 'Torah' (θωρα) in Hebrew, the first letter of
which is taw. The third interpretation, which comes specifically from a Jewish Christian,
states that in Paleo-Hebrew script (τὰ ἀρχαῖα στοιχεῖα) the taw resembles the form of the
cross and is thus prophetic. It may seem far-fetched that a third-century CE Christian interpre-
tation could depend on Paleo-Hebrew script, but Origen's comments on the tetragrammaton
in his commentary on Psalms demonstrate that he was familiar with—or had at least heard
about—Paleo-Hebrew script (Selecta in Psalmos, 12.1104, see above).
Finally, in addition to the aforementioned examples, there are also a few places where
Origen discusses the alphabet in relation to acrostic passages. In his opening remarks regard-
ing Psalm 118/119, he points out that "it is written according to the letters of the Hebrews, so
that the first verses of it are eight starting with aleph, which is the beginning of their alphabet,
and the next eight start with beth, and thus in order" (Selecta in Psalmos, 12.1585.38–42;
Fragmenta in Psalmos 1–150, 118p.1–8).38 While it is true that even the Greek of the LXX
indicates that Psalm 118/119 is a Hebrew acrostic, Origen comments on several other pas-
sages whose acrostic nature is not reflected in the Greek. In the same passage (Selecta in
Psalmos, 12.1585.49–53), when discussing the acrostic patterning in Psalms 111 and 112,
Origen states that in those Psalms "the acrostic (στοιχείωσις) is not drawn out, but delivered
38. κατὰ Ἑβραίων στοιχεῖα γεγραµµένος, ὥστε τοὺς µὲν πρώτους αὐτοῦ στίχους εἶναι ὀκτὼ ἀπὸ τοῦ Ἄλεφ· ὅ
ἐστιν ἀρχὴ τῶν παρ᾽ αὐτοῖς στοιχείων· τοὺς δὲ δευτέρους ὀκτὼ ἀπὸ Βὴθ, καὶ οὕτω καθεξῆς.
- 25 -
briefly."39 Finally, in Fragmenta in Lamentationes (1.1–8), Origen provides a detailed de-
scription of the acrostic patterning of Lamentations.40
In sum, while there is no doubt that Origen obtained some of his information about
the alphabet (e.g., Paleo-Hebrew taw) from Jewish sources, the evidence suggests that Origen
did indeed know the Hebrew letters. This does not necessarily mean that he could have read a
Hebrew text, which would have required vocalization—in fact, there is evidence that such a
task would have been difficult for him—but his recognition of Hebrew letters not conveyed
in transcription (e.g., ˁayin in ‫)ישעי‬, his description of the shape of letters (e.g., beth, kaf, yod,
taw), and his familiarity with the order of the alphabet (e.g., acrostics) seem to indicate that
he had at least learned the alphabet.

2.2.2.2. Memorized Meanings and Etymologies
Beyond the alphabet, there is evidence that Origen knew the meaning of many Hebrew words
and names (for the examples below, see ELLIOTT 1877–1887, 858; HANSON 1956; HANSON
1959, 166, 168, 170; DE LANGE 1976, 153). There are many instances in his writings where
he will reference the Hebrew behind a Greek translation, such as Λευϊαθαν for δράκων (Con-
tra Celsum, 6.25), Ἀζαζήλ for ἀποποµπαῖος (Contra Celsum, 6.43), ις/ζαχαρ for ἀνήρ/ἄρσεν
(Commentarium in evangelium Matthaei, 14.16), and ανιη αρς for πραεῖς τῆς γῆς (Selecta in
Psalmos, 12.1060). It would be reasonable to assume that he used the text of the second col-
umn in such cases.41 There are also many instances in his writings where he will reference the
39. Ἐκεῖ µὲν οὖν ἡ στοιχείωσις οὐκ ἐκτέταται, ἀλλὰ διὰ βραχέων παραδέδοται.
40. Jeremiah ... laments, delineating several sections and distinct parts, which begin in the order of the Hebrew
alphabet. And after completing the lamentation for every letter and on top of all of them the one beginning with
Taw, which is the last letter of the Hebrews, he goes back to aleph and laments once again. This he does four
times, proceeding through the twenty-two letters of the alphabet.
Ὁ Ἱερεµίας ... θρηνεῖ περικοπάς τινας καὶ διαστολὰς περιγράφων ἀρχοµένας ἑξῆς ἀπὸ τῶν παρ’ Ἑβραίοις
στοιχείων. καὶ µετὰ τὸ τελέσαι τὸν ἐφ’ ἑκάστου στοιχείου Θρῆνον καὶ ἐπὶ πᾶσι τὸν ἀπὸ τοῦ Θαῦ ἀρχόµενον,
ὅπερ ἐστὶ τελευταῖον τῶν Ἑβραίων γράµµα, ἐπανέρχεται ἐπὶ τὸ Ἄλφ καὶ πάλιν θρηνεῖ, καὶ τοῦτο ποιεῖ τετράκις
τὰ εἴκοσι καὶ δύο στοιχεῖα ἐπεξιών.
41. HANSON argues that the sporadic instances of transcribed Hebrew in Origen's writings are his own creation
(1959, 168). However, such transcriptions are generally consistent with the transcription system found in the
Ambrosiana palimpsest. One possible exception could be the transcription βσαιµ ‫' ְבּשֵׁם‬in the name of' in a
quotation of Psalms 118:26 in Origen's commentary on Matthew (see below). The αι/ε interchange common in
contemporary Greek does not occur in the Ambrosiana palimpsest, where the word ‫' שֵׁם‬name' is written with an
- 26 -
meaning of a Hebrew name, such as transeuntes 'those passing through' for Hebraei (Homili-
ae in Librum Numerorum, 19.4), θηρῶντες 'hunters' for Σιδων, ὁρῶν 'seeing one' for Ἀζαῦ
(Selecta in Genesim, 12.117), and συνοχή 'distress' for Σόρ (Commentarium in evangelium
Matthaei, 11.16) (for more, see HANSON 1956). HANSON has demonstrated that such etymolog-
ical explanations derive from Jewish sources (1956). Finally, there are at least a couple in-
stances where Origen provides a Greek explanation of a Hebrew word that is transcribed into
Greek, even in the translation, such as γένοιτο for Ἀµήν (Fragmenta in evangelium Joannis,
120) and ἀντικείµενος for σατανᾶς (Commentarium in evangelium Matthaei, 12.21). It is
possible that such instances reflect a slightly more internalized knowledge of Hebrew, gained
neither through the text of the second column nor the etymological lists.
In sum, even though most of this material is likely derived from secondary textual
sources, there is no doubt that someone with as brilliant a mind and as superb a memory as
Origen (HANSON 1959, 182) would have been able to recall much of this information and ap-
ply it in different contexts. The process of consulting different textual and human sources for
the amount of Hebrew material contained in Origen's writings would have been a process
through which Origen learned a great deal.

2.2.2.3. Grammatical Understanding
While there is plenty of evidence that Origen knew the alphabet and meanings of words, only
two passages point to a more sophisticated grammatical knowledge (for the examples below,
see HANSON 1959, 167, 172). In Homiliae in Librum Numerorum (12.724.15–25), when dis-
cussing the presumably awkward use of the conjunction 'and' (καί) in the LXX translation of
Numbers 24:21 ('and having seen the Kenite and having taken up his parable, he said' [καὶ
ἰδὼν τὸν Καιναῖον καὶ ἀναλαβὼν τὴν παραβολὴν αὐτοῦ εἶπεν]), he writes:
It seems reasonable, according to the explanation which we delivered about
the Kenite, that the conjunction 'and' disrupts the sense. But it should be
known that it is natural for the Hebrew language to use the conjunction 'and'
epsilon (σεµ). On the other hand, Origen's transcription σελ (for ‫ ) ֶסלָה‬strongly supports the idea that he was
quoting the second column (see footnote 56).
- 27 -
frequently, so that when it sometimes is excessive, and in those places in
which it is not necessary, as it would seem in our language, it would seem
thrust in: which should in any case be admitted with pardon. For each and
every language has a particular characteristic, which would seem faulty in oth-
er languages. And in this place, then, the conjunction 'and' ought to be regard-
ed as excessive and superfluous.42
This passage does not necessarily demonstrate that Origen had a sophisticated understanding
of the syntax of the conjunction waw in Hebrew, but merely that he was sensitive to the fact
that the Hebrew use of the conjunction differed from that of Greek.
In Philocalia (14.1.6–32), Origen's comments regarding how different Greek transla-
tions render "nouns" (προσηγορίαι) and "predicates" (κατηγορήµατα) in Genesis 1:16–17
may reflect a more sophisticated understanding of the language. His comments focus on the
difference between the phrases ‫ ַה ַלּיְלָה‬/‫שׁלֶת הַיּוֹם‬

ֶ ‫' ְל ֶמ ְמ‬for rule of the day/night' and ‫ִלמְשׁ ֹל בַּיּוֹם‬
‫' וּ ַב ַלּיְלָה‬to rule the day and the night'. He writes the following:
And it should be investigated if 'for rule of the day' is the same as 'and to rule
the day' and 'for rule of the night' as ... 'and to rule the night'. For Aquila also
preserved the parallel, having made it 'for authority' instead of 'for rule' and 'to
exercise authority' instead of 'and to rule' ... Aquila, who was ambitious to
translate most literally, has done no less than [distinguishing] the noun and the
predicate [forms].43
It is unclear from the passage if Origen actually understood the underlying Hebrew or if he
was merely relying on translations. The fact that Origen regards Aquila to have "preserved
the parallel" supports the idea that Origen was aware of this difference in the Hebrew text;
this would reflect a somewhat sophisticated grammatical understanding of Hebrew. On the
other hand, the belief that Aquila translated the Hebrew literally here does not necessarily
mean that Origen was able to discern this grammatical feature in the Hebrew text himself. It
42. Videtur sane secundum hanc expositionem quam de Cineo edidimus, et conjunctio interturbare sensum. Sed
sciendum est, quod vernaculum est Hebraeae linguae et conjunctione frequenter uti, ita ut interdum abundet, et
in non necessariis, ut in nostra apparet lingua, videatur inserta: quod utique cum venia accipiendum est. Habet
enim unaquaeque lingua aliquid proprietatis, quod apud alias linguas vitiosum videatur. Et hic ergo, et
conjunctio quasi abundans habenda est, et superflua.
43. ζητητέον δὲ εἰ ταὐτόν ἐστι τό· εἰς ἀρχὰς τῆς ἡµέρας τῷ· καὶ ἄρχειν τῆς ἡµέρας· καὶ τό· εἰς ἀρχὰς τῆς
νυκτός τῷ ... καὶ ἄρχειν τῆς νυκτός. καὶ ὁ Ἀκύλας γὰρ τὸ ἀνάλογον ἐτήρησε, ποιήσας ἀντὶ µὲν τοῦ εἰς ἀρχάς, εἰς
ἐξουσίαν· ἀντὶ δὲ τοῦ καὶ ἄρχειν, ἐξουσιάζειν ... ὁ κυριώτατα ἑρµηνεύειν φιλοτιµούµενος Ἀκύλας οὐκ ἄλλο
πεποίηκε παρὰ τὴν προσηγορίαν καὶ τὸ κατηγόρηµα.
- 28 -
is quite possible that, due to his high esteem for Aquila and his reputation as the most literal
translator, Origen simply assumed that such a conspicuous translation choice would not have
been arbitrary.
In sum, Origen's discussion regarding the conjunction waw and the distinction be-
tween "nouns" and "predicates" in the Genesis passage would support the idea that Origen's
knowledge of the language extended beyond the alphabet and etymologies into the realm of
grammar. It is striking, though, that there are so few examples of Hebrew grammatical com-
ments in Origen's writings. The relative paucity of such examples, in light of Origen's facility
in discussing Greek grammar, points to a very limited knowledge of Hebrew grammar for
Origen.
2.2.2.4. Comparing Greek Translations with Hebrew
There are a number of passages in which Origen appears to be able to intelligently compare
the Greek translations to the Hebrew, sometimes correcting the Greek and sometimes merely
citing the differences (for the examples below, see HANSON 1959, 164–66, 172–75).
In Selecta in Psalmos (12.1168.4–14), when discussing the phrase 'from the fruit of
grain' (ἀπὸ καρποῦ σίτου) in the Greek translation of Psalms 4:8, Origen argues correctly that
the Hebrew has 'from the time [of the grain]' (ἀπὸ καιροῦ) (cf. ‫) ֵמעֵת דְּ גָנָם‬.44 In the same book
(12.1116.44–48), when discussing the phrase 'from the right way' in the LXX translation of
Psalms 2:12, he rightly claims that "[the word right] is not added to the noun, neither in the
Hebrew nor in the other translations" and that perhaps "the manuscripts [of the LXX] have
erred."45 In Commentarii in evangelium Joannis (6.6.7–10), Origen argues that the phrase 'the
Lord is just and has loved righteousness' (δίκαιος κύριος, καὶ δικαιοσύνας ἠγάπησεν) in the
Greek translation of Psalms 11:7 is a justified reading because "[he] found it thus in the accu-
44. Aquila, Symmachus, and Theodotion all have ἀπὸ καιροῦ (FIELD 1875, 91).
45. οὔτε ἐν τῷ Ἑβραϊκῷ πρόσκειται, οὔτε ἐν ταῖς λοιπαῖς ἑρµηνείαις τὸ, δικαίας ... µήποτε ... τὰ ἀντίγραφα
ἡµάρτηται. He also admits the possibility of the LXX translators making the change according to a "divine
device" (κατ᾽ οἰκονοµίαν) (HANSON 1959, 164).
- 29 -
rate manuscripts, in the rest of the versions besides the Seventy, and in the Hebrew."46 Else-
where in the same book (10.40.282–83), when using the structure of the temple to make a
spiritual application, he claims that the translators did not know the meaning of the word
dabir (δαβειρ) and that others wrongly conflated it with the temple:
The two cherubim were in the dabir (δαβειρ), which those who translated the
Hebrew into the Greek have not been able to interpret correctly. But through a
misuse of language some have said that the thing which happens to be more
precious than the temple (i.e., the dabir) is the temple itself.47
In Contra Celsum (5.48.15–18), when discussing the phrase 'the blood of the circumcision of
my child stayed' (ἔστη τὸ αἷµα τῆς περιτοµῆς τοῦ παιδίου µου) in the LXX translation of Ex-
odus 4:25, Origen writes that "according to the Hebrew itself" (κατὰ δὲ τὸ ἑβραϊκὸν αὐτό),
the reading is 'you are a bridegroom of blood to me' (Νυµφίος αἱµάτων σύ µοι).48
Even though Origen often cites "the Hebrew" as supporting evidence for a given read-
ing (often against the reading of the LXX), it is possible that he is merely trusting that certain
translators, known for their faithfulness to the Hebrew like Aquila, have accurately represent-
ed the Hebrew. This may be hinted at in a passage in Commentarii in evangelium Joannis
(6.41.212), in which Origen discusses the mistakes in proper names in Greek manuscripts
used by Christians:
And it is possible to see the same type of mistake in many places in the Law
and the Prophets, as we have verified by learning from Hebrews and compar-
ing our manuscripts to theirs, which are witnessed by the never-yet-distort-
ed versions of Aquila and Theodotion and Symmachus.49
46. οὕτω γὰρ ἐν τοῖς ἀκριβέσιν ἀντιγράφοις εὕροµεν καὶ ταῖς λοιπαῖς παρὰ τοὺς ἑβδοµήκοντα ἐκδόσεσι καὶ τῷ
Ἑβραϊκῷ.
47. Τὰ µέντοι δύο χερουβεὶµ ἐν τῷ δαβεὶρ ἦν, ὅπερ οὐ δεδύνηνται ἑρµηνεῦσαι κυρίως οἱ µεταλαµβάνοντες εἰς
Ἑλληνισµὸν τὰ Ἑβραίων. Καταχρηστικώτερον δέ τινες ναὸν αὐτὸν εἰρήκασιν τοῦ ναοῦ τιµιώτερον τυγχάνοντα.
48. It is difficult to determine if Origen's comment was based on an understanding of the Hebrew or on a
particular translation. This verse is variously attested among the versions (FIELD 1875, 85–86): Symmachus:
νυµφίος αἱµάτων σύ µοι. Theodotion: νυµφίος αἱµάτων σύ µοι. Τὸ Ἑβραικόν: νυµφίος αἵµατος σύ µοι. Aquila
and Theodotion (Syro-Hexapla): νυµφίον αἵµατος ἔχω. ὁ Ἑβραῖος: ἐσφράγισε τὸ αἷµα τῆς περιτοµῆς.
49. Τὸ δ᾽ ὅµοιον περὶ τὰ ὀνόµατα σφάλµα πολλαχοῦ τοῦ νόµου καὶ τῶν προφητῶν ἔστιν ἰδεῖν, ὡς
ἠκριβώσαµεν ἀπὸ Ἑβραίων µαθόντες, καὶ τοῖς ἀντιγράφοις αὐτῶν τὰ ἡµέτερα συγκρίναντες, µαρτυρηθεῖσιν
ὑπὸ τῶν µηδέπω διαστραφεισῶν ἐκδόσεων Ἀκύλου καὶ Θεοδοτίωνος καὶ Συµµάχου.
- 30 -
This passage may be interpreted to mean that Origen regarded the versions of Aquila,
Theodotion, and Symmachus as accurately reflecting the Hebrew text of the Jews. Accord-
ingly, it ought to be considered that, if the other versions represented a consensus, Origen
might have regarded them as accurately reflecting the Hebrew without needing to check it
himself.50 After all, in Epistula ad Africanum (11.52.21–24), Origen says that Aquila "is be-
lieved by the Jews to have translated the Scripture most zealously, whom those who do not
know the Hebrew language are especially accustomed to use, as he is more successful than
all"51 (cf. HANSON 1959, 172).52 It is possible that this high esteem for Aquila may have even
misled Origen at times.53 In Homiliae in Canticum Canticorum (13.50.43–46), when compar-
ing the translation of the LXX and Aquila for the word ü ֶ‫' אֵיד‬your calamity' in Proverbs 27:10,
Origen argues that "instead of that which the Septuagint interprets as 'unfortunate' (infelix),
Aquila, expressing the actual Hebrew, has placed 'rustic' (ἀγροικός)."54 Unless Aquila and
Origen were dealing with another Hebrew text, the LXX has the better translation here.55
However, there is also evidence that Origen was indeed able to compare the Hebrew
and Greek columns successfully. In Commentarium in evangelium Matthaei (16.19.72-86),
50. See Origen's statement in In Jeremiam (14.3.4–5): "for in most manuscripts ... but in the most accurate
manuscripts and [those which] agree with the Hebrew ..." (HANSON 1959, 175).
51. φιλοτιµότερον πεπιστευµένος παρὰ Ἰουδαίοις ἡρµηνευκέναι τὴν Γραφήν· ᾦ µάλιστα εἰώθασι οἱ ἀγνοοῦντες
τὴν Ἑβραίων διάλεκτον χρῆσθαι, ὡς πάντων µᾶλλον ἐπιτετευγµένῳ.
52. HANSON entirely misses the point of this statement by Origen, mistranslating the phrase 'those who do not
know the Hebrew language' (οἱ ἀγνοοῦντες τὴν Ἑβραίων διάλεκτον) as 'those who know the Hebrew language'
(1959, 172).
53. HANSON notes that Origen uses Aquila's translation to correct the Hebrew text and the translation of the
LXX for Genesis 2:4 (1959, 172).
54. Pro eo quod Septuaginta, infelix, interpretati sunt, Aquila Hebraeam exprimens veritatem ἀγροικός posuit.
55. Another example of this is found in Origen's comments (Selecta in Psalmos, 12.1064.32–39) on the various
renderings of the Hebrew superscription of Psalm 45 (‫שׁנִּים‬ַ ֹ ‫' עַל־שׁ‬on the lilies'). Aquila has 'on the lilies' (ἐπὶ τοῖς
κρίνοις) and Symmachus has 'about the flowers' (περὶ τῶν ἄνθεων). While Origen claims that the original
Hebrew means either 'concerning the lilies' or 'concerning the flowers', he argues that the LXX translation fits
with the others because the appearance of flowers changes quickly (HANSON 1959, 173). However, it is worth
noting that the LXX rendering may be the result of the translator associating it with the root ‫' שׁנ"י‬to change'.
- 31 -
when discussing the reason why the quotation of Psalms 118:25–26 in Matthew 21:9 seems
to skip over a line, Origen writes:

And it seems to me that the [phrase] that has been placed instead of, 'O Lord,
please save', namely, 'Blessed is the one who comes in the name of the Lord',
is referenced in Hebrew in the phrase, 'Hosanna to the son of David'. And so
the Hebrew text also has, 'ANNA ADONAI, OSIANNA, ANNA ADONAI,
ASLIANNA, BAROUCH ABBA BSAIM ADONAI'. It seems to me, then,
that the gospels, being copied continually by the Greeks, who did not know
the language, became commingled with respect to the placement of these
[words quoted] from the aforementioned psalm.56
Origen further supports his argument by appealing to Aquila's translation (16.19.86–91).
Even though he was helped by Greek translations, this passage demonstrates that Origen was
able to coordinate and compare the Greek translations with the Greek transcription of the sec-
ond column in order to elucidate text-critical issues.
In sum, a survey of these texts demonstrates that Origen was capable of utilizing He-
brew in his text-critical discussions, even to the point of evaluating the accuracy of the Greek
translations. However, these passages also show that his access to the Hebrew may have
largely depended on his utilization of those translations that he regarded as particularly faith-
ful to the original Hebrew.

2.2.3. Origen's Use of the Secunda and the Purpose of Including It in the Hexapla
Up to this point, it has been demonstrated that even though Origen lacked the Hebrew skill
necessary to compose the Secunda himself, he did know Hebrew to some degree. Neverthe-
less, even though Origen did not compose the second column himself, it is clear that he inter-
acted with it in his writings and study. It is fitting, then, to conclude this section on Origen's
Hebrew knowledge with a couple examples that illustrate how Origen used the text of the
second column.
56. δοκεῖ δέ µοι τὰ ἀντὶ τοῦ ὢ κύριε, σῶσον δὴ προτεταγµένα τοῦ εὐλογηµένος ὁ ἐρχόµενος ἐν ὀνόµατι κυρίου
ἑβραϊκῶς ἐκκεῖσθαι ἐν τῷ ὡσαννὰ τῷ υἱῷ Δαυΐδ οὕτω δὲ καὶ εἶχεν ἡ ἑβραϊκὴ λέξις ΑΝΝΑ ΑΔΩΝΑΙ
ΩΣΙΑΝΝΑ, ΑΝΝΑ ΑΔΩΝΑΙ ΑΣΛΙΑΝΝΑ, ΒΑΡΟΥΧ ΑΒΒΑ ΒΣΑΙΜ ΑΔΩΝΑΙ. εἶτα δοκεῖ µοι ὑπὸ Ἑλλήνων
συνεχῶς γραφόµενα τὰ εὐαγγέλια µὴ εἰδότων τὴν διάλεκτον, συγκεχύσθαι ἐν τοῖς κατὰ τὸν τόπον ἔχουσι ταῦτα
ἀπὸ τοῦ προειρηµένου Ψαλµοῦ.
- 32 -
For the first example, we may return to a passage examined above from Epistula ad
Africanum (11.77), in which Origen's comments on the terms for 'man' and 'woman' seem to
betray his modus operandi with the text of the second column:
The Hebrews say that the woman was called essa and that 'I took' is made
clear from the word, as is evident from the [verse], 'Chos isouot essa', which is
translated as, 'I will take up the cup of salvation'. And the man is is, as is ap-
parent from the [verse], 'Esre ais', which is, 'Blessed is the man'.57
In each example, Origen proves the meaning of a transcribed Hebrew word by citing an in-
stance of that same transcription elsewhere in the Bible along with its translation into Greek.
In the case of εσσα ‫ ִאשָּׁה‬, this actually leads Origen to erroneously associate it with the verb
λαµβάνω, which would have been on the same line as εσσα ‫ ֶאשָּׂא‬in the Hexapla for Psalms
116:13. In the case of ις ‫אִישׁ‬, Origen correctly associates it with the transcription ἀΐς ‫ ָהאִישׁ‬,
which would have been on the same line as ὁ ἀνήρ in the Hexapla for Psalms 1:1. If these
passages are representative of Origen's modus operandi, it seems that he learned the meaning
of transcribed Hebrew words by comparing multiple instances of the same transcription with
their various Greek translations in parallel lines of the Hexapla. It is worth noting that Epistu-
la ad Africanum is likely dated after the completion of the Hexapla (HANSON 1954, 26).
The second example is even more telling than the first. In Selecta in Psalmos
(12.1057.42–48), when discussing the phrase διάψαλµα in the Greek translation of the
Psalms, Origen writes:

Having frequently searched for the reason that diapsalma was inscribed/sub-
scribed between the psalms, at last having made a close examination of the
Hebrew (τῷ Ἑβραϊκῷ) and examining the Greek alongside it, I found that
where the Hebrew (τὸ Ἑβραϊστὶ) has sel (σὲλ) and the Greek has aei or some-
thing equivalent to it, there the Seventy, Thedotion, and Symmachos assigned
diapsalma.58
57. Φασὶ δὲ οἱ Ἑβραῖοι ἐσσὰ µὲν καλεῖσθαι τὴν γυναῖκα· δηλοῦσθαι δὲ ἀπὸ τῆς λέξεως τὸ ἔλαβον, ὡς δῆλον ἐκ
τοῦ· Χῶς ἰσουὼθ ἐσσά, ὅπερ ἑρµηνεύεται· ἴς δὲ τὸν ἄνδρα, ὡς φανερὸν ἐκ τοῦ· Ἐσρὴ ἀΐς, ὅπερ ἐστί· Μακάριος
ἀνήρ.
58. Πολλάκις ζητήσας τὴν αἰτίαν τοῦ ἐπιγράφεσθαι µεταξὺ τῶν ψαλµῶν διάψαλµα, ὕστερον παρατηρήσας ἐν
τῷ Ἑβραϊκῷ, καὶ συνεξετάζων αὐτῷ τὸ Ἑλληνικὸν, εὗρον, ὅτι ὅπου τὸ Ἑβραϊστὶ σὲλ, Ἑλληνιστὶ δὲ ἀεὶ, ἤ τι
τούτῳ ἰσοδυναµοῦν, ἐκεῖ οἱ Ἑβδοµήκοντα, καὶ Θεοδοτίων, καὶ Σύµµαχος ἔταξαν τὸ διάψαλµα.
- 33 -
Origen then goes on to illustrate this point by citing the various correspondences of διάψαλµα
and σελ in Psalms 75 and 76 (Selecta in Psalmos, 12.1057.48–12.1060.35).
This passage should remove any doubt as to whether or not Origen himself composed
the Hexapla for a couple of reasons. First, the fact that Origen transcribes Hebrew ‫ סלה‬as σελ
makes it clear that his reference to "the Hebrew" (τὸ Ἑβραϊκόν) in this passage refers to the
text of the second column.59 Second, this passage demonstrates that Origen was ignorant of
the solution to his philological problem until he was able to examine "the Hebrew" (τὸ
Ἑβραϊκόν). It is highly unlikely that someone who had transcribed the Hebrew of Psalms into
Greek would have been unfamiliar with the word ‫ ֶסלָה‬, which appears 72 times in the book.
Origen's need to examine the text demonstrates both that he did not compose it himself and
that he was in possession of a text that could help him.
What emerges from this passage is a clear picture of how Origen used the various
texts that were before him, whether they had yet been compiled into the form of the Hexapla
at this time or not. Origen, who was most familiar with the LXX, noticed that the unusual
word διάψαλµα, presumably innovated by the LXX translators, occasionally intervened in the
text of the Psalms. As he was accustomed to do, he attempted to find some correlation be-
tween διάψαλµα and the parallel renderings in the other Greek translations. However, such a
comparison was unsuccessful for a couple of reasons. First, while the alternative renderings
of the other versions such as ἀεί 'evermore' and εἰς τ(ὸν) αἰῶνα 'forever' might be used to
translate other words and phrases, διάψαλµα is only used to translate Hebrew ‫ ֶסלָה‬. Second,
while the LXX always renders ‫ ֶסלָה‬as διάψαλµα (HATCH and REDPATH 1897, 316), the other
versions are not as consistent. For example, while Aquila usually translates ‫ ֶסלָה‬as ἀεί (HATCH
59. Hebrew ‫ ֶסלָה‬is normally transcribed as σελ in the Ambrosiana palimpsest (Ps. 46:4, 8, 12; 49:14; 89:38, 46,
49) and only once as σελα (Ps. 32:7). The fact that a short form of ‫( סלה‬i.e., σελ) is attested in no Hebrew text
other than the second column makes it highly likely that Origen is referring to the text of the second column
here. However, in Theodotion's translation of Psalm 89 as attested in the Ambrosiana palimpsest, ‫ ֶסלָה‬is
regularly rendered as |ἀεί σελ| (Ps. 89:38, 47, 49). In the Quinta, ‫ ֶסלָה‬is transcribed as σελα in Habakkuk 3:3
(HATCH and REDPATH 1897, 1262).
- 34 -
and REDPATH 1897, 28), in Psalms 39:12 he translates it as ᾆσµα 'song' (FIELD 1875, 149).
Therefore, had Origen relied solely on the Greek translations to elucidate the word διάψαλµα
in the LXX, he would not have been able to find a consistent correspondence. Finding the
Greek translations insufficient, Origen finally took recourse to the text of the second column,
where he found that διάψαλµα in the LXX consistently corresponded with Hebrew σελ.
We may assume that Origen's use of the Secunda was consistent with his motivations
and purposes for including it in the Hexapla. From a survey of how Origen uses the Hexapla
in his writings, and especially in the examples cited here, emerge two primary purposes.
First, the second column served Origen as a concordance in his text-critical work in a
way that neither the Greek versions nor the Hebrew consonantal text could. Origen was ac-
customed to compare the various Greek versions word-by-word when dealing with a passage
textually, sometimes comparing other passages in which the same Greek word appears. That
Origen did this without direct reference to the Hebrew is demonstrated by his text-critical use
of the Greek versions in the early part of his commentary on Psalms, dated to his time in
Alexandria (CLEMENTS 1997, 97–99). However, as the discussion regarding selah demon-
strates, such a method had its limits. If Origen wanted to trace the occurrence of a particular
word throughout the scriptures, he would only be successful if the translations were consis-
tent. When they were not, seeking out a particular pattern would be a fruitless endeavor. For
these sorts of issues, he needed to reference the original. His lack of facility in vocalizing He-
brew would have made the Hebrew consonantal text too difficult and ambiguous for such a
purpose. The solution was found in the text of the second column. Rather than get lost in the
potentially inconsistent renderings of the Greek versions, the second column functioned as a
"key" for tracing certain words through the scriptures. By using the second column in this
way, Origen was able to keep his primary focus on the Greek translations, where he was com-
fortable, but established a method by which his efforts would not lead to false assertions in-
consistent with the Hebrew. For the sake of illustration, Origen's work with the Tetrapla in
- 35 -
Alexandria might be compared to one using four English translations of the Bible each with
its own corresponding English concordance. Origen's work with the second column alongside
the Greek versions in Caesarea might be compared to one using four English translations of
the Bible, but with a keyed-to-Hebrew concordance for each of them. Clearly, the second
circumstance would be far more helpful when comparing textual variants.
Second, the text of the Secunda provided Origen with one more source for increasing
his Hebrew knowledge. By constantly comparing the transcribed words in the second column
with their various translations in the Greek versions, Origen was able to add to his Hebrew
vocabulary. Moreover, it seems that Origen was aware that some words could be translated in
different ways and thus it was important to compare multiple passages before concluding that
a particular (transcribed) Hebrew word could be matched with a corresponding Greek word
(e.g., see the discussion regarding ἴς 'man' and Ἐσρὴ ἀΐς 'blessed is the man').
2.2.4. Conclusions: A Portrait of Origen the Hebrew Scholar
Having thoroughly investigated the evidence for Origen's Hebrew knowledge, we may now
paint a portrait of the early church father qua Hebrew scholar. To begin, it is necessary to see
Origen as a lifelong student of Hebrew. He constantly built upon his limited knowledge of
Hebrew in two ways. First, he made use of Jewish sources, interacting with both written texts
and human interlocutors. Through such interaction, he encountered etymological name lists,
heard exegetical nuggets on a variety of passages, and learned other tidbits regarding the He-
brew language. This resulted in a knowledge of Hebrew that was rather piecemeal, largely
dependent on the information he received from those whom he readily accepted as more ex-
pert in the Hebrew language than himself. Second, Origen seems to have used the text of the
second column and the other translations of the Hexapla as a study tool for growing in his
knowledge of Hebrew. Origen interacted with the texts of the Hexapla, and especially the
second column, as if he were building a keyed-to-Hebrew concordance in his head. There is
- 36 -
no doubt that his long hours working with the Hexapla would have, at the very least, helped
him to build his Hebrew vocabulary and gain a sense of the usage of the language.
The actual objects of Origen's Hebrew knowledge may be described in a few parts.
First, it seems that Origen had taken the time to learn the Hebrew alphabet, but did not go far
enough with the language to be comfortable reading the unvocalized script without errors.
Rather, he preferred to access the Hebrew through a comparison of the Greek translations (es-
pecially Aquila) and the transcribed Hebrew in the second column.60 Second, through the in-
formation he gleaned from Jewish interlocutors, the etymological name lists, and the text of
the second column, Origen knew many Hebrew words. While there is some evidence that
Origen had some basic knowledge of Hebrew grammar, his knowledge of Hebrew was essen-
tially a lexical one; that is, his knowledge of the language consisted primarily of knowing
Greek-Hebrew correspondences in both directions. With the help of the Greek translations
and the second column, Origen had enough of a working knowledge of the language and its
grammar to engage Hebrew in his text-critical and commentary work, yet not without
mistakes.
The passages examined above paint Origen as a very resourceful scholar with a phe-
nomenal memory and a brilliant mind, but one who did not have the sort of familiarity with
Hebrew to compose the second column himself. Rather, he utilized it as a tool for his textual
and exegetical work. If one does not begin with the assumption that Origen knew Hebrew
well (contra Eusebius and Jerome) and instead regards him as a theologian using various
tools at his disposal to learn Hebrew, his achievements are impressive. He was a scholar who
made excellent use of the resources he had, even using, it seems, the Greek transcription text
of the second column to help him discover new aspects of the Hebrew language. Neverthe-
less, his significant relationship with the second column, though an important part of his writ-
60. In fact, as DE LANGE writes, "τὸ Ἑβραϊκόν frequently, if not always, refers to the second, not the first,
column of the Hexapla" (1976, 153).
- 37 -
ings, cannot be conceived of as one of authorship. Origen did indeed labor long hours over
the second column of the Hexapla—not as its author, but as its student.
2.3. THE SECUNDA AND THE COMPOSITIONAL HISTORY OF THE HEXAPLA

2.3.1. Origen and the Secunda in Caesarea
If Origen did not compose the text of the second column himself, he either commissioned
others more expert in Hebrew than himself for the task (see section 4) or made use of an al-
ready existing text. In order to determine which of these two alternatives is more likely, it is
first necessary to understand how the first two columns fit into the overall process of the
composition of the Hexapla.
While there is scholarly consensus that Origen completed the Hexapla after he relo-
cated from Alexandria to Caesarea in 233 CE (CROUZEL 1985; CLEMENTS 2000, 341; GRAFTON
and WILLIAMS 2006, 17), there is debate about how and when the process of composition be-
gan. In order to spare a full review of scholarship on this issue, which is beyond the scope of
the present work, it will suffice to say that the most convincing argument yet put forth is that
of CLEMENTS (1997, 2000). According to her, Origen first compiled the Tetrapla (i.e., Aquila,
Symmachus, LXX, and Theodotion in four parallel columns) in Alexandria. Later, after Ori-
gen relocated to Caesarea and came into contact with new Jewish sources, he added the two
Hebrew columns to make the Hexapla. Some of Origen's assistants must have been skilled
enough in Hebrew to correlate the Hebrew columns with the Greek translations when they
were added (CLEMENTS 1997, 96–100).
Such an explanation facilitates a more persuasive argument regarding the columnar
order of the Hexapla, which has proved difficult for many scholars who have attempted to ex-
plain it (e.g., ORLINSKY 1936). For CLEMENTS, when Origen refers to "our copies" and the "He-
brew copies" or "those of the Jews" in Epistula ad Africanum, he is actually distinguishing
between the different Greek translations of the Tetrapla. The LXX and Theodotion, which
exhibit textual similarity, were considered more "Christian" translations. Aquila and Sym-
- 38 -
machus, which were based on the Masoretic Text, were considered more "Jewish" versions.
Having Aquila and Symmachus next to each other to the left of the LXX allowed Origen to
treat them as a pair to determine the more "Jewish" reading to which the LXX might be com-
pared. Having Theodotion to the right of the LXX allowed Origen to compare the LXX with
the translation most closely related to it textually (CLEMENTS 1997, 93–94, 96–97).
CLEMENTS is correct to dismiss the claim of NAUTIN (1977, 333–43), who argues that
Origen acquired the text of the second column as a liturgical text from the Greek-speaking
Jews in Alexandria. This is unlikely for a couple reasons. First, unless the sages of Palestine
exerted significant influence over the Alexandrian Jews and compelled them to read the scrip-
tures in Hebrew, we may assume that the scriptures would have continued to be read in Greek
in Alexandria. Second, in the wake of the rebellion in 117 CE, it is unlikely that a significant
Jewish community had once again developed in Egypt by the beginning of the third century
CE (CLEMENTS 1997, 95–96; GRAFTON and WILLIAMS 2006, 111).
On the other hand, there are good reasons for postulating that the addition of the He-
brew columns to the Hexapla took place during Origen's time in Caesarea. First, unlike
Alexandria, Caesarea came to house a significant Jewish community by the end of the second
century CE. Out of this Jewish community would emerge one of the most significant rabbinic
schools of Palestine in the third century CE. Second, it is in Origen's writings dated to his time
in Caesarea that references to discussions with Jewish interlocutors increase. The constant de-
bates between Christians and Jews would have motivated both groups to examine and com-
pare their scriptures more thoroughly (e.g., see bˁAvoda Zara 4a). Third, and finally, given the
Jewish scholarly presence there, it is more likely that a text like the Secunda would have de-
veloped in Caesarea than in Alexandria (MURRAY 2000; GRAFTON and WILLIAMS 2006, 111; see
also section 3.4.4 on the "Rabbis of Caesarea").
- 39 -
2.3.2. Did Origen Commission the Second Column?
It is entirely possible that Origen commissioned Hebrew experts to compose the text of the
second column during his time in Caesarea. The best argument in favor of this is the sheer ex-
tent of the Secunda. Although only a very small fraction of it has survived until modern
times, it is assumed that at one point it constituted a Greek transcription of the entire Hebrew
Bible.61 A text-critical endeavor focused on comparing every variant would naturally compel
a comprehensive treatment of the material.62 Moreover, the fact that Origen tends to use the
second column as a sort of working concordance, as shown above, further supports this idea.
On the other hand, it is more likely that Origen obtained a pre-existing text for the fol-
lowing reasons: First, the general consistency of the transcription conventions of the Secunda
should not be taken for granted. It is by no means guaranteed that such a consistent transcrip-
tion system should emerge out of a first attempt at transcribing the biblical text (cf. BRØNNO
1943, 7). Second, every other text found in the Hexapla was a pre-existing text that Origen
had collected. Third, neither in Origen's writings nor in any of the ancient accounts about the
composition of the Hexapla is there ever a mention of Origen composing the second column
or commissioning others to do so. In fact, if anything, their descriptions of the work support
the idea that he obtained the text from another source. These accounts will be examined in
more detail in the following section.

2.3.3. Early Church Fathers on the Hexapla
References to Origen's text-critical work and composition of the Hexapla are found in a num-
ber of church fathers' writings. In addition to his own comments in Epistula ad Africanum,
61. Quotations of the Secunda in the church fathers include the following biblical books: Genesis, Exodus,
Leviticus, Numbers, 1 Samuel, 1 Kings, 2 Kings, Isaiah, Jeremiah, Ezekiel, Hosea, Malachi, Psalms, Proverbs,
Lamentations. Moreover, in the descriptions of the Hexapla in the early church fathers, who made a point to
describe features peculiar to certain books (e.g., extra translations in Psalms), it is nowhere mentioned that the
second column was incomplete.
62. CLEMENTS (1997, 95) argues that only the transliteration of "key terms" would be necessary if the second
column was primarily to be a text-critical tool. However, such a position presumes the text-critical work has
already been done. If Origen had difficulty accessing the Hebrew consonantal text, a transcription of the entire
Hebrew Bible really would have been necessary for him to compare every textual variant.
- 40 -
descriptions of his work are found in Eusebius, Jerome, Epiphanius, and Rufinus. Because of
the relevance of these ancient accounts for the present and upcoming discussion, the passages
are quoted below (phrases related to compositional history in bold):63

Origen: And I say these things not because I shrink from also investigating the Jew-
ish scriptures, and comparing all our scriptures with theirs, and seeing the differ-
ences in them ... this we have already done to a high degree as far as possible, inves-
tigating the sense in all the versions and their variants ...64
Eusebius: So great an exacting study of the divine words was introduced to Origen,
that he thoroughly learned the Hebrew language, and acquired as his own posses-
sion the original scriptures held by the Jews in the very letters of the Hebrews. And
he tracked down the versions of the others who had translated the Holy Scriptures
besides the Seventy ... and having gathered them all together into the same
[book], and having divided them by phrase and having set them opposite each
other along with the Hebrew writing itself, left us with manuscripts of that which is
called the Hexapla.65
Jerome: It was our concern to correct all the books of the old law, which the
learned man Adamantius (i.e., Origen) had arranged (digesserat), copied from
the library of Caesarea, from the original [copies] themselves, in which even the
Hebrew words themselves are copied in their very own characters: and with Greek
letters expressed in the nearby column. Aquila also, and Symmachus, the Septuagint
as well as Theodotion, [each] hold their own order.66
Epiphanius (Panarion): Ambrose provided [Origen], the shorthand writers, and his
assistants with food, along with papyrus and the other costs. Origen, through sleep-
less nights and greatest deprivation completed the task of writing. First, he was ea-
ger to carefully gather and set forth the books of the six, Aquila, Symmachus, that
63. Except for the final text of Epiphanius, these texts are found in GRAFTON and WILLIAMS 2006, 89–95, 316–
20. All translations are my own, but they are made in consultation with those of GRAFTON and WILLIAMS.
64. Epistula ad Africanum (11.60.9–15): Καὶ ταῦτα δὲ φηµὶ οὐχὶ ὄκνῳ τοῦ ἐρευνᾷν καὶ τὰς κατὰ Ἰουδαίους
Γραφὰς, καὶ πάσας τὰς ἡµετέρας ταῖς ἐκείνων συγκρίνειν, καὶ ὁρᾷν τὰς ἐν αὐταῖς διαφοράς ... ἐπὶ πολὺ
τοῦτο, ὅση δύναµις, πεποιήκαµεν, γυµνάζοντες αὑτῶν τὸν νοῦν ἐν πάσαις ταῖς ἐκδόσεσι καὶ ταῖς διαφοραῖς
αὐτῶν. It is not clear if Origen is referring directly to the Hexapla here, but presumably he had completed the
work of the Hexapla before writing this letter (HANSON 1954, 26).
65. Historia ecclesiastica (6.16.1, 6.16.4): Τοσαύτη δὲ εἰσήγετο τῷ Ὠριγένει τῶν θείων λόγων ἀπηκριβωµένη
ἐξέτασις ὡς καὶ τὴν Ἑβραΐδα γλῶτταν ἐκµαθεῖν τάς τε παρὰ τοῖς Ἰουδαίοις ἐµφεροµένας πρωτοτύπους αὐτοῖς
Ἑβραίων στοιχείοις Γραφάς κτῆµα ἴδιον ποιήσασθαι. ἀνιχνεῦσαί τε τὰς τῶν ἑτέρων παρὰ τοὺς ἑβδοµήκοντα
τὰς ἱερὰς γραφὰς ἑρµηνευκότων ἐκδόσεις ... ταύτας δὲ ἁπάσας ἐπὶ ταὐτὸν συναγαγὼν διελών τε πρὸς κῶλον
καὶ ἀντιπαραθεὶς ἀλλήλαις µετὰ καὶ αὐτῆς τῆς Ἑβραίων σηµειώσεως. τὰ τῶν λεγοµένων Ἑξαπλῶν ἡµῖν
ἀντίγραφα καταλέλοιπεν.
66. Commentarii in Epistolam ad Titum, 3.9 (26.734–735): Nobis curae fuit omnes veteris legis libros quos vir
doctus Adamantius in Hexapla digesserat de Caesariensi bibliotheca descriptos ex ipsis authenticis emendare,
in quibus et ipsa Hebraea propriis sunt characteribus verba descripta: et Graecis litteris tramite expressa
vicino. Aquila etiam et Symmachus, Septuaginta quoque et Theodotio suum ordinem tenent.
- 41 -
of the Seventy-Two and Theodotion, and the fifth and sixth versions.67 He set
alongside them every Hebrew word and the Hebrew letters themselves together
with it. And opposite, in parallel, making use of the second column for a compo-
sition of Hebrew words through Greek letters, has made yet another column of
[this] composition. So that these books are also called Hexapla, since in addition to
the Greek translations there were two juxtapositions together, Hebrew naturally
through Hebrew letters and Hebrew through Greek letters, so that [the result was]
the entire Old Testament through that which is called Hexapla and through the two
[columns] of the Hebrew words.68
Epiphanius (Weights and Measures): At that time, he placed together both the
Hexapla69 and the two columns of Hebrew opposite in parallel, one translation fac-
ing the other, having named the books Hexapla, so he could examine upwards and
across the width ... and these four columns, having been joined to the two Hebrew
columns are called Hexapla. And if also the fifth and the sixth translation should be
joined to these in order, they are called Octapla.70 Now I am referring to the six
translations and the other two, the one written with the very Hebrew letters and
words, and the one written with Greek letters but Hebrew words.71
Epiphanius (Weights and Measures): For having placed together the six transla-
tions and the Hebrew writing in Hebrew letters and its own words in the first col-
umn, another column he placed at its side, [which was expressed] through Greek
letters but Hebrew words, so that those who do not know Hebrew letters could ap-
prehend to know, through the Greek letters, the power/meaning of the Hebrew ora-
cles. And thus, by means of that which is called by him Hexapla or Octapla, having
placed the two Hebrew columns and the columns of the six translators opposite
67. The fifth and sixth versions were additional Greek translations of the Psalms found by Origen.
68. Panarion (2.406–408): τοῦ µὲν Ἀµβροσίου τὰ πρὸς τροφὰς αὐτῷ τε καὶ τοῖς ὀξυγράφοις καὶ τοῖς
ὑπηρετοῦσιν αὐτῷ ἐπαρκοῦντος, χάρτην τε καὶ τὰ ἄλλα τῶν ἀναλωµάτων, καὶ τοῦ Ὠριγένους ἔν τε ἀγρυπνίαις
καὶ ἐν σχολῇ µεγίστῃ τὸν κάµατον τὸν περὶ τῆς γραφῆς διανύοντος. ὅθεν τὸ πρῶτον αὐτοῦ ἐπιµελῶς
φιλοτιµησαµένου συναγαγεῖν τῶν ἓξ ἑρµηνειῶν, Ἀκύλα Συµµάχου τῶν τε ἑβδοµήκοντα δύο καὶ Θεοδοτίωνος,
πέµπτης τε καὶ ἕκτης ἐκδόσεως τὰς βίβλους ἐξέδωκεν, µετὰ παραθέσεως ἑκάστης λέξεως Ἑβραϊκῆς καὶ αὐτῶν
ὁµοῦ τῶν Ἑβραϊκῶν στοιχείων· ἐκ παραλλήλου δὲ ἄντικρυς, δευτέρᾳ σελίδι χρώµενος κατὰ σύνθεσιν
Ἑβραϊκῆς µὲν τῆς λέξεως, δι’ Ἑλληνικῶν δὲ τῶν γραµµάτων ἑτέραν πάλιν πεποίηκε σύνθεσιν· ὡς εἶναι µὲν
ταῦτα καὶ καλεῖσθαι Ἑξαπλᾶ, ἐπὶ δὲ τὰς Ἑλληνικὰς ἑρµηνείας γενέσθαι δύο ὁµοῦ παραθέσεις, Ἑβραϊκῆς
φύσει δι’ Ἑβραϊκῶν στοιχείων καὶ Ἑβραϊκῆς δι’ Ἑλληνικῶν στοιχείων, ὥστε εἶναι τὴν πᾶσαν παλαιὰν διαθήκην
δι’ Ἑξαπλῶν καλουµένων καὶ διὰ τῶν δύο τῶν Ἑβραϊκῶν ῥηµάτων. Compare the translation of WILLIAMS (2013,
136).
69. "Tetrapla" would make more sense here. As it stands, the passage demonstrates confusion.
70. Octapla presumably refers to the six columns of the Hexapla plus the fifth and sixth versions of the Psalms.
71. De mensuris et ponderibus (516–518, 528–533): ὅτε καὶ τὰ ἑξαπλᾶ καὶ τὰς δύο τῶν ἑβραϊκῶν σελίδας
ἄντικρυ ἐκ παραλλήλου µιᾶς ἑρµηνείας πρὸς τὴν ἑτέραν συνέθηκεν ἑξαπλᾶ τὰς βίβλους ὀνοµάσας, καθ’ ἅπερ
ἄνω διὰ πλάτους εἴρηται ... τῶν τεσσάρων δὲ τούτων σελίδων ταῖς δυσὶ ταῖς ἑβραϊκαῖς συναφθεισῶν ἑξαπλᾶ
καλεῖται· ἐὰν δὲ καὶ ἡ πέµπτη καὶ ἡ ἕκτη ἑρµηνεία συναφθῶσιν ἀκολούθως τούτοις ὀκταπλᾶ καλεῖται· φηµὶ δὴ
ταῖς ἓξ ἑρµηνείαις καὶ ταῖς ἄλλαις δυσὶ τῇ µὲν ἑβραϊκοῖς στοιχείοις καὶ ῥήµασιν αὐτοῖς γεγραµµένῃ, τῇ δὲ
ἑλληνικοῖς µὲν στοιχείοις ῥήµασι δὲ ἑβραϊκοῖς.
- 42 -
in parallel [with one another], he gave a great aid in knowledge for those who
would desire such a noble purpose.72
Rufinus: Then also those most famous codices [Origen] himself first composed, in
which through narrow individual columns he wrote down (descripsit) the work of
each and every translator separately, such that, first of all, he placed the Hebrew
words themselves in Hebrew letters, in the second place, in order, he wrote down
(describeret) the Hebrew words in Greek letters right next to it, third, he joined the
edition of Aquila, fourth, that of Symmachus, fifth, that of the Seventy translators,
which is ours, sixth, he set in order that of Theodotion. And because of the compo-
sition of this sort he named the exemplar Hexapla, that is, that which is written in a
six-fold order.73
Throughout these seven passages from five separate authors, the language used to describe
Origen's work is that of collecting, compiling, joining, copying, placing, and arranging, but
never composing or commissioning an original work. The statements referring specifically to
the second column give no reason to assume that the general procedure of taking a pre-exist-
ing text and copying it into one of the columns of the Hexapla did not apply in the case of the
text of the second column. This claim may be supported by Jerome's statement that Origen
"arranged" (digesserat) the books of the old law, including the Hebrew words expressed by
both Hebrew characters and Greek characters. Despite the difficult syntax, Epiphanius's state-
ment in Panarion might be the most informative in this regard. He states that Origen,
"making use of the second column for a composition of Hebrew words through Greek letters,
has made yet another column of [this] composition."74 Also, Rufinus uses the same word (de-
72. De mensuris et ponderibus (176-183): Τὰς γὰρ ἓξ ἑρµηνείας καὶ τὴν ἑβραϊκὴν γραφὴν ἑβραϊκοῖς στοιχείοις
καὶ ῥήµασιν αὐτοῖς ἐν σελίδι µιᾷ συντεθεικώς, ἄλλην σελίδα ἀντιπαρέθετο δι’ ἑλληνικῶν µὲν γραµµάτων
ἑβραϊκῶν δὲ λέξεων πρὸς κατάληψιν τῶν µὴ εἰδότων ἑβραϊκὰ στοιχεῖα εἰς τὸ διὰ τῶν ἑλληνικῶν εἰδέναι τῶν
ἑβραϊκῶν λογίων τὴν δύναµιν. Καὶ οὕτω τοῖς λεγοµένοις ὑπ’ αὐτοῦ ἑξαπλοῖς ἢ ὀκταπλοῖς τὰς µὲν δύο ἑβραϊκὰς
σελίδας καὶ τὰς ἓξ τῶν ἑρµηνευτῶν ἐκ παραλλήλου ἀντιπαραθεὶς µεγάλην ὠφέλειαν γνώσεως ἔδωκε τοῖς
φιλοκάλοις.
73. Historia ecclesiastica (6.16.4): Unde et illos famosissimos codices primus ipse composuit, in quibus per
singulas columellas separatim opus interpretis uniuscuiusque descripsit, ita ut primo omnium ipsa Hebraea
verba Hebraeicis litteris poneret, secundo in loco per ordinem Graecis litteris e regione Hebraea verba
describeret, tertiam Aquilae editionem subiungeret, quartam Symmachi, quintam septuaginta interpretum, quae
nostra est, sextam Theodotionis conlocaret, et propter huiuscemodi compositionem exemplaria ipsa nominavit
Exapla, id est sextiplici ordine scripta.
74. The latter part of the line may also be interpreted as, "has made yet another composition." However, by
using the word 'another' (ἑτέραν), it would imply that the second column was of the same nature as the first.
- 43 -
scribo) to refer to the writing down of the Greek translations as he does for the writing down
of the second column.
While the statements of the early church fathers should not be accorded more weight
than is due them, their testimony is not insignificant. Even though the early church fathers
(e.g., Eusebius and Jerome) believed that Origen had superior Hebrew knowledge, none of
them explicitly attributed the transcriptions of the second column to his pen. Moreover, if
Origen had commissioned Jewish Hebrew scholars to transcribe the entire text of the Hebrew
Bible, neither Origen nor the early church fathers mention it as part of the process of compil-
ing the Hexapla.
Therefore, in light of the developed transcription system of the second column, the
fact that the rest of the Hexapla was made up of pre-existing texts, and the testimony of the
early church fathers, we may now make a ruling between the two alternatives put forth earli-
er. It seems more likely that Origen acquired the second column as a pre-existing text than
that he commissioned Hebrew experts to transcribe the entire Hebrew Bible into Greek.
Nevertheless, it remains possible, though less likely, that he did commission the
transcriptions of the second column. In this case, however, it would be better to argue that the
second column is the result of a combination of factors. Origen might have come across por-
tions of the Hebrew Bible transcribed into Greek and desired that such material be expanded
to cover all of the scriptures. He then commissioned those familiar with Hebrew and the
transcription technique to complete the task.75 Such a theory would still be consistent with the
developed transcription system of the Secunda and the fact that it is treated as a pre-existing
text in the early church fathers' comments. Nevertheless, even if such is the case, it demands
that the Secunda, or at least parts of it, had an original purpose and function separate and dis-
tinct from its inclusion in the Hexapla.
75. A similar suggestion is made by GORDON (1968, 289) and DE LANGE (1976, 58).
- 44 -
2.3.4. The Second Column as a Parallel Text
The most probable explanation of the evidence is that the Secunda existed in some form prior
to the composition of the Hexapla. If the text of the second column was not originally com-
posed for the Hexapla, it must be asked what it looked like in its original form. Did it stand
alone, or was it originally composed to be read in parallel with the Hebrew text? While the
evidence is inconclusive, the most likely explanation is that the Secunda was originally paral-
lel to the Hebrew text, if not physically, at least functionally.
This claim is supported by a number of factors. First, following CLEMENTS's recon-
struction outlined above, when Origen added the Hebrew to the Tetrapla, he added both
columns together. Second, when the church fathers discuss the texts that make up the Hexa-
pla, the Greek translations are often treated as a group and the two Hebrew columns are often
treated as a group. Eusebius in particular only speaks of adjoining the Greek translations with
"the Hebrew writing" (τῆς Ἑβραίων σηµειώσεως), with no specific mention of the second
column. Epiphanius refers to the added Hebrew columns as "two juxtapositions together"
(δύο ὁµοῦ παραθέσεις). Although grouping the translations together and the Hebrew columns
together would be intuitive, such a grouping may reflect something about the compositional
history of the Hexapla. Third, much like a translation, the initial composition of the Secunda
was necessarily based on the Hebrew text of the Bible. Fourth, if accurate pronunciation of
the Hebrew text was important, the Greek transcriptions of the Secunda could only be a suc-
cessful tool if used in conjunction with the Hebrew text or by one who already knew Hebrew
(EMERTON 1956; see section 3.2). Fifth, and finally, the fact that two or three transcribed
words are occasionally written on the same line (in the Ambrosiana palimpsest) is indicative
of an originally columnar text (see section 3.4.1).76
76. While it is typical for only one word to be written per line in the Secunda, a number of two- or three-word
phrases, most of which would be connected by a maqqaf or a conjunctive accent in the Tiberian tradition, are
written on the same line in the Ambrosiana palimpsest (e.g., |αληκι| ‫ֵיקי‬ ֥ ִ ‫[ עַל־ח‬Ps. 35:13], |χι!αν̓ωχιε᾽λωειµ̀| ‫ִכּי־אָנ ִֹכ֣י‬
‫הים‬Äֱ
֑ ִ ‫[ א‬Ps. 46:11], |αλ!µα!σαυ| ‫שוְא‬ּׁ ָ ֗ ֝ ‫[ עַל־מַה־‬Ps. 89:48]). A similar feature is present in bilingual columnar
translations used to teach Latin literature to Greek speakers. While it is possible for only one word to be written
per line in such texts, it is also common for the text to be broken up into phrases with multiple words per line
- 45 -
Therefore, in all probability, the Greek transcriptions of the Secunda were originally
parallel with the Hebrew text of the Bible even before they were incorporated into the Hexa-
pla.77 This does not have to mean that they were formatted together on the same scroll, but
that the Greek transcriptions at least would have been read alongside the Hebrew text of the
Bible, not apart from it.

2.3.5. Conclusions
Any theory about the compositional history of the second column, due to the paucity of evi-
dence, will necessarily be speculative. Nevertheless, we are now at a point where we may
summarize a probable interpretation of the process. Sometime after his relocation to Caesarea
in 233 CE, Origen encountered a Greek transcription text in use among the Jews alongside the
traditional Hebrew text of the Bible. It is possible that this text was written on the same scroll
as the Hebrew text of the Bible or that it was merely associated with it in its function. More-
over, while it is likely that the transcription text Origen encountered was comprehensive of
the Hebrew Bible, it is also possible that only portions of the biblical text had a correspond-
ing Greek transcription. In the case of the former, he would have required Jewish assistants to
match the Hebrew columns with the Tetrapla. In the case of the latter, in addition to the task
of correlating the columns, he would have required Jewish assistants to expand the Greek
transcription text to cover the entire Bible.
Origen saw in the Greek transcriptions a potential tool which would help him more
accurately compare the various translations and the LXX. The Greek transcriptions served
him as a concordance in a way that neither the Hebrew consonantal text nor the other Greek
(see section 3.4.1). It should be noted that this principle is applied inconsistently in the Ambrosiana palimpsest.
If such a feature had been the result of text-critical considerations in the composition of the Hexapla, one would
expect more consistency in its implementation. It seems likely that such two- and three-word units could reflect
an element of the original format of the Secunda before it was integrated into the Hexapla (see sections 3.4.1–4).
77. Some scholars (e.g., NAUTIN 1977, 333–39) even claim that the Jews had already made a synopsis of their
own, attaching the Hebrew columns to the translations of Aquila and Symmachus. In light of CLEMENTS's
reconstruction (1997, 97–100), GRAFTON and WILLIAMS point out that it is not necessary to posit such a pre-
existing synopsis (2006, 113). Origen could have employed assistants knowledgeable in Hebrew to coordinate
the Hebrew columns with the Greek columns.
- 46 -
translations could have. His interest in learning Hebrew and the potential of the transcriptions
to aid in his Hebrew learning were also significant motivating factors for adding the second
column to the Hexapla.
In conclusion, this section has analyzed and summarized Origen's relationship with
the second column as far as the evidence allows. In this process, we have concluded that the
Secunda most likely had a life of its own before Origen ever encountered it. With respect to
its life before the Hexapla, the present discussion has only been able to affirm two likely
facts. First, it was in use among the Jews of Caesarea. Second, it was originally parallel—at
least functionally, if not physically—with the Hebrew text of the Bible. Questions regarding
the original date and setting of the Secunda remain open. The following chapter will attempt
to determine, in light of the evidence from the Hellenistic, Roman, and Byzantine Near East,
both the original date and Sitz im Leben of the Secunda.
- 47 -
3. THE SECUNDA IN THE HELLENISTIC/ROMAN NEAR EAST
3.1. INTRODUCTION
In the previous section, it was argued that Origen could not have written the second column
himself, but rather acquired the text from another source and incorporated it in the Hexapla.
Accordingly, the issue of the original date and setting of the Secunda remains an open ques-
tion. After a brief review of previous scholarship, this section will examine the relevant evi-
dence in order to best determine the original date, context, and function of the Secunda. With
regard to the date, it will be argued on the basis of linguistic evidence that the terminus post
quem lies at the beginning of the second century CE. With regard to the original setting, paral-
lel texts in the Hellenistic Near East, the linguistic situation in Roman Palestine, and the de-
velopment of the Jewish education system all indicate that the second column was originally
composed with a didactic or scholastic function in the wake of the decline of spoken Hebrew.
3.2. SURVEY OF SCHOLARSHIP

3.2.1. Previous Suggestions
The best summary of previous scholarship regarding the original background of the text of
the Secunda is found in JANSSENS (1982, 13–20).78 Rather than rehearsing the entire review
78. JANSSENS also discusses the views of those scholars who argue that Origen wrote the Secunda based on his
knowledge of Hebrew (e.g., SPEISER 1925–26; ORLINSKY 1937–38; MERCATI 1947) or that a contemporary wrote
or helped him write the Secunda (e.g., KAHLE 1921; MERCATI 1947).
- 48 -
here, a number of observations may be made. With respect to function, a number of scholars
hold what might be termed the "liturgical" theory. That is, the Greek transcriptions were com-
posed so Jews who did not know Hebrew could read the scriptures in the synagogue (e.g.,
HALÉVY 1901, 338, 341; BERTRAM, 1938, 73, 76, 77; JELLICOE 1968, 106–111; DE LANGE 1976,
22, 57–58; MARTIN 2004; 2007). Other scholars, without explicitly mentioning liturgical use,
argue that the transcriptions were intended for Jews who could not read Hebrew (e.g., BLAU
1894, 80–83; STAPLES 1939; JANSSENS 1982, 22–23). With respect to status, several scholars
argue that the text had a canonical status among the Jews (e.g., HALÉVY 1901, 338, 341;
KAHLE 1927, 7, 44; 1950, 184–85). With respect to origin, a number of scholars claim that
Greek-Hebrew transcribed texts were around long before Origen (e.g., WUTZ 1925–1933;
BERTRAM 1938, 73, 76, 77; BRØNNO 1943, 7; 1956, 242; KAHLE 1956, 150–51; 1959, 159, 161,
187; 1960, 385). With respect to provenance, it has been suggested that transcription texts
like the Secunda developed in Egypt (e.g., HALÉVY 1901, 338, 341), Palestine (e.g., STAPLES
1939; JELLICOE 1968, 106–111), or both (e.g., WUTZ 1925–1933). Dates are proposed from as
early as the time of the LXX (e.g., WUTZ 1925–1933; BERTRAM 1938, 73, 76, 77) to the sec-
ond or third century CE (e.g., KAHLE 1927, 7, 44; 1950, 184–85; BRØNNO 1943, 7; 1956, 242).
EMERTON disagrees with the "liturgical" view, arguing that any listeners who knew He-
brew would have found the reading of a transcribed text by someone who did not know He-
brew incomprehensible.79 Anyone who had gone through sufficient training to be able to read
Hebrew from a transliteration would have been able to read the Hebrew script itself. Alterna-
tively, EMERTON claims that the transcribed text essentially functioned in the same role that
niqqud would later fulfill. The two texts were used side-by-side, the transcribed text serving
to elucidate the vocalization of the consonantal text when it was ambiguous or unknown. The
79. EMERTON states that "it is unlikely that devotion to the sacred tongue was of such a character that it led to the
paradoxical result that a debased pronunciation was used which was nonsense to those proficient in Hebrew no
less than to those who knew only Greek." Moreover, the inability of Greek script to distinguish the sibilants, the
gutturals, the glides, and accentuation would have resulted in an incomprehensible pronunciation (1956, 80–81).
- 49 -
Greek representation of the consonants was not especially important, but its representation of
the vowels was indispensible. The consonants merely served as a vehicle to convey the vocal-
ization. He supports his conclusions with comparative evidence from the Hellenistic Near
East, drawing on examples of transcription in Egyptian and Babylonian texts (1956, 79–82).
MARTIN agrees with the "liturgical" view, yet innovatively approaches the problem by
emphasizing non-referential language theory. He claims that the public reading of Hebrew
was regarded as having intrinsic value, whether or not the reader or the congregation under-
stood it. He compares this to Egyptian and Mesopotamian texts from the same period that
were transcribed into Greek because the mere utterance of them was regarded as powerful.
Greek script was chosen to transcribe these voces magicae because, unlike the hieroglyphic,
cuneiform, and Hebrew scripts, which demanded a prerequisite knowledge of the language
for correct reading, the Greek script provided the phonetic information apart from contextual
semantics. For MARTIN, the text of the second column does not have its origin among rabbinic
circles, but was prepared by a "non-rabbinic" group of Jewish scholars so that those ignorant
of Hebrew could carry out public reading of the Hebrew Bible in synagogue (2004; 2007).
3.2.2. Evaluation
The idea that there were other Greek-Hebrew transcription texts is based on three primary
pieces of evidence. First, the theory that the LXX was originally translated from a Greek-He-
brew transcription text (e.g., WUTZ 1925–1933) demands the early existence of such texts.
This theory is no longer taken seriously (MARCOS 2001, 61–62). Second, a number of rabbinic
statements (mMegilla 1:8, bShabbat115a, bMegilla 18a) have been cited as evidence for the
existence of Greek transcriptions of biblical texts before Origen. PRICE and NAEH have
demonstrated quite convincingly that the texts cited do not refer to the adaptation of the bibli-
cal text into other scripts, but into other languages; the rabbinic statements deal with transla-
tion—not transcription—into other languages (2009, 275–84). Third, the opening line of
Melito of Sardis's (2nd CE) homily is also cited:
- 50 -
The scripture of the Hebrew Exodus has been read and the words of the mys-
tery have been elucidated. How the lamb is slaughtered and how the people
are rescued.80
ZUNTZ has persuasively argued that the phrases, ἡ γραφὴ τῆς Ἑβραϊκῆς ἐξόδου 'the scripture
of the Hebrew Exodus' and τὰ ῥήµατα τοῦ µυστηρίου διασεσάφηται 'the words of the mys-
tery have been elucidated' refer to a scripture lesson in Hebrew being followed by an explana-
tion in Greek (1943). It has been assumed, perhaps too readily, that a Hebrew scriptural read-
ing in a Christian community in Asia Minor must have been conducted from a Greek
transcription text (KAHLE 1956, 151). However, as MARCOS points out, the most that can be
ascertained from Meltio's homily is that the early Christians preserved the practice of reciting
certain pericopes in Hebrew for special occasions (2001, 216). A Greek transcription of He-
brew is not mentioned in Melito's homily. Therefore, there is no direct evidence of or any ref-
erence to Greek-Hebrew transcription texts of significant length other than the Secunda.81
With respect to status, the idea that the text of the Secunda was an official text among
the Jews seems to be entirely speculative. The lack of any explicit reference to transcription
texts makes it impossible to conclude that such a text was held in high esteem among the
Jewish communities of Egypt and Palestine. If anything, the lack of explicit reference argues
against it being a highly regarded or official text. It is difficult to imagine how such an offi-
cially recognized text could be lost without leaving behind any allusions in other material.
With respect to the original function of the Secunda, EMERTON's objections to the
"liturgical" view may be sustained (1956, 81–82). In addition to his objections, it might be
added that in both Jewish and Greco-Roman education, the alphabet is always the initial step
in learning the language (see section 3.4.3).
80. De Pascha (3–6): Ἡ µὲν γραφὴ τῆς Ἑβραϊκῆς ἐξόδου ἀνέγνωσται, καὶ τὰ ῥήµατα τοῦ µυστηρίου
διασεσάφηται. πῶς τὸ πρόβατον θύεται καὶ πῶς ὁ λαὸς σῴζεται.
81. There are, of course, numerous examples of short Greek-Hebrew transcriptions on amulets, funerary
inscriptions, etc. For these, see section 3.4.1.3.4.
- 51 -
MARTIN's appeal to non-referential language theory is worthy of more consideration.
Non-referential language theory claims that language is regarded as intrinsically valuable for
its own sake apart from any meaning (or understanding of meaning). Such a perspective re-
garding Hebrew in Origen's time would render irrelevant EMERTON's criticisms of the liturgi-
cal theory, namely, that recitation of Hebrew from Greek script would be unintelligible.
Though important for the discussion, an emphasis on non-referential language theory over-
looks the details of the specific cultural and historical context of ancient Palestinian Judaism.
First, since it was not obligatory to read the scriptures in Hebrew in ancient Pales-
tine,82 it seems unlikely that the scriptures were necessarily read in Hebrew among the Greek-
speaking Jewish communities. On the contrary, rabbinic literature seems to indicate that
translation into Greek was acceptable and that sometimes the scriptures were read in Greek or
other languages (EMERTON 1971, 17–19).83 If the scriptures were read in Hebrew among
Greek-speaking Jews, they would have been read by one who already knew the language
without need of a transcribed text (PRICE and NAEH 2009, 277). Second, the references to
transcription in rabbinic literature refer to other languages (biblical translations) being
transcribed into the Hebrew-Aramaic script (PRICE and NAEH 2009, 279–84). Third, while
mMegilla 2:1 proves that hearing Hebrew read without understanding could fulfill one's reli-
gious duty (MARTIN 2007, 267), the text nowhere implies that the reader was anything other
than a skilled Hebrew reader reading the Hebrew consonantal text. In fact, script and lan-
guage were intrinsically tied in the ideology of the rabbis (PRICE and NAEH 2009, 283).
Fourth, a statement in Jerome implies that an inferior pronunciation of Hebrew based on
transcriptions was regarded by the Jews to be literally ridiculous.84 Therefore, while MARTIN's
82. See mMegilla 1:8, bShabbat115a, bMegilla 18a and the explanation of PRICE and NAEH (2009, 275–84).
83. mMegilla 2:1 allows Esther to be read in Greek if the hearers do not know Hebrew (EMERTON 1971, 19).
84. When discussing Jews' reactions to Christians who pronounce Hebrew proper names incorrectly based on
the transcriptions from the LXX, Jerome writes that "they are accustomed to mock us for our ignorance"
(Commentarium in Epistolam S. Pauli ad Titum, 3:9) (BRØNNO 1970, 205; HARVIAINEN 1977, 49–50).
- 52 -
claim that the Secunda was the product of "non-rabbinic" scholars may be impervious to ob-
jections based on the Halakhah and language ideology of the rabbis, it should be pointed out
that his theory is merely speculative. The idea that the transcriptions were used for public
reading in the synagogue by those who did not know Hebrew is unlikely.
In the following sections, I will argue that EMERTON's view that the Greek transcrip-
tions functioned as a sort of niqqud or vocalization aid (1956; 1971) is to be preferred. It does
not claim any official status and avoids the pitfalls of proposing a liturgical use for the
transcriptions. Reasonably, it assumes that there were those who knew Hebrew but not well
enough to read the entire Bible correctly. Nevertheless, EMERTON's theory remains somewhat
vague. What was the specific context that gave rise to the transcriptions and how were they
used? After a discussion regarding the date for the composition of the Secunda, the rest of the
chapter will refine and expand on EMERTON's theory, drawing on evidence from the Hellenis-
tic, Roman, and Byzantine Near East.
3.3. DATE OF THE SECUNDA: TERMINUS POST QUEM

3.3.1. Introduction
While a number of factors are relevant for determining the date of the original composition of
the Secunda, a loose terminus post quem may be established on the basis of the representation
of ‫ ח‬and ‫ ע‬in transcription material. A survey of such material leads to the conclusion that the
Secunda was composed at the beginning of the second century CE at the earliest.
3.3.2. Merger of /ḫ /,/h ̣/ > /h ̣/ and /ġ /,/ˁ/ > /ˁ/
The Hebrew graphemes ‫ ח‬and ‫ע‬, which in Tiberian Hebrew represent the pharyngeal voiced
and voiceless fricatives /h ̣/ [ħ] and /ˁ/ [ʕ], respectively, originally represented two phonemes
each. In addition to signifying /h ̣/, ‫ ח‬also served to signify the voiceless uvular fricative /ḫ /
([χ]). In addition to signifying /ˁ/, ‫ ע‬also served to signify the voiced uvular fricative /ġ / ([γ]).
Although these phonemes eventually merged (/h ̣/,/ḫ / > /h ̣/; /ˁ/,/ġ / > /ˁ/), they remained distinct
until a relatively late period in the history of Hebrew. Evidence of their distinct realization in
the biblical reading tradition is exhibited in the Greek transcription of proper names found in
- 53 -
the LXX. For example, Hebrew ‫√( ָרחָב‬rḥb) is transcribed in Greek as Ρααβ, but Hebrew ‫אָחָז‬
(√ˀḫ z) is transcribed as Αχαζ. Also, while Hebrew ‫ ֻעזִּיּ ָהוּ‬/‫√( ֻעזִּיּ ָה‬ˁzz) is transcribed in Greek as
Οζιας/Οζιου, Hebrew ‫√( ַעזָּה‬ġ zz) is rendered as Γαζα (BLAU 2010, 75–76).
3.3.2.1. Diachrony of the Merger in the LXX, Josephus, Aquila, and 2 Esdras
The evidence for the polyphony of Hebrew ‫ ח‬and ‫ ע‬was laid out in BLAU (1982). On the basis
of the distribution of the renderings Ø/χ for ‫ ח‬and Ø/γ for ‫ ע‬in the LXX, BLAU demonstrated
that there is general consistency between Ø and etymological ḥ/ˁ, on one hand, and between
χ/γ and etymological ḫ /ġ on the other (1982, 9–15, 43–48). However, in the transcriptions of
proper names in 2 Esdras this consistency has diminished, with etymological /ḫ / and /ġ / often
being represented by Ø and not χ/γ. BLAU understands this phenomenon to reflect the chrono-
logically late nature of the LXX translation of 2 Esdras (1982, 37, 65–66).
More recently, STEINER (2005) has expanded on BLAU's work, examining the evidence
for the merger in the Hebrew and Aramaic of Egypt and Palestine diachronically. He traces
the representation of names with etymological /ḫ / and /ġ / through the LXX, Josephus, Aquila,
and 2 Esdras in order to understand the mergers in the biblical reading traditions (246–49).
What he finds is that etymological /ġ / is hardly represented at all in Josephus, Aquila, and 2
Esdras (246–47). The representation of etymological /ḫ / with χ decreases over time. In Jose-
phus (37–93 CE), etymological /ḫ / is transcribed by Ø about one third of the time. In Aquila
(ca. 125 CE), etymological /ḫ / is transcribed by Ø almost twice as much as in Josephus. More-
over, many of the names with χ in Aquila are likely imitations of the LXX form. It seems that
the merger of /ḫ / and /h ̣/ in the biblical reading tradition had already begun by the first centu-
ry CE and was complete by the time of Aquila's translation (ca. 125 CE) (250–51). The fact that
transcriptions with χ for etymological /ḫ / are more common than transcriptions with γ for ety-
mological /ġ / in Josephus and Aquila supports BLAU's claim that /ġ / > /ˁ/ occurred before /ḫ / >
/h ̣/ in the history of Hebrew (246–50). Finally, the fact that etymological /ḫ / is realized as Ø
- 54 -
in 2 Esdras eight out of nine times leads STEINER to conclude that it was composed after
Aquila, sometime in the middle or end of the second century CE (261–64).
STEINER goes on to argue that the biblical reading traditions, due to their more formal
nature, likely maintained the distinction between the uvular and pharyngeal fricatives later
than did the spoken languages (2005, 250). In order to determine the terminus ante quem for
the merger of /ḫ / and /h ̣/ in the vernacular, he examines epigraphic evidence from Jaffa,
Masada, Jerusalem, and Gaza. On the basis of his analysis, he concludes that the merger of
/ḫ / and /h ̣/ must have occurred sometime between 100 BCE and 26 CE (253–57).
The distinction between the common vernacular, in which /ḫ / had merged with /h ̣/ by
Josephus's time (37–93 CE), and more formal reading traditions, in which /ḫ / had remained
distinct, is further supported by the curious practice of Josephus in transcribing etymological
/ḫ / in the names of postbiblical figures. When the postbiblical figure is a contemporary of
Josephus, etymological /ḫ / is transcribed with Ø, but when the figure preceded his time, ety-
mological /ḫ / is transcribed with χ (2005, 240, 243, 251).
STEINER suggests that the merger of /ḫ / and /h ̣/ may have been the result of contact
with Phoenician. Aramaic and Hebrew speakers of Upper Galilee, where Phoenician influ-
ence was quite strong, had probably merged /ḫ / and /h ̣/ in their speech at an earlier period.
The Hasmonean conquest at the end of the second century BCE, by connecting Judah with
Galilee politically, provided ample opportunity for the merger to slowly begin to travel south
over the coming century (2005, 259–61, 266).

3.3.2.2. Evidence of the Merger in the Secunda
STEINER's diachronic outline provides a method for dating the second column of Origen's
Hexapla based on its conventions for transcribing etymological /ḫ / and /ġ /. The Secunda rep-
resents both etymological /ḫ / and /ġ / only by means of Ø (or a hiatus between vowels) and
never by χ or γ (STEINER 2005, 245). Examples of etymological /ḫ / include αιλ ‫' ַחי ִל‬strength'
(18:40), σεµα ‫שׂ ְמחָה‬

ִ 'joy' (30:12), χαα ‫' כְּאָח‬like a brother' (35:14), σεωθι ‫' שַׁחוֹתִ י‬I bowed down'
- 55 -
(35:14), λαηριµ ‫' ַל ֲאח ִֵרים‬to others' (49:11), and µεεθθα ‫' ְמחִתָּ ה‬terror' (89:41). Examples of ety-
mological /ġ / include αων ‫' עֲוֹן‬iniquity' (49:6), οσραµ ‫' ָעשׁ ְָרם‬their wealth' (49:7), and
αλµουµαυ ‫' עֲלוּמָיו‬his youth' (89:46).85
The lack of any transcriptions with χ or γ would seem to push the terminus post quem
for the date of the composition of the Secunda after Josephus and probably after Aquila. The
fact that transcriptions of proper names are the center of STEINER's analysis may call into
question the validity of applying his conclusions to the text of the Secunda, which is a
transcription of the language in general. Typically, proper names tend to be more resistant to
linguistic change than the rest of the language. However, liturgical reading traditions are also
more resistant to linguistic change than lower registers of the language. Therefore, on the ba-
sis of the complete merger of /ḫ / and /h ̣/ in the Hebrew of the Secunda, we may reasonably
operate under the assumption that the original composition of the second column of the
Hexapla took place between the beginning of the second century CE and Origen's work on the
Hexapla in Caesarea in the first part of the third century CE. In light of the earlier discussion
regarding Origen's acquisition of the transcription text, a second- or third-century CE date fur-
ther supports Palestine, rather than Egypt, as its original provenance (see 2.3.1).
3.3.2.3. Dialectal Variation
Before concluding, we should also consider the possibility that the lack of /ḫ / and /ġ / in the
Secunda may not necessarily reflect a later date but merely a different dialect. The relevance
of dialectal variation for this issue is illustrated by KHAN, who describes a number of phe-
nomena with respect to the historical development of ḵ , ḥ, ḡ , and ʕ in various dialects of
Northeastern Neo-Aramaic. While some dialects merge *ḵ , *ḥ > x (e.g., Qaraqosh), others
merge *ḵ , *ḥ > ḥ (e.g., Hertevin). However, sometimes the more archaic pronunciation may
be maintained due to either phonetic or semantic factors. In many dialects that generally
85. But note the Secunda transcription βεγαβρωθ ‫( ְבּ ַעבְרוֹת‬Ps. 7:7). The variant in Chrysostom, however, has
βεβαρωθ (FIELD 1875, 94).
- 56 -
merge *ḵ , *ḥ > x, the pharyngeal *ḥ is maintained in the environment of emphatic conso-
nants. In the Qaraqosh dialect, *ḥ is maintained in some words of religious significance. Fi-
nally, sometimes the velars and pharyngeals exist side-by-side in free variation (KHAN 2005,
87–93). Moreover, inscriptional evidence indicates that the gutturals were lost in certain di-
alects of Hebrew (e.g., Beth Sheˀan, Ṭivˁon, Ḥaifa) earlier than in others (MOR 2013, 163).
Accordingly, we should not be too comfortable using the lack of /ḫ / and /ġ / in the Se-
cunda as a measure for dating the text. Nevertheless, the data do seem to be consistent with
assuming a general uniformity with respect to /ḫ / and /ġ / in the Hebrew reading traditions in
Palestine (see STEINER 2008, 245–46).86 Therefore, until we find clear evidence to the con-
trary, such as a transcription reflecting either the lack of /ḫ / and /ġ / at an early date or the re-
tention of /ḫ / and /ġ / at a late date, we may cautiously proceed under the assumption that the
lack of /ḫ / and /ġ / in the Secunda is probably indicative of a late date.
3.4. SITZ IM LEBEN OF THE SECUNDA

3.4.1. Comparative Material
3.4.1.1. Introduction
The text of the second column should be compared with other texts with a parallel format (or
function) and other instances of transcription in the Hellenistic Near East. In the following
survey, a distinction is made between parallel texts, which may include transcription, and in-
stances of transcription that stand alone without any accompanying primary text.
3.4.1.2. Columnar, Parallel, and Interlinear Texts in the Hellenistic/Roman/Byzantine
Near East
Whatever the original format of the text of the second column was, there can be little doubt
that it was at least functionally, if not formally, parallel with the Hebrew consonantal text.
This is suggested both by the discussion in chapter 2 and the evaluation of previous scholar-
ship in the current chapter. Presumably, then, a survey of columnar, parallel, and interlinear
86. Note that the preservation or lack of /ḫ / and /ġ / cannot be attributed to a general difference between Egypt
and Palestine. The book of Maccabees (ca. 100 BCE) preserves the distinction between /ḫ / and /h ̣/ in its
transcribed names, even in names not found in the LXX: e.g., Χαλπι, Χαιδαῖοι, and Ονιας (STEINER 2005, 256).
Moreover, there are no clear examples of /ḫ / preserved in a late text indicative of a more conservative dialect.
- 57 -
texts in the Hellenistic, Roman, and Byzantine Near East should shed some light on the origi-
nal function of the text. Formally parallel texts that also happen to contain transcription will
be treated in this section rather than in the subsequent section. As will be demonstrated, the
nature of parallel transcription texts is quite different from stand-alone transcription texts.
3.4.1.2.1. Latin and Greek Tradition
3.4.1.2.1.1. Glossaries
In the Latin and Greek tradition, such formats are most commonly implemented for bilingual
glossaries and translations. P. Oxy. LXXVIII.5162 (1st/2nd CE), for example, contains Greek
words in the left column and Latin glosses transcribed into Greek in the right column:
Translation Left Column Right Column (Latin Characters)
'ram' κρειοϲ αρ⟦ε⟧ιηϲ (= aries)

'bull' ταυροϲ ταυρουϲ (= taurus)
'lion' λεων λεο (= leo)
Figure 5: Greco-Latin Columnar Glossary (P. Oxy. LXXVIII.5162)
There are many other bilingual glossarial texts from the Roman period and early Byzantine
period.87 Such glossaries are usually organized around a particular theme or context,88
presumably intended to grant its users with a practical vocabulary for a particular social situa-
tion. Additionally, this format is utilized for presenting grammatical information and
paradigms.89
There are also a number of Greco-Latin bilingual glossaries that express the Latin in
Latin characters, rather than in Greek transcription.90 If a glossary is made specifically for a
87. P.Oxy. LXXVIII.5163 (1st/2nd CE) and P.Oxy. XLIX.3452 (2nd CE) are similar to the text above.
88. P.Mich.Inv. 2458 is a list of Greek and Roman deities in two columns from the second or third century CE
(KRAMER 1983, 79–80). P. Strasb. Inv. g 1173 is a third- or fourth-century CE glossary with merchandise and
military terms (KRAMER 2001, 65–76). P.Lund I.5 (2nd CE) contains animal names (CAVENAILE 1958, 379). P.Oxy.
XXXIII.2660a contains the names of vegetables and fishes. P.Laur. IV.147 (3rd CE) contains the names of various
animals. P.Fay. 135v descr. (4th CE) is a list of month names (KRAMER 1983, 77–78). P. Lond II.481 (4th CE)
contains a list of words organized according to professions (KRAMER 1977, 231–32; ADAMS 2003, 41–42).
89. P.Strasb. inv. G 1175 (3rd/4th CE) contains Greek verbal conjugations in the left column with the
corresponding Latin conjugations in the right column transcribed into Greek (KRAMER 1983, 45–52). P.Oxy.
LXXVIII.5161 (3rd/4th CE) contains a list of conjugated verbs in alphabetical order.
90. See P.Oxy. LXXVIII.5161 (3rd/4th CE), a grammatical text, and P. Vindob. Inv. L 27 (3rd/4th CE) and C.Gloss.
Biling. 1 10 (4th CE), Greco-Latin columnar glossaries (KRAMER 1983, 73–76; KRAMER 2001, 53–56).
- 58 -
Latin literary text, the Latin is always represented in Latin characters. For example, PSI VII
756 (4th/5th CE) contains a bilingual glossary for Vergil's Aeneid (CAVENAILE 1958, 15–18):
'they press/climb' nituntur εριδονται 'they set themselves'

'by/on steps' gradibus τοις βαδει[σµο]ις 'by/on the steps'
'with left (hands)' sinis[tr]is τες αριστε[ρ]ες 'with left (hands)'
'protected' protecti προβεβληµενοι 'having been thrown'
'they oppose' obiciunt αντιτιθεασιν 'they oppose'
'they press' praessant δρασσονται 'they press' (?)
Figure 6: Greco-Latin Columnar Glossary of Vergil's Aeneid (PSI VII 756)
The fact that this is a glossary and not a bilingual translation is evident from the full quotation
of the lines (words from glossary in bold): Nituntur gradibus clipeosque ad tela sinistris |
Protecti obiciunt, praessant fastigia dextris (Aeneid II.443–44). It seems that only certain
key words needed a Greek gloss.91 Such texts, however, are not necessarily bilingual. P.Oxy.
XXIV.2405 (2nd/3rd CE) is an example of a Greek-Greek glossary for Homer's Iliad, with the
more archaic Homeric Greek on the left and a more familiar Greek gloss on the right:92
'until' οφρα οπως 'until'

'he completes' τελεσση τελειωση 'he completes'
'you will save' σαωσις σωσεις 'you will save'
Figure 7: Greek-Greek Glossary for Homer's Iliad (P.Oxy. XXIV.2405)
The full quotation shows that only those words that would be difficult for a Koine speaker are
glossed: ὄφρα τελέσσῃ δὲ σὺ φράσαι, εἶ σαώσεις µε (Iliad I.82–83).93 While Greco-Latin
glossaries are typically in columnar form, other formats are also attested.94
91. P.Oxy. VIII.1099 (5th CE) represents the same sort of text.
92. DICKEY finds that glossaries of this format (word pair in same column) are the most common for Greek-
Greek glossaries. Of the thirty-nine examples she found, thirty-four have this format (2015b, 819).
93. There are approximately eighteen word-lists for Homer's Iliad in the Egyptian papyri (GAEBEL 1970, 298).
94. BKT IX 150 (1st BCE) IS single-column Greco-Latin glossary with the transcribed Latin gloss indented one
line below the Greek lemma (SCAPPATICCIO 2015, 464–66). P.Sorb. inv. 2069 (3rd CE) alternates Greek and Latin
on the same line. However, DICKEY argues that it was originally columnar (2010, 189, 206). Chester Beatty
codex AC 14999 (4th CE), a glossary to the epistles of the Apostle Paul, separates the Greek lemma from its Latin
gloss with double points (:), with a quotation-like symbol (") separating each entry. DICKEY suggests that such a
text might have been rearranged from an originally columnar format (2015b, 818).
- 59 -
Although each text's specific purpose may have varied, from aids for practical com-
munication to glossaries for a literary text, scholars generally agree that such glossaries were
used by Greek speakers to learn Latin. In the earliest period, they were the most common ma-
terial utilized for such a purpose. The topical, rather than alphabetical, arrangement facilitated
practical use (DICKEY 2012, 11–12). Grammatical information (e.g., P.Oxy. LXXVIII.5161)
was not presented in a distinct format in bilingual material at an early period (DICKEY 2012,
14). The outlier above, a word-list for Vergil's Aeneid, has been compared to the Greek word-
lists for Homer's Iliad, the oldest attestation of which predates the Aeneid glossary by a few
centuries. The purpose of both word-lists was to aid the Greek-speaking student in under-
standing the literary text before them (GAEBEL 1970, 298). Such a word-list, which only con-
tains select words of the classical text, would have been read alongside another copy of the
text. They may have been produced by teachers as aids for their students or by the students
themselves, who looked up the words in lexica and copied them in order (DICKEY 2012, 15).
3.4.1.2.1.2. Translations
P.Ryl. III.478, a fourth-century CE papyrus, contains a Greco-Latin columnar translation of
Vergil's Aeneid (I.252–53) (CAVENAILE 1958, 8–9):
'and Italy' adque Italis και των [Ι]ταλιωνων

'far am I kept' longe disiungimur µακραν διαξευγνυµεθα
'is this, of piety, the reward?' hic pietatis ho[nos] αυτη της ευσεβιας τι[µη]
Figure 8: Greco-Latin Columnar Translation of Vergil's Aeneid (P.Ryl. III.478)
This columnar translation differs from the bilingual glossaries of Vergil cired above in that it
translates almost every word, rather than just providing glosses for important words.95
Columnar translation may also have more of a tendency to include two or three words per
line instead of just one. Also, a comparison with the traditional word order of the Latin origi-
nal shows that word-order changes, presumably for facilitating the learning of the text, were
not uncommon in these columnar translations: ... atque disiungimur longe oris Italis. Hic
95. Other examples of Greco-Latin columnar translations of Vergil include BKT IX.39 (4th CE), P.Fouad 5 (4th/
5th CE), P.Oxy. L.3553 (5th CE), and P.Vindob. inv. L 24 (5th CE).
- 60 -
honos pietatis? Sic reponis nos in sceptra? (Aeneid, I.252–53). While the Aeneid is the most
common columnar translation found, understandably, other texts also take this form,96 such as
the Hermeneumata Pseudodositheana, a group of bilingual pedagogical materials, which de-
scribe everyday life in the ancient world (2012, 3–4).97 Finally, columnar translation is attest-
ed in exemplary epistles (e.g., P.Bon. 5 [3rd/4th CE]) (CAVENAILE 1958, 386–92; DICKEY 2015b,
816).
Greek translation of Latin literary texts is not always in columnar form. Palimps. Am-
bros. (4th/5th CE) contains a portion of the Aeneid (I:588–748) first written out in Latin and
then translated into Greek, preserving the original form of the Latin (CAVENAILE 1958, 23–27):
Restitit Aeneas claraque in luce refulsit

Os umerosque deo similis; namque ipsa decoram
...
Ἀπέστη ὁ Αἰνείας καὶ ἐν καθαρῷ τῷ φωτὶ ἀντέλαµψεν
Τὸ πρόσωπον καὶ τοὺς ὤµους θεῷ ὁµοῖο[ς] καὶ γὰρ αὐτὴ εὐπρεπῆ
Figure 9: Greco-Latin Same Column Translation of Vergil's Aeneid (Palimps. Ambros.)
Presenting the primary text and its translation in the same column is the case for a number of
texts.98 In some later biblical codices, Latin translations of the original Greek are set forth in a
"facing-page" translation format (DICKEY 2015b, 817).99
Bilingual translations are the most commonly attested form of ancient material for
learning language. The teachers likely produced these texts to aid students in their reading.
The degree of similarity between separately attested translations points to the use of lexica,
written or memorized, in their production. Such texts could be read on their own or alongside
96. Α palimpsest fragment (5th CE) contains a Greco-Latin columnar translation of Vergil's Georgics (HUSSELMAN
1957, 454–55). Cicero is found in Greco-Latin columnar translation in P.Rain.Cent. 163 (4th/5th CE) (CAVENAILE
1958, 71–74), PSI Congr.XXI 2 (4th/5th CE), P.Ryl. I.61 (5th CE), and P.Vindob. inv. L 127 (5th CE). Isocrates is
translated in BKT IX.149 (4th CE). One of Aesop's fables is translated in PSI VII.848 (4th CE) (DICKEY 2015b,
816).
97. Ancient attestations of this genre are found in P.Prag. II. (4th/5th CE), which contains a scene from a school,
and P.Berol. inv. 21860 (4th CE) (DICKEY and FERRI 2012, 129-31).
98. P.Mich. VII 457 (3rd CE) is one of Aesop's fables (CAVENAILE 1958, 23, 163; DICKEY 2015b, 818). P.Amh.
II.26 (3rd/4th CE) is a Latin translation with the Greek text of Babrius (XI.1–5) (CAVENAILE 1958, 118–20).
99. Examples of this include PSI XIII.1306 (4th/5th CE), a fragment of Paul's letter to the Ephesians, and Codex
Bezae (DICKEY 2015b, 817).
- 61 -
a different copy of the text. The colloquia served as conversation manuals by which a student
could learn short dialogues and common phrases for certain situations (DICKEY 2012, 14).
Finally, a couple observations can be made about the relationship between columnar
glossaries and columnar translations. First, bilingual columnar translation, whose earliest at-
testations go back only to the fourth century CE, is attested later than bilingual glossaries.
Columnar translation seems to have developed out of the tradition of columnar glossaries.
DICKEY explains the phenomenon as essentially "[treating] a continuous text like a glossary"
(2015b, 814). Glossaries and word-lists that only contain key words of a classical literary
work may reflect a transition period. Second, while Greek transcription of Latin is quite com-
mon in bilingual glossaries, it rarely appears in translation texts. DICKEY argues that this is be-
cause in the earlier period the focus was more on oral proficiency, whereas in the later period
the focus was more on literacy (2012, 10).

3.4.1.2.1.3. Transcription
Pure transcription is attested in abecedaries. O.Max. inv. 356 (1st/2nd CE), a bilingual
abecedary, contains the names of the Latin letters transcribed in Greek (FOURNET 2003, 445):
] γη ˫ ι κα ιλ • µ εν ω πη κου ρ εϲ τη ου ξη
]G•HI K L•M• N • O • P • Q • R • S • T • U • XYZ

Figure 10: Latin Abecedary in Greek Transcription (O.Max. inv. 356)
There are a number of similar abecedaries attested in the papyri, one of which is accompa-
nied by a line from Vergil.100
Examples of Greek transcription were cited earlier in the context of columnar glos-
saries. In each case, the text of both columns is represented in Greek script. In other words,
the transcription is not serving to help one read Latin characters elsewhere on the papyrus,
but is merely the most appropriate vehicle for carrying the Latin text (for a Greek speaker).
Presumably, if one could read Latin script at all, there would be no need for transcription to
100. P.Antinoë I fr. 1 (4th/5th CE) and P.Oxy. X.1315 (5th/6th CE) are additional examples of bilingual abecedaries
(CAVENAILE 1958, 136–37; KRAMER 2001, 33–44; ADAMS 2003, 41–42).
- 62 -
help with pronunciation. This is because, unlike Semitic scripts, Latin script indicated both
the consonants and the vowels. In the case of learning the alphabet, neither the phonetic val-
ues of the letters nor the names of the letters would be apparent to a language learner. Ac-
cordingly, the phonetic representation of their names in Greek would be appropriate.
Ancient students learned the Latin alphabet by copying the letters from a model text
in their proper order and repeating their names (CRIBIORE 2005, 132; DICKEY 2012, 10–11). In
order to practice the letters, a student might also copy out a line of verse (DICKEY 2012, 10–
11). The names of the Latin letters being phonetically represented in Greek is consistent with
the fact that an instructor would teach the names of the letters when teaching schoolboys the
alphabet (ADAMS 2003, 41–42).

3.4.1.2.2. Egyptian Tradition
Similar to the Greco-Latin tradition, it is common to find Egyptian glossaries, translations,
and transcription. The main difference between the former traditions and the latter is that
most bilingual Demotic texts do not use the columnar format (DICKEY 2015b, 819–20).
3.4.1.2.2.1. Glossaries
There is only one attested example of a Greek-Demotic columnar glossary. P. Heid. Inv.-Nr.
G 414 (3rd CE) presents a Greek-Demotic word pair on each line with no separation between
the words. The Demotic is represented in Greek transcription (QUECKE 1997, 72–73):
ταλλαντον = 'talent' ⲧ<ⲁ>ⲗⲗⲁⲛⲧⲟⲛⲕⲱⲣⲓ κωρι = 'talent'

αξινη = 'axe' ⲁⲝⲓⲛⲏⲕⲟⲗⲉⲃⲉⲓⲛ κολεβειν = 'axe'
σιδηρος = 'iron' ⲥ[ⲓ]ⲇⲏⲣⲟ<ⲥ>ⲃⲉⲛⲓⲡⲓ βενιπι = 'iron'
µαχαιρα = 'sword' ⲙⲁⲭⲁⲓⲣⲁⲥⲏⲫⲓ σηφι = 'sword'
υποποτιον = 'footstool' [ⲩ]ⲡⲟⲡⲟⲧⲓⲟⲛⲧⲁⲝ ταξ = 'footstool'
Figure 11: Greek-Demotic Columnar Glossary (P. Heid. Inv.-Nr. G 414)
- 63 -
This format is also found in Greek-Coptic glossaries, with the exception that the Greek-Cop-
tic glossaries divide the words with a double point (:) (DICKEY 2015b, 819).101 Word-lists for
literary works, like the word lists for Vergil's literature, also exist in the Egyptian tradition.102
P. Heid. Inv.-Nr. G 414 bears a resemblance to those Greco-Latin glossaries in which
the Latin column was written in Greek characters. Its function was also parallel. The
transcription of the Egyptian words into Greek and the columnar format seems to point to
Greek-speakers learning how to pronounce certain Egyptian words (RICHTER 2009, 411). Cop-
tic-Greek word lists presumably helped Coptic speakers to access the Greek text more fully.
3.4.1.2.2.2. Translations
P.Berol. inv. 10582 (5th/6th CE), a Latin-Greek-Coptic trilingual colloquium, presents Latin
transcribed into Greek characters, Greek, and then Coptic on each line. Each word is separat-
ed by a double point (:) (DICKEY 2015a, 66):
(= si omnes) σι:οµνης:ειπαντεϲ:ⲉⲥϫⲉⲛⲧⲟ 'If all'

(= biberint) βιβεριντ:επιαν:ⲁⲩⲥⲱ 'have drunk'
(= terge) τεργε:καταµαξον:ⲃⲱⲧⲓ 'wipe'
(= mensam) µενσαµ:τηντραπεζαν: 'the table'
Figure 12: Latin-Greek-Coptic Trilingual Colloquium (P.Berol. inv. 10582)
This text likely has roots similar to the colloquia of the Hermeneumata Pseudodositheana
(DICKEY 2015a, 71). It is more common, however, for bilingual translation of Egyptian to oc-
cur all in the same column, with the translation following the original.103 Other formats are
also attested, such as each translation being on each side of one page, the facing-page format,
and parallel columns without line-for-line correspondence (DICKEY 2015b, 820). The trilin-
gual colloquium, according to DICKEY, was a mix between a dialogue and a phrasebook, de-
signed for Coptic speakers to learn Latin, Greek, or both (2015a, 65, 73).
101. For example, see P.Rain.UnterrichtKopt. 257a (3rd/4th CE).
102. Chester Beatty Papyrus VII is a third century CE example of Coptic glosses written in Greek characters in
parallel with a Greek translation of the book of Isaiah (KENYON and CRUM 1937; RICHTER 2009, 413).
103. This is common in biblical and Christian texts such as P. Osloensis 1661 (4th CE) and P.Köln IV.169 (5th CE).
- 64 -
The Demotic magical papyrus of London, a third-century CE Demotic text containing incanta-
tions, invocations, instructions for divination, and other such voces magicae (GRIFFITH and
THOMPSON 1904, I.1, I.10, I.14–18) sometimes utilizes Greek transcription above a Demotic
word (EMERTON 1956, 86; GRIFFITH and THOMPSON 1904, II.16.8, see also II.23.28):
Figure 13: Supralinear Greek Transcription in the Demotic Magical Papyrus of London
THOMPSON and GRIFFITH's sketch shows the first word, Egyptian nyptwmykh, is glossed with
the Greek transcription ⲛⲓⲡⲧⲟⲩⲙⲓⲭ above it (EMERTON 1956, 86). Old-Coptic was also used
to gloss rare words written in hieroglyphics in late Hieratic manuscripts (RICHTER 2009, 413).
EMERTON points out that the rare words given Greek transcription in this text are not
actually Egyptian, but are almost all special terms invented by the magicians. Because the
Demotic script was almost entirely consonantal, the Greek transcriptions provided a guide for
pronunciation which was essential for the various invocations and spells to work (1956, 86).
3.4.1.2.3. Babylonian Tradition
The corpus of Graeco-Babyloniaca is comprised of around seventeen tablets from
Mesopotamia written in cuneiform and Greek from the Hellenistic and Roman periods. WEST-
ENHOLZ argues that the corpus should be dated between 50 BCE and 50 CE (2007, 274). Akkadi-
an or Sumerian is written in cuneiform on the obverse and Greek transcription of that same
cuneiform on the reverse. Rather than turn end-over-end like a typical cuneiform tablet, these
tablets turn side-to-side (GELLER 1997, 47). This same practice was implemented when the
text on the reverse contained a translation, rather than a continuation of the text on the ob-
verse (GELLER 1983, 114). This is good evidence that the Greek transcription was conceived
of as parallel to the cuneiform (GELLER 1997, 47). SOLLBERGER classifies much of the material
as either lexical texts or literary texts (SOLLBERGER 1962, 63).
- 65 -
BM 34797 is a Sumerian-Akkadian lexical text containing four columns. The obverse con-
tains parallel columns of Sumerian and Akkadian written in cuneiform and the reverse con-
tains parallel columns of the same text in Greek transcription (GELLER 1997, 68):
OBVERSE: REVERSE:
˹e i˺-ki [ε] ειχ

pa5 [p]al-gu [φα] φαλαγ
pa5-˹lal˺ [a]-tap-pi φα λα˹λ˺ [α]˹θ˺αφ
˹pa5˺-sig [a-tap-pi] φα σε˹k˺ αθ˹α˺φε˹ι˺
Figure 14: Sumerian-Akkadian Lexical Text with Greek Transcription (BM 34797)
Other lexical texts include BM 34781, BM 35727, and BM 34799 (SOLLBERGER 1962, 64–67).
There are also a number of examples of literary texts. Rm IV 327 (1st BCE), which con-
tains a portion of the Š amaš Hymn, is expressed in cuneiform on the obverse and Greek
transcription on the reverse (GELLER 1997, 78). The tablets also contain an Akkadian incanta-
tion (HSM 1137 [1st CE]) in cuneiform with Greek transcription and other literary texts (e.g.,
BM 34798, BM 34816, and VAT 412) (SOLLBERGER 1962, 67–71; GELLER 1983, 114–16).
Most scholars seem to agree that the Graeco-Babyloniaca tablets were the texts of an-
cient students of cuneiform (e.g., GELLER 1997, 47–48; WESTENHOLZ 2007, 262, 274). Even
though only a small number of these texts have survived, the transcriptions bear the marks of
a developed system and thus indicate that there were probably many such texts originally
(274). Those trained in cuneiform during this time would most likely have been native Baby-
lonians aspiring to work in astronomy or the liturgy of the cult (275).
Both GELLER (1997, 44) and WESTENHOLZ (2007, 276) set the Graeco-Babyloniaca
texts against the backdrop of the demise of Akkadian as a spoken language. By the first cen-
tury BCE, the last native speaker of Akkadian had passed away (WESTENHOLZ 2007, 276). Ara-
maic or Syriac would have taken the place of Akkadian in most contexts. The sort of
cuneiform tablets that were still written contained temple-related texts, such as astronomical
- 66 -
diaries, incantations, and prayers (GELLER 1997, 64). Formal education would have been nec-
essary for anyone wanting to learn Akkadian and the cuneiform script (276).
After students had learned some Akkadian, they would proceed to the traditional cur-
riculum for the beginner scribe, which included such material as word lists and syllabaries.
Eventually, they advanced to the classical Akkadian literary works. At this point, they were
able to dismiss with the Greek transcriptions and function as true scribes (2007, 276–77).
That the Greek transcriptions were meant to be an essential part of the learning process is
demonstrated by the fact that they were written when the clay was not yet dry (WESTENHOLZ
2007, 277). Because contemporary pronunciation differed in significant ways from its repre-
sentation in cuneiform (see WESTENHOLZ 2007, 283–91), the transcriptions were a helpful tool
for the beginning scribe. The reason that the Greek alphabet was utilized over Aramaic script
for such a pedagogical aid lay in its facility to express the precise vowel qualities and quanti-
ties (277). The Greek alphabet is also easier to learn than cuneiform. GELLER argues that the
Greek transcription on the reverse was an aid for correct reading (likely aloud) (1997, 47).
In sum, the Graeco-Babyloniaca tablets represent an attempt of teachers to transmit
the Babylonian culture, its language and literature included, to the next generation of stu-
dents. Being able to read and write the cuneiform script was an essential part of this endeav-
or. The Greek transcriptions could never have substituted for the cuneiform script and were
never intended to take the place of reading the Akkadian in its original representation. Rather,
they were intended by the teachers who composed them to serve as a tool for training begin-
ning scribes in cuneiform. Once a beginner had advanced far enough, the transcriptions had
served their purpose and were no longer needed (WESTENHOLZ 2007, 280).
3.4.1.3. Stand-Alone Transcription in the Hellenistic/Roman/Byzantine Near East
3.4.1.3.1. Introduction
The first part of this section surveyed the comparative material that exhibited a columnar,
parallel, or interlinear layout. A significant number of these texts also contained transcription.
Most of these texts constituted material for learning another language. Because of the convic-
- 67 -
tion that transcription in columnar, parallel, or interlinear form generally has a different func-
tion than stand-alone transcriptional texts, the two types of transcription have been separated
in our analysis. In the second part of this section, some of the more common instances of
stand-alone transcriptional material will be surveyed by language or language group.

3.4.1.3.2. Latin and Greek Tradition
Transcription is attested in a number of legal or economic texts, perhaps due to common for-
mulae being Latin-specific. SB III.I.6304 (2nd CE), a receipt of a slave trader from Miletus,
originally penned in Italy but carried to Fayû m, constitutes Latin in Greek transcription
(ADAMS 2003, 53–54, 63):
σκριψι µη ακκηπισσε α Τιτω Μεµµιω Μοντανω µιλιτε πεντηρω Αυγιστι

δηναριους σεσκεντους βιγεντι κινκυε πρετιουµ πουελλαι Μαρµαριαι βετρανε.104
I wrote that I have received from Titus Memmius Montanus, soldier of the quin-
quireme "Augustus," 625 denarii as the price of a Marmarian girl, a "veteran."
Figure 15: Latin Receipt of Slave Trader in Greek Transcription (SB III.I.6304)
The reason transcription was used in this text is probably a combination of factors: the legal/
economic formulae may have been Latin-specific, it may have been important for the author
to write in his own hand, and the recipient may have only been able to read Greek script
(ADAMS 2003, 55, 60, 62). Other examples of similar texts are attested in the papyri.105
Transcription is also attested in Greco-Latin funerary inscriptions in the western empire.
There, a Latin funerary inscription might have been expressed by means of Greek transcrip-
tion due to the identity of the deceased (ADAMS 2003, 89–92).106 Transcription is often used in
magic and defixiones (curse tablets). The idea behind this is that the magical "spell" or
"curse" should be obscured in some way; ADAMS calls this "obfuscation" (2003, 43–44, 47).107
104. = scripsi me accepisse a T. Memmio Montano milite pentero Augisti denarious sescentous vigenti cinque
pretium puellae Marmariae vetrane.
105. P.Oxy. XXXVI.2772, a letter to a banker, contains Greek in Latin transcription (ADAMS 2003, 65–66).
106. For example, see IG XIV.698 (ADAMS 2003, 89–92).
107. An example of a Greek transcription of a Latin defixio is found at Hadrumetum (ADAMS 2003, 44).
- 68 -
3.4.1.3.3. Egyptian Tradition
Greek transcription of Egyptian is unique because Greek would later become the basis of the
Coptic alphabet. An Egyptian dating formulary of Pharaoh Hor-Wenefer is transcribed into
Greek script in a 202/201 BCE graffito from Abydos (LACAU 1934; RICHTER 2009, 411–412):
Lε ΠΟΡΩ ΥΡΓΟΝΑΦΟΡ (= ḥ:.t-sp 5 Pr-ˁ: Ḥr-Wn-nfr)

ΜΗΙ ΕΣΙ ΝΟΜ ΟΥΣΙΡΕ ΜΗΙΕ (= mrj- ˀIs.t nm Wsir mrj - ˀI-)
ΜΟΥΝΑΑ ΣΟΝΤΗΡ ΠΝΟΤΩ (= mn - Rˁ nswt - nt̲ r.w p: nt̲ r ˁ:)
Reg.-year 5 of Pharaoh Horwenefer, beloved by Isis and Osiris, beloved

by Amun-Ra, king of gods, the great god
Figure 16: Egyptian Dating Formulary of Pharaoh Hor-Wenefer in Greek Transcription
Also, in a second-century BCE inscription on a stela, the various names of the Egyptian god
Thot are transcribed into Greek (GIRGIS 1965; RICHTER 2009, 411–12).
"Old-Coptic" texts differ from the earlier attempts at transcription in that, while they
are made up mostly of Greek letters, they add a number of Egyptian signs to supplement a
number of Egyptian phonemes lacking in Greek (RICHTER 2009, 412–23).108 For the sake of il-
lustration, a portion of the Schmidt papyrus is quoted below (SATZINGER 1975, 39–40):
ⳗⲉ ⲡⲁⳗⲟⲓⲥ ⲟⲩⲥⲓⲣⲉ ⲛⳍⲁⲥⲣⲱ ⲧⲓⲥⲙⲙⲉ ⲛⲁⲕ ⲁⲣⲓ ⲡⲁ ⲁⲧ ⲛⲙ ⳍⲿⲣ
My lord Osiris, (Lord) of Ḥasrō ! I complain to you, do justice to me and Ḥō r

Figure 17: Old-Coptic in Greek Transcription (Schmidt Papyrus)
Most of the letters are Greek, but a number of Egyptian signs are utilized to represent those
phonemes not present in Greek: ⳗ for t̲ /d̲ , ⳍ for ḥ, for h, ⲿ for ō , etc. (SATZINGER 1975, 38).
Egyptians utilized the Greek script in magical formulae, archaic language, and in-
stances in which precise pronunciation was important (RAY 2007, 813). The "Old-Coptic"
corpus is comprised largely of ritual texts. Transcription is used to make sure the text was
pronounced with precision in a ritual context. From the perspective of non-referential lan-
108. See P.Lond. 98 (95 CE), a horoscope (CERNY, KAHLE, and PARKER 1957); BM 10808 (2nd CE), a spell (CRUM
1942; SEDERHOLM 2006); the Schmidt papyrus (ca. 100 CE), a petition (SATZINGER 1975; RICHTER 2002).
- 69 -
guage theory, one could accomplish the recitation of the archaic ritual texts from the Greek
transcription even if one did not understand the original Demotic (MARTIN 2007, 257–59).
3.4.1.3.4. Hebrew Tradition
Transcription on funerary inscriptions is an identity-motivated choice (ADAMS 2003, 22–23).
It is not uncommon for Jewish funerary inscriptions composed in Greek to have a Hebrew
"tag," such as ‫' שלום‬peace' or ‫' שלום על ישראל‬peace upon Israel', at the beginning or end of the
inscription.109 Such tags are common in the Caesarea area (CIIP 2, 1517, 1549, 1602, 1662,
2098). Hebrew tags like ‫' שלום‬peace' could also be represented in Greek transcription (NOY
1993, no. 72; ADAMS 2003, 22–23, 66),110 as in the following funerary inscription from Beth
She'arim (SCHWABE and LIFHISTZ 1974, no. 91, see also nos. 21, 25, 72):
Σαλλὸµ 'Farewell,
Ἀταν Athan,
Δώρου son of Doron.'
Figure 18: Beth She'arim Funerary Inscription (no. 91) with Transcribed Tag
ADAMS calls such a phenomenon "tag-switching" or "formula-switching," which was motivat-
ed by a sense of the (in this case Jewish) identity of the deceased (ADAMS 2003, 22–23).
A possible example of transcription in a defixio context is found on an ossuary from
the Jerusalem area dated to the first century BCE or CE (CIIP 1/I, 451). The name and identity
of the deceased is written in both Hebrew and Greek script. However, the final line of the in-
scription, a typical warning against any who would disturb the grave, reads as follows: 'who-
ever moves these [bones], blindness will strike him!' υ αντικσκινησ(ας) | αυτα παταξει
αυτο(ν) | ουρουν. According to RAHMANI (1994, no. 559), the word used for blindness
(ουρουν) is apparently a transcription of the Hebrew ‫' עורון‬blindness' (CIIP 1/I, 451).
109. For example, R. Samuel's (3rd–6th CE) epitaph in Jerusalem area: Ῥαββὶ Σαµουήλ[ος] | ἀρχησυν[άγωγος --]
... | ‫' שלום ע]ל מישכבך‬Rabbi Samuel, synagogue ruler ... Peace on your resting place' (CIIP 1/II, 1001).
110. For example: τάφος Ἄνα διὰ βίου σάλωµ (NOY 1993, no. 72).
- 70 -
The western part of the empire also contains a number of examples of religious or
liturgical texts in Greek transcription.111 The Hebrew Shema (Deut. 6:7) was found
transcribed on an amulet buried with a child in what is modern day Austria dated to the sec-
ond or third century CE (BAR-ASHER 2010). The use of Greek transcription for a Hebrew verse
in such a context relates to the importance and power of the words themselves. If the author
did not know Hebrew, transcription was preferred over translation to preserve the original. If
the author did know Hebrew, transcription was chosen to ensure the words were pronounced
correctly. If the amulet had a ritualistic use, Greek script may have been used for obfuscation.
3.4.1.3.5. Semitic Tradition
Transcription appears in a number of religious and liturgical contexts. At Dura-Europos, the
following inscription is found on the wall of the pronaos of the temple (MILIK 1967, 289–90):
θαρθην γοβνιν δααβ Αβιδσαλµα Two cheeses of gold, ˁAbidsalma

βα-νισαν α´ α-Βαρζακικη112 on the first of Nisan, a-Barzaqiqe.
Figure 19: Aramaic in Greek Transcription (Dura-Europos Inscription)
This inscription was likely a votive offering of a shepherd, who offered two small gold mod-
els of cheese to thank the god for protecting his flocks. Having knowledge of the Greek script
but not the language, he composed the inscription in his native Aramaic (MILIK 1967, 291).
Greek transcription of Aramaic is also found in a pair of inscriptions from Nah ̣al Dimonah,
which is regarded as a writing exercise (KIRK 1938; PRICE and NAEH 2009, 268–69).
P. Amherst 63 (2nd BCE), an Aramaic text of considerable length transcribed into De-
motic script, constitutes the New Year's liturgy of a group of exiles from Mesopotamia. A lin-
guistic analysis of the text has demonstrated that the scribe did not know Aramaic himself.
Rather, he seems to have recorded an oral tradition (NIMS and STEINER 1983; STEINER 1997;
PRICE and NAEH 2009, 263–64; STEINER and NIMS 2017). Since the use of Aramaic had greatly
111. A Greek transcription of Hebrew is found on an amulet from Sicily (NOY 1993, 159) and on a phylactery
from Wales (ADAMS 2003, 272).
112. = ‫ברזקיקא‬-‫ א‬1 ‫תרתין גבנין דהב עבדשלמא בניסאן‬.
- 71 -
diminished in Egypt by this time, the text was composed so that a priest, whose knowledge of
Aramaic was poor, could still perform the ritual (STEINER and NIMS 1983, 272; 2017).
A neo-Punic inscription from El-Hofra is written in Greek letters (GORDON 1968, 289).
Like the Aramaic example, it was probably also a votive offering. There are a number of oth-
er Greek transcriptions of Phoenician and Punic, which ADAMS connects to the eventual death
of Punic. It is possible that the presence of deities' names or the cultic connection of certain
inscriptions prompted the use of Punic in this text, even though the script was not used (2003,
240–45). There are also Punic inscriptions written in Latin from the third and fourth centuries
CE from Tripolitania (KERR 2010). PRICE and NAEH argue that such texts arose in an environ-
ment where literacy in Punic had declined even though people still spoke the language (2009,
264–65).
Finally, like many Safaitic inscriptions, a Graeco-Arabic inscription from Jordan (3rd/
4th CE) contains a simple record of a shepherd's activity (AL-JALLAD 2015b, 52). The use of
Greek may reflect that the author was literate in Greek script, rather than the Safaitic script.
3.4.1.3.6. Summary
There is no singular thread of continuity that ties all these various uses of stand-alone
transcription together, but a few trends are discernable. Transcription is used in legal or eco-
nomic contexts, funerary inscriptions, defixiones, magical texts, ritual texts, and religious
texts. Aside from the Greco-Latin tradition, transcription usually arises out of a circumstance
in which the transcribed language is on the decline. Typically, an inscription would have been
written in its "native" script unless the author was insufficiently skilled in that script or made
a conscious choice to use a different script for another reason (PRICE and NAEH 2009, 274).
In some cases, the authors seems to be more (or only) proficient in Greek, but regard
their text as inextricably linked to another language whose script they did not know. This
seems to be the case in the legal or economic texts transcribed into Greek, in which the legal
formulae are regarded as inherently Latin phrases. In the case of religious texts transcribed
- 72 -
into Greek, the names of deities and cultic elements in the inscriptions are difficult to separate
from the language in which the people were accustomed to carry out their religious duties
(e.g., Aramaic, Punic). In each of these instances, because of an inherent quality of the con-
tent of the inscription, transcription was regarded as more appropriate than translation. While
the authors would have written the inscriptions in their "native" scripts if possible, they pre-
served the original languages even though they had to resort to writing them in another script.
In other cases, it seems that the author utilizes transcription as a conscious choice. In
the case of funerary inscriptions, this is done for the sake of carving out an identity. In the
case of defixiones, transcription is used for obfuscation. Finally, in the case of magical or ritu-
al texts, the text was transcribed because the precise pronunciation of the words, which were
regarded as inherently powerful, was important for the success of the utterance and the origi-
nal script required that one know the language in order to read it correctly.
3.4.1.4. Synthesis
While numerous scholars have compared the different practices of transcription in the Near
East with the Secunda (e.g., EMERTON 1956; MARTIN 2007; PRICE and NAEH 2009), none of
these approaches have separated parallel transcription and stand-alone transcription in their
analysis. Moreover, appreciating the Secunda as an originally parallel text has allowed us to
expand the corpus of comparative material to include non-transcribed parallel texts. As a
result of this multi-faceted analysis, a number of conclusions may be drawn about parallel
texts and transcriptional texts in the Hellenistic, Roman, and Byzantine Near East.
First, it was consistently the case that columnar, parallel, and interlinear formats with
or without transcription were implemented most for learning a foreign language or for
learning classical literature. If transcription was found in a parallel context, it almost certainly
indicates that the text was language-learning material. Second, transcription often arises when
the transcribed language is dying. While use and knowledge of the script often fade first, the
language continues to be spoken for some time. Transcription, in such cases, reflects an at-
- 73 -
tempt to preserve the language (e.g., Graeco-Babyloniaca, P. Amherst 63, Latino-Punic) and
thus indicates its decline (PRICE and NAEH 2009, 262–266; see also ADAMS 2003, 66). Third,
transcription may be the result of practical necessity or conscious choice. In the case of the
former, the author is unskilled in the script of the target language but regards the text as inex-
tricably linked to that language. In the case of the latter, the author may choose to implement
transcription for the sake of identity, obfuscation, or ensuring the correct pronunciation of
powerful and efficacious words. The final purpose, namely, ensuring correct pronunciation, is
particularly common when a language is fading away.
In the case of the Secunda, the author must have known the Hebrew script and lan-
guage quite well. Therefore, the utilization of transcription in the second column must have
been the result of conscious choice and not practical necessity. It seems unlikely that a Jewish
Hebrew expert would have utilized Greek script to emphasize his identity and less likely that
he would desire to obfuscate the biblical text. Accordingly, we must rule that the purpose of
transcription in the Secunda was to ensure the correct pronunciation of words regarded as ef-
ficacious or powerful.113 This motivation for transcription, in light of the evidence surveyed
above, would also point to the fading away of the Hebrew language. Finally, the parallel na-
ture of the Secunda would point to its use as material for learning language or literature.
In sum, the comparative material would suggest that the transcriptions of the second
column should be viewed as Hebrew-learning material composed at a time when the lan-
guage was fading away and it was becoming more and more important to ensure an accurate
recitation of the powerful and efficacious words of the scriptures. Accordingly, the remaining
sections of this chapter will examine the relevant evidence to determine if such a portrait fits
for second- and third-century CE Palestine. First, a summary of the linguistic situation in
113. Although it was admitted earlier that the Greek transcriptions themselves would not have ensured correct
pronunciation, the theory espoused below need not make such a claim. If used as a teaching tool, the
transcriptions would have been used in conjunction with other means of Hebrew instruction (see 3.4.4).
- 74 -
Palestine will demonstrate that the date for the composition of the Secunda correlates with
the fading away of Hebrew. Second, it will be shown that the composition of the Secunda
also coincides with the rising importance of education in the recitation of the Hebrew scrip-
tures and the development of schools. Finally, these conclusions will be synthesized to form a
hypothesis regarding the original context for the composition of the Secunda.
3.4.2. The Linguistic Situation in Roman Palestine
In Origen's Epistula ad Africanum (11.61), he describes his attempt to ascertain whether two
plays on words found in the Greek History of Susanna—πρῖνος 'kermes-oak' and πρῖσις 'saw-
ing', σχῖνος 'mastich' and σχίσις 'cleavage/parting'—would also be present in a hypothesized
Hebrew original. He asks a number of Jews how they would translate the Greek words
πρῖνος/πρῖσις and σχῖνος/σχίσις into Hebrew. They respond by saying that they do not know
those Greek words, but request that Origen bring them pieces of the different trees. The ac-
count that follows provides insight into the linguistic situation in third-century CE Palestine:
And (for the truth is precious), I did not hesitate to place before them in their
sight the [pieces of] the trees. One of them claimed that it was not possible to
ascertain with certainty how something is said in Hebrew if it is not mentioned
by name in the scriptures. And when at a loss, one is prone to use the Syriac
word instead of the Hebrew [word]. He went on saying, "even among the
wisest sometimes certain words are lacking." "If then," he said, "you can set
forth the schinos, mentioned by name in some scripture, or the prinos, we are
liable to find there that which is sought and its pair that provides the pun. But
if it is nowhere mentioned by name, then such a word escapes us also.114
This conversation is quite instructive for inferring a number of facts about the linguistic situa-
tion in Palestine (or perhaps just Caesarea) during Origen's time. First, the conversation was
presumably conducted in Greek, which demonstrates that it was not unusual for Palestinian
Jews in the Caesarea area to converse in Greek. Second, the fact that Origen's Jewish infor-
114. Καὶ (φίλη γὰρ ἡ ἀλήθεια), οὐκ ἠπόρησα αὐτοῖς ὄψει παραστῆσαι τὰ ξύλα. Ἄλλος δὲ ἔφασκε τὰ µὴ
ὀνοµασθέντα τῶν Γραφῶν ποὺ οὐκ ἔχειν διαβεβαιώσασθαι, ὅπως Ἑβραϊστὶ λέγεται· προπετὲς δὲ εἶναι, τὸν
ἀπορήσαντα φωνῇ τῇ Συριακῇ χρήσασθαι ἀντὶ τῆς Ἑβραΐδος· καὶ ἔλεγε, καὶ παρὰ τοῖς πάνυ σοφοῖς ἐνίοτε
λέξεις τινὰς ζητεῖσθαι. Εἰ µὲν οὖν, φησὶ, ἔχεις τι παραστῆσαι τὴν σχῖνον ὅπως ποτὲ ὀνοµασθεῖσαν ἔν τινι
Γραφῇ, ἢ τὴν πρῖνον, ἐκεῖθεν ἂν εὕροιµεν τὸ ζητούµενον, καὶ τὴν παρ’ αὐτὰ παρωνυµίαν· εἰ δὲ µηδαµοῦ
ὠνοµάσθη, καὶ ἡµᾶς διαλανθάνει τὸ τοιοῦτον. Translation in consultation with CROMBIE in ROBERTS et al. (1885).
- 75 -
mants are confident that they can come up with the Hebrew name for the trees if Origen will
just provide them with a specimen demonstrates that they were comfortable with Hebrew on
some level. Finally, the fact that one would default to Syriac (i.e., Aramaic) when ignorant of
a Hebrew term shows that Aramaic had overtaken Hebrew as the Jews' dominant language—
or second language, if Greek was their primary language.
The subtleties of this passage, however, demand a more nuanced look at the status of
Hebrew for Origen's Jewish informants. On one hand, his informants tell him that their He-
brew knowledge is essentially limited to that which is attested in scripture. This seems to in-
dicate that Hebrew was no longer a vernacular language. On the other hand, the fact that his
informants conceive of a situation where one might be at a loss (ἀπορήσαντα) for a Hebrew
word and thus need to resort to the more familiar Aramaic term indicates that there existed a
context in which Hebrew was the target of linguistic production. It is likely that such Hebrew
usage was confined to religious discourse (biblical/halakhic discussions, instruction in
̣
schools, synagogue sermons, etc.) and liturgical use (prayers, songs, eventually piyyutim,
etc.). Nevertheless, although Hebrew continued to be used in limited contexts, even the most
knowledgeable among the Jews were ignorant of some of the rarer words.
3.4.2.2. Greek, Latin, Hebrew, and Aramaic
This passage is consistent with what is known about the linguistic situation in third-century
CE Palestine. Greek was widely used in Palestine both during and before Origen's time. While
scholars differ regarding the numbers and distribution of Greek-speaking Jews, there is con-
sensus that Greek would be strongest among the upper classes, such as the rabbis and the ed-
ucated, as well as among those dwelling in Hellenistic urban environments. Caesarea, espe-
cially, would have been a prime location for Jews with a strong grasp of Greek.115 In a city
like Caesarea, the capital of the Roman province Iudaea/Syria Palaestina, Latin also would
115. For the use of Greek among Palestinian Jews, see MUSSIES (1976, 1056–59); HESZER (2001, 237–47); WISE
(2015, 345). For the knowledge of Greek among the rabbis, see SPERBER (2012, 115–21, 129–31, 135–36, 158).
- 76 -
have played a significant, if limited, role. It would have been used among the administrative
elite and enjoyed social prestige.116
For a long time, scholars wrongly assumed that Hebrew was no longer spoken in
Palestine by the first century CE. However, advances in the field of Mishnaic Hebrew and new
epigraphic discoveries over the course of the twentieth century have made it abundantly clear
that both Hebrew and Aramaic were spoken by Palestinian Jews in the first two centuries CE.
Rather than resembling the Hebrew of the Bible, however, the dialect of Hebrew that was
spoken seems to be a form of Mishnaic Hebrew. This constituted the Hebrew vernacular. (Re-
cently, it has been suggested that Mishnaic Hebrew is the product of a high degree of linguis-
tic interference resulting from many native Aramaic speakers learning a form of Biblical He-
brew as a second language during the Hasmonean period.) There existed, at the same time,
the standard literary register of Hebrew, which had to be learned. While it seems possible that
Aramaic might have had a slight edge over Hebrew in the first century CE, they were both
vernaculars of the Palestinian Jews. During the second century CE, probably largely due to the
Bar Kokhba revolt in 135 CE, Aramaic began to replace Hebrew as the common vernacular in
Palestine. By the end of the second century CE, Aramaic had become the vernacular of the
Palestinian Jews, with Hebrew surviving only in religious and liturgical contexts. It is possi-
ble, however, that Hebrew continued to be a vernacular among some small pockets of Jews.117
3.4.2.3. Concluding Remarks
Origen's residence in Palestine comes at the tail end of the period in which Hebrew was still a
vernacular for Palestinian Jews. It is unlikely that Origen encountered any native Hebrew
116. For the use of Latin in Caesarea, see LEHMANN and HOLUM (2000); ECK (2009, 34–40); ISAAC (2009, 55–60).
117. For a history and critque of the "exclusive Aramaic" view, see BALTES (2014a). For a cogent argument
regarding the prevalence of spoken Hebrew in the first two centuries CE, see FASSBERG (2012). For the
relationship of spoken Hebrew and Aramaic to literary Hebrew and Aramaic and Hebrew diglossia, see WISE
2015 (7–12, 317, 330). For the epigraphic evidence for the use of Hebrew and Aramaic in the first two centuries
CE, see NAVEH (1992a; 1992b); BALTES (2014b); TURNAGE (2014). For Judaean Hebrew, see MOR (2015). For
Jewish Palestinian Aramaic as the form of Aramaic spoken in Roman Palestine, see GZELLA (2015, 296–304).
For Mishnaic Hebrew as the result of native Aramaic speakers learning Hebrew as a "non-hybrid
conventionalized second language," see COOK (2016).
- 77 -
speakers, but it is entirely possible that he interacted with the grandsons and great-grandsons
of Hebrew speakers. In third-century CE Caesarea, he was surrounded by (mainly) Greek and
(some) Aramaic. Nevertheless, those Jews who had been to school would have been familiar
with Biblical Hebrew and, if they advanced far enough in their schooling, Mishnaic Hebrew
as well (see next section). Apparently, it was not too difficult for Origen to find such Jews fa-
miliar with Hebrew. This brings the discussion full circle back to the passage in Origen's
Epistula ad Africanum. He conducted a conversation with Jewish informants in Greek, who
seemed to exhibit a relative comfort with Biblical Hebrew, but were native Aramaic speakers.
It should be noted that, like the examples of transcription in the Near East surveyed
above, the description of the linguistic situation in Palestine places the composition of the Se-
cunda during the period when the use of Hebrew as a vernacular was fading away. This
would have resulted in a situation in which the lack of competence in Hebrew, especially
Biblical Hebrew, would have been especially felt.

3.4.3. Learning Hebrew in Tannaitic/Amoraic Palestine
The Babylonian Talmud relates a curious story in Aramaic about sages learning the meaning
of rare Hebrew words from the handmaid of Judah the Prince (2nd/3rd CE):
The rabbis did not know what [the meaning of] ‫ סֵירוּגִין‬was. They heard the
maidservant of Rabbi's house, when she saw the rabbis coming into the house
in intervals, saying, "How long are you coming in ‫"?סֵירוּגִין סֵירוּגִין‬118
...
The rabbis did not know what [the meaning of] ‫שׁמֵד‬ ְ ‫[ ְוטֵאטֵאתִ י ָה ְבּ ַמ ְט ֲאטֵא ַה‬Isa.
14:23] was. One day they heard the maidservant of Rabbi's house saying to
her workmate, "Take a ‫'[ טָאטִיתָ א‬broom'] and ‫'[ ַט ֲאטִי‬sweep!'] the house."119
This second- or third-century CE passage, in light of the discussion regarding the linguistic sit-
uation in ancient Palestine, demonstrates precisely the sort of issues that would have been rel-
118. bRoš Haš ana 26b: ‫לא הוו ידעי רבנן מאי סירוגין שמעוה לאמתא דבי רבי דחזתנהו רבנן דהוו עיילי פסקי פסקי אמרה להו עד‬
‫מתי אתם נכנסין סירוגין סירוגין‬. Translation in consultation with William Davidson Talmud.
119. bRoš Haš ana 26b: ‫ כג( וטאטאתיה במטאטא השמד יומא חד שמעוה לאמתא דבי רבי דהוות‬,‫לא הוו ידעי רבנן מאי )ישעיהו יד‬
‫אמרה לחבירתה שקולי טאטיתא וטאטי ביתא‬. Translation in consultation with William Davidson Talmud.
- 78 -
evant when the text of the second column was composed. The handmaid, a rare surviving na-
tive speaker of Hebrew, has a clear advantage in Hebrew over the rabbis, whose native
language would have been Aramaic. Even though her dialect, resembling Tannaitic Hebrew,
would have been markedly different from that of Biblical Hebrew, it meant that she knew
"rare" words both in Mishnaic and Biblical Hebrew. However, whatever advantage native
speakers of Hebrew might have had in reading the Hebrew Bible, the number of that group
was diminishing throughout the second century CE. If the high literary Hebrew of the Bible
had been difficult for native Hebrew speakers before (see WISE 2015), how much more diffi-
cult it would have been for native speakers of Aramaic or Greek.
Competence in reading and understanding the Hebrew Bible demanded formal educa-
tion. The necessity of such learning becomes all the more poignant when one considers the
fact that certain Hebrew words are said to escape even the knowledge of the sages, both in
this passage from the Babylonian Talmud and in Origen's Epistula ad Africanum. Despite the
challenge in learning it, the course of Jewish history demonstrates clearly that the rabbis
learned and knew Hebrew very well, even when it was no longer their native tongue
(ALEXANDER 1999, 71). The question, then, of how they learned Hebrew is quite relevant.
3.4.3.2. Torah Education and Jewish Elementary Schools
Scholars who have written on Jewish education in ancient Palestine have approached the top-
ic in different ways. One approach takes the Amoraic statements about the existence of a Jew-
ish elementary school system in the first few centuries CE at face value (e.g., SAFRAI 1976).
Another approach questions whether rabbinic texts from the Amoraic period are reliable wit-
nesses of the Tannaitic period (e.g., HEZSER 2001). When it comes to the actual methods of
teaching Hebrew, different scholars draw on different sources to support their views. These
various sources include rabbinic texts (e.g., SAFRAI 1976), parallels with other societies in the
Graeco-Roman world (e.g., HEZSER 2001), and documentary evidence in ancient Palestine
- 79 -
(e.g., WISE 2015). A synthesis of the various perspectives on Hebrew learning and literacy in
ancient Palestine may be summarized in terms of the method, context, and extent of learning.
With respect to the method, there is consensus that learning to read Hebrew began
with learning the alphabet. The teacher would write out the letters for the students and repeat
their names and sounds until the student learned to recognize them. It is possible that the stu-
dents also learned to write the letters and continued to copy texts as part of their education.
After learning the alphabet, the student would begin to read short texts from Leviticus (possi-
bly also Numbers) and Genesis. Finally, the student would progress to read the Torah itself,
which involved two main tasks. The first was memorizing the reading tradition of the Torah.
This was accomplished by the teacher reciting verses and the students repeating them; a
teacher who had learned to recite the reading tradition with precision was highly valued (e.g.,
bGittin 36a). The second was learning and understanding the meaning of the Torah. This was
accomplished by the teacher's explanatory comments that accompanied the recitation of the
Torah. Some scholars also think that a phrase-by-phrase translation, such as the Targumim
and the Greek translations of Aquila and Symmachus were also used for such purposes (see
3.4.3.3). Familiarity with the reading tradition of the Torah and its translation was reinforced
by hearing them read regularly in the synagogue. After learning the reading tradition of the
Bible, students with the opportunity might advance to study Halakha in the ‫בֵּית מִדְ ָרשׁ‬.120
With respect to context, the responsibility to teach a boy to read the Torah originally
fell on the shoulders of his father. Outside of the familial context, a potential teacher might be
found in the local ‫ ַחזָּן‬or ‫סוֹפֵר‬. Jewish education was not yet formalized at this point, but rather
took shape based on the given needs and circumstances. At some point, however, Jewish ele-
mentary schools began to develop. Even then, rather than having their own designated build-
120. For a synthesis of the relevant rabbinic evidence on the methods of learning Hebrew, see SAFRAI (1976,
945–49, 951–53). For an approach based on epigraphic material, comparative material, and only rabbinic texts
dated to the Tannaitic period, see HESZER (2001, 76–80, 83). For the use of parallel texts/translations in learning
Hebrew, such as the Aramaic Targumim and Aquila's Greek translation, see ALEXANDER (1999, 79–84).
- 80 -
ings, lessons were conducted in the synagogue, in study houses, and even in the homes of
teachers. Schools were present in many, but not all, Palestinian towns. The establishment of
Jewish elementary schools happened no later than the third century CE, and possibly as early
as the first century CE, though a date closer to the former should be favored. It may be accu-
rate to view the time between these dates as a transition period of growth in systematic Torah
education. It is perhaps no coincidence that this period was witness to the development of the
Targumim, Aquila's translation, and Symmachus's translation, potential tools for learning the
text of the Hebrew Bible and all prepared under the sponsorship of the rabbis (see 3.4.3).121
With respect to the extent of learning, it was the elite class who were able to avail
themselves of the opportunity to learn to read. Under the assumption that the term "elite" is
synonymous with the top quartile of society, about one in three elite males and one in fifteen
adult males could read the Torah. While they would have grown up with the privilege of edu-
cation, it also would have been possible to learn to read Hebrew as an adult like R. Aqiba.122
3.4.3.3. The Place of the Targumim and Greek Translations in Learning Hebrew
While the Targumim are most commonly discussed for their role in the synagogue, both in
rabbinic literature and in modern scholarship, they were also used in other settings. Accord-
ing to Targum scholars, the rabbis refer to three contexts for the use of the Targumim: in the
synagogue, in schools, and in private study (e.g., ALEXANDER 1985, 21; FLESHER and CHILTON
2011, 285). For our purposes here, we may focus on the role of the Targumim in the ‫בֵּית ֵספֶר‬.
In the Tosefta, a text regarding a list of passages not to be read in the synagogue ser-
vice (tMegillah 3:31–38) concludes with the following statement: ‫מעשׂה דוד ובת שׁבע לא נקרא‬
‫' ולא מיתרגם והסופר מלמד כדרכו‬the story of David and Bathsheba is not to be read and is not to
be translated, but the teacher (‫ )סופר‬teaches as he is accustomed'. According to FLESHER and
121. For a synthesis of the relevant rabbinic evidence regarding the context of learning Hebrew, see SAFRAI
(1976, 952–61). For an approach based on epigraphic material, comparative material, and only rabbinic texts
dated to the Tannaitic period, see HESZER (2001, 40–69). See also ALEXANDER (1999, 71–78, 85–86).
122. For an analysis of the extent of literacy among first century CE Palestinian Jews, see WISE (2015, 53–60,
311, 345–555). For evidence of adults learning Hebrew in ancient Palestine, see HESZER (2001, 76).
- 81 -
CHILTON, the mention of the ‫ סופר‬in this passage not only indicates that the Targumim were
used in the schools apart from the synagogue, but that they were used specifically in chil-
dren's schools (2011, 321). There are also references to Targum in sequential descriptions of a
study curriculum (e.g. Sifré Deuteronomy, 161) (ALEXANDER 1985, 22; FLESHER and CHILTON
2011, 319–320). Finally, there is a reference in the Jerusalem Talmud to a schoolmaster who
owned a written Targum (yMegillah 74d) (ALEXANDER 1985; 22).
In light of such examples in rabbinic literature and comparative pedagogical material
in the Roman world, ALEXANDER suggests that the Targumim may have originally developed
in the ‫ בֵּית ֵספֶר‬context (1999, 81). In fact, it is unlikely that the original Sitz im Leben of the
Targumim, at least in their most primitive stage, was the synagogue.123 Regardless of whether
the Targums developed in the ‫בֵּית ֵספֶר‬, however, it is sufficient for our argument to affirm that
they were used at the elementary level to teach Hebrew. Presumably, similar to the function
of the columnar translations of Vergil, a student learned the meaning of a verse of the Bible
by matching up the words of the Targum with the Hebrew original. In each case, a literal
word-for-word translation facilitated learning (ALEXANDER 1999, 80–82). It should also be
noted that such a hypothesis does not necesssarily entail that a written Targum was in use in
the ‫בֵּית ֵספֶר‬, but merely that the teacher regularly recited a phrase-by-phrase Aramaic transla-
tion after reciting the Hebrew text so that the students could learn to understand the Hebrew.
One might object to such a reconstruction on the grounds that the Targumim are not
verbum e verbo translations but rather exemplify a much freer or even "midrashic" style.
Such a characterization, however, is neither entirely accurate nor sufficiently nuanced. Al-
though the Targumim often contain highly expansionist commentary, there is typically a "hy-
per-literal" core that is characteristic of a verbum e verbo style. FLESHER and CHILTON put it
123. According to Z. SAFRAI, the earliest clear references to the practice of antiphonal Aramaic translation in the
synagogue date to the mid-second century CE. Moreover, a number of early references to synagogue meetings
conspicuously omit any account of Aramaic translation following the Torah reading. At the same time, however,
there are instances of Aramaic translation (e.g., the Job "Targum" at Qumran) and references to Targum (e.g.,
bMegillah 3a) dating to an earlier period (1992).
- 82 -
best by defining Targum as "a translation that combines a highly literal rendering of the origi-
nal text with material added into the translation in a seamless manner." Because the addition-
al commentary elucidated the meaning, the literal core did not necessarily have to convey the
meaning in the clearest manner. Accordingly, it was often "hyper-literal" (2011, 19–37, 378–
80). Whether first developed for the synagogue, for schools, or perhaps for both simultane-
ously, a hyper-literal word-for-word rendering followed by explanatory comments would be
conducive to learning the original Hebrew, provided that it was recited phrase-by-phrase
rather than verse-by-verse.124 The fact that the Targumim stand alone among ancient transla-
tions in this respect (i.e., a hyper-literal rendering followed by extensive explanatory com-
ments) (FLESHER and CHILTON 2011, 379–80) may be indicative of their distinct purpose,
namely, that of helping the hearers learn the text in its original language (i.e. Hebrew).
Some scholars suggest that Aquila's Greek translation of the Old Testament (ca. 125
CE) was used in a similar way (e.g., VERMES 1966; ALEXANDER 1999, 83–84) (see discussion in
MARCOS 2001, 110). This hypothesis is largely based on the "hyper-literal" and (excessively)
word-for-word character of Aquila's translation technique. Particularly noteworthy are those
instances in which he translates a phrase in Hebrew so literally that it no longer constitutes
grammatically acceptable Greek. For example, when the Hebrew definite direct object mark-
er ‫ את‬is followed by the definite article -‫ה‬, Aquila "translates" ‫ את‬with Greek σύν 'with', as
demonstrated by his rendering of Genesis 1:1: ἐν κεφαλαίῳ ἔκτισεν ὁ θεὸς σὺν τὸν οὐρανὸν
(καὶ) σὺν τὴν γῆν ‫ָאָרץ‬

ֶ ‫שּׁ ַמי ִם ְואֵת ה‬
ָ ‫הִים אֵת ַה‬Äֱ‫בּ ְֵראשִׁית בּ ָָרא א‬. MARCOS lists eleven such "hyper-liter-
al" characteristics of Aquila's translation style (2001, 116–17). It is difficult to imagine that
such a translation technique, which produced highly awkward or even ungrammatical Greek,
was not intended to help its readers learn the Hebrew original. For a Greek-speaking Jew, it
would have been highly conducive for learning Hebrew. Although not quite as literal as
124. There actually seem to be hints in rabbinic literature that disputes arose between schoolteachers who
wanted the text divided into phrases and the rabbis who wanted the text divided into verses in the synagogue
recitation (e.g., yMegillah 75b, bMegillah 22a) (see SAFRAI 1976, 951; ALEXANDER 1999, 81–82).
- 83 -
Aquila's translation, the Symmachus's translation could also be adequately matched up with
the Hebrew word-for-word (ALEXANDER 1999, 84).
In sum, while there is insufficient evidence to determine if either the Targumim or
Aquila's translation were originally developed in and/or for pedagogical purposes, a strong
case can be made that they were used by some to learn Hebrew (ALEXANDER 1999, 80–84).
These points will be picked up below, but it is worth emphasizing here that our overall argu-
ment about the original Sitz im Leben of the Secunda does not depend on assuming that the
Targumim or Aquila's translation were originally developed for teaching Hebrew. They are
merely cited as parallel examples of potential tools for learning the Bible. It should also be
noted that, at least in some Jewish communities in ancient Palestine, there may not have been
a sharp categorical distinction between studying the Torah and studying (Biblical) Hebrew.
Before proceeding to our final section, a few observations about the ancient Jewish education
system and its relevance for understanding the Secunda are in order. First, the Jewish educa-
tion system was centered around transmitting a correct pronunciation and understanding of
the Hebrew Bible. Second, a good argument can be made that parallel "texts" such as the Tar-
gumim and Aquila's translation were sometimes utilized as means for achieving the latter
goal (ALEXANDER 1999). Third, while there is debate about when elementary schools were es-
tablished, they were certainly around during Origen's time and possibly began to develop in
the preceding century.

3.4.4. Conclusions: The Original Context of the Secunda
When discussing the possibility of nullifying a vow in order to fulfill a mitzvah, the Babylon-
ian Talmud cites the following example (SAFRAI 1976, 950):

For it is [like the case of] a teacher of children upon whom Rav Aḥa imposed a
vow, based on the consensus of public opinion [to depose the teacher], since
he had acted wrongly towards the children [i.e., using severe discipline with
- 84 -
them]. And Ravina [Rav Ah ̣a] restored him [to his post] since no one was
found who was as precise as he.125
This passage highlights precisely the sort of need that could have given rise to the transcrip-
tions of the Secunda. Although Rav Ah ̣a (4th CE) had dismissed this Hebrew teacher because
of his harshness, he eventually reinstated him because there was no one else who was as
skilled and precise in reciting the reading tradition as he was (SAFRAI 1976, 950). The word
used to describe the Hebrew teacher, ‫' דייק‬precise', often refers to being exact or precise in an
argument (JASTROW 1926, 287–88). In the case of a Hebrew teacher, it probably refers to pre-
cision in vocalization.126 This passage, in addition to other rabbinic statements,127 demon-
strates just how much value was placed on transmitting an accurate and exact reading of He-
brew. It also calls attention to the fact that expert teachers were not found in abundance. It is
in just such a context that the original composition of the Secunda makes sense.
In the final section of this chapter, it will be argued that the text of the Secunda was
originally composed to meet a didactic need in the Jewish community. This didactic need
should be set against the backdrop of the decline of spoken Hebrew during the second centu-
ry CE, on one hand, and the growing prevalence of education from the Tannaitic period to the
Amoraic period on the other. Such a hypothesis is consistent with the trends of transcribed
texts in the Hellenistic Near East, the linguistic and social context in the Jewish community,
the methods of learning Hebrew, the specific scholastic context in Caesarea, and Jewish
views of the biblical text in second- and third-century CE Palestine.
125. bGittin 36a: ‫כי ההוא מקרי דרדקי דאדריה רב אחא על דעת רבים דהוה פשע בינוקי ואהדריה רבינא דלא אישתכח דדייק‬
‫כוותיה‬. Translation in consultation with William Davidson Talmud.
126. These meanings may overlap in a talmudic passage regarding utterances effecting divorce (bGittin 65b):
R. Natan says: [If he says] pattruha, [then] his words stand. [If he says] pitruha, he has said nothing. Rava said:
R. Natan, who is Babylonian, made a distinction (‫ )דייק‬between pitruha and patruha (‫רבי נתן אומר פטרוה דבריו‬
‫ ;קיימין פיטרוה לא אמר כלום אמר רבא רבי נתן דבבלאה הוא ודייק בין פיטרוה לפטרוה‬translation in consultation with William
Davidson Talmud).
127. mBerakhot 2:3: One who has read the Shema ... [if] he recited [it] but was not exact (‫ )דִ יקְדֵּ ק‬in
[pronunciation of] the letters, R. Yose says, "He has discharged [his duty]," but R. Yehudah says, "He has not
discharged [his duty]." (‫א יָצָא‬Ä ‫א דִ יקְדֵ ּק ְבּא ֹתוֹתֶ י ָה ְר׳ יוֹסֵה אוֹמֵ׳ יָצָא ְר׳ י ְהוּדָ ה אוֹמֵ׳‬Äְ‫ ק ָָרא ו‬... ‫שמַע‬
ׁ ְ ‫ַקּוֹרא אֶת‬
ֵ ‫)ה‬.
- 85 -
3.4.4.2. Consistent with Function of Transcriptions in Hellenistic Near East
Positing an originally didactic function among the Jewish community for the transcriptions of
the Secunda is supported by the comparative parallel and transcriptional material in the Hel-
lenistic Near East. The parallel format reflects language-learning material and the use of
transcription reflects an emphasis on the powerful or efficacious words of the text.
First, parallel and columnar texts, with or without transcription, are generally used as
material to learn language or literature. While parallel transcription is more often associated
with learning a language (e.g., Greco-Latin glossaries), parallel translation is more often as-
sociated with learning classical literature (e.g., columnar translations of Vergil's Aeneid). In
other words, transcription is more common in non-continuous glossarial texts and translation
in continuous literary texts.
One might object that the Secunda, which makes use of transcription for a continuous
literary text, does not actually conform to these data. In other words, we are unjustified in
taking principles that are valid for translations, and applying them to a transcription text.
Such an objection may be answered by framing our understanding of these parallel texts both
in terms of the need they address and the method by which they address that need. In the case
of the Greek-Greek word lists for Homer, an unfamiliar Homeric word is set in parallel with a
more familiar synonym from Koine. In the case of a Greco-Latin columnar translation of
Vergil, a Greek gloss (written in Greek script) is placed in parallel to an unfamiliar Latin
word (written in Latin script). In each case, the need was to have the meaning of the word ex-
plained. Accordingly, a corresponding (contemporary) Greek gloss was supplied. There was
no need, however, to aid the reader in pronunciation, since the Latin and Greek scripts con-
veyed both consonants and vowels. Had the Greek (in the case of Homer) or Latin (in the
case of Vergil) scripts been insufficient to communicate the necessary phonetic information to
pronounce the text correctly, parallel transcription likely would have developed as well.
- 86 -
In the case of the Hebrew text of the Bible, on the other hand, the need of the learner
lay in both elucidation of the meaning and instruction in the vocalization. While translation,
such as that exemplified in the Targumim, might have served to alleviate the first need, Greek
transcription, even of a continuous literary text, could have been a possible method of dealing
with the second. This seems to be the case with Akkadian in the Graeco-Babyloniaca tablets
and Aramaic in P. Amherst 63, both of whose scripts required knowledge of the language for
correct pronunciation. Therefore, as long as we are sensitive to the need a particular text or
tool is intended to address and the method by which it addresses it, we are justified in apply-
ing the same sorts of principles to parallel transcription texts that we find to be valid for par-
allel translation texts.
Second, a survey of the function of transcription in the Hellenistic Near East also
demonstrated that it was used to ensure the correct pronunciation of words regarded as pow-
erful or efficacious. This is consistent with the extremely high value that the rabbis placed on
a correct and exact pronunciation of the Hebrew reading tradition of the Bible. This is
demonstrated by the passage quoted at the beginning of this section as well as R. Yehudah's
opinion that one who recites the Shema without being exact in their pronunciation of the let-
ters has not fulfilled their religious duty (mBerakhot 2:3).

3.4.4.3. Linguistic and Social Context
An originally didactic function is also consistent with the linguistic and social context in
which it was first composed, namely, the decline of spoken Hebrew during the second centu-
ry CE and growing interest in Hebrew education from the Tannaitic to Amoraic period.
Even before the decline of Hebrew as a spoken language in the second century CE,
there would have obtained a situation of diglossia. While a dialect akin to Mishnaic Hebrew
would have been spoken in everyday life, Standard Biblical Hebrew (SBH) would have been
routinely heard in religious contexts as well. Although learning SBH would have been diffi-
cult even at this point in time, native Hebrew speakers would have had a significant advan-
- 87 -
tage in learning the language, especially considering the fact that they heard it recited regular-
ly and already possessed internalized grammatical categories for much of what they heard.
However, as the number of native Hebrew speakers began to wane in the second century CE,
it was far more difficult to achieve familiarity with SBH in a passive manner. The only way
to attain any kind of proficiency in SBH was to receive formal education.
It seems no coincidence that the establishment of Jewish elementary schools coincides
with the period when Hebrew ceased to be a vernacular of the Jews. Even though scholars
disagree on when exactly the establishment of Jewish schools should be dated, the best syn-
thesis of their views points to a gradually increasing emphasis on education and study of the
Torah beginning after the revolt of 70 CE and reaching its height sometime in the third century
CE; during this century, in which the existence of schools is uncontested, the rabbis began to
encourage an even deeper study of the Torah (see HEZSER 2001, 68–69). It is this period (70
rd
CE–3 CE) of increased emphasis on the study of the Torah and establishment of schools that
produces a number of potential tools for learning Hebrew, such as the Targumim and Aquila's
translation. That a Greek transcriptional text of the Hebrew Bible might also be a product of
this period of increased emphasis on learning the Hebrew Bible is entirely conceivable.
3.4.4.4. Methods of Learning Hebrew
That the development of these Hebrew learning tools is consistent with the methods of
learning Hebrew described earlier also supports the proposed didactic function of the Secun-
da. After learning the alphabet, students proceeded to learn the vocalization of the Torah by
repeating the recitation of the teacher and the meaning of the Torah by both listening to his
explanatory comments and learning a translation of the scriptures into the vernacular, such as
the Targumim. One who wanted to learn Hebrew, yet was more familiar with Greek than Ara-
maic, might have found more adequate help from a Greek translation such as Aquila or Sym-
machus (see 3.4.3.3). Prepared under the patronage of the rabbis, Aquila in Palestine at the
beginning of the second century CE and Symmachus in Caesarea just before Origen's arrival
- 88 -
there (beginning of 3rd CE), places them in a social, geographical, and chronological context
consistent with their use as tools for learning the Hebrew Bible (ALEXANDER 1999, 83–84).128
It is worth noting, however, that suggesting that such texts might have been used to
learn Hebrew by no means implies that each student interacted with a written copy. ALEXAN-
DER compares these "hyper-literal" translations with the Greco-Latin language-learning mate-
rial from Egypt (see 3.4.1.2.1), which he does not regard as mere student exercises. Rather,
he considers the Greco-Latin texts scholarly in nature and classifies them as "school-masters'
textbooks," emphasizing their surprising consistency across the papyri (1999, 82–83).
Presumably, then, it is conceivable that the Greek transcription text was originally de-
veloped as an aid for students learning the vocalization of Biblical Hebrew. Like the transla-
tions mentioned above, it was never meant to replace the Hebrew text; rather, it was meant to
serve as a "crutch," eventually enabling the student to deal with the Hebrew text on its own
(cf. ALEXANDER 1999, 82). It would be a mistake to suggest that the ability to use such a tool
would require extensive training in Greek literacy. Because it is merely a transcription, one
would only need the most basic level of reading ability to utilize the transcription as a parallel
tool for learning the vocalization. It is also possible that the transcription text was not used by
the students themselves, but essentially served as a reference or manual for the teachers. The
quote from the beginning of this section (bGittin 36a) shows that there were clearly teachers
of varying levels, and some surely needed help in vocalizing at least some parts of the Bible.
That the transcriptions were originally developed in the context of the ‫ בֵּית ֵספֶר‬is not
so far-fetched a claim. SAFRAI, for example, accepts the view that "written vocalization was
first practised by the teachers of young children as a teaching aid" (1976, 950–51; cf. BACHER
1904), though he does not mention the second column. We may also compare the idea that
128. Recall the statement of R. Yehudah HaNasi (2nd/3rd century CE): "In the land of Israel, why [should one
speak] in the Syriac language? Rather, [one should speak] either in the holy language or the Greek language"
(‫ אלא אי לשון הקודש אי לשון יוונית‬,‫( )בארץ ישראל לשון סורסי למה‬bSota 49b).
- 89 -
biblical accentuation traces its origins to the segmentation of the Hebrew text for pedagogical
purposes in the ‫( בֵּית ֵספֶר‬ALEXANDER 1999, 82). In fact, a number of features of the biblical
text may have developed originally in a didactic setting.

3.4.4.5. Scholastic Context in Caesarea
A number of factors also suggest that the Greek transcriptions of the Secunda might have
been a product of the Jewish scholastic community in Caesarea. Origen himself refers to the
presence of schools in Caesarea, the city was known to be home to prominent rabbis and
scholars, and a Jewish community in the midst of a Hellenized city would have made a text
like the Secunda linguistically appropriate for the community.
In Commentarii in Romanos (2.14), when discussing Paul's statement that the Jews
were entrusted with the oracles of God (Rom. 3:2), Origen writes, "For we see many Jews
from infancy until old age ever learning but never attaining the knowledge of the truth"129 (DE
LANGE 1976, 59–60, 181–82; translation from SCHECK 2001, 167). In Commentarii in
Canticum Canticorum (1.1), when discussing the particular scriptural passages to be studied
last, he writes that "there is another practice too that we have received from them—namely,
that all the Scriptures should be delivered to boys by teachers and wise men, while at the
same time the four that they call deuterō seis ... should be reserved for study till the last"130
(DE LANGE 1976, 60, 182; translation from LAWSON 1957, 23).131 Origen certainly seems to
have been in contact with Jews who valued teachers accurately transmitting the scriptures to
children. Although Origen does not explicitly mention Caesarea, it seems likely that the ma-
jority of Origen's experiences of Jewish life came from his time in that city.
129. Videmus enim plurimos Iudaeorum ab infantia usque ad senectutem semper discentes, & nunquam ad
scientiam veritatis pervenientes.
130. Sed et illud ab iis accepimus custodiri, quandoquidem moris est apud eos, omnes scripturas a doctoribus et
a sapientibus tradi pueris, simul et eas, quas δευτερώσεις appellant, ad ultimum quatuor ista observari.
131. The four deuterō seis mentioned are the beginning of Genesis, the beginning of Ezekiel, the end of Ezekiel,
and Song of Songs.
- 90 -
While little is known about the Jewish community in Caesarea during the second cen-
tury CE, it would grow to have a notable scholarly and rabbinic presence in the third century
CE. Nevertheless, the sparse evidence from the second century CE does point to some Jewish
presence in the city. That some Jewish scholars resided in Caesarea during this period may be
concluded from the fact that Rabbi Eliezer b. Hyrcanus came to the city for Sukkoth and
stayed in the house of Yohanan b. Ilai (bSukkoth 27b) (MURRAY 2000).132 Moreover, after
Jews were banned from Jerusalem in the wake of the Bar Kokhba revolt (135 CE), many fled
to the north and settled in Caesarea. A number of rabbinic rulings favoring Caesarea further
encouraged its settlement so that by the end of the second century CE, it likely had a sizeable
Jewish community again—though this community was still a minority among a mostly pagan
population. This Jewish minority in Caesarea and the rabbis who delineated Halakhah that
would apply to them would always be navigating the line between preserving peace for their
community and preserving their Jewish identity in the midst of a Hellenistic urban environ-
ment (MURRAY 2000).
It is in the third century CE that the references to the famous "rabbis of Caesarea" (‫רבנן‬
‫ )דקסרין‬begin in the Talmud (MURRAY 2000). LEVINE calls this group "one of the most highly
developed associations within the Palestinian rabbinate" (1975, 97). Constant debates be-
tween the (Jewish) Christians and the rabbis made the rabbis of Caesarea especially invested
and precise in their handling of the scriptures (MURRAY 2000). For example, when a number
of Christians are appalled at the lack of scriptural knowledge possessed by a Babylonian rab-
bi, Rabbi Abbahu (c. 250–320 CE), the leading figure of the Caesarean rabbis, replies, "We,
who are present in your midst, set ourselves to study [these things] thoroughly. One not [in
your midst] [i.e., the Babylonian rabbi], does not study [as thoroughly]."133 (bˁAvoda Zara 4a)
132. There are a number of other rabbinic references to Jewish residents of Caesarea during the second century
CE (LEVINE 1975, 44; MURRAY 2000).
133. ‫אנן דשכיחינן גביכון רמינן אנפשין ומעיינן אינהו לא מעייני‬. Translation in consultation with EPSTEIN (1935–1948).
- 91 -
(MURRAY 2000). The renown and significance of this group is also demonstrated by the fact
that, in the latter part of the third century CE, Caesarea is marked by the continual presence of
prominent Tiberian rabbis. LEVINE explains that the Tiberian rabbis were likely attracted to
Caesarea due its rabbinic academy. The same may be said about some of the Babylonian
sages found in Caesarea during this same period (1975, 90–91).
Even though the Jewish community was one of three minority groups (Christians,
Jews, Samaritans) living in an overwhelmingly Greek pagan city (MURRAY 2000), there is evi-
dence that they maintained some knowledge of Hebrew down into the fourth and fifth cen-
turies CE. Of the nine inscriptions found in connection with the Byzantine synagogue in Cae-
sarea, six of them are in Greek and three are in Hebrew, including a large inscription
containing the twenty-four priestly courses (1 Chr. 24:7–18) dated to the fourth or fifth centu-
ry CE (CIIP 2, 1145–47). At the same time, there is good evidence that much of the Jewish
community in Caesarea was Greek-speaking, some knowing only Greek. Rabbi Abbahu was
quite knowledgeable in Greek and permitted other Jews to teach Greek to their daughters.
Also, Rabbi Bar Ḥaita is said to have gone to Caesarea and heard the Shema recited in Greek
against the stipulation that it should only be recited in Hebrew (ySotạ 29a:3) (CIIP 2, 29–30).
This city of Caesarea, then, which housed some of the most prominent Jewish schol-
ars and rabbis of the period as well as those who could only recite the Shema in Greek, was
likely the location for the original composition of the Greek transcriptions behind the Secun-
da. There are a couple of different perspectives regarding how a transcription text might have
originally functioned in such a community.
First, as suggested earlier, the transcription text may have been developed as a teach-
ing aid in the ‫( בֵּית ֵספֶר‬see 3.4.4.4). In a city like Caesarea, Greek transcription could have
served as a helpful aid to bridge the gap for Greek-speaking Jews approaching the Bible.
Second, it may have been developed in a scholarly context. In this case, the transcrip-
tions would have functioned in the context of deeper and more detailed study of the Hebrew
- 92 -
scriptures. In fact, ALEXANDER suggests that the ‫ בֵּית תַּ לְמוּד‬might have simply been a more ad-
vanced study and commentary on the Bible (1999, 85). Because commentary, interpretation,
and Halakhah all depend on the reading of the biblical text, establishing the most precise de-
tails of a vocalization tradition would be inextricably linked to the exposition of the text and
its halakhic implications (see NAEH 1992a; 1993; 3.4.4.6 below). The Greek transcriptions
underlying the Secunda, then, would be the product of the rabbinic school of Caesarea solidi-
fying a vocalization tradition upon which to base their exegesis, interpretation, and Ha-
lakhah.134 In fact, it was the study of Hebrew grammar that grew out of the discipline of bibli-
cal interpretation and not the other way around.
Whether the Greek transcriptions originally developed as a teaching aid for the ‫בֵּית‬
‫ ֵספֶר‬or as a solidified tradition for biblical interpretation in a more scholarly context, the rab-
binic community of Caesarea was certainly capable of carrying out such a project. Note, for
example, that the translation of Symmachus was probably carried out at Caesarea under the
sponsorship of the rabbis at the turn of the third century CE (SALVESEN 1991; ALEXANDER 1999,
84; MARCOS 2001, 123–26). A clear prerequisite for such a translation project would be a
thorough knowledge of the Hebrew reading tradition and an extensive knowledge of Greek,
both of which were present among the rabbinic community of Caesarea.

3.4.4.6. Transcriptions and Jewish Views of the Biblical Text in Palestine
The idea that a Greek transcription text was originally composed to fulfill a didactic or
scholastic function in the Jewish community of Caesarea can be related to the development of
Jewish views regarding the authority of the biblical text and its vocalization.
According to NAEH, there are essentially two ways in which the rabbis might relate to
the text of the Torah from an exegetical perspective: First, the consonantal text may be re-
garded as a transcription of divine speech. In this case, the standard accepted vocalization tra-
134. One might compare the motivation behind Aquila's translation style. A number of scholars regard his style
as serving the hermeneutical principles of Rabbi Aqiba or some other school (see MARCOS 2001, 110).
- 93 -
dition (the qere) constitutes the one correct realization of this divine speech. Various readings
are only possible because the consonantal text is ambiguous; thus, other readings exist but are
not considered valid. Second, the consonantal text may be regarded as a divine text in itself.
Such a divine consonantal text (the ketiv) validates an array of possible vocalizations and, on
the basis of various potential vocalizations, a multiplicity of interpretations. Moreover, such
an approach gives greater weight to various elements of the consonantal text such as the
shape of the letters (1992a, 402).
A clear example of an interpretation based on the second approach is found in the
commentary on Genesis 3:8 (‫ ַבּגָּן לְרוּ ַח הַיּוֹם‬°ֵ‫הִים מִתְ ַהלּ‬Äֱ‫שׁמְעוּ אֶת־קוֹל י ְהוָה א‬
ְ ִ ‫' ַויּ‬and they heard the
voice of YHWH God walking about in the garden at the cool of the day') in Bereshit Rabbah
(19:8) from the Amoraic period. R. Berekhyah suggests that ‫שׁמְעוּ אֶת־קוֹל‬
ְ ִ ‫' ַויּ‬and they heard the
voice ... ' should be read as ‫שׁמִיעוּ אֶת־קוֹל‬

ְ ַ ‫' ַויּ‬and the [the trees] caused [their] voice to be heard
... ' and R. Levi suggests that °ֵ‫' מִתְ ַהלּ‬walking about' may be read as ‫ לוֹ‬°ַ‫מֵת ָהל‬/°ֵ‫' מֵת הֹל‬a dead
man is going/has gone his way', applying the words to Adam.135 These readings are clearly re-
vocalizations of the consonantal text—not a distinct inherited vocalization tradition. What
NAEH argues it that this sort of interpretation, which is based on regarding the consonantal
text itself as divine, is never the grounds for halakhic exegesis until the Amoraic period. In
the Tannaitic period, while elements of the written text might be utilized to elucidate an ag-
gadic interpretation, halakhic exegesis always depended on a fixed vocalization tradition
(1992a, 402, 443–45).
The Greek transcription text of the Secunda is best understood against the backdrop of
the first approach, characteristic of the Tannaitic period, namely, that it was the qere that was
divine. Even if the transcription text was meant to be used in conjunction with the Hebrew
ketiv, as suggested, its likely functions proposed in this section support the idea that it was re-
135. For an English translation of the passage, see FREEDMAN and SIMON (1961, 154).
- 94 -
garded as an accepted vocalization reflecting the correct reading of the Bible. If it was used
as a didactic aid for learning Hebrew in the ‫בֵּית ֵספֶר‬, then it represented the standard form of
the Torah that students learned. If it was used in a more scholastic setting, it would constitute
a standardized and accepted vocalization tradition on which to base halakhic exegesis, in line
with the Tannaitic approach outlined in NAEH. Accordingly, we may conclude that it is more
likely that the Greek transcription text of the Secunda grew out of an approach that regarded
the qere as more authoritative than the ketiv, rather than the other way around.136
3.4.4.7. Summary
At the beginning of this chapter it was argued on linguistic grounds that the Greek transcrip-
tions underlying the Secunda must have been composed no earlier than the beginning of the
first century CE. Following this, comparative parallel and transcriptional material from the
Hellenistic, Roman, and Byzantine Near East was examined. The relevant conclusions
emerging from this analysis were that a parallel transcription text would be associated with
both the fading away of the transcribed language and the learning of a language or literature
whose words were regarded as especially powerful. These conclusions were then applied to
Palestine. It was demonstrated that during the chronological window for the composition of
the Secunda, spoken Hebrew was fading away, there was a gradual move toward the estab-
lishment of Jewish schools, and the correct recitation of the Hebrew words of scripture was
regarded as very important. All of this led to the likely hypothesis that the Greek transcrip-
tions underlying the Secunda were originally developed and used in a didactic and/or scholar-
136. At the same time, there is evidence that Origen interacted with Jewish interpretations and exegesis that
rested on the ketiv, rather than the qere. For example, he cites three separate interpretations of Ezekiel 9:4–6 that
depend on the shape of the letter taw in the Jewish script (Selecta in Ezechielem, 13.800–801; see also 2.2.2.1).
However, this does not contradict what has been suggested for a few reasons. First, such an interpretation does
not constitute a revocalization but merely an explanation based on the "material data" (NAEH 1992a, 444–45) of
the Bible. Second, the text in Ezekiel itself contains the word ‫' תָּ ו‬sign', which seems to be understood by
Origen's Jewish interlocutors as a reference to the letter taw. Third, the explanations cited by Origen are better
regarded as aggadic interpretations, rather than halakhic interpretations. Therefore, such examples in Origen's
writings pose no contradiction to the idea that the Secunda fits within the Tannaitic interpretive context and
perspective.
- 95 -
ly context in the schools of Caesarea Maritima toward the end of the second century CE or
beginning of the third century CE.
While similar to the view expounded by EMERTON (1956; 1971), this chapter has built
upon, expanded, and gone deeper than his original articles. The composition of the Secunda
has been better established in the wider context of the Hellenistic Near East and the more
specific context of the linguistic and social world of Palestine and, more specifically, Cae-
sarea. Such a theory is to be preferred over hypothesizing an originally liturgical function for
the Secunda (see section 3.2.2).
- 96 -
4. PRONUNCIATION OF GREEK IN ROMAN PALESTINE
4.1. INTRODUCTION
In the preceding two chapters, it was argued that the original text of the Secunda was com-
posed in Palestine at some point during the second or third century CE. Therefore, understand-
ing the relationship between the orthography and phonology of the Greek of this period will
elucidate the orthography of the Secunda. The focus of the present chapter is a comprehen-
sive analysis of the phonology of Palestinian Koine Greek from the Hellenistic period to the
Byzantine period. First, methodology will be outlined (4.2), including a summary of a prelim-
inary phonemic system (4.3), second, in order to provide proper background, previous re-
search on Egyptian Koine phonology will be summarized (4.4), third, the data for the pronun-
ciation of Palestinian Koine will be presented (4.5), finally, in order to help lay the
foundation for understanding the phonology of the Secunda, the chapter will conclude by out-
lining the correspondences between the Greek phones and the various Greek graphemes used
to represent them at the time of the composition of the Secunda (4.5.4).
4.2. METHODOLOGY
Unlike modern languages, whose phonology can be analyzed through real speech, ancient
languages are only preserved by means of written texts. Apart from ancient grammatical trea-
tises in which phonology is described explicitly, the actual pronunciation of an ancient lan-
- 97 -
guage can only be discerned through occasional spelling interchanges. For example, a mis-
spelling of English perceive as percieve** indicates that ei and ie are equivalent in the writing
system, both representing the phoneme /i/ (GIGNAC 1976, 57). In the same way, the common
"misspelling" of κιτε for κειται 'lies' in Greek funerary inscriptions of the Byzantine period
(e.g., variation 33.15) indicates that the pair ι/ει represented one phoneme (/i/) and the pair ε/
αι represented one phoneme (/ε/) in contemporary pronunciation. This method, which is typi-
cally implemented in studies of Koine Greek phonology (e.g., GIGNAC 1976; TEODORSSON
1977), will be adopted here for the analysis of Palestinian Koine phonology.
The corpus for the study of Palestinian Greek phonology in this chapter is comprised
of the Corpus Inscriptionum Iudaeae Palaestinae (CIIP) (2010–2014)137 and the electronic
database of the Greek Judean Desert Manuscripts (G-JUDEAN-T) (2015).138 At the time of
writing (May 2017), the former includes two volumes on Jerusalem (2010, 2012), a volume
on Caesarea and the Middle Coast (2011), and a volume on the South Coast (2014).139 The
latter includes all the non-biblical non-Qumran Greek documents provenanced to the Judaean
Desert. These two sources account for approximately 28,000 Greek words attested in about
2,000 Greek inscriptions and texts. Approximately 1,400 spelling interchanges have been col-
lected from these sources.
Although such numbers may seem high, they are only sufficient for a near compre-
hensive reconstruction of the phonology of the language. Accordingly, we may utilize the
Egyptian material, which could fill more than 100 volumes (GIGNAC 1976, 50), to fill in some
of the Palestinian material. The work of TEODORSSON (1977), who covers Egyptian Koine in
137. The goal of this six-volume series is to document every inscription in Palestine from Hellenism to the
Islamic conquest.
138. The electronic database (G-JUDEAN-T) is the source of the non-biblical non-Qumran Greek section of
The Dead Sea Scrolls Concordance (2015).
139. Volumes on Judaea/Idumea, Galilee, and the Negev are expected to be published by 2020. It is my
intention to update the present work as the coming volumes are published.
- 98 -
the Ptolemaic period (332 BCE–31 BCE), and the work of GIGNAC (1976), who covers Egyptian
Koine in the Roman and Byzantine periods (30 BCE–735 CE), will serve as our sources for the
phonology of Greek in Egypt. Because the findings of these scholars are valuable not only for
filling in gaps but also for comparison, their work will be summarized briefly before we pro-
ceed to treat the Palestinian material.
4.3. PRELIMINARY PHONEMIC SYSTEM
Because the endeavor of determining ancient phonology depends on spelling interchanges, it
is necessary to have a standard system of correspondences between graphemes (e.g., α, ε, η,
γ, δ, etc.) and phonemes (e.g., /a/, /e/, /εː/, /g/, /d/, etc.) by which spelling interchanges in the
Koine period will be measured. TEODORSSON calls this a tentative phonemic system (1977,
43–47). The basic state of affairs that obtained in Attic Greek around 350 BCE, out of which
the Koine developed (TEODORSSON 1977, 25–35, 44), will function in this way and serve as
our point of departure.
The Attic Greek vocalic system essentially consisted of eight different qualities: /i/,
/y/, /e/, /ε/, /a/, /ᴐ/, /o/, /u/. Three of these were always long: /εː/, /ᴐː/, /uː/. The other five qual-
ities could be long or short: /i(ː)/, /y(ː)/, /e(ː)/, /a(ː)/, /o(ː)/. In addition to simple vowels, the
vowels /y/, /e/, /o/, and /a/ could be combined with /i/ and /u/ to produce diphthongs. The
resulting vocalic system and its graphemic representation are as follows (TEODORSSON 1977,
44–46; PETROUNIAS 2007b; JOSEPH 2014; HORROCKS 2014, 164):

Short Vowels Long Vowels Short Diphthongs Long Diphthongs
/i/ ι /iː/ ι
/y/ υ /yː/ υ /yi/ υι
/e/ ε /eː/ ει /eu/ ευ /εːu/ ηυ
/εː/ η /ei/ > /eː/ ει /εːi/ ηι
/a/ α /aː/ α /au/ αυ
/ᴐː/ ω /ai/ αι /aːi/ αι
/o/ ο /oː/ ου /oi/ οι /ᴐːi/ ωι
/uː/ ου /ou/>/οː/ ου
Chart 1: Preliminary Vocalic Phonemes: Attic Greek Vowels
- 99 -
The Attic Greek consonantal system is made up of fifteen distinct phonemes, includ-
ing three voiceless stops (/p/, /t/, /k/), three voiced stops (/b/, /d/, /g/), three voiceless aspirat-
ed stops (/ph/, /th/, /kh/), two fricatives (/s/, /h/), two liquids (/l/, /r/), and two nasals (/m/, /n/).
The following consonants could be doubled: γγ = /ng/, κκ = /kː/, λλ = /lː/, µµ = /mː/, νν = /nː/,
ππ = /pː/, σσ = /sː/, ττ = /tː/. Additionally, a few graphemes (ζ, ψ, ξ) represented a combina-
tion of two consonantal phonemes (/zd/, /ps/, /ks/) (TEODORSSON 1977, 43–47; PETROUNIAS
2007b, 568–69):
Voiceless unaspirated stops: /p/ = π, /t/ = τ, /k/ = κ

Voiced unaspirated stops: /b/ = β, /d/ = d, /g/ = γ
Voiceless aspirated stops: /ph/ = φ, /th/ = θ, /kh/ = χ
Fricatives: /s/ ([z])140 = σ, /h/ = ῾
Liquid resonants: /l/ = λ, /r/ = ρ
Nasal resonants: /m/ = µ, /n/ = ν
Combinations: /dz/ /zd/ = ζ, /ps/ = ψ, /ks/ = ξ
Chart 2: Preliminary Consonantal Phonemes: Attic Greek Consonants
4.4. EGYPTIAN KOINE GREEK

4.4.1. Vowels
4.4.1.1. Ptolemaic Period
TEODORSSON divides the developments of Ptolemaic Egyptian Koine into four periods. The
first phonological stage, which Teodorsson sets in the mid-third century BCE, is marked by
four sound changes. On the front vowel axis, the diphthong /εːi/ = ηι monophthongizes to /iː/,
as demonstrated by spelling interchanges such as κελεύσις (for κελεύσηις) (1977, 122). The
long vowel /eː/ = ει shifts to /iː/, except before a vowel, where it remains /eː/. This is demon-
strated by spelling interchanges such as χαίριν (for χαίρειν), but σκαφῆα (for σκαφεῖα) (62,
114). The vowel /εː/ = η is raised to /eː/, as demonstrated by interchanges such as ὑγιές (for
ὑγιής) and ὧδη (for ὧδε) (103, 109). On the back axis, the long vowel /ᴐː/ is raised to /oː/,141
140. Preceding a voiced consonant such as /m/, /s/ = [z] (e.g., κόσµος [kozmos]).
141. It should be noted that TEODORSSON assumes that /oː/ = ου shifted to /uː/ = ου already before this period.
- 100 -
as demonstrated by interchanges such as ἔχο (for ἔχω) and τώπους (for τόπους) (152, 156).
The resulting vocalic system and its graphemic representation are as follows (changes in
bold):
/i/ ι /iː/ ι, ει, ηι
/e/ ε /eː/ ει, η /eu/ ευ /eːu/ ηυ
/a/ α /aː/ α /au/ αυ
/ai/ αι /aːi/ αι
/o/ ο /oː/ ω /oi/ οι /ᴐːi/ ωι
/uː/ ου
The second phonological stage, which Teodorsson sets at the beginning of the second
century BCE, is marked by a number of sound changes. On the front vowel axis, /ai/ shifts to
/æː/, as demonstrated by interchanges such as κὰ οὔτε (for καὶ οὔτε) and παλεοῦ (for
παλαιοῦ) (127, 130), and /e/ lowers to /ε/ (254–55). The diphthong /aːi/ shifts to /aː/, as
demonstrated by interchanges such as χώρα (for χώραι) and ἀπὸ βορρᾶι (for ἀπὸ βορρᾶ)
(124, 126). The diphthong /ᴐːi/ monophthongizes to /ᴐː/, as demonstrated by interchanges
such as ἑτοίµω (for ἑτοίµωι) (161). The diphthong /oi/ monophthongizes to /øː/, as demon-
strated by interchanges such as ὀκίαν (for οἰκίαν) and ἀνύγω (for ἀνοίγω) (137, 140). The
resulting vocalic system and its graphemic representation are as follows:

/i/ ι /iː/ ι, ει, ηι
/øː/ οι
/ε/ ε /eː/ ει, η /εu/ ευ /eːu/ ηυ
/æː/ αι /aː/ α, αι /au/ αυ
/a/ α /ᴐː/ ωι
/o/ ο /oː/ ω
/uː/ ου
The third phonological stage, which Teodorsson sets in the mid-second century BCE, is
marked by a number of sound changes accompanied by the neutralization of length. On the
front axis, the vowel /æː/ merged with /ε/ into the vowel /ε/, as demonstraed by the increased
frequency of interchanges such as δέοµέ σου (for δέοµαί σου) and παρακαλεῖται (for
- 101 -
παρακαλεῖτε) (130–31). On the back axis, the vowel /ᴐː/ closed so as to merge with /oː/, as
demonstrated by interchanges such as ἐν οἴκο (for ἐν οἴκῳ) and ὡιµοίως (for ὁµοίως) (168–
69). With respect to the diphthongs, the second element of /au/ and /eu/ had fricativized to cre-
ate the pair [aΦ]/[aβ] and [eΦ]/[eβ], or the pair [aw] and [ew]. The shift of /au/ to [aΦ]/[aβ]/
[aw] is demonstrated by interchanges such as ἀτοῦ (for αὐτοῦ), καθάυπερ (for καθάπερ), and
ῥάυδους (for ῥάβδους). The shift of /eu/ to [eΦ]/[eβ]/[ew] is demonstrated by interchanges
such as κελέουσι (for κελεύουσι), κελύει (for κελεύει), εφ εἴη (for ευ εἴη), and συνεουδοκῶ
(for συνευδοκῶ) (142–43). Interpreting such interchanges as reflecting [aΦ]/[aβ]/[aw] and
[eΦ]/[eβ]/[ew] is grounded in two principles. First, the fact that α and ε may function as
spelling equivalents of αυ and ευ (and vice versa) indicates that the second element of the
diphthongs, originally represented by υ, had ceased to be vocalic. Second, the fact that the
second element of the graphemes αυ and ευ may interchange with β, φ, and ου indicates that
the vowel [u] had not merely elided but came to be realized as a consonant, which was oc-
casionally approximated by β, φ, and ου. The resulting vocalic system and its graphemic rep-
resentation are as follows (changes in bold):

Vowels (Qualitative) Diphthongs (Qualitative)
/i/ ι, ει, ηι
/y/ υ /yi/ υι
/e/ ει, η /εw/ = [εw], [εΦ], [εβ] ευ
/ε/ ε, αι /ew/ = [ew], [eΦ], [eβ] ηυ
/ø/ οι /aw/ = [aw], [aΦ], [aβ] αυ
/a/ α, αι
/o/ ο, ω, ωι
/u/ ου
The only other change to occur before the end of the period was the merger of /ø/ and
/y/ to /y/, as demonstrated by the decreased frequency in the first century BCE of interchanges
such as ποῶν (for ποιῶν) and the increased frequency of interchanges such as σοὶ δέδωκας
(for σὺ δέδωκας) and ἀνύγω (for ἀνοίγω) (137, 140).
- 102 -
4.4.1.2. Roman and Byzantine Periods
GIGNAC has assembled a wealth of data on Egyptian Koine in the Roman and Byzantine peri-
ods along with helpful analyses, but his work lacks the diachronic precision of TEODORSSON.
Accordingly, the data from Gignac's work will be summarized phoneme-by-phoneme, with
only the Roman and Byzantine periods serving as the main diachronic division.142
4.4.1.2.1. Front Axis
The phoneme /i/ remained essentially unchanged, typically being represented by ι and ει, as
demonstrated by interchanges such as ἴκοσι (for εἴκοσι) and µεικρός (µικρός) (189–90).
However, it had an allophonic variant of [ε] before liquid consonants, as demonstrated by in-
terchanges such as χιρός, χερός, χηρός (for χειρός) and καµίλων, καµείλων, καµέλων (for
καµήλων) (261–62). The phoneme /y/ remained essentially unchanged, typically being rep-
resented by υ and οι, as demonstrated by interchanges such as µυ (for µοι), ἥµισοι (for
ἥµισυ), and πυῶ (for ποιῶ) (197–200). Interchanges of υ and υι, such as ὑός (for υἱός) and
ἔγγυιοι (for ἔγγυοι), also indicate that the diphthong /yi/ was simplified to /y/ (202–203, 207).
Expanded orthographies such as υεἱός (for υἱός) and υιεἱοῦ (for υἱοῦ) seem to indicate that υι,
which always precedes a vowel in Greek words, indicated the vowel /y/ followed by a glide
[j] before the following vowel (202–203). The phoneme /e/ was unstable during the period.
By the Byzantine period it had certainly shifted to /i/, as demonstrated by interchanges such
as ἄχρης (for ἄχρις) and ὑµῆν (for ὑµῖν) (238). GIGNAC places the terminus ante quem for the
/e/ > /i/ shift by the second century CE, but admits that /e/ > /i/ might not have been universal
during the Roman period. When the phoneme /e/ was represented, it was typically represent-
ed by η/ηι, rather than ει (191, 330). Like the phoneme /i/, it had an allophone [ε] before liq-
uid consonants, where it could be represented with η, ει, ι, or ε, as demonstrated by the exam-
ples above. The phoneme /ε/ remained essentially unchanged, typically being represented by
ε and αι, as demonstrated by interchanges such as χέρειν (for χαίρειν) and ὑπαίρ (for ὑπέρ)
142. The interchanges which Gignac attributes to the bilingual interference of Coptic are ignored in the
following summary.
- 103 -
(192). However, it had an allophonic variant of [i] before back vowels, nasals, and /s/, as
demonstrated by interchanges such as ἐκθίσται (for ἐκθέσθαι) and ἕνεκιν (for ἕνεκεν) (250).
Like the phonemes /i/ and /e/, it had a lowered allophone ([a]) occurring especially before /r/,
as demonstrated by interchanges such as ὑπάρ (for ὑπέρ) and ἑτάρις (for ἑτέροις) (284).
4.4.1.2.2. Back Axis
The phoneme /a/ remained essentially unchanged, typically being represented by α, occasio-
nally by αι, and rarely by αυ. The latter correspondences are demonstrated by interchanges
such as κατὰι (for κατὰ) and καθαύπερ (for καθάπερ) (194, 229). In unstressed syllables, es-
pecially before /s/ and /n/, /a/ had a tendency to shift to [ə], which was represented by ε. This
is demonstrated by interchanges such as πάντε (for πάντα), συνεγοµένους (for
συναγοµένους), and ἁκουσίως (for ἑκουσίως) (279, 283). This reflects the reduction and cen-
tralization of unstressed vowels to [ə] (285). The phoneme /o/ remained essentially un-
changed, typically being represented by ο, ω, or ωι. The fact that there was no longer a dis-
tinction in length is demonstrated by interchanges such as ἔχο (for ἔχω), λόγο (for λόγῳ), and
αὐτώς (for αὐτός) (276–77). Like the phoneme /a/, in unstressed syllables and especially be-
fore /s/, /o/ had a tendency to shift to [ə], which was represented by ε. This is demonstrated
by interchanges such as τέλες (for τέλος), ἕκαστες (for ἕκαστος), and ὁβδοµήκοντα (for
ἑβδοµήκοντα) (289, 291). As with the /a/ > [ə] shift above, this reflects the reduction and cen-
tralization of unstressed vowels to [ə] (291–92).143 The phoneme /u/ remained essentially un-
changed, typically being represented by ου. While interchanges such as παρὰ σῶ (for παρὰ
σοῦ) and ἐν τῷ νοµοῦ (for ἐν τῷ νοµῷ) demonstrate that ου represented a simple vowel and
not a diphthong, the relative infrequency of interchanges with ω and the possibility of bilin-
gual interference lead to the conclusion that the value of ου was simply /u/ (208, 213–14).
143. It is possible that bilingual interference played factor in the shifts, since Coptic had no /o/ vowel (GIGNAC
1976, 291–92). Nevertheless, even if bilingual interference was a factor, it is significant that in each instance the
reduced centralized vowel is represented with ε.
- 104 -
4.4.1.2.3. Diphthongs
The diphthong /yi/ had simplified to /y/ during the period, as demonstrated by interchanges
such as ὑοί (for υἱοί) and ἀλληλενγυίης (for ἀλληλεγγύης) (202–203, 207). Expanded or-
thographies, such as υεἱός (for υἱός) and υιεἱοῦ (for υἱοῦ), indicate that pre-vocalic υι repre-
sented a combination of the vowel /y/ and the glide [j] before the following vowel (202–203).
There is not much evidence regarding the pronunciation of the diphthong /eu/ in the papyri
from the Roman and Byzantine periods. However, interchanges such as ηοὐχρήστησεν (for
ηὐ- or εὐχρήστησεν) and κουρηούς (for κουρεύς) may indicate that the diphthong was main-
tained, eventually shifting to [ew], then [eβ], and presumably [iβ] after the /e/ > /i/ shift (188–
89). The diphthongs /εu/ and /au/ are often represented simply with ε or α, as in δετέρου (for
δευτέρου) and ἀτοῦ (for αὐτοῦ). While the above interchanges seem to indicate that the final
element of the diphthong had simply been elided, spellings such as ἑρµηνεούς (for ἑρµηνεύς),
δευουδέρου (for δευτέρου), αοὐτοῦ (for αὐτοῦ), and αυοὐτῶν (for αὐτῶν) demonstrate that
the second element of the diphthong was retained. Together, these data point to the fact that
the final υ of the graphemes ευ/αυ had shifted to a sound not typically represented in the
spelling system. During the Byzantine period, the interchanges with ε(υ)ου and α(υ)ου almost
totally disappear, being replaced by interchanges with εβ and αβ, such as προσαγορε͂βσε (for
προσαγορεῦσαι) and ναύιαν (for ναύβιον). This diachronic progression seems to indicate that
the second element of the diphthong was gradually closing, progressing from /u/ > /w/ > /β/.
Because the spellings εβ/αβ are mostly from the Byzantine period and the spellings ε(υ)ου
and α(υ)ου are mostly prior to it, it would be reasonable to assume the following diachrony:
The graphemes ευ and αυ represented /εu/ and /au/ until about the mid-second century BCE,
when they shifted to /εw/ and /aw/. From that point on the second element became gradually
more and more closed until, finally, the Byzantine period serving as the terminus ante quem,
they became /εβ/ and /aβ/ (with likely allophones of [εΦ] and [aΦ] before voiceless conso-
nants) (GIGNAC 1976, 68–70, 226–34; TEODORSSON 1977, 142–43).
- 105 -
4.4.1.2.4. Summary
Because few sound changes distinguish the vocalic system of the Byzantine period from that
of the Roman period, they are represented together in the chart below. Changes from the
Roman to the Byzantine period are marked with an arrow (>) and bolded text:144
Vowels (Qualitative) Diphthongs (Qualitative)
/i/ ι, ει, > ι, ει, η, ηι
/y/ υ, υι, οι /yi/ > /y/ υι
/e/ > /i/ η > η, ει /ew/ = [eΦ]/[eβ] > [if]/[iv] ηυ
/ε/ ε, αι, ει / _l,r /εw/ = [εΦ]/[εβ] > [εf]/[εv] ευ
/a/ α, αι /aw/ = [aΦ]/[aβ] > [af]/[av] αυ
/o/ ο, ω, ωι
/u/ ου
/ə/ ε (ο, α)
4.4.1.3. Summary
A summary of the correspondences between each grapheme and the phonemes it represents is
presented in chart form below. A second chart is added displaying the orthographic inter-
changes which serve as evidence for the phonological developments (changes are marked
with >, every successive change adding an additional >) (charts 3 and 4):
Cl. Attic 250 BCE 200 BCE 150 BCE 50 BCE 30 B.–395 C. 395 C.–735 C.
Graphs
ι /i(ː)/ /i(ː)/ /i(ː)/ > /i/ /i/ /i/ /i/
υ /y(ː)/ /y(ː)/ /y(ː)/ > /y/ /y/ /y/ /y/
η /εː/ > /eː/ /eː/ >> /e/ /e/ /e/ >>> /i/
ε /e/ /e/ /ε/ /ε/ /ε/ > /ε/, [ə] /ε/, [ə]
α /a(ː)/ /a(ː)/ /a(ː)/ > /a/ /a/ /a/ /a/
ο /o/ /o/ /o/ /o/ /o/ /o/ /o/
ω /ᴐː/ > /oː/ /oː/ >> /o/ /o/ /o/ /o/
Digraphs
ου /oː/, /uː/ > /uː/ /uː/ >> /u/ /u/ /u/ /u/
υι /yi/ /yi/ /yi/ /yi/ /yi/ > /y/ /y/
ηι i
/εː / > /iː/ /iː/ >> /i/ /i/ (?) /e/ (?) /i/
ει /eː/ > /eː/, [iː] /eː/, [iː] >> /e/, [i] /e/, [i] >>> /i/ /i/
αι /a(ː)i/ /a(ː)i/ > /æː/, /aː/ >> /ε/, /a/ /ε/, /a/ /ε/, /a/ /ε/, /a/
οι /oi/ /oi/ > /øː/ >> /ø/ >>> /y/ /y/ /y/
144.The changes marked with arrows began in the Roman period and were completed by the Byzantine period.
- 106 -
ωι /ᴐːi/ /ᴐːi/ > /ᴐː/ >> /o/ /o/ /o/ /o/
ηυ u
/εː / > /eː / u u
/eː / >> [ew],[eΦ],[eβ] [ew],[eΦ],[eβ] [ew],[eΦ],[eβ] >>> /iΦ/,/iβ/
ευ /εu/ /εu/ /εu/ > [εw],[εΦ],[εβ] [εw],[εΦ],[εβ] [εw],[εΦ],[εβ] >> /εΦ/,/εβ/
αυ /au/ /au/ /au/ > [aw],[aΦ],[aβ] [aw],[aΦ],[aβ] [aw],[aΦ],[aβ] >> /aΦ/,/aβ/

Chart 3: Egyptian Koine Greek Vowels: Grapheme-Phoneme Correspondences by Period
(x∨ = decreased frequency, x^ = increased frequency, black text indicates a new feature for the period)
250 BCE 200 BCE 150 BCE 50 BCE 30 BCE – 395 CE –

395 CE 735 CE
ι = ηι ι = ηι ι = ηι ι = ηι ι = ηι ι = ηι
ι = ει ι = ει ι = ει ι = ει ι = ει ι = ει
η=ε η=ε η=ε η=ε η=ε η=ε
ω=ο ω=ο ω=ο ω=ο ω=ο ω=ο
αι = α αι = α αι = α αι = α αι = α
αι = ε αι = ε (x^) αι = ε αι = ε αι = ε
ωι = ω ωι = ω ωι = ω ωι = ω ωι = ω
οι = ο οι = ο οι = ο (x∨) οι = ο οι = ο
οι = υ οι = υ οι = υ (x^) οι = υ οι = υ
ωι = ο ωι = ο ωι = ο ωι = ο
αυ = α αυ = α αυ = α αυ = α
αυ = αβ αυ = αβ αυ = αβ αυ = αβ (x^)
ευ = ε ευ = ε ευ = ε ευ = ε
ευ = εφ ευ = εφ ευ = εφ ευ = εφ
ευ = εου ευ = εου ευ = ε(υ)ου ευ = ε(υ)ου (x∨)

αυ = α(υ)ου αυ = α(υ)ου (x∨)
ευ = εβ ευ = εβ (x^)
οι/υ = υι οι/υ = υι
ι=η ι = η (x^)
ε=ο=α ε=ο=α
Chart 4: Egyptian Koine Greek Vowels: Orthographic/Spelling Interchanges by Period
- 107 -
4.4.2. Consonants
4.4.2.1. Ptolemaic Period
Only two consonants changed substantially during the Ptolemaic period. First, ζ = /dz/ /zd/
simplified to /zː/ and eventually to /z/. This is demonstrated by interchanges such as
διαλογίσεσθαι (for διαλογίζεσθαι), κεκόµιζµαι (for κεκόµισµαι), δεσζµοῦ (for δεσµοῦ), and
ἀνασζητήσας (for ἀναζητήσας) (TEODORSSON 1976, 190–91). Second, the voiced velar stop γ
= /g/ shifted to a fricative with two allophonic realizations: [j] / _V[+front] and [γ]. This is
demonstrated by interchanges such as ὀλίον (for ὀλίγον), ὑποτεταµένοις (for
ὑποτεταγµένοις), ἀρχιγερεύς (for ἀρχιερεύς), ἱγεροῦ (for ἱεροῦ), and ὑγιγαίνει (for ὑγιαίνει).
The former shift (/zd/ > /zː/ > /z/) had obtained by the mid-third century BCE, and the latter
shift (/g/ > [γ], [j] / _V[+front]) by the mid-second century BCE (TEODORSSON 1976, 184–87; HOR-
ROCKS 2014, 171).

4.4.2.2. Roman and Byzantine Periods
A number of additional consonantal changes occur during the Roman and Byzantine periods.
First, β = /b/ comes to represent a bilabial fricative [β]. This is demonstrated by the fact that β
is used to transcribe Latin v145 and the fact that β occasionally interchanges with the second
element of the diphthongs αυ/ευ. Examples of the former are βιάτικον (for viaticum),
Σιλβανός (for Silvanus), and Σεβήρου (for Severus). Examples of the latter are προσαγορε͂βσε
(for προσαγορεῦσαι) and ἑυδόµῃ (for ἑβδόµῃ) (GIGNAC 1976, 68–70). Second, the voiced
stop δ = /d/ shifted to a fricative [ð] before [j] during the 1st–3rd centuries CE, and was general-
ized as /ð/ from the fourth century CE onwards. This is demonstrated by interchanges such as
δά (for διά), καρύζα (for καρύδια), ζειαβαλεῖν (for διαβαλεῖν), and ὑβρίδι (for ὑβρίζει) (GI-
GNAC 1976, 75–76). However, HORROCKS argues that there is no reason to suppose an earlier
change of /d/ > [ð] before [j] in the 1st–3rd centuries CE (2014, 170). It should be noted that
even after the fricativization of γ, β, and δ, they maintained their stop realizations after nasals
145. Latin v [w] began to be pronounced as [β] from the first century CE (ALLEN 1978, 40–42).
- 108 -
(e.g., ἀναβαίνω = /anaβeno/, but συµβιος = /symbios/) (GIGNAC 1976, 70). Third, final ν and
final ς have a tendency to be omitted, as demonstrated by interchanges such as τὴν ...
συνήθεα (for τὴν συνήθειαν) and τῆς θυγατρό µου (for τῆς θυγατρός µου). This indicates that
final /n/ and final /s/ had dropped in the speech of some of the writers (111–14, 124–31).
Fourth, aspiration (i.e., the phoneme /h/) had disappeared from speech, as demonstrated by
phrases such as µετ᾽ ὄρκου (for µεθ᾽ὅρκου) and ἐπ᾽ οἶς (for ἐφ᾽ οἷς) (133–38).
Finally, it should be mentioned that GIGNAC finds little evidence for the fricativization
of the aspirated stops (ph > f, kh > χ, th > θ) in the corpus (1976, 75–76). He bases this claim
on the frequency of interchanges with the unaspirated stops, such as πόβον (for φόβον),
χαταβένω (for καταβαίνω), and θιµήν (for τιµήν) (86–98). However, there are a few instances
from the fifth century CE and later in which φ is transcribed as f, as in egrafe ἐγράφη,
Foibammonos Φοιβάµµων, and Epifaniu Ἐπιφάνιος. GIGNAC is careful to point out that there
is far more evidence for the continued plosive pronunciation of the stops in Greek (1976, 99–
100). While GIGNAC does not find evidence for the fricativization of φ, χ, and θ in the Roman
or Byzantine periods, they eventually were fricativized. HORROCKS, based on evidence from
Laconia (σιός for θεός in the 5th CE), Asia Minor (/kh/ > /χ/ in 2nd BCE, /ph/ > /f/ in 2nd CE), and
Attic inscriptions (Ἐφρονίς for Εὐφρονίς in 2nd CE), suggests that fricativization began in the
Hellenistic period outside of Egypt and was fairly widespread by the fourth century CE (2014,
170–71). This conclusion should be held loosely.

4.4.2.3. Summary
In sum, the consonants ζ, γ, β, δ, ν, and ς (and the aspirated stops φ, χ, θ) all underwent
phonological changes from the Hellenistic period to the Byzantine period in Egyptian Koine.
The changes of these consonants are summarized in the following chart (chart 5):
Classical Period Hellenistic Period Roman Period Byzantine Period
ζ = /zd/ > /zz/ > /z/
γ = /g/ > [γ]/[j]
β = /b/ > [β]
δ = /d/ (> [ð] / _[j]) >> [ð]
- 109 -
ν# = /n/ >Ø
ς# = /s/ >Ø
φ, χ, θ = /ph/, /kh/, /th/ > [f], [χ], [θ]
Chart 5: Egyptian Koine Greek Consonants (Classical to Byzantine Period)
4.5. PALESTINIAN KOINE GREEK

4.5.1. Introductory Remarks
The format for the analysis of the Palestinian material follows that of TEODORSSON (1977,
209–56). First, spelling interchanges are tabulated and presented (see appendix A). Second,
the data from the list of spelling interchanges are analyzed phonetically. The interchanges are
treated several at a time, being grouped according to the sound change to which they attest.
Third, based on the phonetic analysis, the phonology of Palestinian Koine will be outlined.
4.5.2. Orthographic Data
See appendix A.
4.5.3. Analysis
4.5.3.1. Graphemic Interchanges
4.5.3.1.1. <ΕΙ> ~ <Ι> (Variations 1–2)
The interchange of ει and ι is by far the most common one in the corpus, occurring 354 times
and approximately 12.45 times per 1000 words. It appears regularly in all regions and times:
≤ 1st CE 2nd CE-3rd CE ≥ 4th CE Undated Total
Var. 1: ει > ι 0/2510 3/4850 44/6626 121/14438 168/28424
Var. 2: ι > ει 25/2510 30/4850 17/6626 114/14438 186/28424
Total per 1K 9.96 6.80 9.21 16.28 12.45
This interchange indicates that in the Roman and Byzantine periods in Palestine both ι and ει
represented [i] (see GIGNAC 1976, 189–91; TEDORSSON 1977, 212–14). This conclusion is fur-
ther supported by the interchanges of ιει/ει and ι/ιε (variations 3–4; see TEODORSSON 1977,
214–15).
4.5.3.1.2. <Υ> ~ <Ι>, <Υ> ~ <ΥΙ> (Variations 39–40, 43–44)
These interchanges occur a total of 10 times and approximately 0.35 times per 1000 words:
Var. 39: υ > ι 0 0 0 3 3
- 110 -
Var. 40: ι > υ 0 0 0 1 1
Var. 43: υ > υι 0 0 0 2 2
Var. 44: υι > υ 2 0 2 0 4
Total per 1K 0.80 0.00 0.30 0.42 0.35
These interchanges, in light of the interchange of υ and οι (variations 53–54), indicate both
that υ was pronounced as /y/ and that the diphthong /yi/ had been simplified to /y/ (see GI-
GNAC 1976, 202–208, 267–73; TEODORSSON 1977, 227).

4.5.3.1.3. <ΟΙ> ~ <Υ>, <ΟΙ> ~ <Ι>, <ΟΙ> ~ <ΕΙ> (Variations 53–54, 58–60)
The interchange of οι and υ occurrs 19 times and approximately 0.67 times per 1000 words:
Var. 53: οι > υ 2 0 8 8 18
Var. 54: υ > οι 0 0 1 0 1
Total per 1K 0.80 0.00 1.36 0.55 0.67
This interchange (variations 53–54) indicates that οι represented a similar or identical quality
to υ = /y/, either [ø] or [y]. The realization of [y] is supported by the 4 instances of the inter-
change of οι and ι (variation 59–60) and οι and ει (variation 58) (see GIGNAC 1976, 197–202,
262–75; TEODORSSON 1977, 225–29).

4.5.3.1.4. <ΟΙ> ~ <Ο>, <ΟΙ> ~ <Ω>, <ΟΙ> ~ <ΟΥ>, <ΟΙ> ~ <ΑΙ>, <ΟΙ> ~ <Ε>, <ΟΙ> ~ <Η>,
<ΟΙ> ~ <ΟΟΙ> (Variations 48–52, 55–57, 61)
On the other hand, οι also interchanges with ο (variation 48), ω (variations 51–52), ου (varia-
tions 49–50), αι (variation 55), ε (variation 56), η (variation 57), and oοι (variation 61). These
combine for a total of 16 times and 0.56 times per 1000 words. While the interchanges be-
tween οι and ι occur mostly in the Byzantine period, the interchanges between οι and ο/ω/ου/
αι/ε/η/οοι occur mostly in the Roman period and mostly in the Judaean Desert. It is possible,
then, that the diphthong οι was not quite realized purely as [y] until the late Roman or Byzan-
tine period, being realized as something like [ø] during the early Roman period (see GIGNAC
1976, 199–202, 215–16; TEODORSSON 1977, 227–29, 234–35, 253–55).
- 111 -
4.5.3.1.5. <Η> ~ <Ι> (Variations 25–26)
The interchange of η and ι is one of the most significant ones covered in this study, occurring
101 times and approximately 3.55 times per 1000 words. As noted above, GIGNAC suggests
that the /e/ > /i/ (= η) shift had occurred in Egypt by the second century CE, but may not have
been universal during the Roman period (1976, 191, 330). In Palestine, on the other hand, the
overwhelming majority of the occurrences are from the Byzantine period. Moreover, it is un-
likely that any of the undated inscriptions would impact the distribution. In fact, most, if not
all, of the undated examples are probably from the Byzantine period:
Var. 25: η > ι 1 0 31 35 67
Var. 26: ι > η 2 1 22 9 34
Total per 1K 1.20 0.21 8.00 3.05 3.55
This interchange indicates that in the Byzantine period in Palestine both η and ι represented
[i], reflecting the /e/ > /i/ (= η) shift (see GIGNAC 1976, 235–242; TEODORSSON 1977, 219–20).
It is unlikely that this shift had occurred in the Roman period. All four attestations of the η/ι
interchange from the Roman period occur in phonetic environments prone to vowel raising.146
Moreover, prior to the Byzantine period, the ει/ι interchange is about 15 times more common
than the η/ι interchange. Therefore, it is probably the case that the general shift of /e/ > /i/ (=
η) was not complete in Palestine until sometime in the Byzantine period. However, /e/ = η
was prone to raising in certain phonetic environments. It is also noteworthy that the inter-
change of υ and η (variations 41–42) is not attested prior to the Byzantine period.
4.5.3.1.6. <Η> ~ <ΕΙ> (Variations 23–24)
The interchange of η and ει is rare, occurring 18 times and approximately 0.63 times per
1000 words. Most of the undated examples should be dated to the Byzantine period:
146. It occurs between a nasal and a liquid in Νατανιλου (255, 1st CE, Jerusalem); it occurs between two nasals
in Βενιαµιν (523, 1st BCE–1st CE, Jerusalem); it occurs between a sibilant and a nasal in Σηµων (210, 1st BCE–1st CE,
Jerusalem); it occurs before a sibilant in Γησχαδαν[ -- ] (Mur92, 100–135 CE, Judaean Desert). See also
variations 28–29.
- 112 -
Var. 23: η > ει 1 1 2 2 6
Var. 24: ει > η 1 0 3 8 12
Total per 1K 0.80 0.21 0.75 0.69 0.63
This interchange indicates that in the Byzantine period both η and ει represented [i] (see GI-
GNAC 1976, 235–242; TEODORSSON 1977, 218–19). Similar to the previous interchange, its few
attestations in the Roman period occur in particular phonetic conditions.147
In sum, the /e/ > /i/ shift occurred in particular contexts during the Roman period (en-
vironment of sibilants/nasals and before vowels) but was widespread in the Byzantine period.
4.5.3.1.7. <Η> ~ <Ε> (Variations 21–22)
The interchange of η and ε is relatively common, occurring 75 times and approximately 2.64
times per 1000 words. It occurs relatively frequently in all periods and regions:
Var. 21: η > ε 6 1 15 19 41
Var. 22: ε > η 2 2 12 18 34
Total per 1K 3.19 0.62 4.07 2.56 2.64
In the earlier periods, in which the η/ι interchange is extremely rare, the η/ε interchange indi-
cates that η = /e/ had not yet merged with /i/ (see GIGNAC 1976, 242–49; TEODORSSON 1977,
216–28). If the interchange of η and αι (variations 31–32) were added to this list, it would in-
crease the frequency in the earlier period. The fact that the η/ε interchange persists after the
fourth century CE indicates that the pronunciation of η as /e/ persisted well into the Byzantine
period. In fact, there is evidence elsewhere that η persisted as /e/ all the way into the 7th cen-
tury CE in the Near East (AL-JALLAD 2015, 13).148 On the other hand, it is possible that η was
pronounced as /e/ in certain phonetic environments during the Byzantine period even though
147. It occurs between a nasal and a sibilant in [Τει]µεισειωνο[ς] (497, 1st BCE–1
st
CE, Jerusalem) and before a
vowel in Ηδηα (243, 1st CE, Jerusalem).
148. AL-JALLAD claims that the interchange of η/ε in Greek renderings of Semitic onomastica such as Ταννε
(PAES III.a 628) for Ταννη prove this point (2015, 13).
- 113 -
it had shifted to /i/ generally. In the environment of sibilants and nasals (e.g., µενι, ηµοι), the
η/ε interchange may be regarded as reflecting a raised allophone ([e] or [i]) of ε = /ε/. Before
a liquid (e.g., σωτεριας, επερωτηµενης), it is likely that the η/ε interchange reflects a lowered
allophone ([ε]) of η = /e/.

4.5.3.1.8. <Ε> ~ <Ι> (Variations 17–18)
The interchange of ε and ι is infrequent, occurring 27 times and approximately 0.95 times per
1000 words. Aside from the Judaean Desert, in which the interchange is rare, it occurs rela-
tively consistently in all regions and periods:

Var. 17: ε > ι 2 0 7 3 12
Var. 18: ι > ε 3 1 3 8 15
Total per 1K 1.99 0.21 1.51 0.76 2.64
This interchange, which occurs almost exclusively in the environment of nasals, sibilants, or
before vowels, points to a raised realization ([e] or [i]) of ε = /ε/ in certain phonetic environ-
ments (see GIGNAC 1976, 249–62; TEDORSSON 1977, 215–16). This is further supported by the
interchange of αι and ι (variations 35–36).

4.5.3.1.9. <Ε> ~ <ΕΙ> (Variations 19–20)
The interchange of ε and ει occurs 13 times and approximately 0.42 times per 1000 words:
Var. 17: ε > ι 1 0 1 2 4
Var. 18: ι > ε 1 0 1 6 8
Total per 1K 0.80 0.00 0.30 0.55 0.42
This interchange, which is found in phonetic environments similar to those of the ε/ι inter-
change, indicates a raised realization ([e] or [i]) of ε = /ε/. It may be assumed that ει > ε be-
fore the liquid ρ (e.g., χερος for χειρός) indicates rather a lowered realization ([ε]) of ει (GI-
GNAC 1976, 261–62; TEDORSSON 1977, 215–16).
- 114 -
4.5.3.1.10. <ΑΙ> ~ <Ε> (Variations 33–34)
The interchange of αι and ε is one of the most common interchanges in the corpus, occurring
103 times and approximately 3.62 times per 1000 words. While it is common in the Byzan-
tine period, it is only meagerly attested in the Roman period. However, it is likely that a sig-
nificant number of the undated inscriptions are from the Roman period.149 Accordingly, the
actual number of occurrences per 1000 words for the earlier period should be higher:
Var. 33: αι > ε 0 3 48 44 95
Var. 34: ε > αι 0 0 1 8 8
Total per 1K 0.00 0.62 7.40 3.60 3.66
This interchange reflects that in the Roman and Byzantine periods in Palestine both αι and ε
represented [ε] (see GIGNAC 1976, 192–94; TEODORSSON 1977, 224).
It should be noted, however, that historical αϊ = /ai/ with trema (¨) on iota was not sub-
ject to this sound change. This is clear from variations 37–38, in which we find τροπαιεικον
(for τροπαϊκον) and Ιηνναη (for Semitic /Yannay/).

4.5.3.1.11. <Ω> ~ <Ο> (Variations 81–83)
The interchange of ω and ο is quite common, occurring 162 times and approximately 6.05
times per 1000 words. It is attested regularly in all regions and periods:
Var. 81: ω > ο 10 5 50 46 111
Var. 82: ο > ω 1 2 25 33 61
Total per 1K 4.38 1.44 11.32 5.47 6.05
This interchange indicates that ω and ο both represented the quality [o] in all periods and that
vowel length had been neutralized (see GIGNAC 1976, 275–78; TEODORSSON 1977, 233–34).
This is also reflected in the ωο/ω interchange in the word υπερων (Xhev/Se64, ?, Judaean
Desert) for ὑπερῴον (variation 83).
149. A significant portion of the undated examples occur in the Judaean Desert texts.
- 115 -
4.5.3.1.12. <Ε> ~ <Α> (Variations 87–88)
The interchange of ε and α occurs 15 times and approximately 0.53 times per 1000 words:
Var. 87: ε > α 1 0 1 5 7
Var. 88: α > ε 3 0 2 3 8
Total per 1K 1.59 0.00 0.45 0.55 0.53
This interchange indicates the tendency for vowels to be reduced or centralized away from
the stress, in addition to the raising and lowering of vowels in certain phonetic environments.
The reduction of vowels is further supported by the interchange of ο > ε (variation 90). The
interchange of ο and α in the environment of ρ, if it does not reflect reduction or centraliza-
tion, may reflect the lowering of /o/ in the environment of a liquid (variations 91–92) (see GI-
GNAC 1976, 278–93).

4.5.3.1.13. <ΕΥ> ~ <Ε>, <Υ>, <ΕΟΥ>, <ΕΟ> (Variations 62–64, 66)
The interchange of ευ and one of these other variations occurs only 6 times and approximate-
ly 0.21 times per 1000 words. It is attested in the earliest and latest periods:
Var. 62: ευ > ε 1 0 1 0 2
Var. 63: υ > ευ 0 0 1 0 1
Var. 64: ευ > εου 0 0 1 0 1
Var. 66: εο > ευ 1 0 0 0 1
Total per 1K 0.80 0.00 0.45 0.07 0.21
The omission of the υ in variations 62–63 indicate that the second element of the diphthong
had either ceased to be pronounced or that it had shifted to a sound not represented in the
writing system. The interchanges with εου and εο (variations 64 and 66), however, demon-
strate that the second element was still pronounced. It is typical to understand the develop-
ment of the diphthongs in Koine as follows: #1 /εu/ > #2 /εw/ > #3 /εβ/, /εΦ/ > #4 /εv/, /εf/.
Variations 62 and 66 indicate that Palestinian Koine had progressed at least to stage 2 by the
first century CE (see GIGNAC 1976, 226–34; TEODORSSON 1977, 229–30; HORROCKS 2014, 169).
- 116 -
4.5.3.1.14. <ΑΥ> ~ <Α>, <ΑΟΥ>, <ΑΥΟΥ>, <Ω>, <Ο> (Variations 67–69, 71–72)
The interchanges of αυ and one of these other variations occurs only 7 times and approxi-
mately 0.25 times per 1000 words. It is attested in all regions and periods:
Var. 67: αυ > α 1 1 0 1 3
Var. 68: αυ > αου 0 0 1 0 1
Var. 69: αυ > αυου 0 0 1 0 1
Var. 71: αυ > ω 1 0 0 0 1
Var. 72: αυ > ο 0 0 0 1 1
Total per 1K 0.80 0.21 0.30 0.14 0.25
Like variations 62–66, the interchanges with αου and αυου (variations 68–69) demonstrate
that the second element of the diphthong /au/ had become consonantal. Variation 67 seems to
indicate that the /au/ diphthong in Palestinian Koine had shifted to /aw/ by the first century CE.
Variation 71 likely reflects the /au/ > /oː/ shift in Latin (see GIGNAC 1976, 226–34; TEODORSSON
1977, 230–31; HORROCKS 2014, 169).

4.5.3.1.15. Latin <EV/AV> ~ <ΕΟΥ/ΑOY>, <ΕΥΟΥ/ΑΥΟΥ>, <ΕΥ/ΑΥ>, <ΕΒ/ΑΒ>, <Ε/Α>
The rendering of Latin v in proper names and Latin loanwords can shed some light on the di-
achrony of the /au/ > [aw] > [aβ]/[aΦ] > [av]/[af] change in Palestinian Greek. While compre-
hensive statistics have not been collected on these interchanges, a number of general trends
are apparent from the data.
First, ου is the regular rendering of Latin v. It may be regarded as the "default"
spelling, attested in all regions and periods: Οκταουιου Octavius (XHev/Se65, 131 CE, Ju-
daean Desert), Γαουιου Gavius (XHev/Se63, 127 CE, Judaean Desert), Νερουα Nervae
(XHev/Se63, 127 CE, Judaean Desert), Σαλουειδιηνου Salvidienus (Judaean Desert-1st/2nd CE),
Σεο[υηρου Severus (Mur114, 171 CE, Judaean Desert), Σιλουανον (2122, ?, Caesarea),
Σιλουανου Silvanus (2535, ≥ 4th CE, South Coast), and Σι]λουανου Silvanus (809, 7th CE,
Jerusalem). Second, the occasional variant υου is attested in early periods: Φλαυουιου Flav-
ius (5/6Heb21–23, ?, Judaean Desert). Third, there are a number of forms with υ from differ-
- 117 -
ent regions and time periods: Φλαυοιου Flavius (5/6Hev20, ?, Judaean Desert), Σευηριναν
Severina (764, Late Roman, Jerusalem), Σευηρου Severus (1266, 500 CE, Caesarea), and
Σευ[ηρου] Severus (1528, 4th CE–7

th
CE, Caesarea). Fourth, there are a number of forms from
early and late periods that signify Latin v by means of hiatus: Γα<ου>ιου Gavius (XHev/
Se62, 127 CE, Judaean Desert), Νοενβριων November (Mur115, 124 CE, Judaean Desert),
Φλα<ου>ιου Flavius (XHev/Se69, 130 CE, Judaean Desert), Φλαια Flavia (2446, 528–530 CE,
South Coast). Fifth, in the Byzantine period, interchanges with β are common in multiple re-
gions: Σεβηρα Severus (1548, 3rd CE–6

th
CE, Caesarea), Φλαβιανον Flavianus (842.67, 4th CE–
6th CE, Jerusalem), Ιουβεναλιου Iuvenalis (962, ≥ 6th CE, Jerusalem), and Σηλβανου Silvanus
(1150, Byzantine?, Caesarea). There is, however, an attestation of Latin v rendered with β pri-
or to the Byzantine period: Βερουταριου Verutarius (221–22, 1st BCE–1st CE, Jerusalem).
The fact that renderings with β are virtually absent until the Byzantine period is sig-
nificant. There is evidence that Latin v had at least begun to shift from /w/ > [β] or [v] by the
first century CE (ALLEN 1978, 40–42).150 Accordingly, at the very least, the increased use of
Greek εβ/αβ to render Latin ev/av in the Byzantine period in Palestinian Koine seems to indi-
cate that Greek β had shifted from /b/ to [β] and perhaps also [v] (see GIGNAC 1976, 68–71).
The fact that the same sequences in similar names may be spelled either with εβ or ευ in the
Byzantine period (e.g., Σευηρου [1266, 500 CE, Caesarea] || Σεβηρα [1548, 3rd–6th CE, Cae-
sarea]) probably indicates that their phonetic values were relatively close, if not equivalent. In
sum, the Greek diphthongs ευ and αυ were pronounced as [εw] and [aw] by the first century
CE in Palestine and as [εβ]/[εv] and [aβ]/[av] in the Byzantine period (see 4.5.3.1.14; GIGNAC
1976, 68–71, 226–34; TEODORSSON 1977, 229–31; HORROCKS 2014, 169).
150. In first century CE inscriptions, Latin v occasionally interchanges with Latin b [β] (< *[b]). In the second
century CE, Velius Longus says that v is pronounced cum aliqua adspiratione 'with some aspiration'. While the
reflex of this sound is /v/ in all the Romance languages, the pronunciation [w] was still around even in the fifth
century CE (ALLEN 1978, 41). For more, see chapter 5 on the transcription of Latin into Greek.
- 118 -
4.5.3.1.16. <ΟΥ> ~ <Υ> (Variations 73–74)
The interchange of ου and υ occurs 15 times and approximately 0.53 times per 1000 words:
Var. 73: ου > υ 3 1 3 4 11
Var. 74: υ > ου 0 0 3 1 4
Total per 1K 1.20 0.21 0.91 0.35 0.53
This change is especially common in the environment of the liquids λ and ρ. It may then re-
flect regressive assimilation and the fronting of the vowel /u/ > [y] in these environments.
The same would apply for the interchanges of ο/υ and ω > υ (variations 75–76, 86) (cf. GI-
GNAC 1976, 293–94; TEODORSSON 1977, 231–32)

4.5.3.1.17. <ΟΥ> ~ <Ο> (Variations 78–79)
The interchange of ου and ο is infrequent, occurring 22 times and approximately 0.77 times
per 1000 words. It is most frequent in the earlier period:

Var. 78: ου > ο 8 0 4 6 18
Var. 79: ο > ου 1 0 1 2 4
Total per 1K 3.59 0.00 0.75 0.55 0.77
This interchange may be viewed in conjunction with the interchange of ου and ω (variations
84–85). As in Egypt, this interchange reflects the monophthongization of the diphthong ου to
a simple vowel /u/ and the neutralization of length (see GIGNAC 1976, 208–214; TEODORSSON
1977, 232–34). A significant portion of these interchanges occur in the environment of the
liquids λ and ρ, which may indicate the lowering of /u/ > [o] in this environment.
4.5.3.1.18. <ΙΟ> ~ <Ι>, <ΕΙ> (Variations 6–10)
The interchange of ιο with ι and ει is fairly common, occurring 32 times and approximately
1.13 times per 1000 words. It occurs most frequently in Jerusalem and the South Coast, less
frequently in the Judaean Desert, and never in Caesarea. It is attested in all periods:
Var. 7: ιο > ει 2 0 0 0 2
Var. 8: ιο > ι 1 4 9 16 30
- 119 -
Total per 1K 1.20 0.82 1.36 1.11 1.13
Variations 6 (ιο > ε) and 9 (ιο > η) are relevant here as well. This interchange reflects the
monophthongization of the vowel sequence /io/ to /i/. It is not attested word-medially, but
rather only in the endings -ιος > -ις and -ιον > -ιν. It has been recognized as a feature particu-
lar of Jewish texts (ROSEN 1963, 66). It may be that speakers of Semitic languages (Hebrew
and/or Aramaic), whose phonotactics did not permit two consecutive vocalic phonemes, natu-
rally simplified these sequences in their Greek. The fact that this interchange also occurs with
the vowel α in the interchange ια > ι (variation 10) seems to support this point.
4.5.3.1.19. <ΙΟ> ~ <Ο>; <ΙΩ> ~ <Ω>; <ΙΑ> ~ <Α> (Variations 11–16)
The interchanges of ιο > o and ιω > ω occur 10 times and approximately 0.42 times per 1000
words. While most of the attestations are late, some of the undated attestations are probably
from earlier periods:

Var. 11: ιο > ο 2 0 2 4 8
Var. 12: ιω > ω 0 0 1 3 4
Var. 14 ια > α 0 0 0 2 2
Total per 1K 0.80 0.00 0.45 0.62 0.49
This interchange most likely reflects the shift of /i/ > [j] after liquids and before another vow-
el (see GIGNAC 1976, 207, 302–306; HORROCKS 2014, 169). If the multiple instances of κυρα
(e.g., 1548, 2086, 2544) attested in Palestinian epigraphy do not represent a lemma distinct
from κυρια, then they may also reflect this phenomenon. The interchanges εω > ω
(αντιληψως), ια > ιεα (Μαριεαµη), and ι > αε (Μαρκαεος) in variations 13, 15, and 16 may
reflect an attempt to represent the glide [j] in a script with no sufficient character for [j]. The
consonantal realization of ι in such contexts is also supported by variations 43–47.

4.5.3.1.20. Vowel Deletion (Variations 95–100)
Because formulaic words are regularly abbreviated in inscriptions, it is difficult to ascertain
when the omission of a vowel in a given word reflects pronunciation. However, there are at
least a few instances in which vowel deletion seems likely. It occurs in foreign words and
- 120 -
names: Βαβθα (for Βαβαθα), Ραββαθµωβοις (for Ραββαθµωαβοις), [ -- ο]υετρανος (for
ουετερανος). It occurs word-initially before a consonant cluster beginning with a sibilant:
σχιων (for ἰσχίων). It most frequently occurs in the environment of liquids and nasals:
θεµντος (for θευµνατος), αιωνς (for αιωνας), θεωφιλστατου (θεοφιλεστατου), µνµα (for
µνηµα), Κορνλιας (for Κορνηλιας), σωττριας (for σωτηριας), ορφαντροφιω (for
ορφαντροφιω), πολποτων (for πολυβοτων), and ευδαιµοσνης (for ευδαιµοσυνης) (see GIGNAC
1976, 302–310).
4.5.3.1.21. Epenthetic Vowels (Variations 93–94)
An anaptyctic vowel is occasionally inserted between consonants: Χαθουσιωνος (for
Χθουσιωνος), Δυστερου (for Δυστρου), and Απερος (for Απρος) (see GIGNAC 1976, 310–
312).
4.5.3.1.22. Vowel Length
It is clear that vowel length has been neutralized in Palestinian Koine from the following in-
terchanges: ι/ει (variations 1–2), ει/ε (variations 19–20), η/ε (variations 21–22), ι/η (variations
25–26), αι/ε (variations 33–34), οι/ο (variation 48), οι/ε (variation 56), ο/ου (variations 78–
79), ω/ο (variations 81–82), and ωο/ο (variation 83) (see GIGNAC 1976, 325; TEODORSSON
1977, 237–38).
4.5.3.1.23. Gemination and Simplification (Variations 104–105, 108, 112–13, 123–24, 129–30,
132, 135–36, 138–39, 148, 150–51, 153, 156, 159–60, 166–67, 171)
Gemination of single consonants occurs with π > ππ (αππαν[τα] for απαντα) (variation 105),
µ > µµ ([α]ν[ο]ιωγµµενον for ἀνεῳγµένον) (variation 112), ν > νν (Βεννιαµιν for Βενιαµιν)
(variation 123), τ > ττ (σωττριας for σωτηριας) (variation 129), δ > δδ (Αινγαδδων for
Ενγαδων/Εγγαδων) (variation 136), σ > σσ (εξεσστω for εξεστω) (variation 139), ζ > ζζ
(τευχιζζει for τευχιζει) (variation 148), κ > κκ (Ιακκωβου for Ιακωβου) (variation 150), χ >
κκ (Ζακκαριας for Ζαχαριας) (variation 156), γ > γγ (ζυγγ[ων] for ζυγων) (variation 159), λ >
λλ (δεσδεκαλλου for διδασκαλου) (variation 167), and ρ > ρρ ([Θεο]δορρω for Θεοδωρω)
(variation 171) (see GIGNAC 1976, 154–65; TEODORSSON 1977, 244–45).
- 121 -
Simplification of double consonants occurs with ππ > π (παπος for παππος) (variation
104), ββ > β (Ραβι for Ραββι) (variation 108), µµ > µ (Αµια for Αµµια) (variation 113), νν > ν
(δεκαενεα for δεκαεννεα) (variation 124), ττ > τ (πιτακιου for πιττακιου) (variation 130),
θθ > θ (Μαθεθ<ος> for Μαθθεθος) (variation 132), δδ > δ (Θαδαιος for Θαδδαιος) (variation
135), σσ > σ (τεσερεσκαιδεκατου for τεσσαρεσκαιδεκατου) (variation 138), κκ > κ (εκλησια
for εκκλησια) (variation 151), χχ > χ (Ζαχαι for Ζαχχαι) (variation 153), γγ > γ (Λογινος for
Λογγινος) (variation 160), and λλ > λ (Εληνεστι for Ελληνιστι) (variation 166) (see GIGNAC
1976, 154–65; TEODORSSON 1977, 244–45).
Taken together, these interchanges are proof that, like vocalic length, consonantal
length (i.e., gemination) was no longer phonemic in Palestinian Koine (see GIGNAC 1976,
154–55; TEODORSSON 1977, 244–45).

4.5.3.1.24. <Φ> ~ <Π> (Variations 101–103)
The interchange of φ and π occurs in the names Σαπιρα (for Σαφιρα) and, unless it is a short
form of Ιωσηπος, Ιωσηπ (for Ιωσηφ) (variation 102). The word πθονε (for φθονε) demon-
strates that φ lost is aspiration before θ ([ph] > [p] / _[th]) (variation 103). Because fricatives
would not be expected after a nasal, the unusual spelling of ενφ (for ἐφ᾽) may reflect a con-
tinued plosive pronunciation of φ into the Byzantine period (variation 101) (see GIGNAC 1976,
98–100; TEODORSSON 1977, 238–39, 244–45). The relatively low frequency of interchanges
between φ and π and the lack of interchanges between φ and β/υ support the pronunciation of
π = [p] and φ = [ph] (see GIGNAC 1976, 93–96; TEODORSSON 1977, 238–39).
4.5.3.1.25. <Β> ~ <Π> (Variations 106–107)
An interchange of β and π occurs in the words πολποτων (for πολυβοτων), βουργος (for
πυργος), and Υβερβερετεου (for Υπερβερεταιου) (variations 106–107) (see GIGNAC 1976,
83–86; TEODORSSON 1977, 238–39). The utilization of Greek β to render Latin v points to a
fricativized realization of β (i.e., /b/ > [β]) by the Byzantine period (see 4.5.3.1.15).
- 122 -
4.5.3.1.26. <ΜΒ> ~ <Β>, <ΜΠ> ~ <Π>, <ΜΣ> ~ <Σ>, <ΝΤ> ~ <Τ>, <ΝΘ> ~ <Θ>, <Ν#> ~ <Ø>,
<Μ> ~ <Ø> (Variations 109–111, 120–22, 126)
It is common for nasals to be omitted word-medially, as in Λαπαδιου (for Λαµπαδιου),
διαφεροτα (for διαφεροντα), προ[σ]ηνεγκοτος (for προσηνεγκοντος), and Μαναηου (for
Μαναηµου) (variations 109–111, 120–22). It is also quite common for ν to be omitted word-
finally, as in αυτο (for αυτον), Μενιαµι (for Μενιαµιν/Βενιαµιν), and βοηθω (for βοηθων).
The omission of nasals may reflect the fact that they were no longer pronounced in speech.
Their elision may have resulted in the nasalization of the preceding vowel. On the other hand,
it is also possible that the nasal assimilated to the following consonant, voicing it in the case
of stops. For example, διαφεροντα would be realized as [diaphεronda] > [diaphεrodda] (see
GIGNAC 1976, 111–14, 116–19, 165–72).

4.5.3.1.27. <Μ> ~ <Ν> (Variations 114, 125)
In several cases, the nasals µ and ν interchange, as in µηνοριων (for µηµοριον), Ναρκελ[λα]
(for Μαρκελλα), and διακοµου (for διακονου). This interchange may also attest to the loss (or
weakening) of nasals in pronunciation (see GIGNAC 1976, 111–14, 116–19).

4.5.3.1.28. <ΜΒ> ~ <ΝΒ>, <ΜΒ> ~ <ΝΜΒ>, <ΜΠ> ~ <ΝΠ> (Variations 115–17)
There is a frequent interchange of ν and µ before stops, as in συνβιον (for συµβιον) and
συνπαροντος (for συµπαροντος). Because nasals are expected to have assimilated, spellings
with ν in these instances probably reflect an orthographic phenomenon. That is, the word was
conceived of in its etymological parts and spelled accordingly (see GIGNAC 1976, 165–72).
4.5.3.1.29. <ΜΨ> ~ <ΜΣ>, <ΜΠΤ> ~ <ΜΤ> (Variations 118-19, 174–76)
When a consonant cluster beginning with a stop follows µ, there is a tendency for the stop to
be omitted, as in πεµτου (for πεµπτου) and επεµσα (for επεµψα) (see GIGNAC 1976, 64–65). It
should be noted that the µ in the interchange of ψ and µψ in variation 174 is probably a lexi-
cal phenomenon connected to the µ in the word λαµβάνω.

4.5.3.1.30. <Τ> ~ <Θ> (Variations 127–28, 133–34)
The interchange of τ and θ occurs in the words θαφος (for ταφος), Βερουτος (for Βερουθος),
and Νατανιλου (for Ναθαναηλ). It also occurs after σ and ν, as in ευφραινεστε (for
- 123 -
ευφραινεσθε), µνιστιτη (for µνησθητι), and επληθυντησαν (for επληθυνθησαν). This indi-
cates that θ lost its aspiration after ν and σ, probably demonstrating that it was also voiced af-
ter ν. The offers no evidence that θ [th] had shifted to [θ] (see GIGNAC 1976, 87; TEODORSSON
1977, 239–40).
4.5.3.1.31. <Τ> ~ <Δ> (Variation 137)
There is only one attestation of an interchange of τ and δ (τρυφακτου for δρυφακτου) (varia-
tion 137), perhaps indicating voicing in the environment of the liquid (see GIGNAC 1976, 80–
85; TEODORSSON 1977, 239–40)

4.5.3.1.32. <Τ> ~ <Ø> (Variations 131, 143)
The voiceless stop /t/ may be omitted following σ (Χρισος for Χριστος) and before ρ
(Κλευπαρος for Κλεοπατρος) (variations 131, 143). The former indicates that the consonant
cluster /st/ was occasionally simplified to /s/ and the latter may indicate that τ and ρ had simi-
lar places of articulation (see GIGNAC 1976, 66–68).

4.5.3.1.33. <Σ> ~ <Ø>, <Σ#> ~ <Ø> (Variations 141–42, 144–45, 157)
It is common for σ to be omitted both word-medially and word-finally. Word-medially, this
occurs before τ (e.g., ωτε for ωστε), θ ([µν]ηθητ[ι] for µνησθητι), and before a vowel
([ε]λεηον for ελεησον) (variations 141–42, 145). It is also erroneously added before κ
(Προβατισκις for Προβατικις) (variation 157). Most commonly, it is omitted at the end of
words, as in αυτη (for αυτης), υιω (for υιως), and αδερφο (for αδελφος). This may indicate
that final /s/ was elided and medial consonant clusters with /s/ were simplified (see GIGNAC
1976, 124–31; TEODORSSON 1977, 245–48).

4.5.3.1.34. <Σ> ~ <Ζ> (Variations 146–47)
The sibilants σ and ζ occasionally interchange, as in πρεζβευτου (for πρεσβευτου),
αγοραζµατος (for αγορασµατος), and Εσκιας (for Εζκιας) (variations 146–47). This indicates
both that /s/ was realized as /z/ before a voiced consonant and that /dz/ had simplified to /z/
(see GIGNAC 1976, 120–24; TEODORSSON 1977, 243–44).
- 124 -
4.5.3.1.35. <Χ> ~ <Κ> (Variations 154–55)
The interchange of χ and κ occurs in Μικαηλ (for Μιχαηλ) and Αντειχουον (for antiquum).
There is no evidence in the material for the shift of [kh] > [χ] (see GIGNAC 1976, 86, 95;
TEODORSSON 240–41).
4.5.3.1.36. <Κ> ~ <Γ> (Variation 158)
The interchange of κ and γ occurs frequently in the preposition εκ before a voiced consonant,
as in εγ µερους (for ἐκ µέρους) and εγ διαταγµατος (for ἐκ διατάγµατος), and in the prefix εκ
attached to a verb beginning with a voiced consonant, as in εγδικησωµεν (for ἐκδικήσωµεν)
and εγδω (for ἐκδῷ). This reflects assimilation of /k/ to the following voiced consonant (GI-
GNAC 1976, 77–80; TEODORSSON 1977, 241–43).

4.5.3.1.37. <ΓΙ> ~ <Γ> (Variation 165)
It is possible that the omission of ι following γ in αγω (for ἁγίῳ) indicates the shift of γ =
[g] > [γ]/[j], but it is inconclusive (see GIGNAC 1976, 71–75; TEDORSSON 1977, 241–43).
4.5.3.1.38. <ΓΓ> ~ <ΝΓ>, <ΓΚ> ~ <ΝΚ>, <ΓΧ> ~ <ΝΧ> (Variations 162–64)
The interchanges of γγ > νγ, γκ > νκ, and γχ > νχ are quite common, occurring a combined 67
times between the three of them, as in παρανγελλω (for παραγγέλλω), ενκληµατι (for
ἐγκλήµατι), and εντυνχανω (for ἐντυγχάνω) (variations 162–64). These interchanges indicate
that the first element of these consonant clusters was realized as /n/ (cf. GIGNAC 1976, 116).
4.5.3.1.39. <Ρ> ~ <Λ>, <Λ> ~ <Ø>, <Ρ> ~ <Ø> (Variations 168–70, 172–73)
The liquids λ and ρ occasionally interchange, as in Γληγοριας (for Γρηγοριας), λιβλαριος (for
λιβραριος), θεοφιρος (for θεοφιλος), and αδερφο (for αδερφος) (variations 168–69). There
are also a few attestations in which they are omitted (e.g., αδεφου [for αδελφου] and
πρεσβευτηο [for πρεσβυτηρου]) (variations 170, 172–73). It would seem that λ and ρ had
similar points of articulation. In Modern Greek, λ > ρ before a consonant is general and dis-
similation between liquids is common (GIGNAC 1976, 102–108).
- 125 -
4.5.4. Summary
4.5.4.1. Introductory Remarks
The spelling interchanges in the Palestinian material indicate that the significant changes in
the phonological system occurred between the Roman and Byzantine periods. Accordingly,
two phonological systems will be described. The first will outline the sound changes that had
become complete by or during the Roman period and the second the changes that had become
complete by or during the Byzantine period. It should be noted, however, that the sound
changes would have occurred gradually and not all at once.

4.5.4.2. Vowels
The main distinctives of the vocalic system against the Attic system are explained on the ba-
sis of the neutralization of length (quantity) and a series of mergers (quality). With regard to
quantity, numerous spelling interchanges attest to the neutralization of phonemic length in the
vocalic system (4.5.3.1.20). With regard to quality, the most important identifications are the
following: ει > [i] = ι (4.5.3.1.1), η > [i] = ι (Byz.) (4.5.3.1.5–6), οι ( > [ø] in Early Rom.?) >
[y] = υ (Byz.) (4.5.3.1.3), αι > [ε] = ε (4.5.3.1.10), and ω > [ο] = ο (4.5.3.1.11).
A few additional observations should be made. First, the distribution of the inter-
changes of η and ε/αι (4.5.3.1.7, 4.5.3.1.10) demonstrate that the pronunciation of η as [e]
may have persisted well into the Byzantine period. Second, reduced or centralized vowels
may be represented as ε, α, and perhaps ο (4.5.3.1.12). Third, a number of interchanges point
to a tendency for sibilants and nasals to raise vowels, on one hand, and for liquids to lower
vowels, on the other (e.g., 4.5.3.1.7–9). Fourth, consecutive vowel sequences have a tendency
to monophthongize, as in -ιος > -ις and -ιον > -ιν (4.5.3.1.17). Fifth, there are a number of
contexts in which the vowel ι [i] seems to shift to the glide [j] (4.5.3.1.19). Sixth, vowel syn-
cope may occur, usually in the environment of liquids and nasals (4.5.3.1.20). Seventh, and
finally, epenthetic vowels may sometimes be inserted between consonants (4.4.3.1.21).
- 126 -
The relationship between the orthography and phonology of the vocalic system of
Palestinian Koine Greek during the Roman period may be summarized as follows (cf.
PETROUNIAS 2007c, 602–605):151

front back
unrounded rounded central/reduced unrounded rounded
high (close) i y u
mid-close e (ø)
mid ([ə]) o
mid-open ε152
low (open) a
Chart 6: Palestinian Koine Greek Vowels (Roman Period)
Sounds Script
[i] ι, ει, (η in Byz.)
[y] υ, υι, οι
[e] η
([ø]) (οι in Early Roman?)
[ε] ε, αι153
[a] α
[o] ο, ω
[u] ου
[ə] ε (α, ο)
Chart 7: Palestinian Koine Greek Vocalic Orthography in (Roman Period)
4.5.4.3. Diphthongs
The second element of the diphthongs αυ [au] and ευ [εu] had become consonantal (i.e., the
semi-vowel [w]) during the Roman period: /εu/ > [εw], /au/ > [aw] (4.5.3.1.13–14). By the
Byzantine period, the second element of the diphthongs had likely become [β]/[Φ] and then
eventually [v]/[f]: /εw/ > [εβ]/[εΦ] > [εv]/[εf], /aw/ > [aβ]/[aΦ] > [av]/[af] (4.5.3.1.15). This
151. My charts are modeled after those of PETROUNIAS (2007c, 602–605).
152. It should be noted that this vocalic quality may not have been a mid-open vowel, but rather a true mid
vowel [e̞ ] (see PETROUNIAS 2007c, 604; HORROCKS 2014, 165–70).
153. But note that αϊ = [ai] (see 4.5.3.1.10).
- 127 -
is demonstrated by the rendering of Latin ev/av [eβ]/[aβ] (or [ev]/[av]) with Greek εβ/αβ [εβ]/
[aβ] alongside Greek ευ instead of the typical εου/αου of earlier periods. The data are insuffi-
cient to determine when precisely this change happened. While the evidence for the change is
essentially confined to the Byzantine period, it is possible that it occurred earlier as well. The
development of the diphthongs up to, during, and after the Roman period may be summarized
as follows:
Roman Period Byzantine Period
[εu] > [εw] [εw] [εw] > [εβ]/[εΦ] > [εv]/[εf] [εv]/[εf]
[au] > [aw] [aw] [aw] > [aβ]/[aΦ] > [av]/[af] [av]/[af]
Chart 8: Palestinian Koine Greek Diphthongs (Roman and Byzantine Periods)
Sounds Script
[εu] ευ
[au] αυ
[εw] ευ, ε, εου,
[aw] αυ, α, αου, αυου, ο
[εβ]/[εΦ] ευ, εου, εβ ?
[aβ]/[aΦ] αυ, (αου), αβ ?
[εv]/[εf] ευ, εου, εβ ?
[av]/[af] αυ, (αου), αβ ?
Chart 9: Palestinian Koine Greek Diphthongal Orthography (Roman/Byzantine Periods)
4.5.4.4. Consonants
There are essentially five types of consonantal changes that occur from Attic Greek to the
Late Byzantine period in Koine Greek in general (4.4.2): the simplification of geminated con-
sonants (/Cː/ > /C/), the simplification of ζ (/zd/ > [zz] > [z]), the fricativization of the voiced
stops γ (/g/ > [γ]/[j]), β (/b/ > [β]), and δ (/d/ > [ð]), the weakening of the nasals and σ, espe-
cially in final position, the fricativization of the aspirated stops φ (/ph/ > [φ]), χ (/kh/ > [χ]),
and θ (/th/ > [θ]), and the loss of aspiration (/h/ > Ø).
There is evidence in the Palestinian material that the simplification of geminated con-
sonants (4.5.3.1.23), the simplification of ζ (4.5.3.1.23, 4.5.3.1.34), and the fricativization of
- 128 -
the voiced stops (4.5.3.1.26, 4.5.3.1.33) had occurred in Palestinian Koine. With respect to
the fricativization of the voiced stops (β, γ, δ), there is only evidence for the fricativization of
β (/b/ > [β]) (4.5.3.1.15). Because it is assumed that the fricativization of γ occurred before
that of β (HORROCKS 2014, 170), we may also assume that /g/ > [γ]/[j] had taken place in
Palestinian Koine as well, for which there may be evidence (variation 165). The shift of /d/ >
[ð], on the other hand, must remain an open question for our period.
There is no evidence for the fricativization of the aspirated stops (φ, χ, θ) in the Pales-
tinian material. While GIGNAC finds only meager evidence for the fricativization of the aspi-
rated stops in the Roman and Byzantine periods in Egyptian Koine (1976, 98–101), HOR-
ROCKS assumes that Egypt reflects a more conservative phonology (2014, 170–71). However,
there are reasons to believe that the Near East in general preserved a more conservative
phonology as well (see 4.6). In fact, the rendering of both Arabic /t̠ / [θ] and /t/ [t(h)] with
Greek θ and the occasional rendering of Arabic /t̠ / [θ] with Greek τ point to a lack of frica-
tivization of the aspirated stops in the east (AL-JALLAD 2015, 13–14, 18–19).154 Therefore,
while speculative, it seems reasonable to posit an aspirated realization of φ, χ, and θ until
some point in the Byzantine period.
I found no evidence in the Palestinian material regarding the loss of aspiration. How-
ever, while spiritus asper (῾) in Greek loanwords in Hebrew is rendered sometimes as ‫א‬,
sometimes as ‫ה‬, and sometimes as ‫ח‬, spiritus lenis is always rendered as ‫( א‬HEIJMANS 2013,
279–81). Accordingly, it may be that aspiration, though weakened, was still preserved to
some degree during the Roman period.
154. For example, Arabic /ġ awθ/ is rendered into Greek both as Γαυθος and Γαυτος (334–35 CE) and the
Nabatean name /h ̣aret̠ ah/ is rendered into Greek as Αρετας. If Greek θ represented [θ] rather than [th], the
alternative representation of Arabic /t̠ / [θ] with Greek τ [t] would not make any sense. However, if Greek θ still
represented [th], then both the general convention and the exceptions make perfect sense. Arabic /t̠ / [θ] and /t/
[th] were both generally rendered by Greek aspirated θ [th] because it most closely approximated both
consonants, but the occasional use of Greek τ [t] to render Arabic /t̠ / [θ] demonstrates that the Arabic interdental
fricative did not perfectly correspond with either Greek θ [th] or τ [t] (AL-JALLAD 2015, 13–14, 18–19).
- 129 -
The relationship between the orthography and phonology of the consonantal system
(excluding the diphthongs) of Palestinian Koine Greek during the Roman period may be sum-
marized as follows (cf. GIGNAC 1976, 178–79; PETROUNIAS 2007c, 606–609) (charts 10 and
11):155
Bilabial Dental Alveolar Palatal Velar Glottal
Plosive p ph ([b]) t th ([d]) k kh ([g])
Nasal m n
Fricative β (ð?) sz γ (h?)
Trill r
Approximant l ([j])
Chart 10: Palestinian Koine Greek Consonants (Roman Period)
Sounds Script (Normal)

[p] π
[t] τ
[k] κ
[ph] φ
[th] θ
[kh] χ
([b]) (ν/µ)β, (ν/µ)π
[d] δ, (ν)τ
([g]) (γ)γ, (γ)κ, κ(δ)
[m] µ
[n] ν, γ(γ,κ,χ)
[β] β
([ð]) δ
[s] σ, ζ(θ)
[z] ζ, σ(δ)
[γ] γ
([h]) ῾
[r] ρ
[l] λ
155. Note that (δ) = voiced consonant and (θ) = unvoiced consonant in the entires for ([g]), [s], [z] below.
- 130 -
[j] γ, ι, Ø
[ks] or [khs]156 ξ
[ps] or [phs]157 ψ
Chart 11: Palestinian Koine Greek Consonantal Orthography (Roman Period)
4.6. CONCLUSION
In sum, our survey of the epigraphic and documentary evidence from the Hellenistic, Roman,
and Byzantine periods has resulted in a picture of Palestinian Koine not drastically different
from that of Egyptian Koine. Unfortunately, it was necessary to fill some gaps in the Palestin-
ian Koine system with evidence from Egypt. This was especially true for the consonantal de-
velopments.158 However, the similarity between the two systems in those areas in which evi-
dence is sufficient should be encouraging. It should be noted that a number of features of
Palestinian Koine, such as the late pronunciation of η as /e/ and the continued aspirated real-
ization of the stops φ, χ, θ, reflect a relatively conservative Greek phonology relative to the
developments in Greece. This is consistent with other data from the Near East (e.g., AL-JAL-
LAD 2015).
This chapter has thus provided a foundation for understanding the orthography and
phonology of the Secunda. The Greek pronunciation and writing conventions of the Roman
period summarized here are what the author(s) of the Secunda would have been familiar with
at the time of the composition of the Secunda. In the following chapter, we will examine gen-
eral Greek practices of transcription of other languages from roughly the same time period.
156. For the aspiration of the first element of ξ and ψ in Greek loanwords in Mishnaic Hebrew, see 5.4.1.3.6.
157. See previous note.
158. This, of course, would be expected. Because the phonological changes in the consonantal system are
primarily rephonologizations—the realization of the phoneme changes without merging with another
phoneme—spelling interchanges are less common (see TEODORSSON 1977, 36–51).
- 131 -
5. GREEK TRANSCRIPTION CONVENTIONS IN THE HELLENISTIC/ROMAN/
BYZANTINE NEAR EAST
5.1. INTRODUCTION
In chapter two, a number of instances of transcription in the Hellenistic, Roman, and Byzan-
tine Near East were analyzed from a historical and social perspective. The present chapter
will examine these same transcription texts from a linguistic and orthographic perspective,
analyzing the transcription conventions utilized when rendering other languages into Greek.
The preceding chapter, in which the historical phonology of Koine Greek in Palestine (and
Egypt) was described, will serve as a foundation for understanding the nature of the transcrip-
tion conventions implemented in these texts. Taken together, these two chapters (4 and 5) will
serve as a foundation for understanding the phonology and orthography of the Secunda.
The first part of this chapter outlines methodology; the second part constitutes a lin-
guistic and orthographic analysis of the transcription texts by language; the third part summa-
rizes relevant research regarding the realization of Greek loanwords in Tannaitic Hebrew.
5.2. METHODOLOGY
For each language analyzed, the corpus of texts and/or inscriptions is outlined. In general,
priority is given to transcription of common words rather than proper names. The phonologi-
cal inventory of the language is briefly summarized. Following this, the Greek transcription
- 132 -
conventions are analyzed in terms of vowels, semi-vowels and diphthongs, and consonants.
The languages chosen for this survey are Latin, Akkadian, Arabic, Aramaic, and Phoenician-
Punic. Latin was chosen because its phonology is the best understood of contemporary lan-
guages attested in Greek transcription.159 The remaining languages where chosen by virtue of
them being Semitic languages in Greek transcription from a roughly contemporary period. Fi-
nally, in order to better understand the cross-linguistic perceptual relationship with respect to
Greek and Hebrew transcription, the evidence regarding how Greek loanwords are realized in
Mishnaic Hebrew will be summarized.
In the case of Latin and Akkadian, I conduct original research based on a comprehen-
sive statistical analysis of a corpus of published editions of Greek transcription texts for the
relevant period. In the case of Arabic, I summarize the work of AL-JALLAD et al. (AL-JALLAD,
DANIEL, and GHUL 2013; AL-JALLAD 2015; AL-JALLAD and AL-MANASER 2015). In the case of
Phoenician and Aramaic, I make general conclusions on the basis of a small number of repre-
sentative continuous-text inscriptions and, in the case of Phoenician, supplement the data
with transcriptional material found in the grammars (e.g., FRIEDRICH and RÖLLIG 1999; KRAH-
MALKOV 2001; HACKETT 2008). In the case of Greek loanwords in Hebrew, I summarize the
work of HEIJMANS (2013) on Greek loanwords in the Mishnah.
5.3. LINGUISTIC AND ORTHOGRAPHIC ANALYSIS: GREEK TRANSCRIPTION

5.3.1. Latin
The corpus for Latin transcribed into Greek includes the following papyri from Egypt:
P.Berol.21246 (1st CE), P.Oxy. XXXIII.XXXIII.2660 (1st–2nd CE), P.Oxy. XLVI.3315 (1st–2nd
CE), P. Oxy. LXXVIII.5162 (1st–2nd CE), P.Oxy. LXXVIII.5163 (1st–2nd CE),
Pap.Laur.Inv.Nr.III–418 (2nd CE), P.Lund I.5 (2nd CE), P.Oxy. XLIX 3452 (2nd CE), SB
III.I.6304 (2nd CE), P.Mich.Inv. 2458 (2nd–3rd CE), P.Oxy. XXXIII.XXXIII.2660a (3rd CE),
159. This is because it is very well attested in inscriptional evidence and there are numerous contemporary
grammatical treatises explicitly describing the phonology of Latin.
- 133 -
P.Strasb.Inv. G 1175 (3rd–4th CE), P.Strasb.Inv. g 1173 (3rd–4th CE), P.Fay. 135 verso (4th CE),
P.Lond. II 481 (4th CE), and P.Berol.Inv.Nr. 10582 (5th–6th CE), P. Vindob. L 91 (6th CE).
Most, but not all (see SB III.I.6304 and P. Vindob. L 91), constitute bilingual Greco-Latin
glossaries. Uncertain readings in the papyri have also generally been excluded.
5.3.1.1. Vowels
The Latin vocalic system is essentially comprised of two systems, a system of short vowels
(a, e, i, o, u) and a system of long vowels (ā , ē , ı̄ , ō , ū ). As is common cross–linguistically, the
long vowels are located more on the periphery of the vowel trapezium and the short vowels
nearer to the center. That is, the long vowels have more tense pronunciations while the short
vowels have more lax pronunciations. With the exception of long /ā / and short /ă /, which
seem to have had essentially the same quality, the long vowels had a significantly more close
quality than that of the corresponding short vowels. Thus, the resulting vocalic system was as
follows: /ı̄ / = [iː], /ı̆ / = [ɪ], /ē / = [eː], /ĕ / = [ε], /ā / = [aː], /ă / = [a], /ō / = [oː], /ŏ / = [ᴐ], /ū / = [uː],
/ŭ / = [ʊ]. It is important to note that the qualities of /ı̆ / and /ŭ / were closer to those of /ē / and
/ō /, respectively, than to those of their long counterparts, /ı̄ / and /ū /160 (ALLEN 1978, 47–50).
At some point in the history of Latin, the quantitative and qualitative oppositions gave way to
only qualitative oppositions (i.e., the neutralization of phonemic length). Latinists date this
change to around the second century CE (MARTÍNEZ 1989, 106).

5.3.1.1.1. a-Vowels
The short vowel /ă / [a] is represented almost exclusively (161/162x or 99.38%) with α:
Greek Transcription Latin Word Manuscript Date
αντε ă ntĕ P.Oxy. XLIX.3452 2nd CE
αρκουϲ ă rcŭ s P.Oxy. LXXVIII.5162 1st–2nd CE
160. The similar quality of /i/ and /ē / is demonstrated by inscriptions in which e is substitued for short i
(trebibos for tribibus) and i for long ē (minsis for mensis). The similar quality of /u/ and /ō / is demonstrated by
inscriptions in which o is substituted for short u (sob for sub) and short u is substituted for long ō (punere for
pō nere). Nevertheless, there were certainly differences between ı̄ /i and ē /e. There was greater palatal contact of
the tongue in the case of ı̄ /i. Likewise, ū /u differed from ō /o in that the lip–rounding was especially close in the
case of ū /u (ALLEN 1978, 47–50).
- 134 -
δεβιττα dē bı̆ tă P. Vindob. L 91 6th CE
The long vowel /ā / [aː] is always (90/90x) represented with α:
ακηταρια ă cē tā rı̆ ă P.Oxy. XXXIII.2660 1st–2nd CE
ουοκαβουλωρουµ vŏ cā bŭ lō rŭ m P.Oxy. XLIX.3452 2nd CE
5.3.1.1.2. e-Vowels
The short vowel /ĕ / [ε] is represented most frequently by ε (206/220x or 93.64%), and rarely
by αι (4/220x or 1.82%), η (3/220x or 1.36%), ι (4/220x or 1.82%), α (2/220x or 0.91%), or
Ø (1/220x or 0.45%):
ρεδδε rĕ ddĕ P.Berol. 21246 1st CE
γενιους gĕ nı̆ ŭ s P.Mich.Inv. 2458 2nd–3rd CE
αικους ĕ cŭ s P. Lund 5 2nd CE
In the environment of r, short /ĕ / may be represented with η: ηρκουλανεους Hĕ rcŭ lā nĕ ŭ s (1st–
2nd CE, P.OXY. XXXIII.2660), κονστηρνατους cō nstĕ rnā tŭ s (5th–6th CE, P.Berol.Inv.Nr. 10582),
[πε]ρηγρινη pĕ rĕ grı̄ nı̄ (5th–6th CE, P.Berol.Inv.Nr. 10582).
The long vowel /ē / [eː] is usually represented by η (75/103x or 72.82%), less fre-
quently by ε (25/103x or 24.27%), and a few times with ι (3/103x or 2.91%):
κλασσης clă ssē s SB III.I. 6304 2nd CE
οµνης ŏ mnē s P.Berol.Inv.Nr. 10582 5th–6th CE
ρεγαλιτερ rē gā lı̆ tĕ r P.Berol.Inv.Nr. 10582 5th–6th CE
There is significant disparity in the representation of Latin /ē / from the fourth century CE:
Latin ē ≤ 1st–2nd CE 2nd–3rd/4th CE ≥ 4th CE Total
as η 26 (96.30%) 28 (80.00%) 21 (51.22%) 75 (72.82%)
as ε 1 (3.70%) 7 (20.00%) 17 (41.46%) 25 (24.27%)
The distribution clearly demonstrates that chronology was a significant factor in the represen-
tation of Latin /ē /. Presumably, after the neutralization of quantitative distinctions in Latin, e
was more prone to be represented by Greek ε. At the same time, the fact that Latin e and
- 135 -
Greek ε were paleographically similar may account for a portion of the instances in which
Latin /ē / is rendered by Greek ε.

5.3.1.1.3. i-Vowels
The short vowel /ı̆ / [ɪ] is represented most frequently by ι (217/236x or 91.95%), rarely by ει
(9/236x or 3.81%), and a number of times by ε (6/236x or 2.54%):
ινετ ı̆ nnĕ t P.Berol. 21246 1st CE
κωγιτατ cō gı̆ tă t P.Strasb.Inv. g 1173 3rd–4th CE
ιουσσειστι jŭ ssı̆ stı̄ P.Berol.Inv.Nr. 10582 5th–6th CE
The long vowel /ı̄ / [iː] is usually represented by ι (60/91x or 65.93%), less frequently by ει
(26/91x or 28.57%), and a few times with ε (4/91x or 4.40%):
φορµικα fŏ rmı̄ că P.Oxy. LXXVIII.5163 1st–2nd CE
ρηγινα rē gı̄ nă P.Mich.Inv. 2458 2nd–3rd CE
ουεντεις vĕ ntı̄ s P.Oxy. XLVI.3315 1st–2nd CE
While ει is used to represent long /ı̄ / only slightly less than one third of the time, the
grapheme ει is far more likely to represent long /ı̄ / than it is short /ı̆ /:
/ı̆ / /ı̄ /
ει 7/29 (24.14%) 22/29 (75.87%)
There are also two instances in which ει represents two distinct vowels, rather than the vowel
quality [i]: πουλειουµ pū lē ı̆ ŭ m (1st–2nd CE, P.Oxy. XXXIII.2660) and ει ĕ ı̄ (SB III.I. 6304, 2nd
CE).
5.3.1.1.4. o-Vowels
The short vowel /ŏ / [ɔ] is usually represented with ο (65/69x or 94.20%) and rarely with ου
(2/69x or 2.90%) or ω (2/69x or 2.90%):
οσπιτ[εµ] hŏ spı̆ tĕ m P.Berol. 21246 1st CE
µορς mŏ rs P.Mich.Inv. 2458 2nd–3rd CE
νωεµ[β]ερ Nŏ vĕ mbĕ r P.Fay. 135 verso 4th CE
- 136 -
The long vowel /ō / [oː] is usually represented by ω (85/107x or 79.44%)161 and less frequent-
ly by ο (22/107x or 20.56%):
ρουβελλιωνης rŭ bĕ llı̆ ō nē s P.Oxy. XXXIII.2660 1st–2nd CE
νωµινα nō mı̆ nă P.Mich.Inv. 2458 2nd–3rd CE
βος vō s P.Lond. II 481 4th CE
The representation of Latin /ō / with ο is more common after the fourth century CE:
Latin ō ≤ 1st–2nd CE 2nd–3rd/4th CE ≥ 4th CE Total
as ω 16 (72.73%) 51 (96.23%) 18 (56.25%) 85 (79.44%)
as ο 6 (27.27%) 2 (3.77%) 14 (43.75%) 22 (20.56%)
Presumably, after the neutralization of quantitative distinctions in Latin, ō was more prone to
be represented by Greek ο. At the same time, the fact that Latin o and Greek ο were paleo-
graphically similar may account for a portion of the instances in which Latin /ō / is rendered
by Greek ο.
5.3.1.1.5. u-Vowels
The short vowel /ŭ / [ʊ] is most frequently represented by ου (169/188x or 89.89%) and oc-
casionally by ο (18/188x or 9.57%):
ινσουλσους ı̄ nsŭ lsŭ s P.Berol. 21246 1st CE
τουσσιτ tŭ ssı̆ t P.Strasb.Inv. G 1175 3rd–4th CE
αρµαριοµ ă rmā rı̆ ŭ m P.Lond. II 481 4th CE
The instances in which /ŭ / is represented by Greek ο seem to be in specific phonetic environ-
ments: in the environment of m (e.g., ινγενιοµ, ı̄ ngĕ nı̆ ŭ m; αρµαριοµ, ă rmā rı̆ ŭ m; µορµορω
[mŭ rmŭ rō ]), in the environment of r (ορσαµ, ŭ rsă m), and before v [w] (πλοουες, plŭ [v]ı̆ s).
Generally, instances in which Latin final nominative -us is represented as Greek -ος have
been excluded from this analysis, due to the fact that it may be a morphological equivalency
natural for a Greek speaker rather than a reflection of pronunciation.
161. The word prı̄ mō , which is transcribed as πριµωι, has been counted in this tabulation.
- 137 -
The long vowel /ū / [uː] is always represented with Greek ου (41/41x):
λακτουκα lă ctū că P.Oxy. XXXIII.2660 1st–2nd CE
σαλουτατ să lū tă t P.Strasb.Inv. G 1175 3rd–4th CE
5.3.1.2. Diphthongs and Semi-Vowels
The two most common Latin diphthongs, ae and au, originally represented the sounds [ai]
and [aw]. While ae was originally written as ai, by the second century BCE it began to be rep-
resented by ae. This may reflect that the diphthong was narrowed and the vowel quality of
the second element had approached that of its beginning point in [a]. In rural dialects, ae
shifted to a mid-open front vowel /ē̞ / and au shifted to a mid-open back vowel /ō /. During the
imperial period, au shifted to a when the following syllable contained an /u/ vowel (e.g.,
Agustus for Augustus). In Late Latin the monophthongization of ae became universal, but au
maintained its diphthongal realization in some regions (ALLEN 1978, 60–62).
The semi-vowels j (or i) and v (or u) were originally pronounced as [j] and [w], re-
spectively. Whenever i was written intervocalically, it represented a geminated consonant [jː]
(e.g., mā iŏ r [majːɔr]). Such a realization obtained throughout the ancient period. Beginning in
the first century CE, v [w] begins to interchange with b in inscriptional evidence, which likely
reflects a bilabial fricative [β] realization of consonantal u.162 Velius Longus describes a frica-
tive pronunciation already in the second century CE. This sound change was essentially uni-
versal by the fifth century CE, though there is some evidence that the pronunciation of [w] re-
mained in some pockets. It should also be noted that the u in the "digraph" qu, which
represented a single labio-velar phoneme [kw], did not become [β] as consonantal u did else-
where. While not as clear, a parallel situation probably obtained with respect to gu [gw]
(ALLEN 1978, 16–20, 25, 37–42).
162. Note how a similar phenomenon occurs in Jewish Palestinian Aramaic (see KHAN 1997, 105), roughly
contemporary Hebrew, and roughly contemporary Greek (see 4.5.3.1.13–15; 6.3.1.1; 6.3.7.1).
- 138 -
5.3.1.2.1. Diphthong ae
The diphthong ae [ai]/[ae] is usually represented by αι (14/20x or 70.00%), but occasionally,
presumably reflecting the aforementioned shift, represented by η (6/20 or 30.00%) or ε
(1/20x or 10.00%):
τερραι tĕ rrae P.Berol. 21246 1st CE
πραιτωριαι praetō rı̆ ae SB III.I.6304 2nd CE
[κ]ηνα cē nă (< caenă ) P.Berol. 21246 1st CE
κηλως cē lō s (< caelō s) P.Lond. II 481 4th CE
<π>ρε<φ>εκτους praefĕ ctŭ s P.Strasb.Inv. g 1173 3rd–4th CE
5.3.1.2.2. Diphthong au
The diphthong au [aw] is almost exclusively represented by αυ (11/12x or 91.67%) and once,
presumably reflecting the au > a / _Cu shift mentioned above, by α (1/12 or 8.33%):
ταυρουϲ taurŭ s P.Oxy. LXXVIII.5163 1st–2nd CE
αυτεµ autĕ m P.Berol.Inv.Nr. 10582 5th–6th CE
αγουστος Augŭ stŭ s P.Fay. 135 verso 4th CE
5.3.1.2.3. Semi-Vowel j
Consonantal i (j) [j] is always represented by ι (12/12x):
µαιορ mā jŏ r P.Berol. 21246 1st CE
α[δ]ιο[υ]τω ă djū tō P.Strasb.Inv. G 1175 3rd–4th CE
ιαµ jă m P.Lond. II 481 4th CE
5.3.1.2.4. Semi-Vowel v
Consonantal u (v) [w] is represented about half the time by ου (28/57x or 49.12%), slightly
less than half the time by β (25/57x or 43.86%), and a few times by Ø (4/57x or 7.02%):
ουεντους vĕ ntŭ s P.Oxy. XLVI.3315 1st–2nd CE
φαουωνιους fă vō nı̆ ŭ s P.Oxy. XLVI.3315 1st–2nd CE
ουιδες vı̆ dē s P.Strasb.Inv. G 1175 3rd–4th CE
βιλα vı̄ llă P.Lond. II 481 4th CE
ναβες nā vı̆ s P.Lond. II 481 4th CE
- 139 -
βιδεω vı̆ dĕ ō P.Berol.Inv.Nr. 10582 5th–6th CE
νωεµ[β]ερ Nŏ vĕ mbĕ r P.Fay. 135 verso 4th CE
It may be assumed that Greek ου represents Latin v = [w] and Greek β represents Latin v =
[β]. Although the statistical quantities of the representation of Latin v are almost evenly split
between ου and β, the distribution may be entirely explained on the basis of diachronic distri-
bution and distribution according to genre:
Latin v ≤ 1st–2nd CE 2nd–3rd/4th CE ≥ 4th CE Total
as ου 23 (95.83%) 14 (82.35%) 0 (0.00%) 37 (56.06%)
as β 0 (0.00%) 3 (17.65%) 22 (88.00%) 25 (37.88%)
as Ø 1 (4.17%) 0 (0.00%) 3 (12.00%) 4 (6.06%)
Latin v is almost exclusively represented by ου up until the fourth century CE, after which
time Greek β, and not ου, is used to represent Latin v. At first glance, this may seem problem-
atic. This is because the Latin sound change v [w] > [β] had already begun in the first century
CE and was widespread enough to be mentioned in a grammatical treatise by the second cen-
tury CE. Moreover, Latin loanwords in Egyptian Koine Greek reflect this change already in
the first and second centuries CE: πρεβέτοις prı̄ vā tŭ s (1st CE), κερβικάριον cĕ rvı̄ cā le (early 2nd
CE), and βιάτικον vı̆ ā tı̆ cŭ m (2nd CE) (GIGNAC 1976, 68–69).
The solution to this problem is to be found in the nature of the text that contains the
only occurrences of Greek β for Latin v prior to the fourth century CE. SB III.I. 6304, a sec-
ond-century CE receipt for the sale of a slave, was originally penned in Italy and subsequently
taken to Egypt. It contains all three examples of Greek β for Latin v prior to the fourth centu-
ry CE: βιγεντι vı̆ gĕ ntı̄ , βετρανε vĕ t(ĕ )rā nĕ , and βενδιδιτ vē ndı̆ dı̆ t. This text differs from the
glossarial texts, of which most of the corpus is comprised, in two ways. First, we know that it
was originally penned in Italy. Accordingly, we could suggest that it reflects a different pro-
nunciation. Even if this is the case, it is inadequate as an explanation since there is evidence
for the v [w] > [β] change in Egypt at an earlier period as well. Second, it differs from the
- 140 -
glossarial texts in genre. Its original function was to record and ratify an everyday sale. There
is no reason why it should reflect anything other than everyday pronunciation. Glossarial
texts, on the other hand, were used to teach someone how to speak Latin. Accordingly, it
would not be surprising if these bilingual glossaries reflected a more standard archaic pronun-
ciation, in which v was still pronounced as [w]. In fact, this is a common phenomenon in for-
mal language instruction. A modern English teacher would read the phrase going to as
[ˈgoʊɪŋ tu] when teaching their students, even though they themselves would pronounce it as
[ˈgɑnə] (i.e., "gonna") among friends. The preservation of the pronunciation [w] in the glos-
sarial texts well after the sound had shifted in colloquial pronunciation strengthens the claim
mentioned in chapter 3 that such texts were actually created by the teachers themselves.
If we assume that this was the case, the diachronic distribution is entirely explained.
Even though Latin v was pronounced as [β] already from the first or second century CE in
Egypt, glossarial and grammatical texts used for language instruction maintained the old pro-
nunciation [w] up until the fourth century CE. During this period, language-learning texts rep-
resented Latin v with ου, reflecting an archaic or standard pronunciation, and non-didactic
texts represented Latin v with β (e.g., SB III.I. 6304), reflecting everyday colloquial pronun-
ciation. After the fourth century CE, it seems that the pronunciation of Latin v as [β] had be-
come so universal that even in grammatical texts Latin v was represented with β.
Finally, it should be noted that the labio-velar phoneme qu [kw] was represented in a
variety of ways. It was most frequently represented by κου (11/20x or 55.00%), but also by
κο (4/20x or 20.00%), κοι (3/20x or 15.00%), and κυ (2/20x or 10.00%) each attempting to
approximate the atypical Latin phoneme [kw]. Its realization is fairly constant in all periods:163
ακουαριους ă quā rı̆ ŭ s P.Oxy. XLVI.3315 1st–2nd CE
εκουειτης ĕ quı̆ tē s P.Strasb.Inv. g 1173 3rd–4th CE
163. The variant κωτ[ι]θιδιανουµ in P.Vindob. L 91 (6th CE) probably reflects the shift of [kwo] > [ko].
- 141 -
κοις quı̆ s P.Berol.Inv.Nr. 10582 5th–6th CE
κοιιδ quı̆ d P.Berol.Inv.Nr. 10582 5th–6th CE
κινκυε cı̄ nquĕ SB III.I.6304 2nd CE
There is only one instance of gu [gw] in the corpus, represented by γου: ουνγουεντουµ
ŭ nguĕ ntŭ m (P.Berol.Inv.Nr. 10582, 5th–6th CE).

5.3.1.3. Consonants
The Classical Latin consonantal system is made up of eighteen distinct phonemes,164 includ-
ing three voiceless stops (/p/, /t/, /k/), three voiced stops (/b/, /d/, /g/), two labio-velars (/kw/,
/gw/), two liquids (/l/, /r/), three nasals (/m/, /n/, /ŋ/), three fricatives (/f/, /s/, /h/), and two
semi-vowels (/w/, /y/). Latin consonants could be doubled, in which case the consonant was
held for a greater duration. The grapheme x represented the combination of /k/ = c and /s/ = s
(ALLEN 1978, 11–46):
Voiceless stops: /p/ = p, /t/ = t, /k/ = c

Voiced stops: /b/ = b, /d/ = d, /g/ = g/c
Labio-velars: /kw/ = qu, /gw/ = gu
Liquids: /l/ = l, /r/ = r
Nasals: /m/ = m, /n/ = n, /ŋ/ = g(n)
Fricatives: /f/ = f, /s/ = s, /h/ = h
Semi-Vowels: /j/ = i/j, /w/ = u/v
Combination: /ks/ = x
5.3.1.3.1. Voiceless Stops
The voiceless stops p, t, and c are represented by their unaspirated Greek counterparts π, τ,
and κ: e.g., πισκης pı̆ scē s (P.Oxy. 3315, 1st/2nd CE), τινγω tı̆ ngō (P.Strasb.Inv. g 1175, 3rd/4th
CE), and κανιϲ că nı̆ s (P.Oxy. LXXVIII.5162, 1st/2nd CE). In the case of τ, it is twice represented
with δ: e.g., ακκεντιδε ă ccĕ ndı̆ tĕ . It is once represented with θ: κωτ[ι]θιδιανουµ165
164. The aspirated stops (/ph/, /th/, /kh/) and the voiced dental fricative (/z/), which enter the language through
Greek loanwords, have been omitted.
165. The sequence τ[ι]θι is probably a form of haplography.
- 142 -
quŏ tı̄ dı̆ ā nŭ m (P.Vindob. L 91, 6th CE). It is once omitted: ε δεµεττε ĕ t dı̄ mı̆ ttĕ (P.Vindob. L 91,
6th CE). In the case of c, it is twice represented by γ: e.g., γονγερ cŏ ngĕ r (P.Oxy. XXXIII.2660,
1st/2nd CE):
Latin p Latin t Latin c
as π 93 as τ 228 as κ 132
as δ 2 as γ 2
as θ 1
as Ø 1
Their representation with the Greek unaspirated series indicates that the Latin voiceless stops
were also unaspirated in their pronunciation.

5.3.1.3.2. Voiced Stops
The voiced stops b, d, and g are represented by their voiced Greek counterparts β, δ, and γ:
e.g., βλιτουµ blı̆ tŭ m (P.Oxy. XXXIII.2660, 1st/2nd CE), δη dē (P.Oxy. 3315, 1st/2nd CE), and
γενιους gĕ nı̆ ŭ s (P.Mich.Inv. 2458, 2nd/3rd CE). In the case of b, it is twice represented by π: e.g.,
δεκεµπερ Dĕ cĕ mbĕ r (P.Fay. 135 verso, 4th CE), in which the π was probably pronounced as [b]
after the nasal m. When preceding s, the combination is represented by ψ: οψιγναν[ς] ŏ bsı̄ g-
nā ns (P.Berol.21246, 1st CE). In the case of d, it is twice represented by τ when following
Greek ν, as would be natural in contemporary Greek orthography: e.g., ουντε ŭ ndĕ (P.Lond.
II. 481, 4th CE). It is once also represented by the combination νδ: µενδιουµ mĕ dı̆ ŭ m
(P.Berol.Inv.Nr.10582):
Latin b Latin d Latin g
as β 53 as δ 228 as γ 46
as π 2 as (ν)τ 2
as ψ / _s 1 as νδ 1
5.3.1.3.3. Labio-Velars
See 5.3.1.2.4.
- 143 -
5.3.1.3.4. Liquids
The liquids l and r are represented by their Greek counterparts λ and ρ, respectively: e.g.,
λεπους lĕ pŭ s (P.Lund 5, 2nd CE) and ρηγινα rē gı̄ nă (P.Mich.Inv. 2458, 2nd/3rd CE). In the case of
r, it is once represented by ν: κοντιναι cŏ rtı̄ nae (P.Oxy. LXXVIII.5163, 1st/2nd CE):
Latin l Latin r
as λ 102 as ρ 204
as ν 1
The nasals m and n are represented by their Greek counterparts µ and ν, respective-
ly:166 e.g., µηνϲα mē nsă (P.Oxy. LXXVIII.5163, 1st/2nd CE). In the case of m, it is transcribed
by ν three times: e.g., κολουνβος cŏ lŭ mbŭ s (P.Lond. II 481, 4th CE), possibly due to the fact
that regressive assimilation can be assumed for the Greek orthography -νβ- = /mb/ (< */nb/).
In the case of n, it is once transcribed by γν after κ: φακ γνοβες fă c nō bı̄ s (P.Lond. II 481, 4th
CE), possibly reflecting the voicing of /k/ as the speaker transitions into the following word. It
is also once transcribed by ϊ: νουϊτιας nū ntı̆ ā s (P.Berol.Inv.Nr.10582, 5th/6th CE), possibly re-
flecting the the weakening of the nasal (KRAMER 1983):
Latin m Latin n
as µ 148 as ν 198
as ν 3 as γν / k_ 1
as ϊ 1
5.3.1.3.5. Fricatives
The fricative f is transcribed as φ: e.g., φορµικα fŏ rmı̄ că (P.Oxy. LXXVIII.5163, 1st/2nd CE).
The fricative s is transcribed as σ: e.g., σπονδαι spŏ ndae (P.Oxy. LXXVIII.5163, 1st/2nd CE).
Latin h is never represented in the transcription: e.g., οµο hŏ mō (P.Berol.21246, 1st CE). In the
case of s, it is once transcribed as νσ: αρ⟦ε⟧'ι'ηνϲ ă rı̆ ē s (P.Oxy. LXXVIII.5162, 1st/2nd CE):
166. The first element of the Latin sequence gn is not treated separately in this analysis. In all 7 instances of the
sequence gn in Latin, Latin gn is transcribed as γν in Greek. The gn sequence in Latin either represented [ŋn] as
in English hangnail or [gn] as a regular "spelling pronuncation" (see ALLEN 1978, 22–25).
- 144 -
Latin f Latin s Latin h
as φ 31 as σ 290 as Ø 12
as νσ 1
5.3.1.3.6. Semi-Vowels
See 5.3.1.2.3–4.
5.3.1.3.7. Combination x
The Latin consonant x [ks] is represented by ξ: e.g., εξ ĕ x (SB III.I.6304, 2nd CE). It is once
represented by ξσ: σεξστουµ sĕ xtŭ m (SB III.I.6304, 2nd CE). The sequence nx, which occurs
once, is represented by ξ: φαλαξ fă lă nx (P.Strasb.Inv. g 1173, 3rd/4th CE):
Latin x Latin nx
as ξ 13 as ξ 1
as ξσ 1
5.3.1.3.8. Consonant Gemination
Consonantal gemination [Cː] is usually represented in the transcription (41/49x or 83.67%):
e.g., ακκηπισσε ă ccē pı̆ ssĕ (SB III.I.6304, 2nd CE), σουπελλεξ sŭ pĕ llĕ x (P.Oxy. LXXVIII.5163,
1st/2nd CE), and σαγιτταριους să gı̆ ttā rı̆ ŭ s (P.Oxy. XLVI.3315, 1st/2nd CE). However, it is not al-
ways indicated (8/49x or 16.33%): e.g., στηλας stē llā s (P.Lond. II 481, 4th CE) and πασαρες
pă ssĕ rē s (P.Lond. II 481, 4th CE):
cc dd ll mm nn rr ss tt Total
CC 7 5 8 1 0 4 9 7 41
C 0 0 3 0 1 1 1 2 8
There is also one instance in which single t is falsely geminated in the transcription as ττ:
δεβιττα dē bı̆ ta (P.Vindob. L 91, 6th CE).

5.3.1.4. Summary
The most common representation of each Latin phoneme is summarized in the charts below.
If a second grapheme is used more than 20.00% of the time, it is included in parentheses
(chart 12):
- 145 -
Vowels
Latin: ă [a] ā [aː] ĕ [ε] ē [eː] ı̆ [ɪ] ı̄ [iː] ŏ [ɔ] ō [oː] ŭ [ʊ] ū [uː]
Greek: α α ε η (ε) ι ι (ει) ο ω (ο) ου ου
Diphthongs Semi-Vowels
Latin: au [aw] ae [ai]/[ae] ei [ei]/[εi] j [j] v [w] v [β]
Greek: αυ αι ει ι ου β
Voiceless Stops Voiced Stops Labio-Velars
Latin: p [p] t [t] k [k] b [b] d [d] g [g] qu [kw] gu [gw]
Greek: π τ κ β δ γ κου (κο) γου
Liquids Nasals Fricatives Combination
Latin: l [l] r [r] m [m] n [n] f [f] s [s] h [h] x [ks] b + s [bs]
Greek: λ ρ µ ν φ σ Ø ξ ψ
Chart 12: Summary of Latin in Greek Transcription: Correspondences
5.3.2. Akkadian
The corpus for Greek transcription of Akkadian is comprised of the Graeco-Babyloniaca
tablets (GELLER 1997): BM 34797 (undated), BM 35727 (1st BCE), BM 34799 (1st BCE), BM
35726 (undated), BM 48863 (1st CE), BM 34781 (undated), BM 77229 (1st CE), HSM 1137 (1st
CE), BM 34816 (1st CE), BM 33769 (1st BCE), VAT 412 (1st CE), BM 38461 (1st/2nd CE), St. Pe-
tersburg tablet (undated) and BM 34798 (1st CE). With GELLER, Ash. Mus. 1937.993 (2nd BCE),
also known as the Ashmolean Incantation, is omitted from the analysis (1997, 83–85). Uncer-
tain readings in the tablets have also generally been excluded.

5.3.2.1. Vowels
Akkadian has four short vowels (/a/, /e/, /i/, /u/) and four corresponding long vowels (/ā /, /ē /,
/ı̄ /, /ū /). The four long vowels, when represented with a circumflex (i.e., /â /, /ê /, /ı̂ /, /û /), are
- 146 -
the result of the contraction of consecutive vocalic phonemes. The difference between the
short and long vowels is essentially duration (HUEHNERGARD 2011, 1).

5.3.2.1.1. a-Vowels
The short vowel /ă / is always represented by Greek α (91/91x):
Greek Transcription Akkadian Word Manuscript Date
αγαµ agam(m) (< *agammu) BM 34781 undated
ναφφαϲ nappaš (< *nappaš u) HSM 1137 1st CE
ραφασθ rapaš t (< *rapaš tu) St. Petersburg Tablet undated
The long vowel /ā / (distinct from â ) is likewise always represented by Greek α (42/42x):
βαβιλ Bā bil (< *Bā bilu) BM 34798 1st CE
µαρ mā r (< *mā ri) HSM 1137 1st CE
ταβ tạ ̄ b (< *tạ ̄ bu) BM 34816 1st CE
Both instances of /â / (distinct from /ā /), which result from the contraction of adjacent vowels,
are represented by α: µαρατ maḫ râ t (BM 34798, 1st CE) and [ν]ακ nâ q (BM 77229, 1st CE).
5.3.2.1.2. e-Vowels
The short vowel /ĕ / is most frequently represented by ε (4/9x or 44.44%) or η (4/9x or
44.44%), and once by ει (1/9x or 11.11%):
ελισ eliš BM 35727 1st BCE
λεµν lemn (< *lemnu) BM 34816 1st CE
µηλω meluḫ ḫ û BM 34799 1st BCE
λα θηροβ lā terrub HSM 1137 1st CE
ιχιλειθ iklet (< *ikleti) HSM 1137 1st CE
According to WESTENHOLZ, the instances in which short /e/ is rendered by Greek η point to
Aramaic influence. The forms µηλω meluḫ (ḫ ) and ηµυκ emū q are the result of pretonic
lengthening in an open unstressed syllable, since a short vowel was not tolerated in such a po-
sition according to the phonotactics of Aramaic. He further argues that the form θηροβ tē rub
(< *terrub), in which lā + preterite is used instead of lā + durative, is the result of morpholog-
- 147 -
ical influence of Aramaic (2007, 288). It is curious, however, that in each of these instances
the unexpected η is adjacent to a liquid or a nasal, the same sort of environment which gives
rise to the η/ε spelling interchange in contemporary Greek orthography. Moreover, the word
meluḫ ḫ û is not even an originally Akkadian word. Accordingly, these forms may simply re-
flect typical conventions of Greek orthography and not necessarily Aramaic influence. The
lack of gemination of ρ is not a difficulty (see 5.3.2.3.8).
The long vowel /ē / (distinct from /ê /) is rendered both by η (4/8x or 50.00%) and ει
(4/8x or 50.00%):167
βηλ bē l VAT 412 1st CE
βη[λ]σον Bē lš un (< *Bē lš unu) St. Petersburg Tablet undated
ζειρ zē r (< *zē ri) BM 38461 1st/2nd CE
ειφισ ē piš St. Petersburg Tablet undated
In all but one occurrence the /ē / vowel is in the environment of a liquid or a nasal. The one
exception is ē piš , in which ει is used to represent long /ē /.
The long vowel /ê / (distinct from /ē /), which is the result of the contraction of adja-
cent vowels, is also represented by both η (3/4x or 75.00%) and ει (1/4x or 25.00%):168
σαυη (2x) š awê (< *š amê ) BM 34798 1st CE
[πετ]ει petê BM 38461 1st/2nd CE
5.3.2.1.3. i-Vowels
The short vowel /ı̆ / is usually represented by ι (40/53x or 75.47%) and less frequently by ε
(7/53x or 13.21%) or ει (6/53x or 11.32%):
ραβισ rā bis ̣ (< *rā bis ̣u) BM 34799 1st BCE
167. Note also the two renderings of the name Nabû Rē manni, in which it is once rendered as ι and once as η:
ναβο[υ]ριυαν (St. Petersburg tablet, undated) and <ναβου>ρη<αν> (St. Petersburg tablet, undated). These
renderings have been omitted from the statistics.
168. GELLER reads λιβει libbê in BM 38461 (1st/2nd CE). However, the reading on the tablet is by no means clear.
Moreover, the supposed form libbê does not exist. Accordingly, the suffix ει has been omitted from my analysis.
- 148 -
ιθθι itti HSM 1137 1st CE
εξερ iḫ ir (< *iḫ r < *iḫ ri) BM 34797 undated
ιθεικ ı̄ tiq BM 34799 1st BCE
Long /ı̄ / (distinct from /ı̂ /) is represented by ι (7/14x or 50.00%) and ει (7/14x or 50.00%):
ιφοσ ı̄ puš BM 34797 undated
λιλι[θ] lilı̄ t (< *lilı̄ tu) BM 34816 1st CE
νιβειθ nibı̄ t BM 34798 1st CE
ρεισαθ rı̄ š ā t (< *rı̄ š ā ti) BM 34798 1st CE
Additionally, a final etymological short /i/ vowel is represented with ι in the word apti αφθι
(HSM 1137, 1st CE). According to GELLER, the apparent retention of a short final case vowel /i/
in the form αφθι is "peculiar." The final ι in the comparable form ιθθι may represent a "fixed
vowel" after the -CC sequence (1983, 117). It is more likely, though, that the final /i/ is a
bound form marker. However, because final short vowels had elided, it is unclear whether
this was a long or short vowel.

5.3.2.1.4. u-Vowels
The short vowel /ŭ / is almost always rendered by Greek ο (28/29x or 96.55%), but is ren-
dered once by ου (1/29x or 3.45%):
φολοξθ puluḫ t (< *puluḫ ta) BM 33769 1st BCE
ιφοσ ı̄ puš BM 34797 undated
οζον uzun (< *uzn) St. Petersburg tablet undated
µορσ murs ̣ (< *murs ̣u) BM 34816 1st CE
σ[ο]υβα[θ] s ̣ubā t (< *s ̣ubā tu) BM 48863 1st CE
Long /ū / (distinct from û ) is represented by ου (5/10x or 50.00%) and ω (5/10x or 50.00%):
τουβ tụ ̄ b BM 38461 1st/2nd CE
[δο]φσαρουθ tupš arrū t (< *tupš arrū ti) St. Petersburg tablet undated
νωρ nū r St. Petersburg tablet undated
- 149 -
ω ū w (< *ū wi < *ū mi) HSM 1137 1st CE
All of the instances of ω for /ū / are either the result of lowering in the environment of /r/ or
the result of assimilation to the semi-vowel /w/. Accordingly, ου should be assumed as the
regular Greek transcription of Akkadian /ū /.
The long vowel /û / (distinct from /ū /), which is the result of the contraction of adja-
cent vowels, is also represented by both ου (3/8x or 37.50%) and ω (5/8x or 57.14%):
ναβο[υ]ριυαν Nabû -Rı̂ manni St. Petersburg tablet undated
[µα]λου malû BM 33769 1st BCE
ναδου nadû BM 48863 1st CE
µαλαλω malallû BM 34797 undated
βιλλοτω[...] pilludû š u BM 34798 1st CE
ασανω asnû BM 34799 1st BCE
µαχανω makkanû BM 34799 1st BCE
µηλω169 ? (< *meluḫ ḫ û ) BM 34799 1st BCE
It is conventional for the long u-vowel in all of these forms to be normalized as û . However,
it is likely that there was actually an /ô / phoneme in Akkadian. Evidence for such a phoneme
is suggested by the correlation between û that results from the contraction of [ă̄ ] + [ŭ̄ ] and the
U and U4 signs, on one hand, and the correlation between û that results from the contraction
of other vowels (e..g, [i] + [u], [u] + [u]) and the Ú sign, on the other. The Greek material
generally supports this distinction. For example, the vowel û in pilludû š u (< *pilluda-ū -š u) is
represented with an ω in Greek, whereas the vowel û in malû (< *mali+u) is represented with
an ου. The transcription ναδου nadû (< *nadā +u), assuming it is the infinitive form, consti-
tutes an exception to this rule (WESTENHOLZ 1991; HUEHNERGARD and WOODS 2004, 233).
Notwithstanding the one exception, these examples, then, do not actually reflect ω for [uː] but
169. It is not clear whether the Greek µηλω is representing meluḫ ḫ û , with retention of the uvular fricative, or
melu(ḫ ḫ )+û > melû , with the elision of the fricative.
- 150 -
ω for /û / [oː] (or /ô / [oː]). Accordingly, ου as the regular representation of /ū / is still
supported.
5.3.2.1.5. Epenthetic Vowels
There are a number of instances (5x) in which apparent epenthetic vowels, not indicated in
the cuneiform, appear in the transcription in non-final consonant clusters:
ασανω asnû BM 34799 1st BCE
σαφαλισ (2x) š apliš BM 35727 1st BCE
ιχιλειθ iklet (< *iklet) HSM 1137 1st CE
οζονει uznı̄ (< *uznı̄ ya) BM 38461 1st/2nd CE
These forms are best explained according to the Syllable Contact Law (SCL), according to
which a fall in sonority is preferred in the transition from the end of one syllable to the begin-
ning of the next. In each instance above (s > n, p > l, k > l, z > n), there is a rise in sonority.
The quality of the epenthetic vowel inserted is identical to the vowel that precedes it.
In the case of original final consonant clusters, epenthesis is common (5x):170
εξξερ iḫ ir (< *iḫ r < *iḫ ri) BM 34797 undated
φαλαγ palag (< *palg < *palgu) BM 34797 undated
χαβαρ qabar (< *qabr < *qabri) HSM 1137 1st CE
οζον uzun (< *uzn < *uznu) St. Petersburg tablet undated
Like epenthetic vowels in word-medial clusters, the epenthetic vowel is identical in quality to
that of the preceding vowel. However, it is just as common (6x), for a final consonant cluster
to go unresolved:
µιτερθ ̣ (< *mitirtu)

mitirt ̣ BM 34797 undated
ιβωρθ (2x) i(b)bū rt (< *ina bū rti) HSM 1137 1st CE
170. Because of the regularity of this rule, each short epenthetic vowel has been tabulated with its
corresponding vowel elsewhere.
- 151 -
µορσ murs ̣ (< *murs ̣u) BM 34816 1st CE
θηροφσ terrubš (< *terrubš u) BM 34816 1st CE
ραφασθ (2x) rapaš t (< *rapaš tu) BM 33769/St. Petersburg tablet 1st BCE/undated
There are a couple of non-phonetic factors to be considered with respect to these two distinct
treatments of final -CC# clusters. First, the first group is comprised entirely of monosyllabic
forms and the second group is comprised almost entirely of bisyllabic forms (with the excep-
tion of µορσ). Second, all of the forms in the first group actually had an old bisyllabic
allomorph with epenthesis as its bound form. Beyond the non-phonetic factors, it is worth
noting that the presence or lack of an epenthetic vowel to resolve a consonant cluster may be
related to relative sonority. Generally speaking, with the exception of φαλαγ, clusters remain
when there is falling sonority and are resolved by an epenthetic when there is rising sonority.
This may be regarded as an instantiation of the Sonority Sequencing Principle (SSP), accord-
ing to which final clusters should exhibit a fall in sonority.
The SSP may also explain the final /i/ vowel in the transcription apti/aptı̄ αφθι (HSM
1137, 1st CE). We might expect a transcription such as αφθ (or αφαθ), but the lack of falling
sonority and the plosive nature in both elements of the cluster might have made neither of
those realizations tenable. Thus, the final ι might have been preserved as a sort of paragogic
epenthetic vowel.171
5.3.2.2. Semi-Vowels
The Akkadian semi-vowel /j/ is only attested twice in the prohibitive particle ajj: αι ιθεικ ajj
ı̄ tiq (BM 34799, 1st BCE) and αι ειθ[ικ] ajj ı̄ tiq (St. Petersburg tablet, undated). The Akkadian
171. Compare, for example, the relationship between sonority and the resolution of final consonant clusters in
Tiberian Hebrew. While final consonant clusters ending in a sonorous consonant were resolved with a word-
internal epenthetic (e.g., *wayyibn > *wayyibεn > ‫[ ַויִּבֶן‬vajˈjiːvεn]), final consonant clusters ending in a
consonant at the bottom of the sonority scale were allowed to remain (e.g., *wayyibk > ‫[ ַויּ ֵ ְב ְךּ‬vajˈjeːvk]). While
most grammarians preserved the final consonant cluster in words like the latter, it was the opinion of some
grammarians that a final consonant cluster like ‫ ַויּ ֵ ְב ְךּ‬may actually be resolved by adding an epenthetic short
vowel to the end of the word, rather than between the consonants (i.e., ‫[ ַויּ ֵ ְב ְךּ‬vajˈjē vka]) (KHAN 2013b, 669).
- 152 -
semi-vowel [w], which in most cases results from the shift of intervocalic [m] > [w], is repre-
sented by υ (6/11x or 54.55%), Ø (5/11x or 45.45%), and perhaps ου:172
σαυη (3x) š awê (< *š amê ) BM 34798 1st CE
ϲαυαϲ Š awaš (< *Š amaš ) HSM 1137 1st CE
ναυαρ nawā r (< *namā ri) HSM 1137 1st CE
ηουκ ewū q (< *emū q) BM 34798 1st CE
ωει ū wı̄ (< *ū mı̄ ) St. Petersburg tablet undated
There is also a very curious transcription in BM 34816 (1st CE). The Greek transcription
[θ]αλαµ[ισ] presumably reflects a normalized Akkadian talammı̄ š (u) from the verb lawû m 'to
surround'. Although the middle radical of this word is w, due to the shift of intervocalic m to
w by this late period, it was common for intervocalic w to be written with m, even though it
was still pronounced as w (e.g., amı̄ lu 'man'). (HUEHNERGARD 2011, 260). Accordingly, even
though the word might be spelled ta-lam-mi-š ú , it would still have been pronounced as /ta-
lawwı̄ š (u)/, for which we would expect the following transcription: θαλαυισ. The fact that the
Greek transcribes the m [w] with a µ is difficult to explain. It may be the result of treating
each sign in isolation, rather than in connection with the wider context of the entire word
(i.e., transliteration and not transcription). Alternatively, because there are no other clear
examples of such a phenomenon, it is possible that the Akkadian word talammı̄ š (u) was actu-
ally pronounced by the scribe of this tablet with an [m] as it is transcribed.

5.3.2.3. Consonants
The Akkadian consonantal system is made up of twenty distinct phonemes, including three
voiceless stops (/p/, /t/, /k/), three voiced stops (/b/, /d/, /g/), three emphatic/glottalic conso-
nants (/tˀ/, /kˀ/, /sˀ/), a glottal stop (/ˀ/), a uvular fricative (/ḫ /), two liquids (/l/, /r/), two nasals
172. In BM 34816 (1st CE), GELLER reads the Greek transcription of lem-nu as λεµν (1997, 76–77), whereas
WESTENHOLZ suggests [λ]εουν for the same word (2007, 269–270). It is unclear from the picture of the tablet
what the correct reading is, though GELLER's reading is unlikely.
- 153 -
(/m/, /n/), three plain sibilants (/s/, /š /, /z/), and two semi-vowels (/w/, /y/) (HUEHNERGARD
2011, 1–3, 586–591).

The voiceless stops /p/, /t/,/ and /k/ are represented by the aspirated Greek stops φ, θ, and χ,
indicating that the Akkadian stops were aspirated: e.g., φολοξθ puluḫ t (BM 33769, 1st BCE),
θαλ tā l (BM 35726, undated), and χο[µβολ] kumbul (BM 34781, undated). There may be one
instance of voicing of /p/ in the transcription βιλλοτω[...] pilludû š u (BM 34798, 1st CE), but it
may simply be a variant form. In one instance /t/ is voiced: [φω]δ pû t (BM 35727, 1st BCE). In
one instance it is realized as τ: µαρατ maḫ ră t (BM 34798, 1st CE):
Akkadian /p/ Akkadian /t/ Akkadian /k/
as φ 19 as θ 30 as χ 7
as β 1 as δ 1
as τ 1
The voiced stops /b/, /d/, and /g/ are represented by β, δ, and γ, respectively: e.g., βαβιλ Bā bil
(BM 34798, 1st CE), δοσ[ω] duš š û (BM 34781, undated), and αγαµ (BM 34781, undated).
Progressive assimilation results in the devoicing of /b/ in the transcription θηροφσ terrubš
(BM 34816, 1st CE). Devoicing may also occur in the transcription βιλλοτω[...] pilludû š u (BM
34798, 1st CE). Finally, ν appears in the transcription [εσ]ανγι<λ> Esagil/É sangil (VAT 412, 1st
CE), reflecting the original phonology inherited from Sumerian:
Akkadian /b/ Akkadian /d/ Akkadian /g/
as β 38 as δ 7 as γ 3
as φ 1 as τ 1 as νγ 1
5.3.2.3.3. Emphatic/Glottalic Consonants
The emphatic (or glottalic) consonants /tˀ/, /kˀ/, and /sˀ/ are represented by τ, κ, and σ, respec-
tively: e.g., ταβ tạ ̄ b (BM 34816, 1st CE), κ[ε]ιµ qē m (BM 48863, 1st CE), and σ[ο]υβα[θ] s ̣ubā t
(BM 48863, 1st CE). Akkadian /tˀ/ is once represented by δ: δοφ tup(p)
̣ (St. Petersburg tablet,
- 154 -
undated). Akkadian /kˀ/ is once represented by χ: χαβαρ qabar (HSM 1137, 1st CE). Akkadian
/sˀ/ is once represented by ζ: ζα[ραρ] s ̣arā r (BM 34781, undated).
Akkadian /tˀ/ Akkadian /kˀ/ Akkadian /sˀ/
as τ 6 as κ 7 as σ 4
as δ 1 as χ 1 as ζ 1
The representation of the emphatic stops with τ and κ demonstrates that it was the unaspirat-
ed nature of the glottalic consonant that identified them with the Greek unaspirated series.
5.3.2.3.4. Gutturals
There is one attestation of the glottal stop /ˀ/ in the corpus: ιβα ibaˀ(ˀ) (ΒΜ 33769, 1st BCE).173
The uvular fricative /ḫ / [χ] is represented by Ø (4/8x or 50.00%), ξ (3/8x or 37.50%), and
once as ξξ (1/8x or 12.50%):
µαρατ maḫ râ t (< *maḫ râ ti) BM 34798 1st CE
σοοσ [2x] suḫ uš (š ) (< *suḫ uš š i) BM 34799 1st BCE
[ν]αφα[ρ] napḫ ar (< *napḫ aru) BM 48863 1st CE
σιξ[ιθ] s ̣ı̄ ḫ it (< *s ̣ı̄ ḫ t < *s ̣ı̄ ḫ tu) BM 34781 undated
εξξερ iḫ ir (< *iḫ r < *iḫ ri) BM 34797 undated
In Greek loanwords in Hebrew, ξ is represented with ‫כס‬, indicating some aspiration of the
first element: i.e., [khs]. Accordingly, it seems that the decision of the Akkadian scribes to use
ξ to represent /ḫ / [χ] may be explained by the combination of two factors: The first element of
ξ, namely [kh], would have been the nearest consonant to Akkadian /ḫ / [χ] in terms of place of
articulation. The second element of ξ, namely [s], would have combined the previous sound
with a fricative element, thus approximating the fricative nature of the Akkadian sound.
173. GELLER reads this word as ιβα˹υ˺[ω] (1997, 78) and WESTENHOLZ (2007, 273) reads this word as ιβα
followed by a blank space and then the remains of a few letters. It seems to me that there is a blank space
followed by an υ. In any case, the most likely reading for the word under discussion is ιβα. This is presumably
the subjunctive form of bâ ˀum after final short vowels have elided: *ibaˀˀu > ibaˀ(ˀ).
- 155 -
5.3.2.3.5. Liquids
The liquids /l/ and /r/ are represented by λ and ρ, respectively: e.g., λα lā (HSM 1137, 1st CE)
and ρατ rā t ̣ (BM 34797, undated). In a couple instances, λ is omitted or assimilated: α
ρεισα[θ] ā l rı̄ š ā t (BM 34798, 1st CE) and α σα φαρ[...] ā l š a pars ̣ū š (BM 34798, 1st CE). In one
instance, ρ is realized as σ: µασθ mā rt (VAT 412, 1st CE). This may reflect a shift of r > s̆ / _t
(VON SODEN 1995, 44):
Akkadian /l/ Akkadian /r/
as λ 49 as ρ 32
as Ø 2 as σ / _θ 1
The nasals /m/ and /n/ are represented by µ and ν, respectively: e.g., µαρ mā r (HSM
1137, 1st CE) and ναρ nā r (BM 48863, 1st CE). Akkadian /m/ is once represented with ι:
Θιλιον[ω] tilmunnû (BM 34799, 1st BCE). It is once omitted in representation: οϲειρ muš ir(r)
(HSM 1137, 1st CE).
Akkadian /m/ Akkadian /n/
as µ 12 as ν 26
as ι 1
as Ø 1
5.3.2.3.6. Sibilants
The sibilants /s/ and /š / are both represented by σ: e.g., σοοσ suḫ uš (š ) (BM 34799, 1st BCE)
and σειρ š ē r (BM 34816, 1st CE). The fricative /z/ is represented by ζ: e.g., ζειρ zē r (BM
38461 (1st/2nd CE). There is one instance in which /s/ is rendered as θ: µαρχαθ markas (BM
34798, 1st CE):
Akkadian /s/ Akkadian /š / Akkadian /z/
as σ 5 as σ 40 as ζ 5
as θ 1
See 5.3.2.2.
- 156 -
Consonantal gemination [Cː] is sometimes represented in the transcription (5/18x or
27.78%): e.g., ναφφαϲ nappaš (HSM 1137, 1st CE) and ορριχ urrik (St. Petersburg tablet, un-
dated). More frequently, however, a geminated consonant in Akkadian is represented by only
one Greek consonant (13/18x or 72.22%): e.g., αθαφει atappı̄ (BM 34797, undated) µαχανω
makkanû (BM 34799, 1st BCE), and λιβει libbê (BM 38461, 1st/2nd CE).174
pp tt kk bb ll rr nn ss š š Total
CC 1 1 0 0 1 1 0 1 0 5
C 2 1 1 2 2 3 1 0 1 13
There is one instance in which ḫ is falsely geminated in the transcription: εξξερ iḫ ir (BM
34797, undated). Etymological final gemination is never represented: e.g., ασαχ aš ak(k) (BM
34816, 1st CE), βερ bir(r) (HSM 1137, 1st CE), and αγαµ agam(m) (BM 34781, undated):
p(p)# t(t)# k(k)# b(b)# d(d)# ḫ (ḫ )# r(r)# m(m)# n(n)# š (š )# Total
C# 2 2 1 2 1 1 2 1 2 2 16
5.3.2.4. Summary
The most common representation of each Akkadian phoneme is summarized in the charts be-
low. If an additional grapheme is used more than 20.00% of the time, it is included in paren-
theses (chart 13):
Vowels
Akkadian: a ā /â e ē /ê i ı̄ /ı̂ u ū û [i]+[u] û /ô [ă̄ ]+[ŭ̄ ]
Greek: α α ε (η) η, ει ι175 ι, ει176 ο ου ου ω
Diphthongs and Semi-Vowels
174. Final gemination C(ː)#, which is always represented with a single consonant, has been excluded from the
analysis because it is not clear if final gemination had been simplified or not.
175. It is also represented by ε (7/53x or 13.21%) and ει (6/53x or 11.32%).
176. The vowel /ı̂ / only occurs twice, once represented by ι and once by η (two variants of the same word).
- 157 -
Akkadian: (a)j [(a)j] w [w]
Greek: (α)ι υ, Ø
Voiceless Stops Voiced Stops
Akkadian: p [ph] t [th] k [kh] b [b] d [d] g [g]
Greek: φ θ χ β δ γ (νγ)
Emphatic/Glottalic Consonants Gutturals
Akkadian: t ̣ [tˀ] q [kˀ] s ̣ [sˀ] ḫ [χ] ˀ [ˀ]
Greek: τ κ σ (ζ) ξ, Ø -
Liquids Nasals Sibilants
Akkadian: l [l] r [r] m [m] n [n] s [s] š [ʃ] z [z]
Greek: λ ρ µ ν σ σ ζ
Chart 13: Summary of Akkadian in Greek Transcription: Correspondences
5.3.3. Arabic
The relevant material for Greek transcription of ancient Arabic has been analyzed by AL-JAL-
LAD in three separate articles (AL-JALLAD, DANIEL, and GHUL 2013; AL-JALLAD 2015; AL-JAL-
LAD and AL-MANASER 2015). While most of the transcription material is comprised of proper
names rendered into Greek, a third or fourth century CE inscription from north-eastern Jordan
provides an example of a continuous transcription text. The corpora examined are the epi-
graphic and papyrological evidence from the Roman and Byzantine Near East, specifically in
southern Syria, central and southern Jordan, and Israel. Because the material has already been
analyzed, we will forego a comprehensive statistical analysis and merely summarize the find-
ings of AL-JALLAD.
- 158 -
5.3.3.1. Vowels
The Arabic vocalic system is made up of three short vowels (/a/, /i/, /u/) and three corre-
sponding long vowels (/ā /, /ı̄ /, /ū /). In addition to these vowels, the contraction of the diph-
thong /ay/ may result in a monophthongs of the e-vowel class.

5.3.3.1.1. a-Vowels
Arabic short /a/ is transcribed with Greek α: e.g., Αλαβδος /al-ˁabd/ (208 CE) and Αλαχβαρ
/al-ˀakbar/ (505–520 CE). Arabic long /ā / is likewise transcribed with Greek α: e.g.,
Μοσαλεµου /mosā lem/ (179–180 CE) and Μαλ /mā l/ (505–537 CE). When short /a/ is raised
pretonically it is transcribed with ε: e.g., Σεουαδος /sewā d/ (< */sawā d/) (undated). When
short /a/ is rounded before a labial, it is transcribed with ο: e.g., Ασλοµου /ˀaslom/ (< */
ˀaslam/) (434 CE) (AL-JALLAD 2015, 31–33).

5.3.3.1.2. i-Vowels
Arabic short /i/ is transcribed most commonly with Greek ε: e.g., Αλεσου /ḫ ā les ̣/ (179–80 CE)
and Κεσεβ /qes ̣eb/ (505–537 CE). Less commonly, /i/ is transcribed by η: e.g., Νασηρος
/nā s ̣ir/ (IGLS XXI 59). Transcribing etymological /i/ with η is common in the environment of
liquids and nasals.177 In very rare cases it is transcribed by ι in stressed closed syllables: e.g.,
Σιθρο /sitrō / (undated) and Ιννου /h ̣inn/ (undated). AL-JALLAD regards these occurrences as
too rare to be meaningful. Arabic long /ı̄ / is transcribed almost always with ι: e.g.,
Αβδαλµιθαβου /ˁabd al-mı̄ t̲ ab/ (434 CE) and Μοκιµος /moqı̄ m/ (undated). Less frequently,
long /ı̄ / may be transcribed with Greek ει: e.g., Μοκειµος /moqı̄ m/ (undated) and
Ουασειχαθος /waś ı̄ kat-/ (undated) (2015a, 32, 34). Short /i/ is rendered as ι in the continuous
text from north-eastern Jordan (3rd/4th CE): e.g., βι-Χανου[ν] /bi-kā nū n/ (AL-JALLAD 2015, 32,
34; AL-JALLAD and AL-MANASER 2015, 52–53).
177. I would like to thank Ahmad Al-Jallad for providing me with this example and observation.
- 159 -
5.3.3.1.3. u-Vowels
Arabic short /u/ is most commonly transcribed with Greek ο: e.g., Οσνη /h ̣osn/ (318 CE), and
and Μοσλεµος /moslem/ (undated). In stressed closed syllables, short /u/ is sometimes
transcribed with ου: e.g., Αλσουφλη /al-sufley/ (505–537 CE) and Ουββος /h ̣ubb/ (undated).
Unstressed short /u/ is transcribed as ου only twice: Αλουλαιφ /al-ḫ ulayf/ (undated) and
Νουµερος /Numeyr/ (undated). Arabic long /ū / is regularly transcribed with Greek ου: e.g.,
Αλουφαθη /ḫ alū fat-/ (5th CE) and Αβου /ˀabū / (565 CE). In the rare instance that long /ū / is
lowered by a following /r/ it is transcribed by ω: e.g., Μεφωρ /meh ̣fō r/, reflecting the lower-
ing of ū > ō before r (AL-JALLAD 2015, 32–34).

5.3.3.2. Diphthongs and Semi-Vowels
Arabic has two diphthongs, namely, /aw/ and /ay/. It is common in the dialects for the diph-
thong /ay/ to monophthongize to /e/. The semi-vowels /w/ and /y/ are fairly stable in Arabic.
5.3.3.2.1. Diphthong /aw/
The diphthong /aw/ is regularly represented by αυ: e.g., Αυσαλλας /ˀaws/ (157 CE), Αυµου
/ˁawm/ (213 CE), and Ζαυανου /zawˁā n/ (6th CE). In one instance, where /aw/ has been raised to
/ew/, the diphthong is represented as εου: Σεουδα /sewdā / (411 CE) (AL-JALLAD 2015, 31, 35).
A few examples are also found in the continuous text from north-eastern Jordan (3rd/4th CE):
Αυσος /ˀAws/, α-δαυρα /ˀad-dawra/, and ειραυ /yirˁaw/ (AL-JALLAD and AL-MANASER 2015,
52–53).
5.3.3.2.2. Diphthong /ay/
The diphthong /ay/ is represented in two ways. First, it is represented with an e-class vowel,
namely, ε or η: e.g., Οβεδου /ˁobeyd/ (undated) and Ονηνος /h ̣oneyn/ (undated). Second, it is
represented with the digraph αι, and less frequently ει: e.g., Χαιρου /ḫ ayr/ (164 CE), Βαιθ
/bayt/ (505–537 CE), and Ζονειννος /ź ̣onayn/ (undated). AL-JALLAD argues that even though
contemporary Greek pronunciation rendered αι as [ε], the distribution of the spellings indi-
cate that the Arabic dipthong persisted as [ai], represented by αι, with a raised allophone of
[ei], represented by ε, η, and ει (2015, 35–36).
- 160 -
5.3.3.2.3. Semi-Vowel /w/
Arabic /w/ is typically represented by ου: e.g., Ραουαου /rawā h ̣/ (233 CE) and Ουαελος /wā ˀel/
(293/4 CE). It may also be represented by Ø (i.e., a hiatus between two vowels): e.g., Ροεος
/ro(w)eyh ̣/ (undated) and Ζοεδαθος /zo(w)eydat/ (undated) (AL-JALLAD 2015, 29–30). In the
continuous text from north-eastern Jordan, the representation of /w/ is inconsistent: αθαοα
/ˀatawa/, ζαθαοε /ś atā w/, ωα /wa/, and αουα /wa/ (AL-JALLAD and AL-MANASER 2015, 52–53).
5.3.3.2.4. Semi-Vowel /y/
Arabic /y/ is typically represented by ι: e.g., Τοβαιαθη /ṯ obayyat/

̣ (undated) and Αλαγιαθ /al-
h ̣ag(i)yā t/ (undated). It may also be represented by Ø: e.g., Μοεαρος /moġ e(yy)ar/ (undated)
(AL-JALLAD 2015a, 29–30). In the continuous text from north-eastern Jordan, word-initial /yi/
is represented with the digraph ει (3rd/4th CE): ειραυ /yirˁaw/ (AL-JALLAD and AL-MANASER
2015, 52–53).
5.3.3.3. Consonants
The reconstructed proto-Arabic consonantal system is made up of twenty-eight distinct
phonemes, including three voiceless stops (/p/, /t/, /k/), three voiced stops (/b/, /d/, /g/), two
interdentals (/ḏ /, /ṯ /), two uvular fricatives (/ḫ /, /ġ /), two pharyngeal fricatives (/h ̣/, /ˁ/), a glot-
tal fricative (/h/), a glottal stop (/ˀ/), two liquids (/l/, /r/), two nasals (/m/, /n/), five emphatic
(or glottalic) consonants (/t/,̣ /ṯ /,̣ /s ̣/, /ś ̣/, /q/), three plain sibilants (/s1/, /s2/, /z/), and two semi-
vowels (/w/, /y/) (AL-JALLAD 2015).

Arabic etymological /p/, which may or may not have shifted to /f/ during the period of the
transcriptions, is represented with φ:178 e.g., Φοσεα /fos ̣eyyah ̣/ (505–537 CE) and Ασαφιρ
/ˁas ̣ā fı̄ r/ (505–537 CE). Arabic /t/ is regularly represented by Greek θ: e.g., θιεµου /tiyeim/
(330 CE) and Γανναθ- (505–537 CE). Arabic /k/ is regularly transcribed by χ: e.g., Χασετος
178. Loanwords such as fars (<*pars) and firdaws (<*paradeisos) suggest that these loanwords were borrowed
into Arabic when /f/ was pronounced as /p/. It is unclear if this realization was so during the period of the
inscriptions. Al-Jallad acknowledges that the transcription of the Nabatean name ‫ חליפו‬as Χαλιπος might point to
an attempt at transcribing /f/, just as τ sometimes attempts to represent /t̲ /. However, representations of /p/ (or
/f/) with π are far more rare than those of /t̲ / with τ.
- 161 -
/kā set/̣ (undated) and Αλαχβαρ /al-ˀakbar/ (505–520 CE). The fact that Arabic /k/ is
transcribed with χ and not κ, seems to indicate that χ was still representing /kh/ and not /χ/ in
eastern Greek at the time of the material (AL-JALLAD 2015, 11–14, 19–20–23, 28, 31–32, 48).
Arabic /b/ is consistently transcribed with Greek β: e.g., Αλαβδος /al-ˁabd/ (208 CE). Arabic
/d/ is regularly represented by δ: e.g., Ζειεδος /zeyeyd/ (315 CE). Arabic /g/ (likely not [d͡ ʒ])179
is regularly transcribed as γ: e.g., Αβγαρ /abgar/ (108/109 CE) (AL-JALLAD 2015, 29, 31, 50).
5.3.3.3.3. Interdentals
The interdental voiced fricative /ḏ / [ð] is represented by δ, as in Δουσαρεος /ḏ ū -ś arey/ (164
CE) and Αουιδου /ˁawı̄ d̲ / (569 CE). The voiceless interdental fricative /t̲ / [θ] is transcribed by
θ: e.g., Αυθου /ġ awθ/ (380 CE) and Αυιθου /ġ awı̄ θ/ (undated). However, Arabic /ṯ / may be
transcribed by τ as well: e.g., Γαυτος /ġ awθ/ (334–335 CE) and Αουιτος /ġ awı̄ θ/ (undated).
The occasional representation of Arabic /ṯ / [θ] with τ is evidence for the continued realization
of Greek θ as [th] at the time of the Graeco-Arabica material (2015, 18–19, 24, 38).180
Arabic /ḫ / ([x] or [χ]) may be transcribed as χ: e.g., Χαιρου /ḫ ayr/ (164 CE) and Χαµσα /ḫ am-
sah/ (undated). It may also be represented by Ø: e.g., Ηρανου /ḫ eyrā n/ (327 CE) and Αλδου
/ḫ ald[ē ]/ (undated). It is more common for /ḫ / not to be represented in the transcription, on
account of which AL-JALLAD argues that Greek χ was still /kh/ in the east even at the time of
the Graeco-Arabica material (AL-JALLAD 2015, 14–18).181
179. In the Islamic period there are Greek transcriptions such as Νεσζιδ */neǧ ı̄ d/ and Γιαφαρ */ǧ aˁfar/ to
approximate the Arabic [d͡ ʒ]. The absence of such representations in the corpus suggest that etymological /g/
was realized as [g] (AL-JALLAD 2015, 20–21).
180. Greek τ is used for emphatic /t/̣ is because they both share the lack of aspiration. On the other hand,
because aspiration by itself is not phonemic, but the opposition between stop and fricative is, the author would
have been more likely to prioritize that distinction. Therefore, if θ had already become [θ] in Greek of the time,
then it would have made more sense for the transcriptions to align τ with Arabic /t/ and θ with Arabic /ṯ /. The
fact that this does not occur seems to indicate that θ represented /th/ in the Greek of the time and region.
Moreover, the fact that Arabic interdental /ṯ / [θ] is occasionally transcribed as τ, rather than with what would
have been an obvious transcription if θ had represented [θ], further supports the theory that Greek θ [th] had not
yet shifted to [θ] (AL-JALLAD 2015, 11–14, 18–19).
181. He provides an example where neither Safaitic /h ̣/ or /ḫ / are represented in the Greek (Αλιζου for ḫ ls ̣ and
Αδδιδανου for ḥddn) in the same text, even though there is no evidence for a merger in Safaitic. He concludes
- 162 -
Arabic /ġ / ([γ] or [ɰ])182 may be transcribed by Greek γ: e.g., Μογεαιρος /moġ eyyir/
(386 CE) and Αλγεβ /al-ġ ebb/ (undated). It may also be transcribed by Ø: Μοεαρος
/moġ e(yy)ar/ (372 CE) and Αυθου /ġ awṯ / (380 CE) (AL-JALLAD 2015, 14–18).
The remaining gutturals (/h ̣/, /ˁ/, /ˀ/, /h/) are represented by hiatuses vetween vowels or
Ø: e.g., {Β}εακκος /be-h ̣aqq[oh]/ (undated), Χααµµος /ka-ˁamm-[oh]/ (undated), Ουαβαλας
/wahb-(ˀ)al(l)ā h/ (undated), and Ουαελος /wā ˀel/ (233 CE) (AL-JALLAD 2015, 30, 44).
5.3.3.3.5. Sonorants
The Arabic liquids /l/ and /r/ are transcribed by Greek λ and ρ: e.g., Ζαγλος /zagl/ (315 CE)
and Ραουαου /rawā h ̣/ (223 CE). The nasals /m/ and /n/ are represented by µ and ν: e.g.,
Θιεµου /tiyeim/ (330 CE) and Οσνη /h ̣osn/ (318 CE) (AL-JALLAD 2015, 21, 29, 32–33).
The emphatic stops /t/̣ and /q/ are transcribed by Greek unaspirated τ and κ: e.g., Χασετος
/kā set/̣ (undated) and Ζαιδοκιµα[ς] /zaydo-qı̄ ma/ (517 CE). Etymological /q/ is transcribed
once by χ in Χαυµος /qawm/ (undated). AL-JALLAD is unsure if the glottalic realization of the
emphatic stops had fronted to pharyngealization in Arabic by the time of the transcriptions.
However, it was the lack of aspiration common to both the Arabic "emphatics" and the Greek
unaspirated stops that led to the association between the two (2015, 11–13, 21–22, 27–28).183
Arabic /s ̣/ is transcribed by Greek σ: e.g., as in Φοσεα /fos ̣eyyah/ (505–537 CE), and
Ασαφιρ /ˁas ̣ā fı̄ r/ (505–537 CE). AL-JALLAD is undecided as to whether or not etymological /s ̣/
was realized as [tsˀ], [sˀ], or [sˁ] during the period. Arabic /ṯ /̣ is usually transcribed by τ: e.g.,
Ναταµος /naṯ am/

̣ (undated). However, in the Nessana papyri it is transcribed by ζ: e.g.,
that the author simply did not regard Greek χ as a suitable representation of the phoneme /ḫ /. He reminds us that
even though we typically think of /ḫ / as representing the uvular fricative [χ], numerous dialects have a velar
fricative [x]. Similarly, it is a possibility that /ḫ / was actually a front velar fricative (or even a palatal fricative) in
the dialects of the material. He suggests that either a front velar fricative [x̟ ] or a palatal fricative [ç] would be
regarded closer to the Greek spiritus asper than the velar stop χ [kh] (AL-JALLAD 2015, 14–18).
182. AL-JALLAD suggests that the representation of */ġ / with Ø may indicate that in Old Arabic */ġ / was realized
as a velar approximant [ɰ] rather than a velar or uvular fricative (2015, 17).
183. It is also worth noting, however, that there are some Arabic dialects in which both non-emphatic /t/ and
emphatic /t/̣ exhibit at least some aspiration (BELLEM 2007, 60–63, 203–204).
- 163 -
Ζαννος /ź ạ nn/ (undated). This likely reflects an emphatic lateral fricative [ɮˁ]. Arabic /ś /̣ ,
which reflects an emphatic lateral [ɫˀ], is transcribed with σ: e.g., Ρασαουαθος /raś ạ ̄ wat/ (un-
dated). In the Nessana papyri, where it is assumed to have merged with /ź /̣ , it is represented
with ζ: e.g., Μαζεκα /maź ẹ ̄ qah/ < */maś ı̄ ̣ qah/ (undated) (2015, 22–27).
5.3.3.3.7. Plain Sibilants
Arabic /s/ is regularly transcribed by Greek σ: e.g., Σεουδα /sewdā / (411 CE) and Αλσουλλαµ
/al-sullam/ (505–538 CE). Arabic /ś / (<*/ɬ/, Modern Arabic /š /), similarly, is regularly
transcribed by σ: e.g., Σαιαθη /ś ayˁat/ (316–396 CE) and Αλσαρκια /al-ś arqiyyah/ (6th CE). Be-
cause Safaitic s1 is used to transcribe Aramaic [ʃ], /ś / had probably not yet shifted to [ʃ] and
was still realized as the lateral [ɬ]. Arabic /z/ is regularly transcribed by Greek ζ: e.g., Ζειεδος
/zeyeyd/ (315 CE) and Αλλουζα /al-lowzah/ (505–537 CE) (AL-JALLAD 2015, 28–29).
See 5.3.3.2.
Consonantal gemination [Cː] is usually indicated: e.g., Ουαβαλλας /wahb-(ˀ)allā h/. However,
it may also be omitted.: e.g., Ουαβαλας /wahb-(ˀ)al(l)āh/. Gemination of the glides /y/ ι/Ø and
/w/ ου/Ø is never indicated: e.g., Αουαθω /ġ awwā ṯ ō /, Ουµαυατ /ˀum(m)-ġ awwaθ/ and
Τοβαιαθη /ṯ obayyat/.

̣ Final gemination is not represented: e.g., Αλγεβ /al-ġ ebb/ (AL-JALLAD
2015, 6, 16, 24, 29, 33, 54).

5.3.3.4. Summary
The most common representation of each Arabic phoneme is summarized in the charts below.
Secondary graphemes are listed in parentheses (chart 14):
Vowels
Arabic: a ā i ı̄ u ū e
Greek: α α ε (η) ι (ει) ο (ου) ου ε, η
Diphthongs Semi-Vowels
- 164 -
Arabic: ay [aj] aw [aw] y [j] w [w]
Greek: αι, ε, η αυ ι, Ø (#yi- = ει) ου, Ø (αοα, αοε, ωα)
Arabic: *p [ph] > f [f] t [th] k [kh] b [b] d [d] g [g]
Greek: φ θ χ β δ γ (νγ)
Interdentals
Arabic: ḏ [ð] ṯ [θ]
Greek: δ θ (τ)
Emphatic Stops Emphatic Sibilants/Fricatives
Arabic: t ̣ [kˀ] q [kˀ] s ̣ [sˀ] or [tsˀ] ś ̣ [ɫˀ] ṯ ̣ [ɮˁ]
Greek: τ κ σ σ ζ
Gutturals
Arabic: ḫ [χ] or [x] ġ [γ] ḥ [ħ] ˁ [ˁ] h [h] ˀ [ˀ]
Greek: χ, Ø γ, Ø Ø Ø Ø Ø
Liquids Nasals Plain Sibilants
Arabic: l [l] r [r] m [m] n [n] s [s] š [ʃ] z [z]
Greek: λ ρ µ ν σ σ ζ
Chart 14: Summary of Arabic in Greek Transcription: Correspondences
- 165 -
5.3.4. Aramaic
While a comprehensive treatment of Aramaic in Greek transcription lies beyond the scope of
this chapter,184 a general description will be carried out based on the limited corpus of the
Nah ̣al Dimona inscription and the Dura-Europos inscription (see 3.4.1.3.5).
5.3.4.1. Vowels
The Aramaic vocalic system varies over time, space, and dialect. However, in Jewish Pales-
tinian Aramaic, which is probably relevant at least for the Nah ̣al Dimona inscription,185 the
vocalic phonemic inventory consisted of /a/, /e/, /i/, /o/, /u/, and "shewa" (KHAN 1997, 107–
111; SOKOLOFF 2011, 613). In our limited corpus, /a/ is transcribed with Greek α: e.g., θαρθην
tarten, σιµαθα simaṯ a, and καιαµα qayyama. Aramaic /e/ is represented by η or ε: e.g., σαβη
s ̣aḇ e, θαρθην tarten, δαελαα daelaha, ιαεβ yaheḇ . Aramaic /i/ is transcribed with ι: e.g., νισαν
nisan, σιµαθα simaṯ a, γοβνιν goḇ nin, and Αβιδσαλµα ˁAbidš alma. Aramaic /o/ (< */u/) is
transcribed with ο: e.g., γοβνιν goḇ nin. Aramaic /u/ is transcribed as ου: e.g., ου hu. A "she-
wa" vowel seems to be realized with the quality of [a], since it is transcribed by Greek α: e.g.,
λαµαν laman, βα-νισαν banisan, δαελαα daelaha, and δααβ dahab.

5.3.4.2. Semi-Vowels
The Aramaic semi-vowel /j/ is transcribed as ι: e.g., καιαµα qayyama and ιαεβ yaheḇ .
5.3.4.3. Consonants
The Aramaic consonantal inventory is made up of twenty-two consonants, including three
voiced stops with fricative post-vocalic allophones (/b/ > [v], /g/ > [γ] or [ʁ], /d/ > [ð]), three
voiceless stops with fricative post-vocalic allophones (/p/ > [f], /k/ > [χ], /t/ > [θ]), three em-
phatic consonants (/t/,̣ /k ̣/, /s ̣/), a glottal stop (/ˀ/), a glottal fricative (/h/), two pharyngeal
fricatives (/h ̣/, /ˁ/), two liquids (/l/, /r/), two nasals (/m/, /n/), three sibilants (/s/, /š /, /z/), and
two semi-vowels (/w/, /j/) (KHAN 1997, 104–107; SOKOLOFF 2011, 612–13).
184. Most of the rest of the material for Aramaic transcription is comprised of proper names.
185. For a transcription of the Nah ̣al Dimona inscription, see PRICE and NAEH (2009, 269).
- 166 -
The voiceless stops (/p/, /t/, /k/) are not distinguished from their fricative allophones
([f], [θ], [χ]) in transcription. Aramaic /t/ [t] and /ṯ / [θ] are transcribed by θ: e.g., θαρθην
tarten and σιµαθα simaṯ a. The voiced stops (/b/, /d/, /g/) are not distinguished from their frica-
tive allophones ([v], [ð], [γ] or [ʁ]) in transcription. Aramaic /b/ [b] and /ḇ / [v] are transcribed
by β: e.g., βα-νισαν banisan and σαβη s ̣aḇ e. Aramaic /d/ [d] and /ḏ / [ð] are transcribed by δ:
e.g., δααβ dahab and Αβιδσαλµα ˁAbidš alma. Aramaic /g/ is transcribed by γ: e.g., γοβνιν
goḇ nin. The emphatic stop /q/ is represented by Greek unaspirated κ: e.g., καιαµα qayyama.
Emphatic /s ̣/ is represented by σ: e.g., σαβη s ̣aḇ e. The attested gutturals are represented by a
hiatus between vowels: /ˀ/ in δαελαα daelaha and /h/ in δααβ dahab. The sonorants (/l/, /m/,
/n/, /r/) are represented as expected: /l/ with λ in λαµαν laman, /m/ with µ in καιαµα qayyama,
/n/ with ν in γοβνιν goḇ nin, and /r/ with ρ in Βαρζακικη barzaqiqe. The voiceless sibilants /s/
and /š / are both represented with σ: e.g., βα-νισαν banisan and Αβιδσαλµα ˁAbidš alma. The
voiced sibilant /z/ is represented with ζ: e.g., Βαρζακικη barzaqiqe.

5.3.4.4. Summary
Although the corpus is limited and the attestation of the phonological inventory is incom-
plete, the correspondences in the Nah ̣al Dimona inscription and the Dura-Europos inscription
may be summarized as follows:
Vowels and Semivowels
Aramaic: a e i o u "ə" j
Greek: α ε, η ι ο ου α ι
Consonants
Aramaic: b ḇ [v] d ḏ [ð] g l m n q s ṣ š t ṯ [θ] z h ˁ
Greek: β β δ δ γ λ µ ν κ σ σ σ θ θ ζ Ø Ø
Chart 15: Summary of Aramaic in Greek Transcription: Correspondences
- 167 -
5.3.5. Phoenician-Punic
The Phoenician data is primarily taken from the instances of Greek transcription cited in
FRIEDRICH and RÖLLIG's Phönizisch-Punische Grammatik (1999).

5.3.5.1. Vowels
The Standard Phoenician vocalic systems was originally made up of three short vowels (/a/,
/i/, /u/) and five long vowels (/ā /, /ē /, /ı̄ /, /ō /, /ū /). As early as the seventh century BCE, original
stressed short /a/ shifted to /o/, which is known as the "Phoenician Shift." This change proba-
bly proceeded as follows: á > ā , ā > ɔ, ɔ > o. When the result of the Phoenician shift was still
/ɔ/, the /o/ vowel resulting from the Canaanite shift (/ā / > /o/) shifted to /u/. The result of this
/o/ > /u/ shift did not merge with original /u/, which instead was pushed forward to /ü/. At the
period of the transcritions, the vowels /a/, /e/, /i/, /o/, /u/ and /ü/ are present. It is not clear if
length was still phonemic during this period, but there does seem to be a phonetic length-
ening of stressed vowels (FOX 1996; FRIEDRICH and RÖLLIG 1999, 36; HACKETT 2008, 87–89).
5.3.5.1.1. a-Vowels
Phoenician short /a/, preserved in doubly-closed syllables, is transcribed by Greek α: e.g.,
λασουναλφ /laš un-ʔalp/, βαλ /baʕl/, and Καδος /kad(d)/. Occasionally, short /a/ in these posi-
tions is represented by either ε or η, likely reflecting a raised allophone of [æ]: e.g., Ηµουνος
/h ̣amō n/ and Μερβαλος /maharbaʕl/ (cf. Μααρβαλ). Unstressed /a/ in an open syllable is also
transcribed by α: e.g., Σαµηµ /š amē m/ and σαδε /š adē̆ /. In the environment of emphatics, /a/
can have a rounded articulation (perhaps [ɔ]?) represented by ο: e.g., Σοφωνιβας /s ̣apan-baʕl/
(FRIEDRICH and RÖLLIG 1999, 38–40).

5.3.5.1.2. o-Vowels
The result of the Phoenician shift, probably /o/ but perhaps still /ɔ/ (< */á /), is represented by
ο or ω, ο presumably reflecting the quality and ω representing the quality and phonetic
lengthening under the stress: e.g., λαβον /labó n/, ναδωρ /nadó r/, σαµω /š amó ʕ/, and Χουσωρ
/kuš á r/. The result of the Canaanite shift, when realized as /ō /, is represented by ω or ο: e.g.,
Αβδαλωνυµος /ʕabd-ʔalō nı̄ m/, Οζερβαλος /ʕō zer-baʕl/, and Ειρωµος /(ʔah ̣ı̄ )-rō m/. Long /ō /
- 168 -
resulting from original */aw/ > /ō / or */ahu/ > */au/ > /ō / can be transcribed as ω: e.g., Μωθ
(< */mawt/) and Ιωµιλκου (< */yah ̣u-milk/) (FRIEDRICH and RÖLLIG 1999, 40–41, 44–45).
5.3.5.1.3. u-Vowels
When the /ō / resulting from the Canaanite shift has further developed to /u/, it is represented
by ου: e.g., αδουν /ʔadun/, λασουν /laš un/, κουλω /qulo/,186 and σαφουν /s ̣apun/. Long /ū /
resulting from the original */aw/ > */ō / > /ū / is represented by ου: e.g., κουλω /qulo/ and
µουθ (< */mawt/). The vowel /ü/ is presumably reflected in Greek transcriptions by υ: e.g.,
ρυβαθων /rübbaton/ (< */rabbaton/) and βυν /bün/ (FRIEDRICH and RÖLLIG 1999, 41, 45–46).
5.3.5.1.4. i-Vowels
Original /i/ in a doubly-closed syllable or an unstressed closed syllable is usually represented
by ι: e.g., Μιλκιατων /milk-yaton/, -µιλχαρ /-milqar/, and Βαλσιλληχ /baʕl-š illek/. Occasion-
ally, it is represented by ε: e.g., Μελκαθρος (read Μελκαρθος) and Εσυµσεληµου /eš mū n-
š illem/. Originaly long /ı̄ / is represented with ι or ει: e.g., Αβιβαλος /ʔabı̄ -baʕl/ and Ειρωµος
/ı̄ -rō m/ (< */(ʔa)h ̣ı̄ -rō m/), and Αβδηλιµος /ʕabd-ʔilı̄ m/) (FRIEDRICH and RÖLLIG 1999, 43–44).
5.3.5.1.5. e-Vowels
In a stressed syllable, etymological */i/ is realized as /e/, represented by η (reflecting phonetic
lengthening) or ε: e.g., σιλληχ /š illek/, σεληµου /š illem/, and Οζερβαλος /ʕō zer-baʕl/. Long
/ē / resulting from the original diphthong */ay/ is represented with η: e.g., Σαµηµ /š amē m/
(FRIEDRICH and RÖLLIG 1999, 43–44).

5.3.5.1.6. Shewa and Epenthetic Vowels
There are a few different realizations of a potential parallel to Hebrew shewa, namely, a short
or reduced vowel in an open syllable away from the stress. Historical /a/ may be preserved as
α: e.g., Φανη/Φανε /panē / and λασουναλφ /laš un-ʔalp/. It may also be reduced and represent-
ed by ε: e.g., Φενη. There seems to be some evidence that shewa was in some instances (or at
some stage) realized as /ü/ and represented by υ: λυ βαλ /l-baʕl/ (FRIEDRICH and RÖLLIG 1999,
45). A variable quality of "shewa" may also be demonstrated by the transcription σιθιλ (< */
186. Note that the original pattern is actually probably *qā l (HUEHNERGARD 2015, 32).
- 169 -
satı̄ l/) (KRAHMALKOV 2001, 35). An epenthetic to resolve final -CC# may assimilate to the
quality of the preceding vowel: e.g., Συδυκ and Σεδεκ (< */s ̣idq/). On the other hand, the
epenthetic may be of a more neutral quality: e.g., Συδεκ (< */s ̣idq/) (FRIEDRICH and RÖLLIG
1999, 53). Finally, an epenthetic may assimilate to an adjacent pharyngeal /ʕ/: e.g., Βααλ
/baʕl/ and ζερα /zerʕ/ (FRIEDRICH and RÖLLIG 1999, 53; KRAHMALKOV 2001, 31).
5.3.5.1.7. Semi-Vowels /w/ and /y/ and Diphthongs
It is only in the conjunction /w-/ that we have a possible attestation of a transcription of con-
sonantal /w/: ου λυ ρυβαθων /w-l-rübbaton/. However, it is not clear if ου is intended to rep-
resent vocalic [u] or consonantal [w] in this instance. The semi-vowel /y/ is represented by
Greek ι: e.g., Μιλκιατων /milk-yaton/. It also seems that the sequence of vowel + /y/ may be
represented by ι: e.g., βινιω /biniyō /. The diphthong /ay/ seems to be maintained in the word
Βαιτυλος (< */bayt-ʔil/) (FRIEDRICH and RÖLLIG 1999, 33, 41, 169, 185).
5.3.5.2. Consonants
The Standard Phoenician consonantal system is made up of twenty-two distinct phonemes,
including three voiceless stops (/p/, /t/, /k/), three voiced stops (/b/, /d/, /g/), three emphatic
consonants (/t/,̣ /k ̣/, /s ̣/), a glottal stop (/ʔ/), a glottal fricative (/h/), two pharyngeal fricatives
(/h ̣/, /ʕ/), two liquids (/l/, /r/), two nasals (/m/, /n/), two affricates (/dz/, /ts/), one sibilant (/š /
[s]), and two semi-vowels (/w/, /j/) (HACKETT 2008, 86–87).

At an early period, /p/, /t/, /k/ are represented by the Greek unaspirated series: Σαραπτα (for
‫)צרפת‬, Βαλιτων /baʕl-yaton/, and Αβδιµιλκων (for ‫)עבדמלך‬. After the second century BCE, they
are represented with the Greek aspirated series φ, θ, χ: e.g., Φανε/Φανη /panē /, Σαρε/αφθα
(for ‫)צרפת‬, and σιλληχ /š illek/. It is not clear if φ, θ, χ also represented fricative allophones in
this late period, but Punic /p/ is transcribed by f in Latin (FRIEDRICH and RÖLLIG 1999, 18–24).
The voiced stops were typically represented by Greek β, δ, γ: e.g., Βαλ /baʕl/, ναδωρ /nador/,
and λαδουν /l-ʔadun/. It is not clear if β, δ, γ also represented fricative allophones in this late
- 170 -
period, but there are instances of post-vocalic Punic /b/ represented by Greek φ: e.g.,
Αφεσαφουν (for ‫ )?עבדצפון‬and αφδε (for ‫( )עבדי‬FRIEDRICH and RÖLLIG 1999, 18–24, 40; KRAH-
MALKOV 2001, 21).

The guttural consonants (/ʔ/, /h/, /ʕ/, /h ̣/), which are never represented directly in transcrip-
tion, are either inferred or implied by a hiatus between vowels: e.g., Μερβαλος/Μααρβαλος
/mahr-baʕl/, σαµω /š amoʕ/, and βαλ Αµουν /baʕl-hamun/. Elision of /ʔ/ may be indicated by
the transcription λαδουν (< */l-ʔadun/) (FRIEDRICH and RÖLLIG 1999, 12, 16–18).
5.3.5.2.4. Sonorants
The sonorants /r/, /l/, /m/, /n/ are represented by ρ, λ, µ, ν: e.g., ναδωρ /nador/, κουλω /qulo/,
σαµω /š amoʕ/, and φανε /panē / (KAI 175). In the name ‫גרסכן‬, /r/ is prone to be elided: e.g.,
Γισκων/Γεσκων. A velar allophone of /n/ [ŋ] is also represented by ν: e.g., Βωνχαρ (cf. Bom-
car in Latin script) (FRIEDRICH and RÖLLIG 1999, 28–31).

The emphatic stops /t/̣ and /k ̣/ are represented by the Greek unaspirated series τ and κ: e.g.,,
Μοτυη (for (‫ )מטו)א‬and κουλω /qulo/. Emphatic /s ̣/, probably realized as an affricate [ʦʔ], is
usually transcribed as σ: e.g., Σιδων (for ‫)צדן‬, Συδεκ/Συδυκ/Σεδεκ (for ‫)צדק‬, and ασιρ (for
‫)חצר‬. It may also be transcribed as στ, τ, and perhaps ζ (in one instance): e.g., ατιρ/αστιρ (for
‫ )חצר‬and Ζωφασηµιν (cf. Hebrew ‫שׁ ַמי ִם‬

ָ ‫( )צֹפֵי‬FRIEDRICH and RÖLLIG 1999, 20–21, 26).
5.3.5.2.6. Affricates and Plain Sibilants
Both original /ʦ/ ([ʦ] > [s]) and original /š / ([s]) are represented by σ: e.g., Γισκων (for ‫)גרסכן‬,
υς (for ‫)אש‬, σαµω /š amoʕ/, and σαδε (for ‫)שד‬. A voiced allophone may be represented by ζ:
e.g., Αβδυζµουνος (for ‫)עבדאשמן‬. Original /ʣ/ is usually represented by ζ: e.g., Οζερβαλος
(for ‫ )עזרבעל‬and ζερα (for ‫)זרע‬. Before /r/, it may also be transcribed as σ or σδ: e.g.,
Ασρουβας/Ασδρουβας (for ‫( )עזרבעל‬FRIEDRICH and RÖLLIG 1999, 24–28).

See 5.3.5.1.7.
- 171 -
Consonantal gemination [Cː] is not represented consistently: e.g., ρυβαθων (< */rabbaton/),
Θεννειθ/Θινιθ (for ‫)תנית‬, Εσυµσεληµου /ʔeš mū n-š illem/, Βαλσιλληχ /baʕl-š illek/, and
Σαδυκος (< */s ̣addı̄ q/) (FRIEDRICH and RÖLLIG 1999, 20–21, 54–55).
5.3.5.3. Summary
The most common representation of each Punic phoneme is summarized in the charts below.
Secondary graphemes are listed in parentheses:
Vowels Semivowels Diphthongs
Punic: a e i o (< *ɔ) u ü "ə" w [w] y [j] ay [aj]
Greek: α η, ε ι, ε ο/ω ου υ α, ε, υ (ου) ι αι
Punic: *p [ph] (> [f]?) t [th] (> [θ]?) k [kh] (> [χ]?) b [b] (> [v]?) d [d] (> [ð]?) g [g] (> [γ]?)
Greek: φ θ χ β (φ) δ γ
Emphatic Stops Emphatic Affricate Plain Sibilants
Punic: t ̣ [tˀ] q [kˀ] s ̣ [ʦʔ] ś , š [ʃ] z ([ʣ] or [z])
Greek: τ κ σ (στ, τ) σ ζ
Sonorants Nasals Gutturals
Punic: l [l] r [r] m [m] n [n] ḥ [ħ] ʕ [ʕ] h [h] ʔ [ʔ]
Greek: λ ρ µ ν Ø Ø Ø Ø
Chart 16: Summary of Phoenician-Punic in Greek Transcription: Correspondences
5.3.6. Summary
The various Greek transcription conventions during the Hellenistic, Roman, and Byzantine
periods in the ancient Near East are summarized in the following chart (chart 17):
Vowels
- 172 -
Latin Akkadian Arabic Aramaic Phoenician
ă α α α α α
ā α α α -
ĕ ε ε (η) (ε, η) η, ε -
ē η (ε) η, ει (ε, η) η, ε
ı̆ ι ι ε (η) ι ι, ε
ı̄ ι (ει) ι, ει ι (ει) ι
ü - - - - υ
ŭ ου ο ο (ου) ου -
ū ου ου ου ου
ŏ ο - - ο -
ō ω (ο) ω (ω) ο/ω
ə - - - α α, ε, υ
Diphthongs and Semi-Vowels
ai/ae/aj αι αι αι, ε, η - αι
au/aw αυ - αυ - -
j ι (α)ι ι, Ø ι ι
w ου υ, Ø ου, Ø - ου
Voiceless Stops
p π [p] φ [ph] φ [ph] - φ
t τ [t] θ [th] θ [th] θ θ
k κ [k] χ [kh] χ [kh] - χ
- 173 -
Voiced Stops
b β [b] β [b] β [b] β β
d δ [d] δ [d] δ [d] δ δ
g γ [g] γ [g] γ [g] γ γ
Sonorants (Liquids and Nasals)
l λ [l] λ [l] λ [l] λ λ
m µ [m] µ [m] µ [m] µ µ
n ν [n] ν [n] ν [n] ν ν
r ρ [r] ρ [r] ρ [r] ρ ρ
Fricatives
f φ - φ - (φ?)
v β ([β] or [v]) - - β (β, φ?)
ṯ - - θ (τ) θ (θ)
ḏ - - δ δ (δ)
s σ σ σ σ σ
z - ζ ζ ζ ζ
š - σ σ σ -
Emphatic/Glottalic Consonants
ṭ - τ τ - τ
ḳ - κ κ κ κ
- 174 -
ṣ - σ (ζ) σ σ σ, (στ, τ)
ś ̣ - - σ [ɫˀ] - -
ṯ ̣ - - ζ [ɮˁ] - -
Gutturals (Liquids and Nasals)
ḫ - ξ, Ø χ, Ø - -
ġ - - γ, Ø - -
ḥ - - Ø - Ø
ˁ - - Ø Ø Ø
h Ø - Ø Ø Ø
ˀ - - Ø - Ø
Chart 17: Summary of Latin and Semitic in Greek Transcription: Correspondences
5.4. LINGUISTIC AND ORTHOGRAPHIC ANALYSIS: GREEK IN HEBREW
The primary evidence for Hebrew transcription of Greek from the Roman period is attested in
the various vocalized manuscripts of the Mishnah, especially Kaufmann and Parma A, as ana-
lyzed by HEIJMANS in his work on Greek and Latin loanwords in the Mishnah (2013). A num-
ber of relevant epigraphic examples are also cited below.

5.4.1. Greek Loanwords in the Mishnah
5.4.1.1. Vowels
5.4.1.1.1. α
Greek α is represented with Hebrew pataḥ or qamas ̣. The distribution of pataḥ/qamas ̣ is ac-
cording to the Tiberian Hebrew rules, with pataḥ in a closed unstressed syllable and qamas ̣ in
an open unstressed syllable or a closed stressed syllable. Far from the stress, a ḥataf pataḥ
may be used. Examples from the Kaufmann manuscript include ἀήρ ‫אָוֵיר‬, γάµµα ‫ ַגּמָּה‬,
σάνδαλον ‫ ַסנְדָּ ל‬, κάµπτρα ‫ ַק ְמט ְָרה‬, and ἄσηµον ‫ ֲאסִימוֹן‬. In Palestinian-pointed manuscripts, there
is no consistency in the distribution of /a/ vowel signs. Some words exhibit an /a/ > /i/ shift
- 175 -
(e.g., ἀσσάριον ‫)אִיסָּר‬, but these are regarded by HEIJMANS as reflecting a later internal Hebrew
development (i.e., attenuation). In the environment of sonorants (λ, µ, ν, ρ) or κ, Greek α may
be rendered as /o/ or /u/: e.g., πάλλιον ‫ פּוֹלְיוֹן‬and κάψα ‫( קוֹ ְפסָא‬HEIJMANS 2013, 259–61).
5.4.1.1.2. ε/αι
Greek ε is mostly rendered with Hebrew /a/ vowels, usually pataḥ. Perhaps due to nothing
more than statistical coincidence, almost all the instances of Hebrew /a/ for Greek ε occur in
closed syllables.187 About half of these occur word-initially, transcribed with an initial ‫א‬: e.g.,
ἐµβατή ‫אַ ְמ ָבּטֵי‬, ἐνθήκη ‫הָאַנ ְִֿתיקֵי‬, σπεκλάριον ‫ ְס ֿפַק ל ְַרי ָיה‬, ἐπίτροπος ‫ ַאפִּיטְרוֹפּוֹס‬, and κεφαλωτόν
ֿ‫ ַה ַקּפְּלוֹטוֹת‬. HEIJMANS concludes that rendering Greek ε with Hebrew /a/ is due to an open pro-
nunciation of Greek ε in contrast with Hebrew /e/, at least in closed syllables, during the peri-
od when the word was borrowed. In a minority of instances Greek ε is rendered as /e/, /i/, or
/o/. Due to the paucity of data, no conclusions may be drawn regarding such apparent shifts.
The grapheme αι, which interchanged frequently with ε in the Roman period, is transcribed
with s ̣ere in the word καῖρος ‫ קֵירוֹ ֿס‬in the most reliable manuscripts (2013, 262–63, 275).188
5.4.1.1.3. η
Unlike ε, Greek η is usually rendered with Hebrew /e/ vowels, usually s ̣ere: e.g., ἀήρ ‫אָוֵיר‬,
ἀσθενής ‫אַ ְס ְטנֵס‬, and διαθήκη ‫ָיתקֵי‬

ֵֿ ‫דְּ י‬. Greek η is occasionally rendered with Hebrew ḥiriq.
However, HEIJMANS makes a distinction between cases which are only attested in some manu-
scripts and thus reflect a later tradition, and those which are ḥiriq across the manuscript tradi-
tion and thus reflect the pronunciation at the time when Mishnaic Hebrew was spoken. To the
former class belong words such as βῆµα ‫( ֿבִימָא‬but cf. ‫ בֶימָה‬,‫ )בֵימַה‬and δηνάριον ‫( דִּ ינָר‬but cf.
‫)דֵּ ינָר‬. To the latter class belong words such as ἄσηµον ‫ ֲאסִימוֹן‬and σηµεῖον ֿ‫( סִימְיוֹנוֹת‬HEIJMANS
187. The one example which does not reflect a closed syllable in Hebrew is ἐπίτροπος ‫( ַאפִּיטְרוֹפּוֹס‬assuming the
dagesh in the peh only reflects a stop pronunciation).
188. There is also an instance in which Greek αι is rendered by Hebrew /a/ in πραιτώριον ‫ָטוֹרין‬
ִ ‫ ָפּל‬, a development
consistent with the omission of the second element of the diphthong that occurs in the papyri. This may also be
reflected by the transcription κυαίστωρ ‫ ָקסְטוֹר‬in Parma A.
- 176 -
2013, 264–66).189 The fact that Greek η is usually rendered with Hebrew /e/ is evidence that
even the vocalization of the Mishnah reflects pre-Byzantine Greek pronunciation.

5.4.1.1.4. ι/ει
Greek ι is usually rendered with Hebrew ḥiriq: e.g., ὀµφάκινον ‫אַנְפֹּקִינוֹן‬, λίτρα ‫לִיט ְָרא‬, and
κινάρα ‫קִינ ְָרס‬. In closed unaccented syllables, Greek ι may also be rendered by Hebrew /e/
vowels: e.g., ἰδιώτης ‫ה ְֶדֿיוֹט‬, ἐµπίλια ‫אַנְפֵּי ְלי ָא‬, and µίνθα ‫מֵינְתָּ ה‬. In open unaccented syllables,
Greek ι may be rendered as /e/. This reflects the reduction of the vowel and thus the represen-
tation of vocal shewa with Hebrew s ̣ere or seghol.190 As would be expected, Greek ει is
transcribed by Hebrew ḥiriq: e.g., χεῖ ‫ כִּי‬and δεῖγµα ‫ ִ ֿדיגְֿמָא‬. However, in Yemenite manuscripts
there is one example in which the /e/ pronunciation of ει before a vowel is preserved, namely,
φορειαφόροι ‫פַּרי ֵיפ ְִרין‬. Finally, ι in the sequence ια can be rendered consonantally in Hebrew:
e.g., σπεκλάριον ‫ ְספַק ל ְַרי ָיה‬and ἐµπίλια ‫( אַנְפֵּי ְלי ָא‬HEIJMANS 2013, 262, 267–68, 275–76).
5.4.1.1.5. ο/ω
Greek ο is usually rendered by Hebrew ḥolam-waw: e.g., ὄρυζη ‫אוֹרז‬,

ֶ νοταρικόν ‫נוֹט ֵָריקוֹן‬, and
κισσός ‫קִיסוֹס‬. In a number of words with the Greek suffixes -ος and -ον, the ο of the suffix is
rendered as Hebrew /a/: e.g., ἄµυλον ‫ ֲאמֵילָן‬, βωµός ‫בּוֹמַס‬, and λευκόν ‫ ַלוְוקָן‬. This reflects the
reduction and centralization of the suffixes to /əs/ and /ən/.191 This is supported by epigraphic
evidence, in which Greek names ending in -ος have various representations: e.g., θεόφιλος
‫תיפילס‬, Ἀντίοχος ‫אנטיכס‬/‫אנטיכיס‬/‫אנטיוכוס‬, and Ἡλιόδωρος ‫הלידורס‬/‫הלידרוס‬. Only a few words
render Greek ο with shuruq: e.g., ὀλεάριος ‫ אוּ ְלי ִָירין‬and ἐµπορία ‫ְפּוּרי ָא‬
ְ ‫אַנ‬. Greek ω is rendered
by Hebrew /o/ in the majority of instances: e.g., ἐξώστρα ‫כְּסוֹ ְסט ְָרא‬, δράκων ‫דְּ ָרקוֹן‬, and
χαράκωµα ‫כּ ְַרקוֹם‬. There are a few instances in the environment of labial consonants /m/ and
189. The only instance of ηι in a Greek loanword, λῃστής ‫ליסטים‬/‫ליסטיס‬, is rendered in Hebrew by /e/ or /i/. The
former represents the earlier pronunciation and the latter rendering represents the Byzantine pronunciation.
190. Additionally, there are a number of words in which ι is rendered with s ̣ere even in closed stressed syllables
(but not necessarily closed in Greek) (e.g., σµίλη ‫) ֻאזְמֵל‬. In a few instances, ι is rendered with /a/ before a liquid
consonant (e.g., ἄσιλλα ‫) ֲאסָל‬. Finally, ι is rendered by /o/ or /u/ a number of times in the environment of labials
and κ (e.g., µιλιάριον ‫ ֻמ ְלי ָיר‬, κίστη ֿ‫) ַק ְוסְתּוּת‬.
191. Variants of the same word in the same manuscript support this (e.g., γύψος ‫גִּי ְפּסֵיס‬/‫גִּי ְֿפסַס‬, φανός ‫ ָפנָּס‬/‫) ָפּנֵיס‬.
- 177 -
/p/, in which Greek ω is rendered by Hebrew /u/: e.g., µονοπώλης ‫ ַמנְפּוּל‬and πήλωµα ‫ֿפִילוּמָא‬
(HEIJMANS 2013, 269–74).192

5.4.1.1.6. υ/οι
Greek υ is rendered by Hebrew ḥiriq about half the time and by Hebrew /o/ or /u/ about half
the time: e.g., ḥiriq for υ in γύψος ‫גִּי ְפּסֵיס‬, λάγυνος ‫ ָלגִֿין‬, and σύµφωνον ֿ‫ ;סִימְפוֹנוֹת‬/o/ or /u/ for υ
include βυρσεύς ‫בּוּרסִי‬,

ְ γρύτης ‫גְּרוֹטֵי‬, and χρυστουµῖνος ‫קְרוּ ְס ָטמִילִים‬. The distribution points to
Greek υ being realized as /y/ during the period.193 Similarly, the diphthong οι is rendered in
Hebrew with ḥiriq: e.g., κοινωνία ‫ קִינוֹנִייא‬and κοιλία ‫( ִכּלְיי ַא‬HEIJMANS 2013, 272–73, 276).
5.4.1.1.7. ου
Greek ου is rendered in Hebrew both by /u/ and /o/: e.g., βοῦργος ‫בּוֹר ְגּנִין‬/‫ִין‬
ְ ‫בּוּרגָנ‬,
ְ φοῦνδα /‫פֿוּנְדָּ ה‬
‫פּוֹנְדָּ ה‬, κουκούµιον ‫קּוֹ ֿמְקּוֹם‬/‫קוֹמְקוּם‬/‫קוּמקוּם‬, and βούλιµος ‫בּוּ ְלֿמּוֹ ֿס‬/‫בֿוֹלְמוֹס‬.194 HEIJMANS admits that
there is no simple explanation for such interchanges. Similar interchanges occur in the Papyri
from Egypt and in Jewish Palestinian Aramaic. There could be any number of factors reflect-
ed in this interchange, including transmission history. However, HEIJMANS does maintain that
at least some of the interchanges reflect an old pronunciation (2013, 277–78).

5.4.1.2. Diphthongs
5.4.1.2.1. ευ/αυ
Only two or three words, which are certain to have been borrowed from Greek and not
Latin,195 attest to the rendering of the Greek diphthongs ευ/αυ: εὐθέως ‫ְותיאוֹס‬
ִֿ ‫אֵיו‬, λευκόν ‫ ַלוְוקָן‬,
and possibly Νικόλαος ‫( נִ ְק ָל ֿבֵס‬if we assume /nikolaos/ > /nikolaus/ > /nikolavs/). The orthog-
raphy with double waw ‫ וו‬reflects the pronunciation of the Byzantine period ([εv],[av]),
192. There are also instances of /o-o/ dissimilating to /i-o/: e.g., κωνωπεῖον ‫ קִינּוֹף‬and βωµός ‫בִימוֹס‬. Finally, there
are a few instances in which Greek ω is rendered by /e/ (e.g., πραιτώριον ‫ָטוֹרין‬
ִ ‫ ) ָפּל‬or /a/ (e.g., κατωφερής ‫) ְק ַט ְפ ֵֿרס‬.
193. Additionally, there are four instances in which υ is rendered by Hebrew /e/: ζῦθος ‫זֵיתֿ וֹ ֿם‬, ἄµυλον ‫ ֲאמֵילָן‬,
συνέδριον ‫ ֶסנְה ְֶד ִֿרין‬, ζωµάρυστον ‫זוֹ ָמנֵי צט ְָרה‬, and ὄρυζα ‫אוֹרז‬.
ֶ This corresponds with a similar interchange of υ/ε in
Greek papyri, reflecting that υ was sometimes realized as /e/. HEIJMANS suggests that such may be the case in the
famous transcription συµα for Hebrew ‫שׁמַע‬ ְ . There are also a number of examples in which υ is rendered by
Hebrew /a/ (e.g., ὑποθήκη ‫ַפּוֹתיקִי‬ ִֿ ‫)א‬.
194. The same interchange is attested in the transcription of Latin crustuminus ‫קְרוּ ְֿס ָֿט ֿמֵילֿין‬/‫קְרוֹ ְס ָטמִילִים‬.
195. The word claustrum ‫קְלוֹ ְסט ְָרה‬, caulis ‫וְקוֹ ְלסֵי‬, and paragauda ‫ פּ ְַרגֿוֹ ֿד‬seem to attest to the contraction /au/ > /o/
characteristic of Latin, but not Greek.
- 178 -
which is supported by the spellings ‫ אביתוס‬and ‫( לבקן‬HEIJMANS 2013, 276–77). While it is
correct that spellings with ‫ וו‬and ‫ ב‬indicate the Byzantine pronunciation, they do not therefore
determine that this was not already the realization of ευ/αυ in the Roman period (276–77). A
Jerusalem inscription from the first century BCE/CE renders the Greek name Εὐπτόλεµος as
‫( אפטלמוס‬CIIP I/1, no. 407), representing the second element of the diphthong with ‫פ‬.
5.4.1.3. Consonants
5.4.1.3.1. π, τ, κ
The unaspirated voiceless stops π, τ, κ are rendered in Hebrew by ‫ ק‬,‫ ט‬,‫פ‬: e.g., βασιλική
‫ ָבּסִי ְלקִי‬, σπλήνιον ֿ‫אַ ְס ְפּ ָלנִית‬, and ἰσάτις ‫ ֶא ְסטֵס‬. While Greek π could only be represented by ‫פ‬, the
choice to represent unaspirated τ and κ with ‫ ט‬and ‫ ק‬indicates that it is likely that the "em-
phatic" consonants in Hebrew also lacked aspiration. A shift of /p/ > /b/ occurs in a number of
instances: e.g., σπάθη ‫ אצבתי‬and πάρδαλις ‫ברדלס‬196 (HEIJMANS 2013, 235–40).

5.4.1.3.2. β, δ, γ
The voiced stops β, δ, γ are rendered in Hebrew by ‫ ג‬,‫ ד‬,‫ב‬: e.g., βιβάριον ‫ ַהבִּי ֿב ִָרין‬, µαργαρίτης
ֿ‫מ ְַר ָגּלִית‬, µάλαγµα ‫מְלוֹגְֿמָא‬, δηνάριον ‫דִּ ינָר‬, and γράδος ‫וּגֿ ְָרדֿוֹן‬. The dagesh seems to operate ac-
cording to Tiberian rules: e.g., γενέσια ‫ ְגּנֶי ְסי ָה‬and τράγος ‫ט ְָרגֿוֹס‬. It is also worth noting that
when voiceless alternatives of γ and δ occur, they are rendered as the emphatics ‫ ק‬and ‫ ט‬in
Hebrew: e.g., γλωσσόκοµον ‫ ִמקְּלוֹ ְס ָקמָא‬and Καλένδαι ‫ ָק ַלנְטס‬. This indicates that the Hebrew
voiced stops were probably unaspirated (HEIJMANS 2013, 241–44).

5.4.1.3.3. φ, χ, θ
The aspirated voiceless stops φ, θ, χ are rendered in Hebrew by ‫ כ‬,‫ ת‬,‫פ‬: e.g., δίφορος ‫דִּ י ְפ ָֿרא‬,
κοχλιάριον ‫כּוּ ְֿכ ְלּי ַֿיר‬, and ‫ תֵּ יק‬θήκη. Representing aspirated θ and χ with ‫ ת‬and ‫ כ‬indicates that
the Hebrew non-emphatic voicless stops were aspirated. The dagesh seems to operate accord-
ing to Tiberian rules: e.g., µέλαθρον ֿ‫ ַמלְתְּ ָריוֹת‬and καθέδρα ‫( ק ֶָֿת ְ ֿד ָרה‬HEIJMANS 2013, 245–48). A
196. In addition to these changes, Hebrew ‫ כ‬renders the first part of the sequence κτ in Greek, reflecting a κτ >
χθ shift in the Koine pronunciation. Also, primarily in oriental manuscripts, there are attestations of ‫ד‬
representing τ, an attempt at reflecting the lack of aspiration.
- 179 -
Jerusalem inscription from the first century BCE/CE renders Greek βοηθός as ‫בוטון‬, represent-
ing Greek θ with ‫( ט‬CIIP I/1, no. 119–20).

5.4.1.3.4. λ, µ, ν, ρ
The sonorants λ, µ, ν, ρ are usually represented with Hebrew ‫ ר‬,‫ נ‬,‫ מ‬,‫ל‬: e.g., ἄλφα ‫אַ ְל ַפֿא‬,
µηχανή ‫ ֶמ ְֿכנֶה‬, νόµος ‫נִימוֹס‬, and ὄρυζα ‫אוֹרז‬.

ֶ Just as in Koine, λ and ρ may interchange: e.g.,
ὀλεάριος ‫ אוריירין‬and πραιτώριον ‫ָטוֹרין‬

ִ ‫ ָפּל‬. Interchange of µ and ν is less common, limited to
the environment of preceding labial stops, as in ἔµβολον ‫ ֶענְבּוּל‬. The initial aspirated rho ῥ is
reflected in the word ῥοδοδάφνη ‫( ה ְָרדַ ְפנֵי‬HEIJMANS 2013, 248–52).

5.4.1.3.5. σ, ζ
Greek σ is usually realized as Hebrew ‫ס‬: e.g., σέλλα ‫ ַסלָּה‬and σῆµα ‫סִימָן‬. Before voiced conso-
nants, it undergoes voicing and is represented with ‫ז‬: e.g., προσβολή ‫פְּרוֹזְבֿוֹל‬. Before τρ, Greek
σ becomes emphatic (or glottalic) in Hebrew: e.g., γάστρα ‫ ָגּ ְצט ְָרה‬. However, this phenomenon
is more common in Babylonian manuscripts. Greek ζ is realized as Hebrew ‫ז‬: e.g., ζεῦγος ֿ‫זוֹג‬
(HEIJMANS 2013, 253–56).

5.4.1.3.6. ξ, ψ
Greek ξ is represented by Hebrew ‫כס‬: e.g., ἐξέδρα ‫ אַ ְֿכס ְַד ָֿרה‬and ξένος ‫אַ ְכ ְסנַי ִים‬.197 Greek ψ is
represented by Hebrew ‫פס‬: e.g., ὀψώνιον ‫ ֶא ְֿפ ֶסנְי ָיא‬and ψῆφος ‫ ְפּ ַס ְֿפסִין‬. This indicates that the
first element of each must have been aspirated: [khs] and [phs] (HEIJMANS 2013, 257–58).
ֿ5.4.1.4. Summary
The most common realization of each phoneme in Greek loanwords in Mishnaic Hebrew is
summarized in the chart below (chart 18):
Vowels Diphthongs
Greek: α ε/αι η ι/ει ο/ω υ/οι ου ευ αυ
Hebrew: ‫ ַב‬,ָ‫ב‬ (‫ַב )בֵי‬ ‫ֵב‬ ‫בִי‬ ‫בוֹ‬ ‫בוּ‬/‫ בוֹ‬,‫בִי‬ ‫בוֹ‬/‫בוּ‬ (‫ אַוְו )אב‬,ְ‫אֵיו‬ ‫אָב‬
197. Word-finally, Greek ξ is represented by either ‫ קס‬or ‫גס‬: e.g., πίναξ ‫ ִפּנְקֵס‬and πάλλαξ ‫ ַֿפ ְלגָֿס‬. This is best
explained as an internal-Greek development (πίναξ > πίνακος).
- 180 -
Voiceless Stops Voiced Stops Aspirated Stops
Greek: π τ κ β δ γ φ θ χ
Hebrew: ‫פ‬ ‫ט‬ ‫ק‬ ‫ב‬ ‫ד‬ ‫ג‬ ‫פ‬ ‫ת‬ ‫כ‬
Sonorants Sibilants Combination
Greek: λ µ ν ρ σ ζ ξ ψ
Hebrew: (‫ל )ר‬ (‫מ )נ‬ (‫נ )מ‬ (‫ הָר‬,‫ צ]טר[( ר )ל‬,‫ס )ז‬ ‫ז‬ (‫ גס‬,‫כס )קס‬ ‫פס‬
Chart 18: Summary of Greek Loanwords in Mishnaic Hebrew: Correspondences
5.5. CONCLUDING REMARKS
The preceding chapter outlined the correspondences between the phonology and orthography
of the local Greek during the Roman period. The present chapter has surveyed and tabulated
the Greek orthographic conventions utilized in transcribing Latin, Akkadian, Arabic, Arama-
ic, and Punic as well as summarizing how Greek loanwords are realized in Hebrew. Taken to-
gether, the findings of these two chapters provide a foundation for understanding the phonol-
ogy and orthography of the Secunda. First, the work on Greek pronunciation provides an
expectation of what the most natural phonemic value for any particular Greek grapheme
might be at the time of the composition of the Secunda. Second, the work on Greek transcrip-
tion conventions provides an expectation of how the Greek script might be used to represent
various foreign phonemic values with an imperfect overlap. Third, the representation of
Greek loanwords in Hebrew aids our understanding of how the two phonological systems
corresponded to one another. To avoid redundancy, the data presented in this and the previous
chapter will be referred to repeatedly in the discussion of the phonology and orthography of
the Secunda in the following chapter rather than in a concluding synthesis here. Generaliza-
tions about transcription practices will be made in passing as the data are applied to the
transcriptions of the second column.
- 181 -
6. THE PHONOLOGY AND ORTHOGRAPHY OF THE SECUNDA
6.1. INTRODUCTION
Chapters 4 and 5 investigated Greek pronunciation in Roman Palestine and Greek transcrip-
tion conventions in the Roman Near East. The present chapter applies the findings of the pre-
vious two chapters to the phonology and orthography of the Secunda. The ultimate objective
is to arrive at a phonemic and phonetic transcription of the Hebrew reflected in the Secunda.
6.2. METHODOLOGY
Current research on historical Hebrew linguistics will serve as a point of departure for discus-
sing the phonemes and phonemic features of Secunda Hebrew. We will operate under the as-
sumption that each phoneme or phonemic feature (e.g., quantity, gemination, vowel quality)
generally falls on some point of the spectrum between its (reconstructed) realization in Proto-
Northwest Semitic and its realization in one of the later reading traditions of Hebrew (e.g.,
Tiberian, Babylonian, Palestinian, Samaritan). Each phoneme will be addressed from the per-
spective of Greek historical phonology and orthography, Greek transcription conventions,
and Hebrew dialectology. Analysis will be informed by modern linguistic studies in the fields
of phonetics and phonology with a special emphasis on (cross-language) perception studies,
since we are dealing with transcriptions into another language's script (for a summary of our
methodology as it relates to cross-language perception studies, see 1.3.3).
- 182 -
In light of all these data, each transcription will be assigned a phonemic and phonetic
transcription. The phonetic transcription, though informed by the evidence, is not necessarily
intended to be understood as indisputable. Rather, though quite tentative, it is included as an
educated approximation to better elucidate the nature of the Secunda transcriptions.
6.3. CONSONANTS
Consonants are fairly stable throughout the history of Hebrew, in the various reading tradi-
tions of Hebrew, and in the consonantal text of the Hebrew Bible. In most cases, it is merely
the reflex of a particular consonant that is under discussion.

6.3.1. Stops (/b/, /p/, /d/, /t/, /g/, /k/)
In the earliest stages of Hebrew, the stops were probably unaspirated, as is demonstrated by
the fact that the Greek alphabet adopted the Semitic voiceless stops to represent the Greek
unaspirated series (e.g., ‫ = תָּ ו‬ταυ, ‫ = כַּף‬καππα, ‫ = פֵּא‬πι). Relatively early in the history of He-
brew, the voiceless stops acquired aspiration: /k/ > [kh], /p/ > [ph], /t/ > [th]. Eventually, proba-
bly as a result of Aramaic influence, each stop developed a spirantized post-vocalic allophone
(i.e., /b/ > [v], /g/ > [γ]/[ʁ], /d/ > [ð], /kh/ > [x]/[χ], /ph/ > [f], /th/ > [θ]) (KUTSCHER 1965; STEIN-
ER 2007, 54; YUDITSKY 2017, 21).
There is no consensus regarding the precise timing of spirantization in Aramaic and
Hebrew, but there is good reason to believe that not all the stops were affected at once, but
shifted in stages. Though spirantization is attested at a relatively early stage (ca. 7th BCE) in
Aramaic in Mesopotamia, it did not reach the west until later. STEINER has argued compelling-
ly that while the labials (/b/, /p/) and the dentals (/d/, /t/) underwent spirantization prior to the
loss of */ḫ / (ca. 1st BCE/CE), only after the loss of */ḫ / did the velar stop /k/ develop a spiran-
tized allophone, because, prior to such a loss, the spirantized allophone of /k/ ([x] or [χ])
could have been confused with */ḫ / ([χ]). That the velar stops developed spirantization at a
later stage is further supported by the lack of a spirantized /k/ in the Egyptian Aramaic of P.
- 183 -
Amherst 63 (ca. 4th/3rd BCE) and the fact that the velar stops /k/ and /g/ never acquired spiran-
tized allophones in Samaritan Hebrew (2007, 56, 64–65).198
In light of STEINER's argument that spirantization of the labials and dentals occurred
prior to the loss of the uvular fricatives (*/ḫ / and */ġ /), it is likely that at the time of the com-
position of the Secunda (2nd/3rd CE) the labial and dental stops had spirantized post-vocalic
allophones: /b/ > [v], /g/ > [γ], /d/ > [ð], /t/ > [θ] (see 3.3.2.2). It is unclear whether or not spi-
rantized allophones of the velar stops /k/ and /g/ had yet developed before the Secunda.
6.3.1.1. Voiced Labial Stop: /b/ = β
In the Secunda, both syllable-initial and post-vocalic /b/ are represented with β:
Greek Phonemic Representation Phonetic Representation Gloss Verse

καρβαµ /qerbam(m)/ [kˀaʀba(̃ m)] 'within them' Ps. 49:12
βαναυ /bɔ̄naw/ [bɔːnaw] 'his sons' Ps. 89:31
αβδω /ʕabdō / [ʕaβdoː] 'his servant' Ps. 35:27
αβι /ʔɔ̄bı̄ / [ʔɔːβiː] 'my father' Ps. 89:27
There is epigraphic evidence that already by the first century CE, Greek β in Palestinian Koine
represented a bilabial fricative [β] and not the bilabial stop [b] (4.5.3.1.15).199 It was only af-
ter a nasal that β represented the historical plosive pronunciation (e.g., φόβος [phoβos], but
λαµβάνω [lambano]; 4.5.3.1.26). In fact, the rare utilization of µβ to transcribe /b/ in the
LXX in names such as Ζαµβινα ‫( זְבִינָא‬Ezra 10:43) and Αµβακουµ ‫( ֲחבַקּוּק‬Hab. 1:1) may re-
flect an attempt to represent a plosive pronunciation of /b/ ([b]) (see ROSÉN 1963, 68). This
convention (µβ), however, is by no means the normal mode of transcribing a plosive [b] into
Greek. Both Latin /b/ and Semitic /b/ are regularly transcribed into Greek as β (5.3.6).
Presumably, the bilabial fricative [β] was the nearest Greek consonant to the bilabial plosive
[b], differing only in manner of articulation (fricative vs. plosive).
198. Samaritan Hebrew grammarians discuss the allophones of ‫ בפדו"ת‬and not ‫( בג"ד כפ"ת‬STEINER 2007, 54).
199. At some point during the Byzantine period, the bilabial fricative [β] shifted to a labio-dental fricative [v].
- 184 -
If we assume that there was a post-vocalic fricative allophone of /b/ ([β] or [v]), as the
history of Hebrew would suggest, it is no surprise that it was also transcribed by β. This is
precisely what happens in Greek transcriptions of Latin when attempting to represent the
newer fricative pronunciation of consonantal v, namely, [β] or [v] (< *[w]) (5.3.1.2.4). Ara-
maic syllable-initial and post-consonantal /b/ are also both transcribed by β (5.3.4.3).
There are also alternative ways of representing a bilabial fricative [β] or labio-dental
fricative [v] in Greek orthography and transcription. In the sequence αυ and ευ, the second el-
ement came to represent [β]/[ɸ] by the Byzantine period, during which it ultimately shifted to
[v]/[f] (4.5.3.1.13–14). Latin consonantal v in proper names is occasionally represented in
this way in Palestinian Koine (4.5.3.1.15). In fact, Hebrew ‫ ב‬is used in a few instances to ren-
der the second element of the diphthongs αυ/ευ in loanwords from Greek in the Mishnah
(5.4.1.2.1). In transcriptions of Phoenician-Punic, there is one instance in which post-vocalic
/b/, perhaps reflecting a fricativized allophone, is represented with φ (αφδε /ʕabde/) (5.3.5.3).
In light of these last few points, it is worth mentioning a few examples of interest in
external witnesses to the Secunda. First, in scholia on Genesis 34:2, Hebrew ‫שׁכַּב‬
ְ ִ ‫' ַויּ‬and he lay
down' is transcribed as ουεσχαυ (Cod. 127). If the reading is original, the rendering of post-
vocalic /b/ by υ would indicate a fricative pronunciation ([β] or [v]). However, it is likely that
the reading of Cod. 344, which has ουεσχαβ, is more reliable.200 Second, the feminine impera-
tive of ‫ לב"שׁ‬is transcribed as λαφσι ‫( ִל ְבשִׁי‬Isa. 51:9) (see YUDITSKY 2017, 81). Because the He-
brew voiced stops were probably unaspirated (5.4.1.3.1–2), an unvoiced allophone of a plo-
sive /b/ should be represented with π instead of φ.201 The fact that an unvoiced allophone is
represented with Greek aspirated φ probably points to a post-vocalic fricative realization of
/b/.202 Third, a similar example may be found in the transcription (α)ζαφθανει ‫' ֲעזַבְתָּ נִי‬you have
200. I would like to thank Peter Gentry for discussing with me the attestations of this word in the manuscripts.
201. For this principle in the transcriptions of the LXX, see KNOBLOCH (1995, 175).
202. Note that devoicing of /b/ before an unvoiced consonant also occurs in Greek transcriptions of Akkadian
- 185 -
forsaken me' (Ps. 22:2) in Codex Bezae's (folio 99v) reading of Matthew 27:46, which is like-
ly to be a quotation of Psalm 22:2 from the Secunda (or another transcription tradition).203
Relatively contemporary Hebrew evidence also supports a post-vocalic fricative real-
ization of /b/. Interchanges of ‫ו‬/‫ ב‬are attested in Mishnaic Hebrew (BREUER 2013, 111; BAR-
ASHER 2015, 61–62) and at Qumran (REYMOND 2014, 70–71). Interchanges of ‫פ‬/‫ ב‬are also at-
tested in Mishnaic Hebrew (BREUER 2013, 111; BAR-ASHER 2015, 61–62) and in the Judaean
Desert texts (MOR 2015, 121–22).
Contemporary Greek and Hebrew evidence is consistent with positing a syllable-ini-
tial realization of /b/ as a plosive [b] and a post-vocalic realization as a fricative [β] or [v] in
the Secunda. It should be noted that the process of spirantization in Hebrew and Aramaic, at
least in the case of the bilabial stops, involves a change both in the manner of articulation
(plosive > fricative) and in the place of articulation (bilabial > labio-dental). However, it is
more sensible linguistically if /b/ was first fricativized and only later realized as a labio-den-
tal: [b] > [β] > [v]. This is precisely what happens in the development of Greek β ([b] > [β] >
[v]) and is attested cross-linguistically (e.g., spirantized /b/ in Spanish is [β] and not [v]).
Neo-Aramaic dialects also commonly realize post-vocalic /b/ phonetically as [β].204
(e.g., θηροφσ terrubš [5.3.2.3.2]).
203. The entire line in Codex Bezae reads: ΗΛΕΙΗΛΕΙΛΑΜΑΖΑɸΘΑΝΕΙ = ηλει ηλει λαµα ζαφθανει. The
reading ζαφθανει is clearly secondary; most manuscripts have σαβαχθανι, reflecting Mishnaic Hebrew or
Aramaic ‫( שבקתני‬BUTH 2014, 395–96). Because ζαφθανει is not original, yet reflects the Biblical Hebrew of
Psalm 22:2, it is entirely possible that the original reading of σαβαχθανι was later amended to ζαφθανει
according to a text like the Hexapla. Alternatively, the manuscript may have been amended based on another
text or oral tradition. However, a strong case can be made that it is a quotation from the Secunda. In Jerome's
letter to Pammachius (Epistula LVII), he cites the words of Psalm 22:2 as Eli Eli lama zapthani. When Jerome
transliterates Hebrew, he often, but not always, seems to transcribe an already existing Greek transcription of
Origen (or perhaps Theodotion) into Latin letters. The fact that his transliteration here agrees with that of Codex
Bezae in two unique respects (omission of a in [a]zapthani and interchange of p for b in zapthani) almost
certainly indicates that his Latin transliteration comes from the same Greek source as Codex Bezae, which is
most likely the Secunda (or perhaps another source such as Theodotion or Aquila).
204. In the dialect of Qaraqosh, post-vocalic /b/ is occasionally realized as [β] (KHAN 2002, 26). In the dialect of
Urmi, the reflex of */b/ (/w/) is realized phonetically as [β] or [v] (KHAN 2008a, 20). In the dialect of the Jews of
Arbel, post-vocalic /b/ may be realized as [β] (KHAN 1999, 17).
- 186 -
Positing a bilabial rather than a labio-dental realization of post-vocalic /b/ actually
better explains the interchanges of ‫ ב‬and ‫ו‬. A bilabial fricative would be more similar in artic-
ulation to a labio-velar approximant ([w]) than a labio-dental fricative ([v]) would. The devel-
opment of Latin consonantal v (*/w/ > *[β] > [v]) supports this. It is likely that Hebrew waw
developed in the same way (see SHARVIT [2016, 290–91] on the development in Aramaic).
Also, post-vocalic Hebrew /b/ is consistently transcribed by Jerome as b in Latin, even
though consonantal v would better approximate a [v].205 Therefore, in phonetic transcription
of the Secunda, post-vocalic /b/ will be represented as a bilabial fricative [β], with the under-
standing that it was in the process of becoming or had already become [v] as in Tiberian He-
brew. A bilabial fricative [β] realization of ‫ ב‬is also attested in some Jewish communities in
Yemen (YA'AKOV 2012), Aleppo, and Tunisia (HENSHKE 2013a, 537; HENSHKE 2013b, 861).206
6.3.1.2. Voiceless Labial Stop: /p/ = φ
In the Secunda, both syllable-initial and post-vocalic /p/ are represented with φ:
φανη /pnē / [phaneː] 'the face of' Ps. 18:43

θεσφιλ* /teš pı̄ l/ [t ɪʃp iːl]
h h
'you bring down' Ps. 18:28
αφαχθ /hɔ̄pakt/ [hɔːɸaχt ] h
'you turned' Ps. 30:12
χοφρω /koprō / [k ʊɸʀoː]
h
'his ransom' Ps. 49:8
At the time of the Secunda, Greek φ represented an aspirated voiceless bilabial stop [ph]. It
was distinguished from Greek π, which represented an unaspirated voiceless bilabial stop [p]
(4.5.3.1.24). It was not until the Byzantine period that φ represented a labio-dental fricative
[f]. While the grapheme φ is used consistently to transcribe the phoneme /p/ in Semitic lan-
guages, π is used to transcribe the phoneme /p/ in Latin (5.3.6). This is best explained by
positing an unaspirated realization of Latin /p/ ([p]) and an aspirated realization of Semitic /p/
([ph]), which would have corresponded perfectly with φ. In Greek loanwords in Mishnaic He-
brew, both φ and π are rendered with Hebrew ‫פ‬, though the unaspirated π is sometimes ren-
205. Jerome actually distinguishes ‫ ֿב‬from consonantal ‫ו‬: e.g., aven ‫ אָוֶן‬but azuba ‫( עֲזוּבָה‬SIEGFRIED 1884, 39).
206. The bilabial fricative pronunciation is not merely a post-vocalic allophone, but a regular realization of ‫ב‬.
- 187 -
dered by Hebrew ‫ב‬, which probably reflects that lack of aspiration of Hebrew /b/ was priori-
tized over the lack of voicing yet presence of aspiration in Hebrew /p/ (5.4.1.3.1–3).
If Hebrew /p/ had a post-vocalic spirantized allophone, it would likely be transcribed
by φ. Latin, Punic, and Arabic /f/ are represented by φ in Greek (5.3.1.3.5; 5.3.3.3.1; 5.3.5.3).
In addition to exhibiting a perfect correspondence with Hebrew [ph], Greek φ [ph] would have
been regarded as the closest approximation of a fricative allophone of /p/ ([ɸ] or [f]). In an
Aramaic inscription from first-century BCE/CE Jerualem, the diphthong ευ is rendered by ‫ אפ‬in
the name ‫( אפטלמוס‬for Εὐπτόλεµος) (CIIP I/1, no. 407). During the Roman period, the diph-
thong ευ developed from [εw] > [εɸ] > [εf] (before voiceless consonants). Therefore, the ren-
dereding with ‫ אפ‬is best explained by assuming that ‫ ֿפ‬was near to [ɸ] or [f] in its realization.
There is also evidence for a post-vocalic spirantized allophone of /p/ in contemporary
Hebrew evidence. Interchanges of ‫פ‬/‫ ב‬are attested in Mishnaic Hebrew (BREUER 2013, 111;
BAR-ASHER 2015, 61–62) and in the Judaean Desert texts (MOR 2015, 121–22). Further, there
is at least one potential interchange of ‫ פ‬and ‫ ו‬at Qumran (REYMOND 2014, 70–71).
The contemporary Greek and Hebrew evidence is consistent with positing a syllable-
initial realization of /p/ as an aspirated voiceless bilabial plosive [ph] and a post-vocalic real-
ization of /p/ as a fricative [ɸ] or [f] in the Secunda. On the basis of similar reasoning for
positing a bilabial rather than labio-dental realization for ‫ ֿב‬, it is worth considering whether or
not ‫ ֿפ‬may also have been realized as a bilabial fricative [ɸ] before it was realized as a labio-
dental fricative [f] as in Tiberian. This may also better explain the interchange of ‫ פ‬and ‫ו‬
found at Qumran. It would also explain why ‫ ֿפ‬is sometimes transcribed by Jerome as ph (e.g.,
basupha ‫בְּסוּפָה‬, hasephataim ‫שּׂפָתַ י ִם‬

ְ ‫ ) ַה‬and sometimes (less frequently) by f (e.g., afar ‫ ָעפָר‬,
Efron ‫( ) ֶעפְרוֹן‬SIEGFRIED 1884, 36, 38, 40, 63–64). A labio-dental fricative [f] could have been
adequately transcribed by f in Latin.207 In the Egyptian papyri, there are a few instances in
207. However, BARR makes the point that speakers conceive of their language phonemically and thus it is no
wonder that Jerome would transcribe both a syllable-initial and post-vocalic realization of the phoneme /p/ in
- 188 -
which Greek φ was transcribed by Latin f (e.g., egraf[e] for ἐγράφη) (GIGNAC 1976, 100),
perhaps after φ [ph] had shifted to [f]. Therefore, in phonetic transcription of the Secunda,
post-vocalic /p/ will be represented as a bilabial fricative [ɸ], with the understanding that it
was in the process of becoming or had already become [f] as in Tiberian Hebrew. In both
Tunisian and Moroccan reading traditions of Hebrew, ‫ פ‬may be realized as a voiceless bilabi-
al fricative [ɸ] post-vocalically (HENSHKE 2013b, 861; AKUN 2013, 704). In the Neo-Aramaic
dialect of the Jews of Arbel, post-vocalic /p/ is also sometimes realized as [ɸ] (KHAN 1999,
18).
6.3.1.3. Voiced Dental Stop: /d/ = δ
In the Secunda, both syllable-initial and post-vocalic /d/ are represented with δ:
δαµµου /dammū / [damːuː] 'they were silent' Ps. 35:15

**µαγδιλιµ /ham-magdı̄ lı̄ m/ [hamːaʁdiːlıː̃ (m)] 'those who exult' Ps. 35:26
χαβωδ /kɔ̄bō d/ [khɔːβoːð] 'honor' Ps. 29:1
ιδαθι /h ̣ı̄ dɔ̄tı̄ / [ħiːðɔːθiː] 'my riddle' Ps. 49:5
In Palestinian Koine during the Roman period, Greek δ most likely represented a voiced
alveodental stop [d], but it is also possible that it had shifted to a fricative [ð] (4.5.3.1.31;
4.5.4.4). At some point during the Roman or Byzantine period, the grapheme δ came to repre-
sent a voiced interdental fricative [ð], but such a realization may not have obtained at the time
of the Secunda. After δ had come to represent [ð] in Egypt, it occasionally interchanged with
ζ (e.g., ζειαβαλεῖν for διαβαλεῖν). Following a nasal, the grapheme (ν)τ also represented the
value [d] (4.5.3.1.26), which was even implemented to transcribe the sequence nd in Latin
(e.g., ουντε ŭ ndĕ ) (5.3.1.3.2). Both Latin /d/ and Semitic /d/ are regularly transcribed by
Greek δ. In Greek loanwords in the Mishnah, δ is rendered by ‫ ד‬in Hebrew (5.4.1.3.2). If
Greek δ had not yet shifted to a fricative by the time of the Secunda, then δ would overlap
perfectly with Hebrew ‫ד‬. If it had shifted to a voiced interdental fricative, voicing was priori-
tized over the manner of articulation in the choice of δ rather than τ to transcribe Hebrew ‫ד‬.
the same way (1967, 9–16).
- 189 -
If we assume that Hebrew /d/ had a post-vocalic fricative realization of [ð], it is
reasonable that it would be represented with δ. Although HORROCKS argues that δ did not be-
come fricativized until the third or fourth century CE (2014, 170), it is possible that the shift
had begun earlier (GIGNAC 1976, 75–76). Regardless, Arabic /ḏ / is transcribed with δ already
in 164 CE at Ḥawrā n (Δουσαρεος /ḏ ū -ś arey/) (5.3.3.3.3). Moreover, Aramaic post-vocalic /d/
is also transcribed by δ (5.3.4.3). If Hebrew post-vocalic /d/ were to be transcribed by Greek
ζ, it would unambiguously attest to a fricative pronunciation. Nevertheless, the regular
transcription of post-vocalic Hebrew /d/ with δ is consistent with a fricative realization of [ð].
A couple phenomena in the Secunda require further explanation with respect to a
post-vocalic fricative realization of /d/. In one instance, post-vocalic /d/ is represented by θ(ι):
αϊεγγιθι /hayeggı̄ d/ [hajɪgːiːθ] 'will it tell?' Ps. 30:10
YUDITSKY corrects this transcription to αϊεγγιδ (2017, 300), but it may be possible to explain
the transcription phonetically. First, final devoicing may have caused [ð] to shift to [θ].208
Second, the sequence of θ [th] + ι [i], which might have indicated something like a paltalized
[tʲ], could have been an attempt to approximate a fricative [ð]. This would be similar to the
instances in the Egyptian papyri in which δι, δ, and ζ interchange (GIGNAC 1976, 75–76).
Also, because the Hebrew voiced stops were likely unaspirated (5.4.1.3.2), the fact that a
voiceless allophone is rendered with Greek aspirated θ instead of unaspirated τ likely points
to a spirantized realization of /d/ (see KNOBLOCH 1995, 175). Mishnaic Hebrew also attests to
occasional interchanges of /d/ and /t/, such as ‫ מות‬for ‫ מאד‬in the phrase ‫( והנה טוב מות‬BAR-ASH-
ER 2015, 183; SHARVIT 2016, 152). According to BAR-ASHER, the interchange of ‫ ד‬and ‫ ת‬is un-
common, but when it occurs, it usually happens in word-final position (2009, 151–52).
The second phenomenon requiring further discussion is the epenthetic vowel inserted
to break up the final consonant cluster /dt#/ in the Secunda:
208. This is attested sporadically in both Greek and Latin transcriptions of Hebrew in other sources: e.g.,
Hebrew ‫( דּוֹאֵג‬Ps. 52:2) is transcribed as Δωηκ in the LXX and as Doec in Jerome.
- 190 -
εεµεδεθ /heʕmedt/ [hɪʕɪmɪðɪθ] 'you established' Ps. 30:8, 31:9
βρεδεθι /b-redtı̄ / [b(a)ʀɪðɪθiː] 'in my going down' Ps. 30:10
These transcriptions seem to argue against a fricative allophone of /d/, since [dt#] would be
more prone to require an epenthetic than [ðt#] (YUDITSKY 2017, 76–77).209 However, accord-
ing to the Obligatory Contour Principle (OCP), homorganic consonants, irrespective of
manner of articulation, are sufficient to occasion an epenthetic (see 6.5.1.5.1; 6.5.2.2). The
fact that a similar phenomenon seems to occur in Babylonian Hebrew, in which post-vocalic
/d/ and /t/ were pronounced as [ð] and [θ], suggests that the presence of an epenthetic be-
tween /d/ and /t/ does not necessarily indicate a plosive pronunciation /d/.210 Rather, the
voiced dental fricative [ð] and voiceless dental plosive [th] may have been regarded as similar
enough phonetically so as to warrant an epenthetic. In fact, even in Tiberian Hebrew, ‫ ֿד‬was
not an interdental fricative but a post-dental fricative, which would have made it more pho-
netically similar to ‫( ת‬KHAN 2013a, 87, 93). Alternatively, the epenthetic vowel may have
been inserted when spirantization was not yet in effect and remained thereafter. This would
also account for the rafeh sign on the ‫ ת‬of the suffix in Babylonian yIqtdgp'ri rippaḏ əṯ ı̄ 'I spread
out' (Job 17:13). This phenomenon is also attested in a Secunda quotation from Epiphanius:211
ιελεδεθεχ* /yledtek(k)/ [jəlɪðəθɪkh] 'I have begotten you' Ps. 110:3
In sum, the contemporary Greek and Hebrew evidence is consistent with positing a
syllable-initial realization of /d/ as a plosive [d] and a post-vocalic realization of /d/ as a frica-
209. YUDITSKY also draws on the Babylonian evidence cited here in support of his claim (2017, 76–77).
210. In the suffix conjugation of Old Babylonian, a shewa sign may be marked on the consonant preceding the
suffixes ‫ תי‬and ‫ ת‬when that consonant is ‫ ד‬or ‫ט‬: e.g., yItd]m'l;w and yItf]q'ç;. Although it is not certain, YEIVIN regards
the distribution of shewa in these instances as possibly indicative of a mobile shewa due to the phonetic
similarity between ‫ט‬/‫ ד‬and ‫ ת‬of the suffix. However, the fact that the rafeh sign appears on the ‫ ת‬of the suffix
after ‫ ד‬in another instance (yIqtdgp'r)i demonstrates that an epenthetic is entirely compatible with a fricative
pronunciation of both ‫ ד‬of the root and ‫ ת‬of the suffix (1985, 427, 515). For a further discussion of the OCP as it
relates to the /dt/ cluster in the Secunda, see 6.5.1.5.1.
211. The form ιελεδεθεχ reflects a 1cs verb in the suffix conjugation from the root ‫( יל"ד‬corresponding with MT
ü‫ )יְלִדְ תִּ י‬against the MT reading of ü‫יַלְדֻּ תֶ י‬. This reading is supported by the ketiv of the MT, the LXX, the
quotation of this verse in the New Testament (Acts 13:33; Hebrews 1:5; 5:5), and possibly also the allusion to
the verse and the reaction of the Sanhedrin in Jesus's trial (see Luke 22:69–70; FLUSSER 1988).
- 191 -
tive [ð] in the Secunda. The epenthetic vowel inserted to resolve /dt#/ consonant clusters in
transcriptions like εεµεδεθ does not necessarily reflect a post-vocalic plosive pronunciation of
/d/ at the time of the Secunda (see above), but may only indicate that fricative /d/ was not an
interdental but more of a post-dental fricative (and thus nearer to alveolar /t/).
6.3.1.4. Voiceless Dental Stop: /t/ = θ
In the Secunda, both syllable-initial and post-vocalic /t/ are represented with θ:
θαµιµ /tɔ̄mı̄ m/ [thɔːmıː̃ (m)] 'blameless' Ps. 18:26

αµαρθι /ʔɔ̄martı̄ / [ʔɔːmaʀthiː] 'I said' Ps. 30:7
αννωθην /han-nō tē n/ [hanːoːθeːn] 'he who gives' Ps. 18:48
βαραθα /bɔ̄rɔ̄tɔ̄/ [bɔːʀɔːθɔː] 'you created' Ps. 89:48
At the time of the Secunda, Greek θ represented the aspirated voiceless alveodental stop [th].
It was distinguished from Greek τ, which represented an unaspirated voiceless alveodental
stop [t] (4.5.3.1.30–32).212 At some point in the Byzantine period, θ came to represent the
voiceless interdental fricative [θ]. While the grapheme θ is used consistently to transcribe the
phoneme /t/ in Semitic languages, τ is used to transcribe the phoneme /t/ in Latin. This is best
explained by positing an unaspirated realization of Latin /t/ ([t]) and an aspirated realization
of Semitic /t/ ([th]), which would have corresponded perfectly with Greek θ. This correspon-
dence is also present in Greek loanwords in Mishnaic Hebrew, where Greek θ is rendered by
Hebrew ‫ ת‬and Greek τ is rendered by ‫ ט‬in Hebrew (5.4.1.3.1; 5.4.1.3.3).
If Secunda Hebrew had a post-vocalic spirantized allophone of /t/, we would expect it
to be transcribed by θ. The Arabic interdental /ṯ / [θ] is usually transcribed by θ, but may be
transcribed by τ on occasion (5.3.3.3.3). Aramaic post-vocalic /t/ is also transcribed by θ
(5.3.4.3). This shows that, in addition to Greek θ [th] corresponding perfectly with Hebrew
syllable-initial /t/, Greek θ also would have been the closest realization of the fricative allo-
phone of /t/ ([θ]). A parallel to Arabic /ṯ / being transcribed with Greek τ may be found in the
transcription Νατανιλου (for ‫ )נתנאל‬from first-century CE Jerusalem (4.5.3.1.30). If post-vo-
212. For an alveodental rather than dental stop in ancient Greek, see PETROUNIAS (2007b, 561).
- 192 -
calic /t/ were a plosive in ‫נתנאל‬, there would be no reason to represent it with anything other
than Greek θ. However, because Greek scribes had no precise equivalent of fricative [θ], they
had to choose between θ [th] and τ [t]. While they generally preferred θ for rendering Hebrew
ֿ‫ת‬, the transcription Νατανιλου may be an example of a scribe preferring τ over θ as an imper-
fect approximation of Hebrew ֿ‫ת‬.213 In Tiberian Hebrew, the fricative allophone of /t/ was real-
ized as an alveolar fricative (KHAN 2013a, 93). There is one instance of post-vocalic ‫ ת‬being
represented with ‫( ס‬before it was corrected to ‫ )ת‬in MS Kaufmann of the Mishnah (‫ בֵּיס הלל‬for
‫)בית הלל‬, but it probably only reflects an Ashkenazi pronunciation (BAR-ASHER 2015, 65, 92).
In sum, while the evidence from the Secunda transcriptions themselves is inconclu-
sive, it is consistent with contemporary Greek and Hebrew evidence to posit a syllable-initial
aspirated realization of /t/ [th] and a post-vocalic fricative allophone of [θ].

6.3.1.5. Voiced Velar Stop: /g/ = γ
In the Secunda, both syllable-initial and post-vocalic /g/ are represented with γ:
µισγαβ /maś gɔ̄b/ [misgɔːβ] 'a fortress' Ps. 46:8

γαµ /gam(m)/ [ga(̃ m)] 'also' Ps. 49:3
µαγεν /mɔ̄gen(n)/ [mɔːʁɪn] 'a shield' Ps. 18:31
εδαλλεγ /ʔdalleg/ [ʔɪðalːɪʁ] 'I will leap' Ps. 18:30
At the time of the Secunda, Greek γ represented a voiced velar fricative [γ] with an allophone
of [j] in the environment of high vowels (4.5.3.1.37). After nasals, it would have been real-
ized as a voiced velar stop [g]. Before another velar (γ, κ, χ), γ represented a nasal ([ŋ] or [n])
(4.5.3.1.38). Both Latin /g/ and Semitic /g/ are consistently transcribed by Greek γ (5.3.6). In
Greek loanwords in the Mishnah, γ is rendered by ‫ ג‬in Hebrew (5.4.1.3.2). Although no
longer a perfect correspondence with Hebrew /g/, γ was regarded as the nearest approxima-
tion of plosive ‫ג‬. The plosive realization seems to be apparent in the following transcription:
βαµεθγε /bɔ̄-metg/ [bɔːmɪθgə] 'with a bit' Ps. 32:9
213. AL-JALLAD explains the Greek transcription of Arabic /t/ and /ṯ / in the same way (2015, 13–14).
- 193 -
An epenthetic vowel is added in βαµεθγε to resolve the final consonant cluster /tg#/ (see YU-
DITSKY 2017, 79). The fact that we find paragogic epenthesis (CC# > CCv#) rather than anap-
tyctic epenthesis (CC# > CvC#) seems to indicate that there was no unrepresented vowel be-
tween /t/ and /g/ and that /g/ was thus realized with a plosive pronunciation (cf. BLAU 1998a,
8–9; see 6.5.2.1.1).
If Hebrew /g/ had a post-vocalic fricative allophone, it likely would have been real-
ized somewhere between a velar [γ] and a uvular [ʁ] as in Tiberian (KHAN 2013a, 86–87).
Greek γ, representing a velar fricative [γ], may have corresponded better with either of these
phones than it would with a plosive /g/. This is the case with transcribing Arabic /ġ / ([γ̞] or
[ɰ]), the Semitic consonant nearest to ֿ‫ג‬, which is rendered by Greek γ (5.3.3.3.4). At an earli-
er period, Hebrew /ġ / was also transcribed by γ (e.g., Γοµορρα ‫ֲמוֹרה‬

ָ ‫ ע‬in the LXX). Evidence
for a uvular post-vocalic /g/ in the Secunda may be present in the following transcription:
θαµωγ /tɔ̄mū g/ [thɔːmoːʁ] '(it) will melt' Ps. 46:7
An expected long /ū / vowel is realized as [oː]. While a morphological explanation is entirely
possible (see YUDITSKY 2017, 60, 137), there may also be a phonetic explanation. In Akkadi-
an, both /r/ and /ḫ / have a lowering effect on adjacent vowels, as is demonstrated by the Grae-
co-Babyloniaca texts (5.3.2.1.3–4). This may be because both /r/ and /ḫ / had a similar place
of aritculation, namely, uvular.214 Therefore, if the explanation of the transcription θαµωγ is
actually phonetic rather than morphological, the lowering of the vowel before /g/ may be evi-
dence of a post-vocalic spirantized allophone of /g/ realized as a voiced uvular fricative [ʁ].
There may also be evidence of a spirantized ֿ‫ ג‬in contemporary Hebrew epigraphy.
The name of the Roman emperor Trajan, Trajanus ([trajanus]) in Latin, is rendered as ‫ טרינס‬in
214. Akkadian /ḫ / is a reflex of what may be reconstructed as either a voiceless uvular fricative [χ] or a
voiceless velar fricative [x] in Proto-Semitic (HUEHNERGARD 2004, 142; KOGAN 2011, 54). Although the reflexes
of PS /ḫ / in various Semitic languages vary between the velar fricative [x] and the uvular fricative [χ], the fact
that /ḫ / eventually merges with the pharyngeal fricative /h ̣/ [ħ] in Akkadian (see STEINER 2011) seems to indicate
that Akkadian /ḫ / may have been a uvular fricative [χ] before the merger. This would better explain why /ḫ / and
/r/, but not /k/ and /g/, lower vowels in Akkadian. Accordingly, /r/ was likely a uvular roll or uvular trill.
- 194 -
the Judaean Desert texts (5/6Hev 8), as ‫ טרוגינוס‬in the Mekhilta DeRabbi Shimon Bar Yochai
(21:13) and as ‫ טרוגיינוס‬in the Jerusalem Talmud (Sukkah 23a). Τραιανος is typical in Greek
orthography, but spellings with γ such as Τραγιανου are not uncommon, especially in inscrip-
tions (GIGNAC 1976, 72). While it is possible that ‫טרוגיינוס‬/‫ טרוגינוס‬is a slavishly literal translit-
eration of a variant Greek spelling, the spelling Τραιανος is the norm in Palestine.215 If the
Hebrew spelling ‫טרוגיינוס‬/‫ טרוגינוס‬is not a literal transliteration, it is likely evidence of a spi-
rantized /g/, just as the same variant spelling is evidence of such in Greek. Finally, the omis-
sion of ‫ ר‬in the word ‫( מגש‬for ‫ )מגרש‬at Qumran (QIMRON 1986, 26–27) may be explained by
positing a similar realization of both ֿ‫ ג‬and ‫ר‬: i.e., /magraš / [maʁʀaʃ] > [maʁ(ː)a(ː)ʃ].
In sum, while the evidence is inconclusive, it is consistent with both contemporary
Greek evidence and Hebrew dialectology to posit that Hebrew /g/ was realized as [g] sylla-
ble-initially and as [ʁ] (or [γ]) post-vocalically in the Secunda. Because of the possible lower-
ing of the vowel in θαµωγ, post-vocalic /g/ will be transcribed as a uvular [ʁ], but it may have
been realized as a velar [γ] if the transcription θαµωγ is better explained morphologically.
6.3.1.6. Voiceless Velar Stop: /k/ = χ
In the Secunda, both syllable-initial and post-vocalic /k/ are represented with χ:
χι /kı̄ / [khiː] 'because' Ps. 18:28

οσχι /h ̣oš kı̄ / [ħʊʃkhiː] 'my darkness' Ps. 18:29
βαχ /bɔ̄k/ [bɔːχ] 'in you' Ps. 18:30
νηχιµ /nē kı̄ m/ [neːχıː̃ (m)] 'wretches' Ps. 35:15
At the time of the Secunda, Greek χ represented the aspirated voiceless velar stop [kh]. It was
distinguished from Greek κ, which represented the unaspirated voiceless velar stop [k]
(4.5.3.1.35). At some point in the Byzantine period, χ came to represent the voiceless velar
fricative [x]. While the grapheme χ is used consistently to transcribe the phoneme /k/ in Se-
mitic languages, κ is used to transcribe the phoneme /k/ in Latin. This is best explained by
215. In the Greek Judaean Desert texts, Τραιανος is spelled normally (without a γ) in all fourteen of its
occurrences. Also, in an inscription from Ashkelon the same name is spelled Τραιανε (CIIP 2395).
- 195 -
positing an unaspirated realization of Latin /k/ ([k]) and an aspirated realization of Semitic /k/
([kh]), which would have corresponded perfectly with Greek χ. This correspondence is also
present in Greek loanwords in the Mishnah, where Greek χ is rendered by Hebrew ‫ כ‬and
Greek κ is rendered by ‫ ק‬in Hebrew (5.4.1.3.1; 5.4.1.3.3).
If we assume a post-vocalic fricative allophone of Hebrew /k/, it likely would have
been realized somewhere between a velar [x] and a uvular [χ] as in Tiberian (KHAN 2013a,
89–90).216 Greek χ, representing an aspirated velar stop [kh], would have been the nearest ap-
proximation of this phone. The nearest Semitic consonant to [x]/[χ] is /ḫ /. Akkadian /ḫ / is
transcribed by ξ [k(h)s],217 perhaps because of the combination of the voiceless velar [kh] and
the fricative [s] (5.3.2.3.4). Arabic /ḫ / is usually transcribed by χ (5.3.3.3.4). In each case,
however, /ḫ / may also be transcribed by Ø, perhaps reflecting a shift to /h ̣/. Nevertheless, the
data indicates that Greek χ best approximated Semitic /ḫ /.
The following form may also be relevant for the discussion of a potential post-vocalic
spirantized allophone of /k/ in the Secunda:
λαµαλχη /l-malkē / [lamalχeː] 'to the kings of' Ps. 89:28
In Tiberian Hebrew, the ‫ כ‬in this word is rafeh (i.e., spirantized) even though it is syllable-ini-
tial. A rafeh ‫ בג"ד כפ"ת‬letter following a post-vocalic shewa, referred to as shewa medium in
the literature, is generally explained as the result of spirantization operating at a time when
the ‫ בג"ד כפ"ת‬letter was still post-vocalic: *la-malakay > (contraction of diphthongs) > *la-
malakē > (spirantization) > *la-malaḵ ē > (syncope) > l-malḵ ē (see KHAN 2005, 86–87).218 If
spirantization had operated after vowel syncope in l-malkē , the ‫ כ‬would not have been spiran-
tized since it would not have been post-vocalic. Babylonian Hebrew exhibits the same appar-
216. See note 8 for the description of the place of articulation of spirantized ‫ כ‬in Hidā yat al-Qā riʔ (ELDAR 1994,
59–61; KHAN 2013a, 89–90).
217. There is evidence that ξ represented [k(h)s] in Palestinian Koine (5.4.1.3.6), but it may have represented [ks]
in Mesopotamian Greek.
218. This is presumably the explanation for the regular plural in Biblical Aramaic ‫( ַמ ְלכִין‬Dan. 2:21) as well:
*malakı̄ n > *malaḵ ı̄ n > malḵ ı̄ n.
- 196 -
ent rule ordering, though there are a small number of exceptional forms in which the third
radical has a dagesh in a comparable pattern (e.g., µk'yEKlm# and µh'yEKrd') (YEIVIN 1985, 342).219
Because such rule ordering, namely, that vowel syncope in the form *qatalē > *qatlē
operated after the spirantization of post-vocalic /k/, is evidenced in both Tiberian and Baby-
lonian, an argument can be made that both developments had already occurred in a common
Hebrew ancestor. The Secunda form λαµαλχη with syncope may indicate that such a common
ancestor, in which spirantization and syncope had already operated, predates the composition
of the Secunda. However, it should be noted that such a line of argumentation is entirely de-
pendent on how closely Secunda Hebrew is related to Tiberian and/or Babylonian Hebrew,
which is by no means clear.220 Unfortunately, a comprehensive treatment of this topic lies be-
yond the scope of the present work. Finally, it should also be noted that because Greek χ
could represent either [kh] or [x]/[χ], the form λαµαλχη cannot be taken as direct evidence of
either pronunciation.
On the other hand, transcriptions of the 2ms verbal object suffix may reflect a non-
fricative realization of post-vocalic /k/. Normally, it is represented by -εχ:
ιζαµµερεχ /yzammerek(k)/ [(ʔ)i(ː)zammɪʀɪkh] 'I will sing to you' Ps. 30:13

ωδεχ /ʔō dek(k)/ [ʔoːðɪkh] 'I will praise you' Ps. 35:18
In one instance, however, this 2ms verbal suffix is rendered with a κ (-εκ) rather than a χ:
ουωρεκ /w-ʔō rek(k)/ [(ʔ)uʔoːʀɪkh] 'and I will show you' Ps. 32:8
While scholars have explained the -εχ suffix in various ways, it is best regarded as a histori-
cal development of what is a pausal form in Tiberian Hebrew: -inka > ikka > ikk > ik(k) >
ek(k) (e.g., ‫[ י ַ ֶכּכָּה‬Isa. 10:24]). The ε (and not η) in the suffix points to etymological final gem-
219. YEIVIN suggests that dissimilation with the 2mp suffix may be a factor (1985, 342).
220. Note even the apparent dialectal variation in Jerome. In his commentary on Isaiah, Jerome states: praeter
unam litteram aleph, quae in angelorum vocabulo addita est, eodem reges et angeli apud hebraeos appellantur
nomine, id est malache 'except for one letter aleph, which is added in the noun "messengers," kings and
messengers are called by the same name among the Hebrews, that is malache'. The form malache for ‫ ַמ ְלכֵי‬may
indicate a lack of syncope (see KHAN 2013h, 551; YUDITSKY 2013, 818), but Jerome might also have confused a
text-critical issue with a phonological one.
- 197 -
ination (6.3.8.6), which is further supported by the one instance in which the final vowel is
preserved: αϊωδεχχα /ha-yō dekkɔ̄/ [hajoːðɪkhːɔː] (Ps. 30:10).
If κ for χ in ουωρεκ is not a scribal error—these two letters were written similarly
throughout the history of Greek (THOMPSON 1966, 154–55)—then the spelling with κ seems to
reflect both a non-fricative and non-aspirated realization of post-vocalic ‫ כ‬in this word. The
non-fricative realization of ‫ כ‬may be explained in a few ways. First, it is possible that spi-
rantization may not (yet) have applied to /k/ as in Samaritan Hebrew. Second, if post-vocalic
spirantization did normally apply, it is possible that ‫ כ‬was defricativized in word-final posi-
tion. In fact, despirantization of a fricative in syllable-final position is common cross-linguis-
tically (HONEYBONE and SALMONS 2015, 418–19). Third, etymological final gemination may
still have been preserved (at least on a phonological level) when the spirantization rule ceased
to operate. Accordingly, the plosive realization of the consonant was maintained. Note also
the word ְ‫[ אַתּ‬ʔaːt] (not **[ʔaːθ]) in Tiberian Hebrew (see 6.3.8.6).221 The non-aspirated real-
ization of ‫כ‬, then, according to YUDITSKY, is explained as the loss of aspiration in word-final
position, which is common cross-linguistically. An interchange of ‫ק‬/‫ כ‬is also attested in the
Secunda transcription λακταλ ‫ טַל‬üְ‫( ל‬Ps. 110:3) quoted in Epiphanius (2017, 22, 25, 104–
106).
This example presents no problem in light of Greek evidence. In Rabbi Abbahu's
clever Greek pun, Greek ὀκτώ is transcribed as ‫( אוכט"א‬Bereshit Rabbah 14:2), indicating that
the Greek sequence κτ was equivalent to ‫ כט‬and not ‫קט‬, ὀκτώ being pronounced as either
[ʔokhto] or [ʔoxto] (BUBENIK 2007, 633). In Medieval Greek, voiceless obstruent clusters
comprised of either [stop] + [stop] or [fricative] + [fricative] underwent dissimilation to be-
come [fricative] + [stop]: e.g., κτίζω [ˈktizo] 'I build' becomes χτίζω [ˈxtizo] (HORROCKS 2014,
281–82). A similar realization of κτ is found in Modern Greek. While the reflex of Classical
221. For a similar phenomenon with respect to spirantization in Syriac, see EDZARD (2001).
- 198 -
Greek κτ is pronounced as [kt] in the logio ("learned") pronunciation, it is pronounced as [xt]
in the laiko ("popular") pronunciation, resulting in persistent doublets: e.g., logio κτίζω [ˈkti-
zo] 'I build' but laiko χτίζω [ˈxtizo]; logio κτίσιµο [ˈktisimo] 'a building' but laiko χτίσιµο [ˈx-
tisimo]; logio κτυπώ [ktiˈpo] 'I knock' but laiko χτυπώ [xtiˈpo]; logio κτύπος [ˈktipos] 'a
knock' but laiko χτύπος [ˈxtipos] (PARADIA and MITSIS 2013, 381).
In Mishnaic Hebrew, ‫ק‬/‫ כ‬interchanges are common in the environment of ‫( פ‬e.g.,
‫ אסכופא‬for ‫ <[ אסקופה‬σκοπος]), the liquid ‫( ר‬e.g., ‫ לבכר‬for ‫)לבקר‬, a guttural consonant (e.g.,
‫ עקבת‬for ‫)עכבת‬, and an emphatic consonant (e.g., ‫ טקסיס‬for ‫ <[ טכסיס‬τάξις] and ‫ לכתיקה‬for
‫ <[ לקטיקא‬lectica]) (HENSHKE 2010, 430; SHARVIT 2016, 116, 118–20, 125–26, 133, 137–38).
Curiously, these environments apply to both transcription variants: in ουωρεκ, χ > κ is in the
environment of /r/,222 and in λακταλ, χ > κ is in the environment of an emphatic. It is not
entirely clear why these environments might condition ‫ כ > ק‬or ‫ק > כ‬, but it may be due to the
fact that the distinction between emphatic and non-emphatic can become blurred in the envi-
ronment of consonants associated with rounding or emphasis (e.g., labials, /r/, pharyngeals).
In sum, the evidence regarding a potential spirantized pronunciation of /k/ at the time
of the Secunda is inconclusive. One may argue, depending on one's view of the relationship
between Secunda Hebrew, Tiberian Hebrew, and Babylonian Hebrew, that the transcription
λαµαλχη indicates that the post-vocalic spirantization of /k/ had already occurred in a com-
mon Hebrew ancestor. On the other hand, the transcription ουωρεκ seems to reflect a non-
fricative realization of post-vocalic /k/. The variant spelling (χ > κ), however, is susceptible to
phonetic explanations whether a spirantized allophone of /k/ is posited for the Secunda or not.
Accordingly, it is consistent with contemporary Greek and Hebrew evidence to posit a real-
ization of syllable-initial /k/ as [kh] and post-vocalic /k/ as [x] or [χ] in the Secunda.223
222. See also χορσελαϊ, in which κ > χ before ρ (see 6.3.2).
223. It may be preferable to posit [χ] since a uvular realization is more likely for post-vocalic /g/ due to the
apparent lowering of vowels in θαµωγ (* < tā mū g).
- 199 -
It should be noted, first of all, that there is not enough evidence to determine with certainty
whether or not the ‫ בג"ד כפ"ת‬consonants had post-vocalic spirantized allophones in Secunda
Hebrew. Nevertheless, assuming that the Hebrew reading tradition reflected in the transcrip-
tions is not an outlier in the trends of the history of the language, we may conclude that the
evidence of the Secunda transcriptions and contemporary Hebrew is consistent with positing
post-vocalic fricative allophones for the Hebrew stops.
Another important point that has emerged from this discussion is that fricativization
did not necessarily change the place of articulation immediately. For example, spirantized ‫ב‬
was likely first realized as a bilabial fricative [β] before it shifted to a labio-dental fricative
[v]. Similarly, /d/ and /t/ probably developed fricative allophones in the same places of articu-
lation before they shifted to interdentals in modern reading traditions. Even as late as Tiberian
Hebrew, ‫ ֿד‬and ֿ‫ ת‬have the same places of articulation as their plosive counter parts ‫ ד‬and ‫ת‬, re-
spectively (KHAN 2013a, 87, 93).

6.3.2. Sibilants (/s/, /š /, /ś /, /z/)
Standard Biblical Hebrew originally distinguished three non-emphatic sibilants and a lateral
fricative: a voiceless sibilant fricative /s/ [s] = ‫ס‬,224 a voiced sibilant fricative /z/ [z] = ‫ז‬,225 a
voiceless palato-alveolar sibilant /š / [ʃ] = ‫שׁ‬,226 and a voiceless lateral fricative /ś / [ɬ] = ‫שׂ‬. In-
terchanges of ‫ס‬/‫ שׂ‬in Late Biblical Hebrew, the Judaean Desert texts, and Mishnaic Hebrew
show that /s/ and /ś / had merged to [s] by the Second Temple period in most Hebrew tradi-
tions (RENDSBURG 2013a, 102; MOR 2015, 97–105; SHARVIT 2016, 181). The resulting conso-
nant-phoneme correspondences (/s/ = ‫ שׂ‬,‫ ;ס‬/z/ = ‫ ;ז‬/š / = ‫ )שׁ‬are as in Tiberian. Unlike the other
224. The sibilants /s/ (= ‫ )ס‬and /z/ (= ‫ )ז‬are reflexes of the original Proto-Semitic affricates /ʦ/ and /ʣ/.
However, these affricates were likely simplified to the sibilant fricatives [s] and [z] in ancient Hebrew.
225. See previous note.
226. Hebrew /š / is the result of the shift of */s/ [s] > /š / [ʃ] and the merger of */ṯ / [θ], /š / [ʃ] > /š / [ʃ]. While other
Hebrew dialects likely maintained /ṯ /, in Standard Biblical Hebrew /ṯ / > /š / (RENDSBURG 2013a, 102).
- 200 -
reading traditions, Samaritan Hebrew /ś / [ɬ] shifted to /š / [ʃ] rather than /s/ [s] (FLORENTIN
2013, 446).
6.3.2.1. Voiceless Sibilant Fricatives: /s/, /ś /, /š / = σ
In the Secunda, each of the voiceless sibilant fricatives is represented with σ:
αωσιµ /ha(h ̣)-h ̣ō sı̄ m/ [haħoːsıː̃ (m)] 'who take refuge' Ps. 18:31
βσεθρ /b-setr/ [b(ɪ)sɪθʀ] 'in the hiding place of' Ps. 31:21
σαµου /ś ɔ̄mh ̣ū / [sɔːmħuː] 'they rejoiced' Ps. 35:15
σαµ /ś ɔ̄m/ [sɔ̃ː(m)] 'is placing' Ps. 46:9
µωσϊ /mō š ı̄ ʕ/ [moːʃiːʕ] 'savior' Ps. 18:42
λαµασαλ /l-mɔ̄š ɔ̄l/ [lamɔːʃɔːl] 'to a proverb' Ps. 49:5
In the Koine Greek of Roman period Palestine, the grapheme σ represented [s]. Before a
voiced consonant, it was prone to represent a voiced allophone [z], sometimes represented
with ζ (4.5.3.1.34). It is presumed that, just as in Modern Greek, ancient Greek /s/ was pro-
nounced with the tip or dorsum of the tongue nearing the alveolar ridge but not closing com-
pletely so that air could pass through (PETROUNIAS 2007b, 562–63). In IPA terms, this may be
regarded as a voiceless laminal sibilant [s̻ ] or a voiceless apico-alveolar sibilant [s̺ ]. It should
be noted that this sound approaches the voiceless palato-alveolar sibilant [ʃ]. Accordingly, be-
cause Greek had no other voiceless sibilants, σ was the most appropriate grapheme to render
both [s] and [ʃ]. In transcription, Latin /s/ and Semitic /s/ and /š / are represented by σ (5.3.6).
Voiced allophones of /s/ are occasionally transcribed by ζ. In Greek loanwords in the Mish-
nah, Greek σ is normally rendered by Hebrew ‫ס‬, but its voiced allophone may be rendered by
‫ ז‬in Hebrew (5.4.3.1.5). There is at least one example of the voicing of /s/ in the Secunda (see
YUDITSKY 2017, 81):
βεεζδαχ /b-h ̣esdɔ̄k/ [bɪħɪzdɔːχ] 'in your mercy' Ps. 31:8
Presumbly, this indicates that Hebrew /s/ tended to assimilate in voice to the following conso-
nant and that Hebrew /s/ and /z/ had identical places of articulation, differing only in voice.
A peculiar characteristic of sibilants in the Secunda is that they seem to bring about
the raising (and perhaps fronting) of adjacent vowels (YUDITSKY 2017, 92–95):
- 201 -
e > i / _C[+sibilant]:
σιµου /š mʕū / > /š emʕū / [ʃimʕuː] 'listen!' Ps. 49:2
νισβαθ /neš baʕt/ [niʃbaʕth] 'you swore' Ps. 89:50
a > e / _C[+sibilant]:
βεσαυει227 /b-š aw(w)ʕı̄ / [bɪʃawʕiː] 'when I cried' Ps. 31:23

λαµεσφατι /l-maš pɔ̄tı̄ ̣/ [lamɪʃphɔːtˀiː] 'to my judgment' Ps. 35:23
a > i / _C[+sibilant]:
χισους228 /k-sū s/ [khisuːs] 'like a horse' Ps. 32:9

µισγαβ /maś gɔ̄b/ [misgɔːβ] 'a fortress' Ps. 46:8
Different patterns and morphological variants have been invoked to explain some of these
forms, but the evidence for vowel raising in the environment of a sibilant is compelling (YU-
DITSKY 2017, 92–95). Vowel raising near a sibilant is also attested in Jerome (e.g., messa ‫ַמשָּׂא‬
[HARVIAINEN 1977, 169]), Mishnaic Hebrew, and even Tiberian Hebrew (e.g., ‫( ) ְוהִתְ קַדִּ שְׁתֶּ ם‬92).
In the Neo-Aramaic dialect of Barwar, the vowel /ə/ is realized especially high when fol-
lowed by a sibilant: e.g., /məš xa/ [mɪʃxæ] ~ [miʃxæ] 'oil' (KHAN 2008b, 77).
Τhis phenomenon does not necessarily constitute a sound change. In a linguistic-per-
ceptual study of fricative-vowel coarticulation, YENI-KOMSHIAN and SOLI have shown that
high vowels are more easily identifiable in the environment of [s] and [ʃ].229 Because of cer-
tain coarticulatory qualities of a fricative sibilant, which associate it with qualities of high
vowels (e.g., similar degree of sonority), [a] is often misidentified as a high vowel near [s]
and [ʃ] (1981).230 Another likely factor is the lack of voice of the sibilant. It is possible, then,
that these transcriptions do not reflect an actual sound change but merely a linguistic-percep-
tual phenomenon (e.g., µισβιθ /maš bı̄ t/ [miʃbiːθ], but cf. **µαγδιλιµ /ham-magdı̄ lı̄ m/
227. Note that the normal vowel on the preposition /b-/ is [a]: βα- (see YUDITSKY 2017, 224–29).
228. Note that the normal vowel on the preposition /k-/ is [a]: χα- (see YUDITSKY 2017, 224–29).
229. With the exception of the sequence [ʃi], for reasons given in YENI-KOMSHIAN and SOLI (1981, 974).
230. The relationship between high vowels and the palato-alveolar sibilant is also illustrated by the Greek
transcriptions Μωυσῆς (for ‫ )מֹשֶׁה‬and συµα (‫שׁמַע‬ְ ) (see BAR-ASHER 2010), in which the adjacent high vowel υ is
likely added to approximate the palato-alveolar realization of ‫שׁ‬.
- 202 -
[hamːaʁdiːliːm]). The perceptual element becomes all the more significant when we consider
that the same phenomenon occurs in Greek. In the environment of σ, raised allophones of
vowels may occur (see 4.5.3.1.5–8; 4.5.3.1.12). This may demonstrate that the place and
manner of articulation of Greek and Hebrew sibilants were similar.
Sibilants may also cause the deletion of adjacent vowels (YUDITSKY 2017, 61–62):
σφτηνι* /š ptẹ̄ nı̄ ?/ [ʃəɸtˀeːniː] 'judge me!' Ps. 35:24

µσ῾χνη /maš knē / [miʃkhneː] 'the dwellings of' Ps. 46:5
φσαµ /peš ʕam(m)/ [phɪʃʕa(̃ m)] 'their transgression' Ps. 89:33
σφωθαϊ /ś pō tay(y)/ [sɸoːθaj]/[sɪɸ(ʔ)oːθaj] 'my lips' Ps. 89:35
In Palestinian Koine, vowel deletion is also attested in at least one instance before a conso-
nant cluster beginning with a sibilant (4.5.3.1.20). The nature of sibilants in both Greek and
Hebrew may have been such that they were perceived as inherently vocalic as well. The study
of YENI-KOMSHIAN and SOLI (1981) cited above may support this point. Additionally, in a sep-
arate perceptual study, YENI-KOMSHIAN and SOLI have shown that when a fricative is excised
from a fricative-vowel sequence, the high vowels [i] and [u], but not the low vowel [a] can be
identified in the fricative itself apart from the following vowel (1982). Cross-linguistically,
the duration of sibilants is generally longer than that of other consonants (BLEVINS 2004;
DMITRIEVA 2012, 20). Ιt is consistent cross-linguistically, then, to suggest that the scribe
identified both the sibilant fricative and the accompanying high vowel in the grapheme σ.231
Finally, there may be evidence for lip rounding with /š / in the following transcription:
µοσαυε /mš awwɛ̄/ [mʊʃawːɛː] 'making meet' Ps. 18:34
While the rounding is typically, and rightly, attributed to assimilation to the bilabial /m/ (YU-
DITSKY 2017, 91), it is curious that such rounding is not attested in other piˁel participles (e.g.,
µαλαµµεδ, λαµανασση, etc.). A few transcriptions of names in the LXX may actually indicate
that such rounding was partly due to the /š /: e.g., µοσφαθαιµ ‫שׁפְּתַ י ִם‬
ְ ‫( ִמ‬Judg. 5:16), Μοσωβαβ
231. YUDITSKY suggests that short /i/ or /e/ was regarded as part of the frication (2017, 62). A similar example of
this may be found in the Leningrad Codex of Psalm 45:3: ü‫שׂ ְפֿתֿ וֹתֶ י‬
ְ ‫( ְבּ‬cf. Aleppo ü‫שׂ ְפֿתֿ וֹתֶ י‬
ִ ‫( ) ְבּ‬BLAPP 2016).
- 203 -
(I Chr. 4:34), and Συχεµ ‫שׁכֶם‬
ְ (Gen. 12:6) (see also KNOBLOCH 1995, 459–60). If such a distrib-
ution is not merely a statistical coincidence, it may indicate that Hebrew /š / was pronounced
with some coarticulatory lip rounding similar to the English or French pronunciation of /ʃ/
(see LADEFOGED and MADDIESON 1996, 148). On the other hand, [a] > [o] may be facilitated by
regressive assimilation due to the following /w/.
In sum, the representation of /s/ and /ś / by σ (and the voiced allophone of /s/ by ζ)
points to a voiceless fricative sibilant [s] realization of /s/ and /ś / in the Secunda. The place of
articulation was probably alveolar as in Tiberian (KHAN 2013a, 90, 93). It is theoretically pos-
sible, though less likely,232 that /ś / merged with /š / in Secunda Hebrew as it did in Samaritan
Hebrew. The representation of /š / by σ in the Secunda is consistent with its expected realiza-
tion as a palato-alveolar fricative [ʃ]. The raising of vowels in the environment of the sibilants
evidenced in the Secunda is attested both in contemporary Hebrew and Greek. Modern lin-
guistic study suggests that this may be a phonetic-perceptual phenomenon.

6.3.2.2. Voiced Sibilant Fricative: /z/ = ζ
In the Secunda, Hebrew /z/ is regularly represented with Greek ζ:
µαζµωρ /mazmō r/ [mazmoːʀ] 'a psalm' Ps. 29:1

οζνι /ʔoznı̄ / [ʔʊzniː] 'my ear' Ps. 49:5
ζαναθ /zɔ̄nah ̣t/ [zɔːnaħth] 'you rejected' Ps. 89:39
In Palestinian Koine, the grapheme ζ represented [z]. Though it had represented [sd]/[zd] at
an earlier stage in its history, by the Roman period it had already shifted from [zd] > [zz] >
[z] (4.5.3.1.34). Before a voiceless consonant, it was prone to represent a voiceless allophone
[s], sometimes represented by σ (4.5.3.1.34). Greek /z/ probably had the same place of articu-
lation as /s/, namely, the tip or dorsum of the tongue nearing the alveolar ridge but not closing
completely so that air can pass through (PETROUNIAS 2007b, 562–63). In IPA terms, this may
be described as a voiced laminal sibilant [z̻ ] or voiced apico-alveolar sibilant [z̺ ]. In transcrip-
232. Note that in a Jerusalem inscription from the Second Temple period the Hebrew name ‫ שׂ ָָרה‬is written with a
‫ס‬, thus indicating the merger of /ś / and /s/ and not /ś / and /š /: ‫( סרה‬CIIP I/1, no. 201).
- 204 -
tion, Semitic /z/ is represented by Greek ζ (5.3.6). In Greek loanwords in the Mishnah, ζ is
rendered by ‫ ז‬in Hebrew (5.4.1.3.5). The Secunda attests to one instance of devoicing of ‫( ז‬see
YUDITSKY 2017, 81):
νεγρεσθι /negraztı̄ / [nɪʁʀæsthiː] 'I was cut off' Ps. 31:23
Presumbly, this indicates that Hebrew /z/ tended to assimilate in voice to the following conso-
nant and that Hebrew /z/ and /s/ had identical places of articulation, differing only in voice.
On the other hand, an interchange of ζ and σ in a Secunda quotation in Eusebius, µασµωρα
‫( ִמזְמוֹר‬Ps. 9:1), is explained on the basis of Greek orthography. Because Greek σ before a
voiced consonant was pronounced as [z] (e.g., κοσµος [ˈkozmos]) (4.5.3.1.34), the σ instead
of ζ in µασµωρα is simply a Greek orthographic variant for representing [z].
The aforementioned transcription also provides a potential example of assimilation of
the vowel to the sibilant (νεγρεσθι ‫נִג ְַרזְתִּ י‬, but cf. νεµσαλ ‫( )נִ ְמשַׁל‬see above 6.3.2.1). It is also
probable that the vowel has been raised in the environment of the sibilant /z/ in the following
transcriptions (cf. YUDITSKY 2017, 94–95):
ουναζερθι /w-naʕzartı̄ / [(ʔ)unaʕzæʀthiː] 'and I was helped' Ps. 28:7

ουαϊαλεζ /way-yaʕloz/ [wajːaʕləz]/[wajːaʕlɪz] 'and it rejoiced' Ps. 28:7
ϊεζεβου /yeʕzbū / [jɪʕzəβuː] 'they will abandon' Ps. 89:31
The voiced sibilant ζ does not bring about vowel raising in Greek. Therefore, the fact that
vowel raising may occur in the environment of /z/ in Hebrew further supports the likelihood
that this is a feature of Hebrew and not merely an element of the Greek accent of the scribe.
Vowel raising in the environment of /z/ is also attested in Jerome (HARVIAINEN 1977, 62).
In sum, the consistent representation of /z/ by ζ in the Secunda supports the expected
realization of a voiced fricative alveolar sibilant [z]. Devoicing of ζ before an unvoiced con-
sonant (represented by σ) suggests a similar place of articulation as that of Hebrew /s/.

6.3.3. Emphatic Consonants (/q/, /t/,̣ /s/)̣
It is generally accepted that the so-called "emphatic" consonants in Semitic were originally
realized as glottalics (i.e., ejectives) (KOGAN 2011, 59–61). With respect to /s ̣/, there is consid-
- 205 -
erable evidence that it was originally an affricate ([ʦˀ]) both in Proto-Semitic and in Hebrew
(STEINER 1982, 11–40; KOGAN 2011, 61–71). This yields the following original realizations for
the Hebrew emphatics: /q/ [kˀ], /t/̣ [tˀ], /s ̣/ [ʦˀ]. In Tiberian Hebrew, however, the emphatics /t/̣
and /s ̣/ were velarized (or pharyngealized) and /q/ was realized as a uvular or post-velar as in
Arabic: /q/ [q], /t/̣ [tˁ], /s ̣/ [sˁ] (KHAN 2013a, 89, 91–92). It is not clear when the Hebrew "em-
phatic" consonants shifted from glottalic to velarized/pharyngealized consonants (and from a
velar to a post-velar/uvular in the case of /q/), though some scholars have suggested that the
emphatics were velarized/pharyngealized already at the time of the Secunda (e.g., YUDITSKY
2017, 24–25). A possible clue for reconstructing the realization of the emphatics in the He-
brew of the Secunda is the behavior of vowels in the environment of the emphatics (see be-
low). While pharyngealized consonants (and uvular [q]) are prone to lower vowels as in Ara-
bic, ejective consonants should have no such effect as in Amharic.233

6.3.3.1. Emphatic Velar Stop: /q/ = κ
In the Secunda, Hebrew /q/ is represented with Greek κ:
κωλ /qō l/ [kˀoːl] 'all' Ps. 28:6

σαδδικιµ /s ̣addı̄ qı̄ m/ [ʦˀadːiːkˀıː̃ (m)] 'righteous ones' Ps. 32:11
(ου)ϊκραηνι /hū yeqrɔ̄ʔē nı̄ / [huː jikˀʀɔːʔeːniː] 'he will call me' Ps. 89:27
At the time of the Secunda, Greek κ represented the unaspirated voiceless velar stop [k]. It
was distinguished from Greek χ, which represented the aspirated voiceless velar stop [kh]
(4.5.3.1.35). Before voiced consonants and after nasals, κ was prone to assimilate and repre-
sent [g] (4.5.3.1.36). In transcription, κ is used to transcribe the /k/ phoneme only in Latin,
whereas Semitic /k/ is transcribed by χ. Emphatic /q/ (or /k ̣/) in Semitic is transcribed by
Greek κ. These three facts are best explained by positing an unaspirated feature both of Latin
/k/ ([k]) and the Semitic emphatic /q/ (or /k ̣/), on one hand, and an aspirated feature of the
233. Note that the emphatics do not lower adjacent vowels in the Tiberian vocalization system. In my view, this
is because the vocalization tradition is older than the velarized/pharyngealized realization of the emphatics,
which likely entered the Tiberian pronunciation of Hebrew as a result of residence in an Arabophone area (see
6.3.3.4). Presumably, at least from the perspective of acoustic phonetics, the Tiberian emphatics probably did
lower adjacent vowels to some degree even though it is not indicated in the niqqud.
- 206 -
regular Semitic voiceless velar stop /k/ ([kh]) on the other. In Greek loanwords in the Mish-
nah, Greek κ is rendered by ‫ ק‬in Hebrew (5.4.1.3.1). Because both glottalic and pharyngeal-
ized emphatic stops in Semitic lacked aspiration, they were associated with the unaspirated
Greek series (AL-JALLAD 2015, 13). Representing ‫ ק‬with κ only indicates that ‫ ק‬was distinct
from ‫כ‬, being either glottalic or pharyngealized, but it does not determine between the two.
The primary evidence for a glottalic realization of ‫ ק‬in the Secunda is its effect on ad-
jacent vowels, which may raise near /q/ (or become more front) (YUDITSKY 2017, 96):234
κεσθ235 /qaš t/ [kˀɪʃth/kˀæʃth] 'a bow' Ps. 18:35

σεκκι /ś aq(q)/; /ś aqqı̄ / [sækˀ]; [sækˀːiː] 'sackcloth'; 'my sackcloth' Ps. 30:12; 35:13
ρεκ /raq(q)/ [ʀækˀ] 'only' Ps. 32:6
βεκορβ236 /b-qorb/ [bɪkˀʊʀb] 'in the midst of' Ps. 36:2
κερου /qɔ̄rʔū / [kˀɛːʀʔuː]? 'they called' Ps. 49:12
λαβεκρ237 /lab-boqr/ [labːəkˀʀ] 'at the morning' Ps. 49:15
βεκοδσι 238
/b-qodš ı̄ / [bɪkˀʊðʃiː] 'in my holiness' Ps. 89:36
Vowel raising in the environment of /q/ supports a velar rather than uvular "pharyngealized"
realization of /q/. Cross-linguistically, in the environment of velars, back vowels tend to be
fronted and low front vowels (e.g., [æ] and [ε]) tend to be raised (HILLENBRAND and CLARK
2001, 754). In the environment of uvular consonants, on the other hand, vowels tend to be
lowered as in Arabic (BROSELOW 2006, 610). If Hebrew /q/ had already shifted from a glottal-
ic velar stop to a uvular stop [q] as in Tiberian, not only would we not expect vowel raising,
but we would expect vowel lowering to be reflected in the transcriptions.
Another relevant piece of evidence occurs in the one interchange of κ > χ = ‫ק < כ‬:
234. YUDITSKY also cites the verbal prefix ι- before κ (as opposed to regular ιε-) as an example of a raised
allophone of /e/ in the environment of κ (2017, 96).
235. Compare the form κασθ in Psalm 46:10.
236. Note that the normal vowel on the preposition /b-/ is /a/: βα-.
237. Compare the form βοκρ in Psalm 46:6.
238. See note 32.
- 207 -
χορσελαϊ239 /qorslay(y)/ [khʊʀsəlaj] 'my ankles' Ps. 18:37
This interchange, which presumably reflects an interchange of ‫ כ‬and ‫ ק‬in Hebrew, may indi-
cate that ‫ ק‬and ‫ כ‬had the same place of articulation, namely, velar. It is also possible that it is
merely a scribal error, since χ and κ were similar paleographically (THOMPSON 1966, 154–55).
YUDITSKY suggests that χ for κ could be an example of the dissimilation of emphatics (i.e.,
Geers' law) (2017, 25),240 but this assumes that the sibilant is an emphatic against the attesta-
tion of the word in Hebrew (‫קורסל‬/‫ קרסל‬and not ‫ קורצל‬as in Aramaic) and that Geers' law,
which refers to Akkadian, also applied to Hebrew (cf. ZEMÁNEK 1996, 51–52).
In Mishnaic Hebrew, the interchange of ‫ כ‬and ‫ ק‬is common in the environment of /p/,
/r/, emphatics, and gutturals (HENSHKE 2010, 430; SHARVIT 2016, 116). BAR-ASHER points out
that the reverse shift of ‫ כ < ק‬may occur as a result of the influence of the back vowel /o/
(2015, 1466).241 Just as labialization (and thus rounded vowels) is associated with emphasis in
modern Semitic languages (KHAN 2013c, 387–88), it may be that the rounded vowel follow-
ing ‫ ק‬blurred the distinction between emphatic and non-emphatic and occasioned the inter-
change. If this is the case, it is curious that labialization only seems to be a feature of ‫ ק‬and
not the other emphatics ‫ ט‬and ‫צ‬. Cross-linguistically, when labialization is a secondary coar-
ticulatory feature of ejectives, it usually occurs on velar and uvular ejectives (FORDYCE 1980,
133–34). Labialization in Geˁez, which occurs with the velars, is naturally only found with
the velar emphatic (e.g., /kwˀ/, but cf. /tˀ/ and /ʦˀ/) (LAMBDIN 1978, 4–5). Finally, the inter-
239. Note the Mishnaic Hebrew form ‫( קוּרסֵל‬see KUTSCHER 1974, 63).
240. YUDITSKY argues that the original pattern of this noun is *qutlub and that the form in the Secunda is the
result of vowel dissimilation (*qurs ̣ul > qurs ̣el) (2017, 206). If, however, the Proto-Hebrew pattern was *qatlub,
one could explain the vowels in the Secunda as a result of an a > o shift before /r/ and an o > e shift in the
environment of a sibilant; the former is attested in Mishnaic Hebrew (see below) and the latter is attested
elsewhere in the Secunda (see below).
241. For velar/uvular ejective variation in the environment of back vowels in a modern language, see FALLON's
work on Proto-Agaw (2009, 15).
- 208 -
change of ‫ ג‬and ‫ ק‬in Mishnaic Hebrew, especially in the environment of liquids (e.g., ‫גלפתרא‬
for ‫( )קלפתרא‬SHARVIT 2016, 115–16), seems to favor a velar realization of ‫[ ק‬kˀ].
In sum, the fact that Hebrew /q/ (= κ) is consistently transcribed distinctly from He-
brew /k/ (= χ) indicates that they represent distinct phonemes in Secunda Hebrew. The one
occurrence of χ for κ in χορσελαϊ is difficult to explain phonetically, but it corresponds with
similar changes known from Mishnaic Hebrew. While κ for ‫ ק‬would fit either a glottalic or
pharyngeal hypothesis, vowel raising (and not lowering) in the environment of ‫ ק‬seems to fa-
vor a non-pharynagealized and thus glottalic realization of /q/ in the Secunda ([kˀ]).
6.3.3.2. Emphatic Dental Stop: /t/̣ = τ
In the Secunda, Hebrew /t/̣ is represented with Greek τ:
ταµνου /tɔ̣̄ mnū / [tˀɔːmnuː] 'they hid' Ps. 31:5

φαλητ /pallē t/̣ [phalːeːtˀ] 'deliverance' Ps. 32:7
ουαββωτη /w-hab-bō tẹ̄ h ̣/ [(ʔ)uhabːoːtˀeːħ] 'and the one who trusts' Ps. 32:10
At the time of the Secunda, τ represented the unaspirated voiceless alveodental stop [t]. It
was distinguished from Greek θ, which represented the aspirated voiceless alveodental stop
[th] (4.5.3.1.30). After nasals, τ was prone to represent the voiced alveodental [d] (4.5.3.1.26).
In transcription, Greek τ is used to transcribe the /t/ phoneme only in Latin, whereas Semitic
/t/ is always transcribed by θ. Moreover, emphatic /t/̣ in Semitic is always transcribed by
Greek τ. Like the other emphatic stop, these facts are best exlpained by positing an unaspirat-
ed feature both of Latin /t/ ([t]) and Semitic emphatic /t/,̣ on one hand, and an aspirated fea-
ture of the regular Semitic voiceless dental stop /t/ on the other. In Greek loanwords in Mish-
naic Hebrew, Greek τ is rendered by ‫ ט‬in Hebrew (5.4.1.3.1).
There is one instance in which an /e/ vowel lowers to /a/ in the environment of /t/̣ (or,
̣
the etymological */a/ vowel is preserved in the environment of /t/):
αττε /ʔetṭɛ̣̄ / or /ʔatṭɛ̣̄ / [ʔatˀːɛː] 'I will incline' Ps. 49:5
YUDITSKY suggests that such lowering may be explained by assuming a pharyngealized pro-
nunciation of /t/̣ ([tˁ]) (2017, 25, 57, 95). However, the fact that etymological */a/ was not
- 209 -
preserved in the imperative form (cf. εττη /hetṭẹ̄ / [hɪtˀːeː] 'incline!' [Ps. 31:3]) calls into ques-
tion such an interpretation. Also, Interestingly, there are no examples of interchanges between
‫ ט‬and ‫ ת‬in Mishnaic Hebrew (HENSHKE 2010, 438–40; SHARVIT 2016, 151–52).
In sum, the fact that Hebrew /t/̣ (= τ) is consistently transcribed distinctly from He-
brew /t/ (= θ) indicates that they represent distinct phonemes in Secunda Hebrew, with the
lack of aspiration of Hebrew /t/̣ indicating either a glottalic or pharyngealized realization.
One example of vowel lowering before /t/̣ would favor a pharyngealized realization of /t/̣
([tˁ]) in the Secunda, but the evidence for "lowering" near /t/̣ is contradicted by a counter-
example; further, there are other reasons for positing a glottalic realization ([tˀ]) (see 6.3.3.4).
6.3.3.3. Emphatic Affricate/Sibilant: /s/̣ = σ
In the Secunda, Hebrew /s ̣/ is represented with Greek σ:
αρους /ʔɔ̄rū s ̣/ [ʔɔːʀuːʦˀ] 'I will run' Ps. 18:30

νωσηρ /nō s ̣ē r/ [noːʦˀeːʀ] 'keeping' Ps. 31:24
σαβαωθ /s ̣bɔ̄ʔō t/ [ʦˀaβɔːʔoːθ] 'hosts' Ps. 46:8
At the time of the Secunda, σ represented /s/ ([s̺ ] or [s̻ ]) in general and [z] before voiced allo-
phones. It caused vowel raising and sometimes deletion (for a fuller discussion of Greek σ at
the time of the Secunda, see 6.3.2.1). In transcription, Semitic emphatic /s ̣/ is usually
transcribed by Greek σ. There is one instance in which emphatic /s ̣/ in Akkadian is
transcribed by ζ (ζα[ραρ] s ̣arā r). STEINER (1982), who investigates remnants of the affricated
s ̣ade in the Semitic languages, has shown that the emphatic affricate in Semitic is represented
in Greek by σ, τι, or στ in Hebrew (40–42), τ (e.g., ατιρ ḥas ̣ı̄ r), στ, or σ in Punic (60–65) (see
also 5.3.5.3),242 and τι in Ethio-Semitic (82). In later Arabic documents, after Arabic /s ̣/ had
shifted from an affricate to a pharyngealized sibilant, /s ̣/ is regularly represented by σ (AL-
JALLAD forthcoming, 20). As far as transcription conventions go, Greek σ could have repre-
sented either [ʦˀ] or [sˁ]. In Greek loanwords in the Mishnah, Greek σ is perceived as emphat-
242. There is also one instance of Phoenician /s ̣/ transcribed with ξ (STEINER 1982, 69).
- 210 -
ic and transcribed by ‫( צ‬not ‫ )ס‬when it precedes τρ, due to the influence of /r/ (HEIJMANS,
2013, 254–55), but this phenomenon is more common in the Babylonian branch (5.4.1.3.5).
The evidence for vowel lowering and raising in the environment of /s ̣/ in the Secunda
is mixed. In one instance, an /e/ vowel is realized as [a] in the environment of /s ̣/ (rather, the
etymological */a/ vowel is preserved in the environment of /s ̣/):
ασµιθηµ* /ʔas ̣mı̄ tē m/ or /ʔes ̣mı̄ tē m/ [ʔaʦˀmiːθeːm] 'I will destroy them' Ps. 18:41
YUDITSKY believes that the [a] vowel is the result of the influence of the pharyngealized em-
phatic consonant /s ̣/. He also compares the example ασλιαννα in Origen's commentary on
Matthew (2017, 25, 57, 95).243 On the other hand, just as was the case with αττε and εττη in
the Ambrosiana palimpsest, no lowering occurs in the hifˁil prefix of the imperative form be-
fore /s ̣/ (e.g., εσιληνι /hes ̣s ̣ı̄ lē nı̄ / [hɪʦˀːiːleːniː] 'rescue me!' [Ps. 31:3]). Moreover, there is one
example of vowel raising before /s ̣/ (YUDITSKY 2017, 92–95):
σερουφα /s ̣rū pɔ̄/ [ʦˀəʀuːɸɔː] 'refined' Ps. 18:31
YUDITSKY sets this example against the wider backdrop of vowel raising in the environment of
sibilants attested both in the Secunda and Jerome (HARVIAINEN 1977, 58–66; YUDITSKY 2017,
94; see also 6.3.2.1). Alternatively, Greek ε may represent a centralized "shewa" vowel away
from the stress (4.5.3.1.12). In external sources, /a/ (or /e/) is once raised to [i] before /s ̣/ in an
environment where it could not represent a shewa vowel: ουµεµµισραιµ ‫( וּ ִמ ִמּצ ְַרי ִם‬Hos. 11:1)
(YUDITSKY 2017, 93). In another place, no vowel at all is represented before /s ̣/ (61–62):
ουµσωθαϊ /w-mas ̣wō tay(y)/ [(ʔ)um(ɪ)ʦˀ(w)oːθaj] 'and my commandments' Ps. 89:32
Short /e/ or /i/ were peceived as part of the hissing of the sibilant and thus not indicated in
transcription (YUDITSKY 2017, 61–62; 6.3.2.1); alternatively, syncope applied (6.5.1.5.1).
Finally, further evidence for the realization of /s ̣/ is provided two or three centuries
later in Jerome. In one place, he describes s ̣ade as a sound which "our ears thoroughly
243. There is a variant reading: Kl has ΑΣΛΙΑΝΝΑ but M and H have ἀσαῒ ἀννὰ (KLOSTERMAN 1935, 541).
- 211 -
dread"244 (GRAVES 2007, 28). Elsewhere, he describes its pronunciation as "between z and s ...
it is shrill (stridulus), and with the teeth clenched (strictis dentibus) it is barely articulated by
pressing the tongue [against them]"245 (STEINER 1982, 43–44). The word Jerome uses to de-
scribe the sound of /s ̣/ (stridulus) is used elsewhere in Latin to describe the notes a war trum-
pet strikes out246 or a whizzing saw.247 Note also Jerome's translation of τρίζει τοὺς ὀδόντας
αὐτοῦ 'he is grinding his teeth' (Mk. 9:18) as stridet dentibus in the Vulgate. The real clue to
the nature of the sound that Jerome's ears "thoroughly dread," however, is found in the phrase
strictis dentibus in the passage above. This precise phrase is perhaps used only a couple
times248 elsewhere in all of Jerome's writings, one of which is as follows (Letter to Damasus):
In sum, just as we in the Latin language also have some interjections, so that
in exulting we say 'ua' and in being amazed 'papae' and in grieving 'heu', when
we want to command silence, with clenched teeth (strictis dentibus) we re-
strict and confine breath to utter the sound 'st' ... 249
The social context (shushing someone), restriction of breath, clenched teeth, and the identifi-
cation of the sound with st seem to indicate an interjection similar to the "dental click" sound
Arabic speakers might make to answer in the negative or the tsk! tsk! sound an English
speaker might make to convey the meaning, "what a shame!" (see KIRCH 1979, 422). It seems
more likely, then, that Jerome's description of s ̣ade indicates an affricate [ʦˀ].250
244. quam aures nostrae pentius reformidant.
245. inter z et s ... est enim stridulus et strictis dentibus uix linguae impressione profertur.
246. Seneca (Oedipus, 732–33): lituusque adunco stridulos cantus elisit aere 'and the war trumpet with a curve
shrill notes strikes into the air'.
247. Marcus Servius Honoratus glosses the phrase argutae serrae 'grading saw' as stridulae 'shrill' (Commentary
on the Georgics of Vergil, 1.143).
248. See also Epistula XXII, 29: Non delumbem matronarum salivam delicata secteris, quae nunc strictis
dentibus nunc labiis dissolutis balbutientem linguam in dimidiata verba moderantur, rusticum putantes omne
quod nascitur. 'And do not, out of affectation, follow the sickly taste of married ladies who, now pressing their
teeth together, now keeping their lips wide apart, speak with a lisp, and purposely clip their words, because they
fancy that to pronounce them naturally is a mark of country breeding' (FREMANTLE, LEWIS, and MARTLEY 1893).
249. Epistula XX, 5.1.4: Ad summam, sicuti nos in lingua Latina habemus et interiectiones quasdam, ut in
exultando dicamus 'ua' et in admirando 'papae' et in dolendo 'heu' et, quando silentium uolumus imperare,
strictis dentibus spiritum coartamus et cogimus in sonandum 'st' ...
250. While analyzing ancient linguistic perception is by no means a simple task, an ejective affricate, rather than
a pharyngealized sibilant, is more likely to "offend" the ears of one whose language has sibilants but no
affricates. English speakers learning Arabic have difficulty distinguishing Arabic [sˁ] from [s] without the help
- 212 -
In sum, the fact that Hebrew /s ̣/ is consistently transcribed by σ does not help deter-
mine whether it had an ejective or pharyngealized realization. Only transcriptions of /s ̣/ by τ,
τι or στ (not before /r/) would remove the ambiguity. Lowering in the environment of /s ̣/
would seem to support a pharyngealized realization ([sˁ]), but it was shown that the evidence
for lowering in the environment of /s ̣/ is inconsistent. On the other hand, the fact that /s ̣/
seems to bring about vowel raising would point to an affricate ejective realization ([ʦˀ]),
since raising would not accompany a pharyngealized [sˁ]. At the same time, Palestinian Koine
Greek is also witness to vowel raising in the environment of σ. Therefore, these phenomena
may merely reflect the Greek accent of the scribe. Nevertheless, the descriptions in Jerome's
writings, in my opinion, favor an ejective affricate realization of Hebrew /s ̣/ ([ʦˀ]). It is possi-
ble that the pronunciation of [sˁ] existed alongside [ʦˀ] in various Hebrew dialects of ancient
Palestine, but it makes more sense to posit [sˁ] entering Hebrew at a later period.
In the Hebrew traditions attested in the Middle Ages, /s ̣/ was realized as an affricate in
all non-Arabic-speaking areas, stretching geographically from Iran to northern Spain (STEINER
1982, 11). It is probably the case that the pharyngealization of glottalic consonants in Semitic
originated in Arabic and was promulgated by the spread of Arabic (ZEMÁNEK 1996, 27).
Therefore, it seems best to explain the pharyngealized realization of /s ̣/ ([sˁ]) as a later He-
brew development as a result of contact with Arabic.251 Nevertheless, the presence of the
vowel α in the prefix of ασµιθηµ* remains a difficulty.252

Although the evidence for the realization of /q/, /t/,̣ and /s ̣/ at the time of the Secunda is in-
conclusive, there are a number of relevant pieces of evidence that argue against hypothesizing
of changes in adjacent vowels (HAYES-HARB and DURHAM 2016). With respect to identifying [sˁ], modern English
speakers and ancient Latin speakers would have been in a similar position, having [s] but no pharyngealized
consonants in their own language. It is difficult to imagine how a foreign sound, often indistinguishable from
one's native [s] without the help of neighboring vowels, would "offend" the ear as Jerome says.
251. Note that potential earlier contact with Arabic is irrelevant, since s ̣ā d was an affricate ejective in early
Arabic as well (AL-JALLAD 2014; forthcoming, 20).
252. However, α also appears in the hiphil prefix in ιαγι* ‫( י ַגִּי ַהּ‬Ps. 18:29).
- 213 -
a pharyngealized realization of the "emphatic" consonants. First, in the Secunda, there is no
general lowering of vowels in the environment of the emphatics. Second, in the Secunda,
both /q/ and /s ̣/ occasion the raising of vowels on some occasions. Third, Jerome's description
of the realization of /s ̣/ favors an affricate pronunciation of /s ̣/ in at least some Hebrew tradi-
tion in the fourth and/or fifth century CE. Fourth, the pharyngealized realization of /q/, /t/̣ and
/s ̣/ in Tiberian Hebrew is best explained as a result of the influence of Arabic. Therefore, the
best synthesis of the data from the Secunda and the history of Hebrew points to an ejective
(and affricate) realization ([kˀ], [tˀ], [ʦˀ]) for the emphatic consonants.
YUDITSKY rejects this claim for two reasons. First, he believes that the glottalic co-ar-
ticulation of an ejective would be perceived as aspiration in Greek and thus [kˀ] and [tˀ]
would have been represented by χ and θ in Greek (2017, 25). However, such a claim runs
contrary both to Greek transcription conventions253 and to the phonetic nature of ejectives.254
In the Jibbali language of Oman, for example, aspiration is actually a significant element that
distinguishes the non-glottalic voiceless consonants from the glottalic consonants (RUBIN
2014, 27). Second, he attributes the "lowering" in the transcriptions αττε and ασµιθηµ* to a
pharyngealized realization of /t/̣ and /s ̣/ (YUDITSKY 2017, 95). While these examples are diffi-
cult to explain, the apparent "lowering" in one morphological category255 in two out of four
instances—if we include external sources, three out of five instances—is hardly enough to
overturn the rest of the data. Nevertheless, in light of such apparent lowering, while /q/ only
raises vowels, it is worth considering the possibility, though unlikely for Hebrew, that per-
253. Note how Geˁez /s ̣/ is represented by τι (not θ) in Τιαµῶ s ̣əyā mo (KOGAN 2011, 62). In the earliest
attestations of Greek transcription of Arabic, the ejectives are represented with κ and τ (AL-JALLAD 2015).
254. While glottalization is a form of aspiration, it is distinct from the sort of aspiration of the Greek and
Hebrew stops that is represented with [h].
255. Note how "lowering" only occurs in the 1cs forms in the Ambrosiana palimpsest, but it is absent in the
imperative. Also, the prefix vowel of hifˁil is etymological */a/ and is realized as /a/ in the main Hebrew reading
traditions. Also, if the "lowering" in the hifˁil prefix was due to the pharyngealized realization of the emphatics,
we would expect it to occur also before pharyngeals, which it does not: e.g., ιεµιδηνι* ‫( י ַ ֲעמִידֵ נִי‬Ps. 18:34), εεζεκ
‫( ַה ֲחזֵק‬Ps. 35:2). In external sources, the imperative also has α: ασλιαννα ‫( ַה ְצלִיחָה נָּא‬Ps. 119:25).
- 214 -
haps not all of the emphatic consonants were pharyngealized at the same time or to the same
degree.256
Finally, it should be noted that what is suggested here runs contrary to FABER's argu-
ment that assimilated forms in the Dt stem such as ‫' ִה ְצטַדֵּ ק‬he justified himself' prove that al-
ready in ancient Hebrew the emphatics were pharyngealized, since pharyngealization spreads
more than glottalization (FABER 1980, 140–41). However, FALLON, in his comprehensive
study of ejectives, cites numerous examples of glottalic assimilation: e.g., Oromo /t͡ ʃˀapˀ-ti/
[t͡ ʃˀapˀtˀi] 'it (f.) breaks' and Northwest Caucasian /t-ʃˀəʁe/ [tˀʃˀəʁe] 'we made' (2002, 43, 48).
6.3.4. Nasals (/m/, /n/)
In ancient Hebrew, /m/ most likely represented a bilabial nasal [m] and /n/ an alveolar nasal
[n]. These are their respective realizations in Tiberian Hebrew (KHAN 2013a, 90).
6.3.4.1. Bilabial Nasal: /m/ = µ
In the Secunda, Hebrew /m/ is normally represented with Greek µ:
µαϊµ /maym/ [majm] 'waters' Ps. 32:6

ραββιµ /rabbı̄ m/ [ʀabːıː̃ (m)] 'great' Ps. 89:51
σαλωµ /š ɔ̄lō m/ [ʃɔːloːm] 'peace' Ps. 35:20
In Palestinian Koine, µ represented a bilabial nasal [m]. In transcription, /m/ in both Latin and
Semitic is transcribed by µ (5.3.6). In Greek loanwords in the Mishnah, Greek µ is regularly
rendered by ‫ מ‬in Hebrew (5.4.1.3.4). In the Secunda, /m/ is once represented by Greek β:
βσεβωθαµ /b-š mō tam(m)/ [b(ɪ)ʃəmoːθa(̃ m)] 'by their names' Ps. 49:12
YUDITSKY corrects βσεβωθαµ to βσεµωθαµ* (2017, 303), but it is possible that [m] became a
fricative in partial assimilation to the preceding sibilant fricative [ʃ] and was realized as [β],
represented by β (= [β]) in Greek (see 6.3.1.1). This may be compared to the transcription
256. In the Mehriyō t dialect of Mehri, for example, /k ̣/ has a glottalic initiation, whereas /t/̣ and /s ̣/ are realized
with pharyngeal contraction and tongue retraction (WATSON and BELLEM 2010). In the Mehreyyet dialect, on the
other hand, each of the emphatics /k ̣/, /t/,̣ and /s ̣/ exhibits a different distribution of showing "ejective tokens." In
both dialects, the emphatics tend to be accompanied by pharyngealization (WATSON 2012, 16). If
pharyngealization began to occur in Hebrew earlier than suggested, /t/̣ and /s ̣/ but not /q/ may have been
pharyngealized by the time of the Secunda. We could then attribute the raising and fronting of vowels in the
environment of /s ̣/ to the influence of the Greek accent of the scribe, since σ brought about raising in Greek also.
- 215 -
Μενιαµι (CIIP III, no. 2223) for the Hebrew name ‫( בנימן‬or ‫)?מנימין‬, exhibiting a shift of /b/ >
/m/ word-initially (4.5.3.1.26). Interchanges of ‫ ב‬and ‫ מ‬are also attested in a few words in
Mishnaic Hebrew (e.g., ‫ ימנה‬for ‫( )יבנה‬SHARVIT 2016, 284). Hebrew /m/ is also once represent-
ed by Greek λ:
µαλλαχωθ /mamlɔ̄kō t/ [mal̃ ːɔːχoːθ] 'kingdoms' Ps. 46:7
YUDITSKY, regarding assimilation unlikely, corrects it to µαµλαχωθ* (2017, 208), though µ
and λ are not especially similar paleographically. Assimilation should not be ruled out. If the
nasals were weakened in Greek or Hebrew (6.3.4.3), the following assimilation is conceiv-
able: [ma(̃ m)lɔːχoːθ] > [ma(̃ n)lɔːχoːθ] > [mal̃ ːɔːχoːθ].

6.3.4.2. Alveolar Nasal: /n/ = ν
In the Secunda, Hebrew /n/ is normally represented with Greek ν:
ιαλιν /yɔ̄lı̄ n/ [jɔːliːn] '(it) will dwell' Ps. 30:6

νααρ /nɔ̄hɔ̄r/ [nɔːhɔːʀ] 'a river' Ps. 46:5
ναθαν /nɔ̄tan/ [nɔːθan] 'he set' Ps. 46:7
At the time of the Secunda, ν represented an alveolar nasal [n]. In transcription, /n/ (in both
Latin and Semitic) is transcribed by ν (5.3.6). In Greek loanwords in the Mishnah, Greek ν is
regularly rendered by ‫ נ‬in Hebrew (5.4.1.3.4).

6.3.4.3. Interchanges of µ/ν
There are a number of examples of µ and ν interchanging in the Secunda. Most significantly,
word-final /m/ in the Secunda is occasionally represented with ν (see YUDITSKY 2017, 23–24):
θαµµιν /tɔ̄mı̄ m/ [thɔːmıː̃ (m)] 'blameless' Ps. 18:31

θεσθιρην /testı̄ rē m/ [t ɪst iːʀeː̃ (m)]
h h
'you hide them' Ps. 31:21
ααµιν /hɔ̄-ʕammı̄ m/ [hɔːʕamːı(̃ m)] 'the peoples' Ps. 49:2
αυωναν /ʕwō nam(m)/ [ʕawoːna(̃ m)] 'their iniquity' Ps. 89:33
In one instance, /#n/ was originally written as µ and then corrected to ν above the line. If
originally µηερθ, this would be another example of an m/n interchange (YUDITSKY 2017, 24):
µνηερθ /nē ʔert/ [neːʔɪʀth] 'you abhorred' Ps. 89:40
- 216 -
At the time of the Secunda, there is evidence for the weakening (or even loss) of nasals in
Greek pronunciation, demonstrated by the omission of µ and ν (especially word-finally)
(4.5.3.1.26) and the interchange of µ and ν (4.5.3.1.27) in contemporary Greek orthography.
The loss of nasals may have resulted in the nasalization of the preceding vowel. Alternatively,
the nasals may have assimilated to a following consonant (GIGNAC 1976, 113–14). In
transcription, there are a number of interchanges of µ and ν. Also, a rare transcription of ι for
/n/ attested in both Latin and Akkadian transcriptions may reflect the weakening of the nasal
(5.3.1.3.4; 5.3.2.3.5). In Palestinian epigraphy, the name ‫( בנימן‬or ‫ )?מנימין‬is once transcribed
as Μενιαµι (4.5.3.1.26), reflecting the elision of final /n#/. The transcriptions Σαλω/Σαλων
(CIIP I/1, no. 134a, 591) for the proper name ‫שלון‬/‫ שלום‬may also attest to this phenomenon.
The occasional interchange of /m/ and /n/ in the Secunda has been explained as the
result of a lack of phonemic distinction of nasals in word-final position as in contemporary
Hebrew (see below), dissimilation due to the presence of multiple nasals/sonorants in a given
word, or the fact that Greek words tend to end in ν and not µ (YUDITSKY 2017, 23–24). There
are a number of examples of this phenomenon in external sources as well.257
A number of parallels to this phenomenon exist in contemporary Hebrew evidence.
The interchange of ‫ ם < ן‬in final position is attested in Mishnaic Hebrew, the Dead Sea
Scrolls, and the Judaean Desert texts. It most frequently occurs when the masculine plural
morpheme ‫ ִים‬- is realized as ‫ ִין‬- or suffix forms ending in ‫ם‬- are realized as ‫ן‬- (i.e., grammati-
cal morphemes): e.g., ‫( עומדין‬for ‫)עומדים‬. However, it also occurs below the morphological
level (i.e., non-grammatical morphemes): e.g., ‫( אדן‬for ‫)אדם‬. Final ‫ ן‬may also be omitted in
spelling: e.g., ‫( למע‬for ‫ )למען‬and ‫( יוחנה‬for ‫)יוחנן‬. Finally, an originally open final syllable may
257. Examples of µ > ν external sources: σουµην ‫שׁ ַמי ִם‬ ָ (Gen. 1:8; Procopius), ακχερουβιν ‫( ַהכְּרוּבִים‬Gen. 3:24;
Heb 400), σιειν ‫( ִציּ ִים‬Jer. 50:39), and ζωην ‫( זֹעֵם‬Ps. 7:12). Finally, in the transcribed list of biblical books from
Origen quoted in Eusebius's history, the name of the book of Chronicles is transcribed as Δαβρηϊαµεὶν ‫דִּ ב ְֵרי ַהיָּמִים‬
(in the same list are present Ἑλεαδδεβαρεὶµ ‫ ֵאלֶּה הַדְּ ב ִָרים‬and Σφαρθεαλείµ ‫) ֵספֶר תְּ ִהלִּים‬. There is one example of
ν > µ: βεδεµ ‫( ְבּעֵדֶ ן‬Gen. 2:8).
- 217 -
be spelled with a ‫ן‬- instead of a ‫ה‬-: e.g., ‫( יודן‬for ‫יהודה‬/‫ )יודה‬and ‫( למטן‬for ‫)למטה‬258 (QIMRON
1986, 27–28; MOR 2015, 106–15; SHARVIT 2016, 226–28). KUTSCHER argues that the data re-
flect a realization of both final ‫ ם‬and ‫ ן‬as [n] (1976, 58–68). BEN-ḤAYYIM, on the other hand,
argues that the elision of the final nasal resulted in the nasalization of the vowel (i.e., ‫= אדן‬
[ʔaːðaː̃ ] or [ʔaːðaːŋ]) (1958, 210–11). The distribution of ‫ן‬/‫ ם‬interchanges in grammatical and
non-grammatical morphemes in Mishnaic Hebrew has been treated comprehensively by
NAEH.259 With respect to grammatical morphemes in the Judaean Desert texts, MOR has
demonstrated that, aside from the dual form,260 the distribution of ‫ן‬/‫ ם‬is a scribal phenomenon.
With respect to non-grammatical morphemes, the historical spelling is always maintained
(NAEH 1992b, 297–306; NAEH 2013, 369–92; MOR 2015, 107–108).
Because the interchange of µ > ν occurs in both non-grammatical morphemes (e.g.,
θαµµιν) and grammatical morphemes (e.g., ααµιν, αυωναν) in the Secunda, it is likely that
the variant spellings in the Secunda reflect a phonetic rather than a morphological reality.
That is, this orthographic phenomenon likely indicates the weakening of the final nasal and
subsequent nasalization of the vowel (i.e., θαµµιν *[tɔːmiːm] > [tɔːmıː̃ (m)]). Although such a
change may seem unusual in Semitic, in which root integrity is important, it should be noted
that a very similar change also occurs in the Jibbali language of Oman.261 The suggestion that
258. When the following word begins with an ‫( מ‬e.g., -‫)למטה מ‬, final ‫ ה‬is not replaced by ‫( ן‬MOR 2015, 112).
259. With respect to non-grammatical morphemes in Mishnaic Hebrew, final ‫ ן‬occurs after low vowels, whereas
final ‫ ם‬occurs after high vowels. This points to a final nasalized vowel. With respect to the grammatical
morphemes (mp endings/suffixes), nominal forms tend to maintain the ‫ ִים‬- ending, whereas verbal participles
tend to take the ‫ ִין‬- ending. For NAEH, this distribution points to morphological change due to the influence of
Aramaic and not a nasalized vowel (NAEH 1992b, 297–306; NAEH 2013, 369–92; MOR 2015, 107–108).
260. The dual is always written with ‫( ם‬e.g., ‫ שתים‬,‫ שנים‬,‫)טפחים‬. This is likely due to the fact that the dual ending
had become lexicalized with the word and was not conceived of as an individual morpheme (MOR 2015, 111).
261. In Jibbali, after a full vowel, final /m#/ and /n#/ often undergo devoicing or are lost. Consequently, the
preceding vowel is nasalized. Additionally, the vowel is followed by a "slight aspiration" or "nasal expiration."
For example, /sεm/ 'poison' is usually pronounced as [sε̃h]. In verbs, however, the final nasal is usually
preserved: e.g., /zəh ̣á m/ 'he came' is pronounced as [zəħˈam
̃ ] (RUBIN 2014, 37–38).
- 218 -
dissimilation of nasals occurs, based on the transcriptions θαµµιν and ααµιν (YUDITSKY 2017,
23–24), is unlikely in light of the transcription αυωναν.
It is unclear whether this was a general change (Vm, Vn > Ṽ / _#) or limited in its dis-
tribution in the Secunda. In Mishnaic Hebrew, it is attested primarily after low vowels, likely
due to their higher sonority (see note 54). This distribution does not apply in the Secunda, but
all four instances of µ# > ν# are preceded by a sonorant consonant, which would increase the
sonority of the segment. The relatively low frequency of these spelling variants does not nec-
essarily correspond to a low frequency in the actual vocalization (contra YUDITSKY 2017, 24),
since writing is usually more conservative than speech. Even though we should not expect the
Greek transcriptions to be as conservative as the Hebrew script itself in preserving historical
spellings, there does seem to be evidence that the scribe was working from the consonantal
text of the Hebrew Bible,262 which likely would have prevented him from making a greater
number of errors in transcribing consonants. It is entirely conceivable that a regular change of
Vm, Vn > Ṽ / _# applied in Secunda Hebrew and was only evidenced in a few spelling vari-
ants, just as is the case at Qumran (QIMRON 1986, 27–28). Accordingly, /Vm#/ and /Vn#/ are
represented in phonetic transcription as [Ṽ (m)]/[Ṽ (n)] to indicate the sound change while
leaving open the possibility (by enclosing m/n in parentheses) of a conservative pronuncia-
tion (i.e., final Vm/Vn were pronounced as [Vm]/[Vn]) for the biblical reading tradition.
The evidence of the Secunda transcriptions and the history of Hebrew is consistent with
positing that Hebrew /m/ was realized as a bilabial nasal [m] (represented by µ) and Hebrew
/n/ was realized as an alveolar nasal [n] (represented by ν). Word-finally after a vowel, both
/m/ and /n/ were weakened with the consequent nasalization of the preceding vowel.
262. The reading ουϊεδαββερ ‫( ַויּ ַדְ בֵּר‬Ps. 18:48) is likely the result of the scribe vocalizaing the consonantal text
with the Hebrew verb most familiar to him for the consonantal frame of ‫וידבר‬.
- 219 -
One final observation worth emphasizing is the parallel between Hebrew and Greek
with respect to the weakening of final nasals. Though numerous Hebraists have tied together
the various data regarding final nasals in Mishnaic Hebrew, Qumran, Judaean Hebrew, Ara-
maic, and the transcriptions, none have turned to the evidence of Koine Greek phonology to
suggest that the weakening of final nasals in Hebrew and Aramaic might actually be an areal
feature resulting from diffusion from Greek. The close contact of Greek and Hebrew/Aramaic
in Palestine during the Second Temple period would certainly support such a hypothesis.
6.3.5. Liquids (/l/, /r/)
Throughout the history of Hebrew, /l/ was probably realized as a voiced lateral [l]. In Tiberian
Hebrew, it is realized as an alveolar lateral continuant [l] (KHAN 2013a, 90), but there is some
evidence for a "dark" velarized [l]̣ in the earliest stages of Hebrew (FABER 1989). It is not
clear how /r/ was pronounced in ancient Hebrew. Proto-Semitic /r/ is generally reconstructed
as a dental resonant (KOGAN 2011, 54). In Tiberian Hebrew, /r/ has two allophonic realiza-
tions: a voiced uvular roll [ʀ] (or uvular frictionless continuant [ʁˎ]) and, in the environment
of alveolar consonants, an emphatic apico-alveolar roll [r ̣] (KHAN 1995; KHAN 2013a, 92–93).
Babylonian Hebrew has only one pronunciation of resh (apico-alveolar trill [r]), with a more
"robust" pronunciation occuring syllable-initially, though apparently they had two realiza-
tions of resh in their vernacular (KHAN 2013c, 385–86).

6.3.5.1. Alveolar Lateral: /l/ = λ
In the Secunda, Hebrew /l/ is normally represented with Greek λ:263
αϊλ /h ̣ayl/ [ħajl] 'force' Ps. 18:40

λανου /lɔ̄nū / [lɔːnuː] 'for us' Ps. 46:2
φλαγαυ /plɔ̄gaw/ h
[p (a)lɔːʁaw] 'its streams' Ps. 46:5
At the time of the Secunda, Greek λ represented [l]. Greek λ occasionally interchanges with ρ
(4.5.3.1.39), indicating that their places of articulation were similar, probably alveodental, λ
263. There are eight instances in which α (6x) or δ (2x) is transcribed instead of an expected λ. These are likely
scribal errors arising from the fact that λ is similar to α and δ in shape.
- 220 -
being a lateral and ρ a trill (PETROUNIAS 2007b, 563–64). In transcription, both Latin /l/ and
Semitic /l/ are transcribed by Greek λ (5.3.6). Twice, /l/ is not represented in transcription:
µεσσω > µεσσωλ* /meš -š (ʔ)ō l/ [mɪʃːoːl] 'from Sheol' Ps. 30:4
µηοδ > µηολδ*264 /mē -h ̣old/ [meːħʊ(l)d] 'from the world/lifetime' Ps. 89:48
These transcriptions are probably scribal errors for µεσσω<λ> and µηο<λ>δ (YUDITSKY 2017,
309–310), but there is precedence in contemporary Greek orthography for the loss of liquids
(4.5.3.1.39) or assimilation to a following stop (µηο<λ>δ) or nasal (µεσσω<λ> νεφσι) (GI-
GNAC 1976, 108). There are also comparable phenomena elsewhere in Semitic.265 There is no
reason not to assume an alveolar lateral realization [l] of Hebrew /l/ in the Secunda. The
transcription µαλλαχωθ may suggest that /l/ and /n/ had the same place of articulation
(6.3.4.1).
6.3.5.2. Uvular Roll: /r/ = ρ
In the Secunda, Hebrew /r/ is normally represented with Greek ρ:
ραµωθ /rɔ̄mō t/ [ʀɔːmoːθ] 'lofty' Ps. 18:28

σιρ /š ı̄ r/ [ʃiːʀ] 'a song of' Ps. 30:1
σουρ /s ̣ū r/ [ʦˀuːʀ] 'the edge of' Ps. 89:44
At the time of the Secunda, Greek ρ represented an alveolar trill [r]. In transcription, both
Latin /r/ and Semitic /r/ are transcribed by Greek ρ. In Greek loanwords in Mishnaic Hebrew,
ρ is regularly represented by ‫ ר‬in Hebrew (5.4.1.3.4).
The precise realization of /r/ in the Secunda may be examined on the basis of its effect
on vowels. It is common for /r/ to lower an adjacent vowel (YUDITSKY 2017, 89–91):266
ραννη /ronnē / /rannē / [ʀanːeː] 'shouts of' Ps. 32:7

ελθαρακ /ʔal terh ̣aq/ [ʔæ thaʀħakˀ] 'do not be far!' Ps. 35:22
264. Other Greek translations support reading this as 'from the world' as phonologically transcribed.
265. The loss of final liquids occurs in Jibbali. The word µηοδ might also be compared to a phenomenon in
Jibbali, in which /l/ is lost and the preceding vowel rounded, especially in monosyllabic nouns of the pattern
CaCC (i.e., #CalC# > #CɔC#): e.g., */gald/ > [gɔd] (cf. MT ‫( ) ָחלֶד‬RUBIN 2014, 35, 37–38). Note also how Proto-
Semitic *kalb is realized in Mehri as /kawb/ (RUBIN 2010, 17).
266. Note that YUDITSKY is not sure if ραννη should be read with α or ο (2017, 177). The correct reading is
ραννη. Lowering of vowels in the environment of /r/ also occurs in external sources: σωρ ‫' צוּר‬rock' (Isa. 26:4).
- 221 -
αρφου /harpū /herpū ?/ [haʀphuː] 'be still!' Ps. 46:11
καρβαµ /qerbam(m)/ [kˀaʀba(̃ m)] 'within them' Ps. 49:12
ζεδαρχαµ /zɛ̄ derkam(m)/ [zɛː dɪʀkha(̃ m)] 'this is their way' Ps. 49:14
ζαρω /zarʕō / [zaʀʕoː] 'his seed' Ps. 89:30
αρφαθ /h ̣arpat/ [ħaʀphaθ] 'the reproach of' Ps. 89:51
αρηµωθ /hrı̄ mō t/ or /hrē mō t/ [haʀeːmoːθ] 'you lifted up' Ps. 89:43
In contemporary Greek orthography, it is also common for vowels to have lower allophones
in the environment of ρ (4.5.3.1.9; 4.5.3.1.12). Two transcriptions may be interpreted in such
a way so as to indicate that /r/ is responsible for the rounding of an adjacent vowel:
χορσελαϊ /qorslay(y)/ [khʊʀsəlaj] 'my ankles' Ps. 18:37

βεκορβ /b-qorb/ [bɪkˀʊʀb] 'in the midst of' Ps. 36:2
Each example may also be explained as deriving from variant patterns (cf. ‫ק}ו{רסל‬267 in
Mishnaic Hebrew and ‫ בקורב‬in the Dead Sea Scrolls).268 The change of α > ο is also common
in Greek in the environment of liquids (GIGNAC 1976, 288; 4.5.3.1.12).269 The rounding of
vowels in the environment of resh is attested at Qumran (e.g., ‫( )והי ֿו ֿרדן‬QIMRON 1986, 39–40),
in the western tradition of Mishnaic Hebrew (e.g., ‫ קורדום‬qordom for ‫)ק ְַרדּוֹם‬, in Palestinian
Aramaic, and in the LXX (e.g., Ιορδανης for ‫( )י ְַרדֵּ ן‬KHAN 2013c, 387–88). Hebrew /r/ may
also cause the lengthening of o > ō / _r# (see YUDITSKY 2017, 67, 120–21):
εσµωρλω /ʔeš mor lō / [ʔɪʃmoːʀ loː] 'I will keep for him' Ps. 89:29
If resh was realized as a uvular, it may have been weakened in final position and thus the du-
ration of the vowel was perceived as (or was actually) longer. At Qumran, there is evidence
that resh has weakened by its omission in spelling, especially in the environment of gutturals
(e.g., ‫[ משע‬for ‫]משער‬, ‫[ מערכת‬for ‫]מערכת‬, and ‫[ מגש‬for ‫( )]מגרש‬QIMRON 1986, 26–27). As the
267. The Mishnaic form may demonstrate the same phonetic change: a > o / _r (but cf. the discussion in
KUTSCHER [1974, 63]).
268. So argues YUDITSKY, though he appeals to the Syriac form ‫ קרצל‬/qurs ̣ul/ instead of the Mishnaic form
‫ק}ו{רסל‬. He cites the example of ‫ בקורב‬from Qumran to suggest a qutl pattern for βεκορβ (2017, 187–88, 206).
For ‫קרסל‬, note the attestations in Mishnaic Hebrew: ‫( ק ְַרסוּ ַלּי ִם‬Hul. 3:7), ‫( ק ְַרסוּלָה‬Bek. 7:6), ‫ַ}ו{רסֻל‬
ְ ‫( ק‬Ohol. 1:8).
The final example has an erased ‫ו‬. It is not clear which pattern, qursVl or qarsVl, is more original.
269. α > ο also occurs in the Greek of Nabatea and Batanea from the fourth century CE (BUBENIK 2007, 632).
- 222 -
third radical in the qVtl pattern, /r/ is the only non-guttural that develops an anaptyctic vowel
(cf. YUDITSKY 2017, 79):
ιεθερ /yetr/ [jɪθɪʀ] 'abundance' Ps. 31:24
The helping vowel here is probably due to the fact that the final consonant cluster begins with
a minimally-sonorous consonant followed by a highly-sonorous consonant, though it may
also reflect a realization of /r/ similar to the gutturals. Also, /r/ is never geminated and
exhibits compensatory lengthening in two or three instances (see YUDITSKY 2017, 39–40):270
ουβαρεχ /w-bɔ̄rek/ [(ʔ)uβɔːʀɪχ] 'and bless!' Ps. 28:9

αρισωνιµ /hɔ̄-rı̄ š ō nı̄ m/ [hɔːʀiʃoːnıː̃ (m)] 'the first' Ps. 89:50
ηρφου /h ̣ē rpū / [ħeːʀɸuː] 'reproached' Ps. 89:52
Compensatory lengthening in the piel stem271 and the lack of gemination after the definite ar-
ticle272 may point to a guttural-like realization of /r/ in the Secunda. On the other hand, lack of
compensatory lengthening is attested in the following transcription (see YUDITSKY 2017, 40):
µερεσθ /me(r)-reš t/ [mɪʀɪʃth] 'from the net' Ps. 31:5
Similar exceptions to compensatory lengthening occur in Tiberian (e.g., üֶ‫ מ ִָרגְז‬,‫מ ְִרד ֹף‬, ‫)מִחוּץ‬,
Babylonian, and Palestinian (KHAN 2013c, 386–87; YUDITSKY 2017, 40). The word µερεσθ re-
flects a time gap between the loss of gemination and compensatory lengthening of the vowel;
after simplification of gemination, a mora slot was left empty (see KHAN 2013c, 386–87):
σ σ
µ µ µ µ
m e ø r e š t
µερεσθ = /(me).(reš ).t/ [mɪʀɪʃt]

Figure 20: Moraic Representation of µερεσθ
270. Compensatory lengthening also occurs in the external attestation µηρεµ ‫( מ ֵֶרחֶם‬Ps. 110:3).
271. Cf. the piˁel imperative φελλετηνι (Ps. 31:2) and the piˁel suffix conjugation ουκεσσες (Ps. 46:10).
272. The definite article is geminated 9/10x instances (excluding /r/ for now) when preceding a non-guttural. In
the one exception, αϊωµ, the trema on the ι may point to gemination. Therefore, the lack of gemination of /r/ in
ἀρισωνιµ probably indicates a phonemic reality, indicating that /r/ was treated more like a guttural consonant.
- 223 -
It is also possible that µερεσθ reflects an intrusion of the spoken language, which preserved
geminated /r/ in some cases, similar to ‫ שֶׁרּ ֹאשִׁי‬in Song of Songs (5:2).273 Gemination of resh is
attested in Mishnaic Hebrew, usually in the eastern tradition. It is likely that resh with dagesh
in medieval manuscripts has roots at a time when Hebrew was still a living language (KHAN
2013d, 502–3). That is, instances of doubled resh in medieval manuscripts are not arbitrary,
but reflect the preservation of a feature that was characteristic of at least some spoken dialects
of Hebrew in which resh could still be doubled during the Second Temple period.
In sum, the evidence for the realization of /r/ could be interpreted to support either a
uvular realization or an emphatic apico-alveolar realization. An emphatic pronunciation of
resh is favored by the fact that it frequently lowers adjacent vowels and may cause the round-
ing of adjacent vowels, both of which are effects of emphasis attested in modern Semitic lan-
guages (KHAN 2013c, 387–88). However, the lowering of adjacent vowels could also be char-
acteristic of a uvular roll. Moreover, the rounding of vowels is attested not only with /r/ in
contemporary Hebrew, but also with the other sonorant consonants.
A uvular pronunciation of resh is favored by the fact that /r/ behaves like gutturals in
the Secunda transcriptions, with respect to both gemination and epenthetic vowels in the qVtl
form. If the behavior or resh in the Secunda was similar to that at Qumran, the weakening of
resh in the environment of gutturals (e.g., ‫ משע‬for ‫' משער‬from the gate of') also seems to sup-
port a uvular realization. In BOLOZKY's work on resh in Modern Hebrew, in which resh is real-
ized as a uvular, he had difficulty distinguishing between the words ‫' שעה‬hour' and ‫' שער‬gate',
because they were both pronounced something like [ʃaː] (2013, 390). However, it should be
noted that ‫ ע‬is no longer realized as a pharyngeal in modern Hebrew.
It may be that both pronunciations of /r/ go back to the Second Temple period. If we
assume that it was phonetic similarity to the gutturals that brought about the degemination of
273. Note also that both of these are examples of what KHAN terms "junctural gemination" rather than
"morphological gemination" (2013c, 387)
- 224 -
resh, it may be possible to draw a correlation between various traditions and the pronuncia-
tion of /r/. Tiberian Hebrew, in which /r/ is normally realized as a uvular [ʀ] with the emphat-
ic [r ̣] as an allophone, would be similar to the western tradition of Mishnaic Hebrew and
Palestinian Aramaic, in which the gemination of /r/ is virtually absent. The Babylonian tradi-
tion, in which /r/ is normally realized as an apico-alveolar [r], would be similar to the eastern
tradition of Mishnaic Hebrew, in which gemination of /r/ is more common. Therefore, while
the evidence is inconclusive, it seems more likely that /r/ was a uvular roll [ʀ] in the Secunda.
A transcription such as µερεσθ, just like ‫( שֶׁרּ ֹאשִׁי‬Song 5:2) in the Bible, may be indicative of
linguistic diversity and the influence of spoken language on the reading tradition.
The transcriptions of Jerome may serve as a test case for this theory. When he
transcribes Biblical Hebrew, r is not geminated (see YUDITSKY 2013, 806): e.g., merehem
‫מ ֵֶרחֶם‬. However, in a quotation of the lost Gospel of the Hebrews (or the Hebrew Gospel of
Matthew), he indicates that the Hebrew corresponding to Ὡσαννὰ ἐν τοῖς ὑψίστοις 'Hosanna
in the highest!' (Matt. 21:9) is osianna barrama (i.e., ‫ )הוֹשִׁיעָה־נָּא בּ ָָרמָה‬in Latin letters (Epistula
XX, 5.45). The quotation from the Gospel of the Hebrew with geminated rr may reflect a
more colloquial pronunciation than the biblical tradition without geminated rr in merehem.
6.3.6. Gutturals (/ʕ/, /h ̣/, /ʔ/, /h/)
At the earliest stage of Hebrew, there were six guttural consonants: a voiced pharyngeal frica-
tive /ʕ/ ([ʕ]), a voiceless pharyngeal fricative /h ̣/ ([ħ]), a voiced velar/uvular fricative /ġ / ([γ]
or [ʁ]), a voiceless velar/uvular fricative /ḫ / ([x] or [χ]), a voiceless glottal stop /ʔ/ ([ʔ]), and a
voiceless glottal fricative /h/ ([h]). Eventually, /ġ / and /ḫ / merged with /ʕ/ and /h ̣/, respective-
ly (i.e., /ʕ/, /ġ / > /ʕ/; /h ̣/, /ḫ / > /h ̣/), though there is evidence that /ḫ / remained distinct at least
in some dialects and/or registers in the Second Temple period (see 3.3.2). Another develop-
ment during the Second Temple period was the weakening of the gutturals, attested in loca-
tions such as Qumran, Beth She'an, and Ḥaifa (MOR 2013, 162–65). While the guttural conso-
- 225 -
nants maintained their pronunciation in the Middle Ages in the Tiberian reading tradition,
they were largely lost in Samaritan Hebrew.

6.3.6.1. Glottals and Pharyngeals: /ʔ/, /h/, /ʕ/, /h ̣/ = Ø
In the Secunda, gutturals are not represented directly. Typically, their presence is inferred:
αδαµ /ʔɔ̄dɔ̄m/ [ʔɔːðɔː̃ (m)] 'man' Ps. 49:3

αφαχθ /hɔ̄pakt/ [hɔːɸaχth] 'you turned' Ps. 30:12
εριµ /herʕı̄ m/ [hɪʀʕıː̃ (m)] 'he thundered' Ps. 29:3
ενναµ /h ̣ennam(m)/ [ħɪnːa(̃ m)] 'without a cause' Ps. 35:19
Their presence may be also be indicated by a hiatus between vowels:
αηλ /hɔ̄-ʔē l/ [hɔːʔeːl] 'the God' Ps. 18:31

νααρ /nɔ̄hɔ̄r/ [nɔːhɔːʀ] 'a river' Ps. 46:5
φααλθα /pɔ̄ʕaltɔ̄/ [p ɔːʕalt ɔː]
h h
'you made' Ps. 31:20
αωσιµ /ha(h ̣)-h ̣ō sı̄ m/ [haħoːsıː̃ (m)] 'those who take refuge' Ps. 18:31
In Greek transcription, the gutturals (Semitic gutturals and Latin h) are usually left unrepre-
sented. Only Semitic /ḫ / and /ġ / are transcribed on occasion, being represented by χ (or ξ in
Akkadian) and γ, respectively (5.3.6). At the same time, both /ḫ / and /ġ / may be left unrepre-
sented in transcription. Nevertheless, in light of general Hebrew transcription conventions,
the fact that neither χ for /ḫ / nor γ for /ġ / occurs in the Secunda is probably an indication that
the following mergers had occurred: /ḫ /, /h ̣/ > /h ̣/ and /ġ /, /ʕ/ > /ʕ/.
The gutturals /h/, /h ̣/ are twice rendered by ι word-initially (see YUDITSKY 2017, 31):
ιεσδι /h ̣esdı̄ / [ħɪsdiː] 'my mercy' Ps. 89:34

ιωσια /hō š ı̄ ʕɔ̄/ [hoːʃiːʕɔː] 'save!' Ps. 28:9
These transcriptions probably reflect an attempt by the scribe to approximate the guttural
sound, which was not present in Greek.
Although other contemporary Hebrew traditions show weakening of the gutturals, YU-
DITSKY has argued quite convincingly that the gutturals were still pronounced in the Secunda
and that they were probably realized similar to the gutturals in Tiberian (2008a; 2008b; 2017,
25–32). Therefore, they will be transcribed as [ʔ], [h], [ʕ], and [ħ] in phonetic transcription.
- 226 -
6.3.6.2. A Note on "Furtive" patah ̣
There is at least one case of an apparent epenthetic before a final pharyngeal in the Secunda:
χρηε /x-rē ʕ/ [kh(a)ʀeːʕ] 'like a neighbor' Ps. 35:14
Previous scholars have correctly concluded that "furtive" pataḥ does not exist in the Secunda,
citing forms such as θωσι ‫( תּוֹשִׁי ַע‬Ps. 18:28) and αββωτη ‫( ְוהַבּוֹ ֵט ַח‬Ps. 32:10). However, this
conclusion leads them to interpret the transcription χρηε as reflecting ‫ כּ ְֵרעֶה‬rather than ‫ כּ ְֵר ַע‬as
in the MT (BRØNNO 1943, 160, 294–95; YUDITSKY 2017, 198). While positing a textual variant
may certainly explain the form χρηε, this theory cannot explain the final ε in Origen's
transcription of the name Joshua (Ιωσουε ‫שׁ ַע‬

ֻ ‫[ י ְהוֹ‬Josh. 1:1]), the form ιαδαε ‫( י ָדַ ע‬MT ‫( )י ֵדַ ע‬Ps.
92:7), and, if correctly emended, the imperative σµαε* ‫שׁמַע‬

ְ (MT ‫שׁמַע‬
ָ ) (Ps. 28:6). I would sub-
mit, rather, that the ε in the forms χηρε, Ιωσουε, ιαδαε, and σµαε is merely the result of a per-
ceptual phenomenon with analogues in perception of modern Semitic.
It is not uncommon for students of Arabic to perceive a final high vowel before /ʕ/ as
a vowel sequence such as [ia] or [ea]. This is illustrated by a nineteenth-century orientalist
grammar of Arabic, notably before modern transliteration conventions, in which final v[+high]ʕ
(but not v[+high]ḥ) is transliterated with an additional lower epenthetic vowel: e.g., baeá for ‫ﺑــﺎﺋــﻊ‬
/bā ʔiʕ/ 'a seller', elbaddhá eea ‫ اﻟــﺒــﻀــﺎﯾــﻊ‬/el-badā yiʕ/ 'merchandise', but cf. rá eh for ‫ راﯾــﺢ‬/rā yih ̣/
(FULTON COMPTON HAYES 1859, 117, 129, 163).274 It is also worth noting that in a study of mis-
perception of Arabic consonants by English speakers, SANKER demonstrated that /ʕ/ was the
most common consonant misperceived, and one of the most common misperceptions of it
was to identify it as a vowel (usually /a/) (2015, 456).
Acoustic studies of Arabic gutturals have shown that at the transitional boundary be-
tween a vowel and /ʕ/, the first formant is especially high (i.e., the vowel is lower) and the
second formant becomes more characteristic of a more central vowel. The change at the V-C
274. Cf. also isra for ‫ اﺳـﺮع‬/israʕ/ 'hurry!' (110) tasma for ‫ﺗﺴﻤـﻊ‬
‫ ـ ـ‬/tasmaʕ/ 'you hear' (113), and errabee for ‫ اﻟـﺮﺑﯿــﻊ‬/er-
rabı̄ ʕ/ (126).
- 227 -
boundary is more pronounced before /ʕ/ than it is before /h ̣/ (BUTCHER and AHMAD 1987, 160;
MCCARTHY 1991, 79). Therefore, it would not be unreasonable to suggest that the ε in
transcriptions like χρηε and Ιωσουε reflects the phonetic reality of the transition to /ʕ/ or the
consonant /ʕ/ itself. This by no means constitutes a fully developed and consistent "furtive"
pataḥ, but rather a representation of how the phonetic reality at the V-C boundary was per-
ceived by the Greek accent. KNOBLOCH points out a similar phenomenon in the LXX
transcriptions in Genesis (1995, 414–16). Jerome also has similar forms with e instead of a:
e.g., thafue ‫( תַּ פּוּ ַח‬Josh. 15:53).

6.3.7. Semi-Vowels (/w/, /y/)
In ancient Hebrew, ‫ ו‬represented a labiovelar approximant [w] and ‫ י‬represented a palatal ap-
proximant [j]. In Tiberian Hebrew, ‫ ו‬came to represent a labio-dental [v] in most positions,
but remained a labio-velar approximant [w] when preceded or followed by a u-vowel (e.g.,
‫[ וּ ֻפוָּה‬ufuwˈwɔː] and ‫ י‬represented a palatal unrounded semi-vowel [j] (KHAN 2013a, 87–89).
6.3.7.1. Voiced Labio-Velar Approximant: /w/ = ου and υ
In the Secunda, when the conjunction waw /w-/ is followed by a vowel (13x) and thus conso-
nantal, it is represented by ουα or ουε (just ου for /w-/ is treated in 6.5.1.6):
ουαλσωνι /w-lš ō nı̄ / [walʃoːniː] 'and my tongue' Ps. 35:28

ουαθθεµας /wat-temʔas/ [wathːɪmʔas] 'and you rejected' Ps. 89:39
ουεβροβ /w-b-rob(b)/ [wɪβʀʊb]/[wəβʀʊb] 'and in the abundance of' Ps. 49:7
Word-medial /w/ is represented by υ or ου. After α, it is always represented by υ (10x):
λδαυειδ /l-dɔ̄wı̄ d/ [l(a)ðɔːwiːð] 'of David' Ps. 29:1

εσθαυου /heš tah ̣wū / h
[hɪʃt aħwuː] 'worship!' Ps. 29:2
αυωναν /ʕwō nam(m)/ [ʕawoːna(̃ m)] 'their iniquity' Ps. 89:33
In two instances, one after η and one after a consonant, word-medial /w/ is represented by ου:
βσαλουι /b-š alwı̄ / [b(ɪ)ʃalwiː] 'in my ease' Ps. 30:7

βγηουαθω /b-gē (ʔ)wɔ̄tō / [b(a)ʁeːwɔːθoː] 'in his pride' Ps. 46:4
The diphthong /aw/ is always represented by αυ (17x):
σαυ /š aw(ʔ)/ [ʃaw] 'vanity' Ps. 31:7

µαυθ /mawt/ [mawth] 'death' Ps. 49:15
- 228 -
βαναυ /bɔ̄naw/ [bɔːnaw] 'his sons' Ps. 89:31
In Palestinian Koine of the Roman period, the digraph ου represented the vowel [u]
(4.5.3.1.16–17). The grapheme υ represented the vowel [y] (or [ø]) (4.5.3.1.2–4). In the se-
quence αυ or ευ, the second element of the diphthong represented a phone somewhere on the
spectrum from [w] > [β]/[ɸ] > [v]/[f] (4.5.3.1.13–15). In transcription, [w] is most commonly
represented by ου both in Latin and Semitic. In Akkadian, however, [w] is transcribed by υ
(5.3.2.2). After Latin v [w] had shifted to [β]/[v], it is represented by β in Greek (5.3.1.2.4–5).
The diphthong [aw] in both Latin and Arabic is represented by αυ (5.3.1.2.2; 5.3.3.2.1). The
latest Latin transcriptions are quite instructive, in which consonantal v [β]/[v] is represented
by β, the diphthong au [aw] is represented by αυ, and the labiovelar qu [kw] is represented by
κου (also κο, κυ) (5.3.1.2.4). In Greek loanwords in the Mishnah, the second element of the
diphthongs αυ/ευ is represented by ‫ וו‬or ‫ ֿב‬in Hebrew (5.4.1.2.1).
In terms of distribution, the transcription of Latin names in Palestinian epigraphy is
instructive. Word-medially, Latin v ([w] > [β] > [v]) is represented by either Greek ου or υ:
e.g., Φλαυοιου Flavius, Σευηριναν Severina [CIIP I, no. 764]). Word-initially, Latin v is only
represented by Greek ου: e.g., Ουεττηνου Vettenus [CIIP I, no. 9], Ουικτορ [CIIP II, no.
1134]). However, once Greek β [β] (< *[b]) had become an adequate representation of Latin
v, Greek β may be utilized at the beginning of a word: e.g., Βερινης Verina [CIIP I, no. 859]
and Βικτωρος [CIIP III, no. 2432/2452]. We may explain this distribution as follows: Greek
ου [u] was utilized to represent the phoneme [w] because the high back rounded vowel [u]
was nearest to the labiovelar approximant [w]. Whether ου appeared word-initially or word-
medially, it retained the value of [u]. In the same way, the grapheme β was realized as a bil-
abial fricative [β] (nearest to Latin v after the shift of [w] > [β]/[v]) in all positions (except af-
ter nasals). The grapheme υ, on the other hand, would have represented the high front round-
ed vowel [y] word-initially and the value [w] > [β]/[ɸ] in the diphthongal sequences ευ and
- 229 -
αυ. However, because the second element of the Greek diphthongs αυ and ευ had become
consonantal, υ would have been an even better representation of [w] than ου after α/ε.
It is also important to remember that the use of Greek ου and (α)υ/(ε)υ to represent
Latin v in Latin proper names has its roots at a time when Latin v was pronounced as [w].
Therefore, after Latin v and Greek β shift to [β] (> [v]), the representation of Latin v in proper
names with Greek ου or (α)υ/(ε)υ instead of β does not necessarily indicate that ου represent-
ed [β] or [v], but that the historical spelling of the name was preserved.
A similar distribution is found in the third-century CE Arabic inscription from north-
eastern Jordan. When preceding a vowel, the semi-vowel [w] is represented by ου, ω, and ο:
αουα /wa/, ωα /wa/, αθαοα /ʔatawa/. The diphthong /aw/, on the other hand, is always repre-
sented with αυ: Αυσος /ʔaws/, α-δαυρα /ʔad-dawra/, and ειραυ /yirʕaw/ (5.3.3.2.1; 5.3.3.2.3).
The data of Greek orthography and transcription conventions for representing [w] is
entirely consistent with the distribution of ου and υ in the Secunda. Word-initially, Hebrew
/w/ is represented with ου, just as in Greek transcription of Latin names. Word-medially, /w/
is represented by υ if it can be represented in a Greek diphthongal sequence (i.e., αυ) but by
ου if it cannot (i.e., after η and after a consonant). Finally, the diphthong /aw/ is represented
by αυ, just as in Greek transcriptions of Latin in the language-learning texts from Egypt. This
likely indicates that ου and υ are different orthographic variants for representing [w] in the
Secunda, since the distribution of ου and υ corresponds to the distribution of representing
Latin v [w] in Greek transcription of Latin.275 Further, the fact that Hebrew /w/ is represented
by ου and not β confirms that Hebrew /w/ had not yet shifted to [v]. While cases of Latin v
(when Latin v = [β] or [v]) corresponding with Greek ου and (α)υ/(ε)υ in proper names at a
late period should be regarded as conservative historical spellings,276 there would be no rea-
275. YUDITSKY comes to a similar conclusion, drawing on the LXX and inscriptions (2017, 34–36).
276. Note that many of these Latin names were first rendered in Greek at a time when Latin v represented [w].
- 230 -
son for Hebrew /w/ to be transcribed by Greek ου unless it was still realized as [w].277 For the
presence or lack of a vowel following the conjunction waw (ου-) in the Secunda, see 6.5.1.61.
In addition to the data explored above, /w/ may be unrepresented in the transcription:
αων /ʕwō n/ [ʕawoːn] 'the iniquity of' Ps. 49:6

ουµσωθαϊ /w-mas ̣wō tay(y)/ [(ʔ)um(ɪ)ʦˀ(w)oːθaj] 'and my commandments' Ps. 89:32
σφωθαϊ /ś pō tay(y)/ [sɸoːθaj]/[s(ɪ)ɸ(ʔ)oːθaj] 'my lips' Ps. 89:35
YUDITSKY interprets these transcriptions as evidence that the semi-vowel /w/ had become
weakened, drawing on comparative examples such as ‫ מצאות‬and ‫ שפאותיכה‬from the Dead Sea
scrolls (2017, 36).278 While YUDITSKY's explanation is entirely possible, especially in light of
contemporary Hebrew evidence, Greek transcription conventions may shed further light.
In Palestinian and Egyptian Koine, one of the indicators that the phone represented by
the grapheme (α)υ/(ε)υ has shifted from [u] to [w] is that it ceases to be represented in the or-
thography (4.4.1.2.3; 4.5.3.1.13–14). In transcription, although /w/ is usually represented,
there are also examples of /w/ left unrepresented in Latin (e.g., νωεµ[β]ερ November)
(5.3.1.2.4), Akkadian (e.g., ωει ū wı̄ ) (5.3.2.2), and Arabic (e.g., Ροεος /ro(w)eyh ̣/) (5.3.3.2.3).
In each example, the /w/ is in the environment of a back rounded vowel. It seems that in these
instances it is the transition between a back rounded vowel and another vowel that approxi-
mates the labio-velar semi-vowel [w]. This theory is supported by transcriptions of Latin
names such as Φλαουβίου Flavius (120 CE) and Οὐβαλέρις Valerius (4th CE) in the Egyptian
papyri (GIGNAC 1976, 69). The fact that consonantal β [β] intervenes between ου and the fol-
lowing vowel indicates that it was the transition between the two vowels that approximated
the semivowel [w] in Greek orthography and not the digraph ου itself. The function of the ου
digraph is merely to provide the back rounded articulation, which is why ου and ω can serve
277. Note how in Nikolaos of Otranto's (12th/13th CE) Greek transcriptions of the Biblical Hebrew reading
tradition in Italy, in which consonantal waw was pronounced as [v] (RYZHIK 2013, 363), -ְ‫ ו‬is transcribed as β,
which represented [v] in contemporary Greek: e.g., βεεθ ‫( ְואֶת־‬Gen. 1:1) (Disputatio contra Judaeos, 5.11).
278. My transcriptions of ουµσωθαϊ and σφωθαϊ are based on YUDITSKY's reconstruction (2017, 36).
- 231 -
that purpose in Arabic (e.g., αουα and ωα for /wa-/). As the articulators move between the
back rounded vowel [u] and the adjacent phones, a (near) glide is created.
It is also entirely consistent with modern linguistic perceptual studies to suggest that a
vowel sequence with a back rounded vowel may be phonetically equivalent to that same se-
quence with the semi-vowel [w] substituted for the back rounded vowel. For example, in a
perceptual study of Romanian, CHITORAN has shown that there is no phonetic difference be-
tween the sequences [wa] and [oa] (2002, 221). Instead of the weakening of the semi-vowel
[w], a transcription like αων may be just as easily regarded as the Greek scribe's perception of
Hebrew [ʕawoːn], especially in light of the variant form αυωναν. One may also compare
Greek transcription of Latin forms like ουας ū vā s 'grapes' (P.Lond. II 481). In fact, all the
above transcriptions exhibit the apparent weakening of /w/ in the environment of a back
rounded vowel. Therefore, in phonetic transcription, [w] in these words is enclosed in paren-
theses, with the understanding that it may have been pronounced.

6.3.7.2. Palatal Approximant: /y/ = ι
In the Secunda, the palatal approximant /y/ [j] is usually represented by ι (or ϊ):
ωϊηβ /ʔō yē b/ [ʔoːjeːβ] 'an enemy' Ps. 31:9

φεδιων /pedyō n/ [phɪðjoːn] 'the redemption of' Ps. 49:9
ϊαδω /yɔ̄dō / [jɔːðoː] 'his hand' Ps. 89:26
The sequence /y/ + vowel is often represented by Greek ι (or ϊ) (see YUDITSKY 2017, 32–33):
αϊθι /hɔ̄yı̄ tı̄ / [hɔːjiːθiː] 'I was' Ps. 30:8

ισρη /yaš rē / or /yiš rē / [jiʃʀeː] 'those upright of' Ps. 32:11
ισροφ /yesrop/ [jisʀʊɸ] 'he will burn' Ps. 46:10
At the time of the Secunda, Greek ι represented the vowel [i] (4.5.3.1.1). Greek ϊ with trema
(ϊ) indicated that the ϊ was to be read as an individual grapheme distinct from the preceding
vowel and not as a digraph including the previous vowel (4.5.3.1.10).279 In transcription, both
Latin and Semitic /y/ [j] are represented by Greek ι. In Arabic, word-initial /yi/ is once repre-
279. It is worth noting that while JANSSENS claims that trema was added only in the eighth or ninth century CE
(1982, 38–39), ι with trema (ϊ) is attested already in the ancient papyri (see THOMPSON 1966, 63).
- 232 -
sented by ει (e.g., ειραυ /yirʕaw/), the contemporary phonetic value of which was [i]
(5.3.3.2.4). This may indicate that, from the perspective of Greek, word-initial [ji] was per-
ceptually equivalent to [i] (or [ʔi]).280 It is actually probable that certain prefix forms of the
piˁel should be interpreted as reflecting [ʔi(ː)] instead of [ji]: e.g., ιµαλλετ ‫( י ְ ַמלֵּט‬Ps. 89:49)
may represent */ymallet/̣ > /ı̄ mallet/̣ [ʔiːmalːɪtˀ]. A similar phenomenon occurs in Old Baby-
lonian manuscripts (YEIVIN 1985, 523–27; KHAN 2013e, 955). However, other transcription
pairs such as ιδαββερ ‫( י ְדַ בֵּר‬Ps. 46:10) and ουϊεδαββερ ‫( * ַוי ְדַ בֵּר‬Ps. 18:48) may suggest that
perhaps it is better to interpret the verbal prefix ι as an allophonic variant of ιε, representing
[ji] and [jɪ], respectively (for a full discussion of these forms, see 6.5.1.4.3).
The diphthong /ay/ is usually represented by αϊ:281
ελωαϊ /ʔlō hay(y)/ [ʔɪloːhaj] 'my God' Ps. 18:29

µαϊµ /maym/ [majm] 'waters' Ps. 32:6
αλαϊ /ʕɔ̄lay(y)/ [ʕɔːlaj] 'against me' Ps. 35:16
In contemporary Greek orthography, it is necessary to distinguish Greek αι from αϊ (with
trema). In Roman Palestinian Koine, the digraph αι signified the vowel [ε], which is reflected
by interchanges such as κε for και (4.5.3.1.10). That αι (without trema) represented [ε] is also
indicated by the fact that Greek καιρός is rendered as ‫ קֵירוֹ ֿס‬in Mishnaic Hebrew (5.4.1.1.2).
The digraph αϊ signified the sequence [ai] throughout the Koine period.282 This is demonstrat-
ed by interchanges such as τροπαιεικον for τροπαϊκόν (4.5.3.1.10). In transcription, the diph-
thong /ay/ is regularly represented by αι (or/for αϊ) in Latin and Semitic (5.3.6). Therefore, it
is entirely consistent with the conventions of contemporary Greek orthography and transcrip-
tion for the digraph αι/αϊ to signify [aj] in the Secunda (contra JANSSENS 1982, 20–21).
280. See note 67.
281. Out of the 44 times that the Hebrew diphthong /ay/ is represented by αι/αϊ in the Secunda, 41 of them have
trema and 3 are without trema. It is highly likely that the 3 examples without trema are due to scribal error. Two
small dots above a letter are very easily obscured, omitted, or erroneously added during transmission. The need
for trema to indicate /ay/ shows that the Secunda was composed at a time when Greek αι (without trema)
signified [ε] and not [ai].
282. The trema (¨) might not always be written in inscriptions, but it would be preserved in pronunciation.
- 233 -
The diphthong /ay/ is also transcribed once by εϊ and twice by η:
ιλεϊ /ʔē lay(y)/ [ʔeːlaj] 'to me' Ps. 31:3

ωεβη /ʔō y(e)bay(y)/ [ʔoː(j̞ )ɪβaj] 'my enemies' Ps. 35:19
σωνη /ś ō n(e)ʔay(y)/ [soːnʔaj] 'those who hate me' Ps. 35:19
In Roman Palestinian Koine, it is necessary to distinguish between the digraph ει, which rep-
resented the same vowel as ι = [i], and the digraph εϊ (with trema) (4.5.3.1.1), which would
have represented the sequence [εi]. The grapheme η would have represented [e] (4.5.3.1.7).
In transcription, while the sequence ει is normally used to represent a vowel of the [i] quality
(usually [iː]), it is used a couple times in Greek transcription of Latin to represent the se-
quence [ei] or [eːi] in texts from the first and second century CE (5.3.1.1.3). It is also used in
Akkadian to represent [eː] (5.3.2.1.2). The Arabic diphthong /ay/, in addition to its typical
rendering as αι, is rendered at times by ει and η/ε (5.3.3.2.2). AL-JALLAD interprets these vari-
ant renderings not as representing monophthongizaion but rather a raised allophone [ei] of the
diphthong [ai]. In light of the fact that /y/ [j] tends to raise adjacent vowels in the Secunda
(see YUDITSKY 2017, 96–98), this explanation is probably valid for the Secunda as well. This
theory may find support in various transcription conventions for names from the base /zon-
ayn/ in Palestinian epigraphy: e.g., Ζονενος, Ζοναινου, Ζονηνωνος (CIIP III, no. 2425, 2443,
2445–6, 2469). Also, in a cross-language perceptual study, it was found that Italian speakers
identified the English diphthong [ei] with Italian [e] (FLEGELAN and MEADOR 1999, 2980).
In addition to the data cited above, /y/ is sometimes not represented by ι:
ουεµιναχ /w-ymı̄ nɔ̄k/ [(ʔ)u(j̞ )ɪmiːnɔːχ] 'and your right hand' Ps. 18:36
ωεβη /ʔō y(e)bay(y)/ [ʔoː(j̞ )ɪβaj] 'my enemies' Ps. 35:19

ουεζρα /w-yeʕzrɔ̄(h)/ [(ʔ)u(j̞ )ɪʕzʀɔː(h)] 'and he will help her' Ps. 46:6
εωσηβ /yō š ē b/ [(j̞ )oːʃeːβ] 'dweller' Ps. 49:2
αεα /hɔ̄yɔ̄/ [hɔː(j̞ )ɔː] 'he was' Ps. 89:42
YUDITSKY argues that these transcriptions attest to the assimilation of the glide [j] to the previ-
ous vowel or the shift of the glide [j] to a glottal stop [ʔ] (2017, 32–33). Such an explanation
- 234 -
is entirely valid and supported by contemporary Hebrew evidence,283 but an examination of
the Greek evidence may offer a more complete perspective.
In Palestinian and Egyptian Koine, the omission of ι in orthography is actually one of
the indications that the vowel [i] has shifted to [j] in a particular word. For example, the
spelling κυρα (for κυρία) is usually interpreted as reflecting the following change: [kyria] >
[kyrja] ( > [kyra]) (4.5.3.1.19; GIGNAC 1976, 302–3; HORROCKS 2014, 169). In addition to the
omission of ι as an indication of [j], there are also examples in which the semi-vowel [j]
seems to be indicated by the sequencing of vowels (e.g., Μαριεαµη [for Μαριαµη = Hebrew
‫)]מרים‬. In Greek transcription of Arabic, the glide [j] may also go unrepresented: e.g.,
Μοεαρος /moġ e(yy)ar/ (5.3.3.2.4). It is unlikely that a geminated /yy/ would be weakened;
rather, it seems that the Greek scribe regarded that particular sequencing of vowels as the best
way to approximate the non-Greek phoneme [j]. Finally, [j] may be omitted in the transcrip-
tion of the Arabic dipthong /ay/ when it is realized as its raised allophone [ey] (5.3.3.2.2).
In Koine Greek of Asia Minor, ε-αι = /ε/ (or /e̞ /) before another vowel is allophonical-
ly realized as /i/: e.g., θιᾶς for θεᾶς. Moreover, ε or ι followed by another vowel eventually
resulted in a shift to the palatal semivowel [j] (BRIXHE 2010, 233). Accordingly, sequences
such as εω and εα, at a certain stage of Greek in Asia Minor, were realized as [jo] and [ja].
The same phenomenon may be attested in the spelling Γειωργιου (for Γεωργιου) in Palestin-
ian epigraphy (CIIP III, no. 2143).
In light of the evidence of Greek orthography and transcription conventions, it is pos-
sible that a vowel sequence such as εα in αεα (for /hɔ̄yɔ̄/ [hɔː(j̞ )ɔː]) was actually perceptually
equivalent to [jaː]/[jɔː] for the Greek scribe. The use of the sequence εα to represent [ja] is ac-
tually supported by modern linguistic perceptual studies. For example, in CHITORAN's work on
283. MOR has an excellent treatment of such forms. In the Judaean Desert texts, /y/ may be represented as ‫א‬
word-initially or word-medially when it precedes /e/ or /ə/: e.g., ‫ שאש‬for ‫' שיש‬that there is' and ‫ אשעיה‬for ‫ישעיה‬
'Isaiah' (2015, 125–26). See also the inscriptions Εισµαηλ ‫( אשמעל‬for ‫( )ישמעאל‬CIIP I/1, no. 543) and ‫( גאוס‬for
Latin Gaius) (CIIP I/1, no. 60) from Jerusalem of the Second Temple period .
- 235 -
glide-vowel sequences in Romanian, she found that although the sequence [ea] is phonetical-
ly distinct from [ja], it is prone to be misidentified as [ja] more than 20% of the time, stating
at the outset that "impressionistically, [ja] and [ea] have very similar pronunciations" (2002,
219–221). Therefore, in our phonetic transcription, [j] in these words is enclosed in parenthe-
ses, with the understanding that it may have been fully pronounced. In those cases for which
YUDITSKY suggests that the glide has assimilated to the previous vowel (e.g., αεα), we might
also render /y/ in IPA transcription as [j] with a downtack ([j̞ ]) (i.e., [hɔːj̞ ɔː]), indicating a low-
er tongue position in the articulation of the palatal approximant.
In light of these points, we may also posit that the reason ι may signify the glide [j]
and a following vowel in the Secunda is because ι is inherently vocalic and the glide element
[j] is actually derived from the behavior of the articulators as they move to and/or from the
high front vowel [i], just as was the case with ου [u] and the glide [w]. Modern linguistic
studies support the concept of the transition from one articulatory position to the next being
interpreted as a glide. For example, for some English speakers, the sound between the [g] and
the [r] in the word guarantee [ˌgærənˈtiː] is identified as a glide [j] because of the transitions
of the F1, F2, and F3 formants from [g] to [r] (ESPY-WILSON 1987, 187). Also, in Greek loan-
words in the Mishnah, the sequence ια is often rendered in Hebrew with a consonantal yod:
e.g., σπεκλάριον ‫ ְספַק ל ְַרי ָיה‬and ἐµπίλια ‫( אַנְפֵּי ְלי ָא‬HEIJMANS 2013, 262; 5.4.1.1.4).
Finally, /y/ is omitted in transcription in two instances without vowel sequencing:
...(ϊ) αγι /yaggı̄ h/ [jagːiːh] 'he will illuminate' Ps. 18:29
βακαρ /b-yqɔ̄r?/ [ba(j)kˀɔːʀ]/[bɔːkˀɔːʀ] 'in luxury' Ps. 49:13
It is possible that each of these transcriptions reflect a pattern different from that of the MT,284
but there are also alternative explanations. The first transcription may be Sandhi writing (con-
text: ελωαι αγι [ʔɪloːhaj (j)agːiːh]). The second transcription may be the result of an /á y/ > /a/
284. αγι may reflect ‫ ָהגִי ַהּ‬and βακαρ may reflect ‫( ָבּקָר‬YUDITSKY 2017, 42, 164–65, 189).
- 236 -
sound change, which is also attested in Targum Onqelos and Babylonian Aramaic (GARR
1991). Because the vowel with the preposition /b-/ in the Secunda is /a/, the transcription
βακαρ might have developed as follows: /b-yqɔ̄r/ [bajəkˀɔːʀ] > [bajkˀɔːʀ] > [baːkˀɔːʀ].
It is worth noting here, with respect to the transcription of Hebrew /w/ [w] and /y/ [j] in par-
ticular, how much the transcriber was working within typical Greek orthographic practices.
For example, he only used υ to signify Hebrew /w/ when it followed α, because Greek υ only
signified [w] (or [β]/[ɸ] > [v]/[f]) in the diphthongal sequences αυ and ευ. Elsewhere, he used
Greek ου for /w/. If one were inventing an entirely new system, we would expect to find one-
to-one consonantal correspondence. However, the distribution of the various transcriptions of
Hebrew /w/ demonstrates that, although the transcriber knew Hebrew well, he was approach-
ing the Hebrew perceptually through his Greek accent and orthography. This comes through
no clearer than in the transcription εσθαυου (Ps. 29:2). Even though /w/ does not immediately
follow /a/ in the Hebrew—the guttural /h ̣/ intervenes—the scribe transcribes /w/ with υ be-
cause it follows α in the Greek. Finally, while it is indeed likely that the glides had weakened
as in other contemporary attestations of Hebrew, it is also possible that their occasional omis-
sion in the Secunda may actually be a feature of Greek orthography, consistent with epigraph-
ic evidence and modern linguistic studies on speech perception.

6.3.8. Consonant Gemination (Cː or CC)
Gemination (doubling) is essentially consonantal length. Acoustically, it involves holding the
closure for a longer duration than the corresponding singleton consonant, sometimes as little
as around 1.2x longer and sometimes as much as 2x or 3x longer. Gemination is a phonemic
category, with varying phonetic durations required to signify phonemic gemination relative to
context (pause, nuclear stress, etc.).285 Early in Hebrew, all consonants could be geminated.
285. See the discussion on gemination in Neo-Aramaic in KHAN (2004, 51–52).
- 237 -
At various points during the history of Hebrew, the gutturals and resh lost the ability to be
geminated: the process began with ‫ א‬and ‫ר‬, then ‫ ע‬and ‫ה‬, and finally ‫( ח‬BLAU 2010, 82–83).
6.3.8.1. Regular Gemination
In the Secunda, double consonants are usually signified by two graphemes in Greek:
αννωθην /han-nō tē n/ [hanːoːθeːn] 'he who gives' Ps. 18:48

λεββι /lebbı̄ / [lɪbːiː] 'my heart' Ps. 28:7
ουµεσσιρι /w-meš -š ı̄ rı̄ / [(ʔ)umɪʃːiːʀiː] 'and from my song' Ps. 28:7
εδαλλεγ /ʔdalleg/ [ʔɪðalːɪʁ] 'I will leap' Ps. 18:30
οκκωθαϊ /h ̣oqqō tay(y)/ [ħʊkˀːoːθaj] 'my statutes' Ps. 89:32
In Palestinian Koine of the Roman period, many spelling interchanges attest to the fact that
consonantal length was no longer phonemic (4.5.3.1.23). Previous minimal pairs, such as ἄλη
/alē / [ˈaleː] 'wandering' and ἄλλη /allē / [ˈalːeː] 'elsewhere' were no longer distinct in pronun-
ciation, both being realized as [ˈale]. Nevertheless, in transcription, gemination is often repre-
sented. In Greek transcription of Latin, gemination is represented about 80% of the time
(5.3.1.3.8). In Greek transcription of Akkadian, gemination is only represented about 30% of
the time (5.3.2.3.8). In Greek transcription of Arabic, while gemination is normally represent-
ed, sometimes it is not (5.3.3.3.9). In a few instances in Latin and Akkadian, singleton conso-
nants are represented as geminates. Therefore, in light of contemporary Greek orthography
and transcription conventions, the inconsistent representation of gemination (see below) in
the Secunda is not surprising.
YUDITSKY conducts a thorough analysis of gemination in the Secunda. In general, he
tends to accept that gemination in the transcription reflects gemination in Secunda Hebrew
and lack of gemination in the transcription reflects lack of gemination in Secunda Hebrew.
He does, however, admit a number of exceptions (see below). Instances of incongruity be-
tween the representation of gemination in the Secunda transcriptions and the presence or lack
of gemination in the history of Hebrew (or in other attestations in the Secunda) he explains
phonetically either as degemination or secondary gemination (2017, 36–44).
- 238 -
While YUDITSKY has provided a valuable and potentially accurate account of how
degemination and secondary gemination might be explained phonetically in the Secunda, I
will proceed by focusing on some perceptual elements that may also have played a role in the
representation of gemination in the Secunda.286 It should be noted that YUDITSKY's approach
finds support in parallel cases of degemination and secondary gemination in Neo-Aramaic
(see KHAN 1999, 57–61; 2002, 58–61; 2004, 52–55; 2008, 40–42; 2016, 195–200; FASSBERG
2010, 29–30). My emphasis on perceptual factors is intended to complement his work.
The light that modern linguistics has shed on the acoustic nature of the singleton/gem-
inate contrast is also worth mentioning here. Although phonologically we tend to regard gem-
inate consonants as C[+long] (/Cː/) and singleton consonants as C[-long] (/C/), the durational
ratio between a geminate and singleton consonant may vary inasmuch as it depends on a
number of factors: type of consonant (stop, nasal, sibilant, etc.), vocalic context, speed of
speech, etc. (AOYAMA and REID 2006). Moreover, in a transcriptional context, we are dealing
with the perception of consonant duration from the perspective of a Greek accent and not
necessarily a native phonological conception of gemination. All these factors need to be taken
into account in an examination of the various representations of gemination in the Secunda.

6.3.8.2. Gemination of /w/, /y/, and /z/
While most instances of gemination or lack thereof in the Secunda are expected,287 there are a
number of contexts in which gemination is not regularly represented. First, the consonants
/w/, /y/, and /z/ are never geminated in transcription:

αϊη /ʔayyē / [ʔajːeː] 'where?' Ps. 89:50
εχαζεβ /ʔkazzeb/ [ʔɪχazːɪβ] 'I will lie' Ps. 89:36
286. See also SPEISER (1932–33) and JANSSENS (1982, 44), albeit with a less sophisticated linguistic framework.
287. Just as gemination is usually represented as expected, the lack of gemination is also usually represented as
expected. YUDITSKY shows that gutturals and /r/ are not geminated in the transcriptions. There are a couple
ambiguous cases in which /h ̣/ and /h/ do not exhibit compensatory lengthening, but even these are parallel with
the same forms in the Tiberian tradition (2017, 39–40). For the one possible case of /r/ doubling, see 6.3.5.2.
- 239 -
Although these consonants were actually geminated in Secunda Hebrew, YUDITSKY suggests
that they were not doubled in the transcription because υ, ι, and ζ were not doubled in Greek
orthography. He points out that both υ and ι represented double consonants after a vowel (-
ayy-, -aww-) and that ζ represented the sequence /zd/ (2017, 40).
Although υ, ι, and ζ are not doubled in standard Greek orthography, there are attesta-
tions of ζζ in Palestinian Greek orthography and transcription: e.g., τευχιζζει (for τευχιζει)
and αζζανα (for ‫) ַחזָּנָא‬. By the time of the Secunda, the grapheme ζ /zd/ had shifted to /zz/ and
then simplified to /z/ (4.5.3.1.23; 4.5.3.1.34). In Greek transcription of Arabic, gemination of
the glides /w/ and /y/ is never indicated in transcription (5.3.3.3.9). Cross-linguistically,
glides typically have the smallest durational difference between singleton and geminate con-
sonants (AOYAMA and REID 2006). In spite of the couple examples of ζζ in Palestinian epigra-
phy, we may conclude that the gemination of /w/, /y/, and /z/ is not represented in the Secun-
da in accordance with standard Greek orthography.

6.3.8.3. Gemination in Sonorous Segments
Second, gemination may not be indicated in segments with higher sonority:
ονηνι /h ̣onnē nı̄ / [ħʊnːeːniː] 'have mercy on me!' Ps. 31:10

µενεγδ /men-negd/ [mɪnːɪʁd] 'from before' Ps. 31:23
ραβιµ /rabbı̄ m/ [ʀabːıː̃ (m)] 'great' Ps. 32:6
φαλητ /pallē t/̣ [phalːeːtˀ] 'deliverance' Ps. 32:7
µεµµενι /mem-mennı̄ / [mɪmːɪnːiː] 'from me' Ps. 35:22
θελαθαχ /thellɔ̄tɔ̄k/ [thɪhɪlːɔːθɔːχ] 'your praise' Ps. 35:28
ιαµιµ /yammı̄ m/ [jamːıː̃ (m)] 'seas' Ps. 46:3
ααµιν /hɔ̄-ʕammı̄ m/ [hɔːʕamːı(̃ m)] 'the peoples' Ps. 49:2
αµιµ /ʕammı̄ m/ [ʕamːıː̃ (m)] 'peoples' Ps. 89:51
Also, in segments with high sonority, singleton /b/ and /m/ may be doubled in transcription:
οϊεββαϊ /ʔō y(e)bay(y)/ [ʔojɪβaj] 'my enemies' Ps. 30:2

λεββαβεχεµ288 /lbɔ̄b(ə)kem(m)/ [lɪβɔːβəχɪ(̃ m)]/[lɪβɔːβɪχɪ(̃ m)] 'your heart' Ps. 31:25
288. It is likely that the form λεββαβεχεµ is a forma mixta, combining **λεββεχεµ with λεβαβεχεµ (see
YUDITSKY 2017, 41).
- 240 -
θαµµιν /tɔ̄mı̄ m/ [thɔːmıː̃ (m)] 'blameless' Ps. 18:31
θαµµιµ /tɔ̄mı̄ m/ [thɔːmıː̃ (m)] 'blameless' Ps. 18:33
σαµµαϊµ /š ɔ̄maym/ or /š ammaym/ [ʃɔːmajm]/[ʃamːajm] 'heaven' Ps. 89:30
YUDITSKY devotes a discussion to the irregular behavior of gemination as it relates to the labi-
als /b/ and /m/. According to him, the representation of gemination or lack thereof around the
labials actually reflects variant forms in the Hebrew. He derives the following rule: the labials
/b/ and /m/ tend to be geminated or lose their gemination especially when they appear at least
twice in the same word (2017, 40–41). YUDITSKY highlights an important phenomenon, but in
my opinion, it is better to classify this group in terms of high sonority.
In Roman Palestinian Koine, similar examples are attested both in general orthogra-
phy and transcription: e.g., [α]ν[ο]ιωγµµενον (for ἀνεῳγµένον), Αµια (for Αµµια), Βεννιαµιν
(for Βενιαµιν ‫) ִבּנְיָמִן‬, δεκαενεα (for δεκαεννεα), and Ραβι (for Ραββι ‫)רבִּי‬
ַ (4.5.3.1.23). In
transcription of Punic, there are similar pairs such as Θινιθ alongside Θεννειθ (5.3.5.2).
The first group of Secunda transcriptions, in which etymological gemination is unrep-
resented, may be explained in light of the phonetic context. Cross-linguistically, the gem-
inate-to-singleton durational ratio tends to be higher with sonorant consonants (with the ex-
ception of semivowels, see AOYAMA and REID 2006) than most other types of consonants.
Because sonorant consonants exhibit greater similarity with vowels, which are inherently at
the top of the sonority hierarchy, a greater duration is presumably necessary to set off the
geminate ([+long]) from the surrounding context. In a study of geminates in Arabic, KAWA-
HARA found that sonorant consonants tend to be more difficult to perceive than obstruent con-
sonants. Moreover, as the sonority of a segment increases, the difficulty of perceiving a gem-
inate sonorant also increases (2007, 1, 57).289 In a perceptual study of geminates, HARDISON
and SAIGO found that a greater difference in sonority between the geminate consonant and the
surrounding vocalic context aided perception. When geminates are misperceived, they may
289. But cf. DMITRIEVA (2012, 137), who argues against the general consensus regarding sonority and geminates.
- 241 -
be perceived as a long vowel and singleton consonant (2010, 81).290 Most of the mispercep-
tions of gemination above occur with a sonorous geminate or singleton consonant in a highly
sonorous context (e.g., adjacent to glide, adjacent to liquid/nasal, adjacent to nasalized vow-
el). The high level of sonority in the segment likely obscured a plain distinction between the
geminate or singleton consonant and the surrounding environment.

6.3.8.4. Gemination in Sibilants
Third, gemination of sibilants is often unrepresented in transcription (see SPEISER 1932, 261):
ουεσιγηµ /w-ʔeś ś ı̄ gē m/ [(ʔ)uʔɪsːiːʁeːm] 'and I will catch them' Ps. 18:38
εσιληνι /hes ̣s ̣ı̄ lē nı̄ / [hɪʦˀːiːleːniː] 'save me!' Ps. 31:3
ιαροσου /yɔros ̣s ̣ū / [jɔːʀʊʦˀːuː] 'they will oppress' Ps. 49:14
χασαµς /kaš -š amš / h
[k aʃːamʃ] 'like the sun' Ps. 89:37
σασουου /š assū hū / [ʃasːuːhuː] 'they plundered it' Ps. 89:42
αλ�µα�σαυ /ʕal maš -š aw/ [ʕal maʃːaw] 'on account of what vanity?' Ps. 89:48
There is also evidence in Palestinian Koine for singleton/geminate alternations of sibilants:
e.g., εξεσστω (for εξεστω) and τεσερεσκαιδεκατου (for τεσσαρεσκαιδεκατου). Cross-linguis-
tically, sibilants tend to be longer than other types of consonants and sibilants also tend to
have a lower geminate to singleton durational ratio than other consonant types (BLEVINS 2004;
DMITRIEVA 2012, 20). In a perceptual study of geminated /tt/, /kk/, and /ss/ followed by the
vowels /a/ or /u/, it was found that /ss/ + /u/ was the hardest sequence in which to identify the
geminate consonant. This is because the sequence /ss/ + /u/ has the smallest difference in
sonority between the consonant and the vowel (HARDISON and SAIGO 2010, 81, 85, 95). At
least in the case of the transcriptions ιαροσου and σασουου, these principles likely apply.
A couple transcriptions in the Secunda, ϊεσσι ‫שׁעִי‬

ְ ִ ‫( י‬Ps. 18:47) and ουνεσσαφου ‫ְונֶ ֶאסָפוּ‬
(Ps. 35:16), may attest to the assimilation of a guttural to an adjacent sibilant: ʕ > C1 / C1[+sibi-
290. The reverse of this, misperceiving a long vowel and a singleton as a short vowel and a geminate, is also
possible. Anecdotally, I may cite my experience learning Arabic, during which I remember hearing the word
/muġ ā marā t/ 'adventures' and misperceiving and misproducing it as /muġ ammarā t/. Curiously, this
misperception also occurred in the environment of a highly sonorous segment (i.e., low vowel and nasal).
- 242 -
lant] _ and ʔ > C2 / _C2[+sibilant]; it is more likely, however, that they should be corrected to ϊεσει*
and ουνεεσαφου* (see YUDITSKY 2017, 40, 81–82).
The transcriptions ασσακερ ‫שׁקֵּר‬

ַ ‫( ֲא‬Ps. 89:34) and ασσανε ‫שׁנֶּה‬
ַ ‫( ֲא‬Ps. 89:35) exhibit a
double σσ /š š / in an unexpected context, making them appear as niphal forms instead of piel
forms. YUDITSKY also mentions the form θεσσαβερ ‫שׁבֵּר‬

ַ ְ‫( תּ‬Ps. 48:8),291 found in a quotation in
Chrysostom. These forms are unusual in two ways: (1) the first radical is doubled in the piel
prefix forms and (2) the vowel of the prefix is α instead of ε (cf. εχαζεβ ‫) ֲא ַכזֵּב‬. YUDITSKY com-
pares these forms to the secondary doubling of the initial radical in Syriac 1cs forms (ʔeqqa-
tel), but does not believe the Secunda has been influenced by Syriac. Nevertheless, his com-
ment that the first radical of all three forms is /š / is significant (2017, 42, 152–53). We may
also add to this list the most recent, though uncertain, reading of Ps. 46:10 in the Ambrosiana
palimpsest, which has ισσαβερ for the piel form ‫שׁבֵּר‬

ַ ְ ‫י‬. Because all of these examples involve
a sibilant and a high vowel, it may be explained perceptually as above. However, the unifor-
mity of this change suggests that it may be a more well-defined phonological phenomenon.
While the Syriac forms that YUDITSKY mentions occur irrespective of the type of con-
sonant, other Aramaic phenomena offer better parallels. HUEHNERGARD has shown that the
doubling of /š / occurs in the environment of #Ci_V, citing the following examples: Common
Aramaic *ʔiš š -a/ā t- 'fire' < *ʔis-ā t; Syriac neš š in 'women' < nis-ı̄ na and qeš š at 'bow' (cstr. of
qeš tā ) (2017?, 7). Because the prefix vowel in the piˁel stem was originally /i/ in Hebrew
(STEINER 1980), all of these forms in the Secunda would correspond with this limited sound
change in Aramaic. The New Testament Greek form Μεσσίας ‫' ָמשִׁי ַח‬Messiah' may also be rel-
evant here: *maš ı̄ ḥ > (vowel raising in environment of sibilant) > meš ı̄ ḥ > (š > š š / #Ci_V) >
meš š ı̄ ḥ. Finally, KHAN points out that in the Neo-Aramaic dialect of the Jews of Arbel, phono-
logical /liš ā na/ 'tongue' is realized variously with respect to gemination: [lɪʃɑːˈnæ],
291. YUDITSKY incorrectly cites this as Psalms 46:8 (2017, 153).
- 243 -
[lɪʃʃɑːˈnæ], [liːʃɑːˈnæ] (KHAN 1999, 58). A form in the Kaufmann manuscript of the Mishnah
also attests to a comparable phenomenon: ‫שּׁמֵּר אֶת גַּגּוֹ‬

ַ ְ ‫שׁיּ‬
ֶ ‫' עַד‬until he keeps watch on his roof'
(Tohar. 9:6, folio 265v). The lack of doubling in the middle radical in the forms mentioned
above (θεσσαβερ, ισσαβερ, etc.) may be due to subsequent confusion with nifˁal after the
doubling of the initial radical. The prefix α in the forms ασσακερ and ασσανε is difficult to
explain. Nevertheless, it is possible that these forms exhibit influence of Aramaic phonology.
6.3.8.5. Gemination of /t/
Fourth, gemination is often left unrepresented in the environment of /t/:
φεθεθα /petteh ̣tɔ̄/ [phɪthːɪħthɔː] 'you released' Ps. 30:12

ουεθαζερηνι /wat-tʔazzerē nı̄ / [wæt ː(a)ʔazːɪʀeːniː]
h
'and you girded me' Ps. 30:12
ϊεθεν /yetten/ [jɪthːɪn] 'he will give' Ps. 49:8

σαθου /š attū / [ʃat ːuː]
h
'they set themselves' Ps. 49:15
Aside from the instances of irregular gemination that he attributes to the presence of /b/ and
/m/, YUDITSKY suggests that degemination occurs after short /e/ in the Secunda. He also ar-
gues that the lack of gemination in the initial syllable of wayyiqtol forms such as ουεθαζερηνι
‫( וַתְּ ַאזּ ְֵרנִי‬Ps. 30:12), ουθεθθεν ‫( וַתִּ תֶּ ן‬Ps. 18:36) and ουθεζορηνι ‫( וַתְּ ַאזּ ְֵרנִי‬Ps. 18:40) is evidence
that the Hebrew of the Secunda reflects a transitional period during which the narrative past
tense form w-yiqtol (< *wa-yaqtul [≠ *wa-yaqtulu]) was gradually shifting to wayyiqtol. Dur-
ing this transitional period, gemination would be present in some past yiqtol verbs and absent
in others (2017, 44, 231–32).292
There is evidence in Palestinian Koine for alternations of τ/ττ and θ/θθ in orthography
and transcription: e.g., σωττριας (for σωτηριας), πιτακιου (for πιττακιου), and Μαθεθ<ος>
(for Μαθθεθος). Cross-linguistically, voiceless stops require a greater duration than voiced
292. These forms may have another explanation. In my view, because the narrative past tense wayyiqtol was not
a part of the spoken language, it was not always identified in the consonantal text, especially in poetry. The
ancient Greek translations also indicate inconsistency in the renderings of w(ay) + yiqtol forms in Psalms.
Therefore, forms without a vowel after ου and without double θθ in the Secunda may indicate that the
transcriber identified them as w + yiqtol non-past tense forms just as the ancient translators sometimes did. Note
that ουϊεδαββερ ‫( ַויּ ַדְ בֵּר‬Ps. 18:48) is translated as present in Aquila, Symmachus, and Theodotion and ουθεθθεν
‫( וַתִּ תֶּ ן־‬Ps. 18:36) as future in Symmachus.
- 244 -
stops to be perceived as geminates (DMITRIEVA 2012, 31–32). Geminates followed by high
vowels are more difficult to discern than geminates followed by low vowels (HARDISON and
SAIGO 2010, 90). Further, there is some evidence that the contrast between singleton and gem-
inate consonants is less between unstressed vowels (DMITRIEVA 2012, 137).293 This is the case
for the prefix consonant /th(ː)/ = θ(θ) of the waw consecutive forms.
6.3.8.6. Gemination and Syllable Structure
Fifth, and finally, a lack of gemination in the Secunda is often precipitated by syllable struc-
ture. Final gemination is never represented, but it is always preceded by a short vowel:
λεβ /leb(b)/ [lɪb] 'heart' Ps. 32:11

σεκ /ś aq(q)/ [sækˀ] 'sackcloth' Ps. 35:13
εµ /ʔem(m)/ [ʔɪ(̃ m)] 'a mother' Ps. 35:14
χολ /kol(l)/ [k ʊl]
h
'all' Ps. 35:28
In Greek transcription of Arabic and Akkadian, final gemination is not represented (5.3.2.3.8;
5.3.3.3.9). A number of modern Semitic dialects (mostly Arabic) have final geminates
(phonologically) which surface phonetically as singletons: e.g., Syrian Arabic /ṃayy/ [ṃaj]
'water' and Modern Mandaic /rabb/ [rab] 'large'. In Arabic, final geminates are allowed in the
coda-maximalizing dialects. In the Ğ ubb'adı̄ n dialect of Neo-Aramaic, final -CC# was simpli-
fied. Cross-linguistically, it is common for final -CC# to appear only after monosyllabic
words with short vowels (CVCC#), since a long vowel followed by a final geminate
(CVːCC#) is not permitted in the phonotactics of most languages (DMITRIEVA 2012, 2, 161–62,
166, 168, 219). These principles may be illustrated by a comparison of the various monosyl-
labic syllable structures in the Secunda:
293. But cf. DMITRIEVA (2012, 139), who suggests that the decreased perceptibility might not be significant.
- 245 -
ω ω ω
σ σ σ
µ µ µ µ µ µ µ µ µ
P e l d e r k l e b b
ηλ = /(ʔee).l/ [ʔeːl] δερχ = /(der).k/ [dɪʀk] λεβ = /(leb).b/ [lɪb]

Figure 21: Moraic Representation of ηλ, δερχ, and λεβ
The short vowel in the qill and qoll patterns above demonstrates that the syllable structure of
words like λεβ resembles δερχ more than ηλ. In light of the cross-linguistic evidence and the
evidence of modern Semitic dialects, we may posit that the Hebrew reflected in the Secunda
had at least underlying final geminates, though they may have surfaced as singletons.
If a final underlying geminate was a ‫ בג"ד כפ"ת‬consonant as in λεβ, it is not clear
whether it would have been realized with the plosive or spirantized realization. Note that in
the Northeastern Neo-Aramaic (NENA) dialects, in which the reflex of post-vocalic */b/ is
generally realized as /w/ and in which synchronic post-vocalic /b/ is sometimes realized as
[β], plosive /b/ is maintained when it originates from */bb/, including in syllable-final con-
texts: e.g., the dialect of Qaraqosh has gib (< *gebb) 'with' and š abṯ a (< *š abbəṯ ā ) 'week'
(KHAN 2002, 26, 31). In light of the interchange of κ for χ for k# (< *kk#) in the transcription
discussed earlier (ουωρεκ [6.3.1.6]), we will tentatively posit the same phenomenon for the
Secunda, though it is highly speculative.
An underlying geminate is also likely for the following C1VC2ːC3V segments (cf. YU-
DITSKY 2017, 43):
ϊεσαυου /yš aw(w)ʕū / [jɪʃawʕuː] 'they will cry out' Ps. 18:42
µεχφιριµ /mek-(k)pı̄ rı̄ m/ [mɪkhɸiːʀıː̃ (m)] 'from young lions' Ps. 35:17
εθνηου /ʔet(t)nē hū / [ʔɪthneːhuː] 'I will set him' Ps. 89:28
Arabic dialects also bear witness to consonant-adjacent geminates that are neutralized on the
surface: e.g., Iraqi Arabic /dabbrat/ [dabrat] 'she arranged' and Syrian Arabic /waʔʔef/
- 246 -
[waʔʔef] 'stop! (masc.)' vs. /waʔʔfi/ [waʔfi] 'stop! (fem.)' (DMITRIEVA 2012, 2, 21, 161). In the
underlying form, a geminate is followed by an unstressed short vowel. Syncope occurs fol-
lowed by a neutralization of the geminate in the new consonant-adjacent context. We may
compare two Secunda forms of similar syllable structure, one in which syncope and degem-
ination occur (εθνηου) and one in which they do not (φελλετηνι) (see WATSON 2007, 352):
ω ω
σ σ σ σ σ σ σ σ
=
µ µ µ
=
µ µ µ µ µ µ µ µ µ µ µ
P e t e n e h u p a l e t e n i
˙
εθνηου = /(ʔet).ø.(nee).(huu)/ [ʔɪtneːhuː] φελλετηνι = /(pal).(le).(tee).(nii)/ [pælːɪtˀeːniː]

Figure 22: Moraic Representation of εθνηου and φελλετηνι
Syncope and degemination do not typically occur in this syllable structure (CVCCVCV́ ː) in
the Secunda (e.g., φελλετηνι ‫[ ַפּ ְלּ ֵטנִי‬Ps. 31:2], ζαµµερου* ‫[ זַמְּרוּ‬Ps. 30:5], ιεφφολου ‫[ יִפְּלוּ‬Ps.
18:39], but cf. βεσαυει ‫שׁ ְוּעִי‬

ַ ‫[ ְבּ‬Ps. 31:23] and ϊεσαυου ‫שׁוְּעוּ‬
ַ ְ ‫[ י‬Ps. 18:42]). Phonetically, degem-
ination and subsequent syncope may have occurred because of the homorganic nature of /t/
and /n/ in Hebrew.294 Because gemination is preserved in other parts of the paradigm (e.g.,
ουϊεθθεν), we may posit underlying gemination that is neutralized on the surface in the
transcription εθνηου. A similar phenomenon is likely reflected also in βεσαυει and ϊεσαυου.
Other potential instances of degemination in relation to syllable structure are covered
in YUDITSKY (2017, 43). However, because their relevance for potential degemination depends
on our interpretation of whether or not the word is intended to be definite, an interpretation
which must remain uncertain, they have been omitted from the discussion here.
A number of the suggestions put forth in this section to explain the irregular representation of
gemination in the Secunda merely constitute possible explanations and are, to a degree, ad
294. Note how in Tiberian Hebrew, /l/ is usually degeminated in the hithpaˁel stem with suffixes if it is followed
by another /l/ but not otherwise: e.g., ‫ י ִתְ ַפּלְלוּ‬but ‫י ִתְ ַהלְּכוּ‬.
- 247 -
hoc. This is because there is no definite way of knowing precisely how geminates were real-
ized in Secunda Hebrew with respect to their relative durations across different consonant
types and phonetic and prosodic environments. Even if some of the suggestions put forth turn
out to be incorrect, the discussion has demonstrated that when dealing with irregular repre-
sentations of gemination in the Secunda, a whole array of factors must be taken into account.
Standard Greek orthographic conventions may limit the ability of the scribe to represent gem-
ination in the most efficient way. The acoustic and articulatory characteristics of the conso-
nant and its immediate context may lessen the geminate-singleton contrast. One must also
consider how the underlying phonology might have been neutralized in the surface forms. Fi-
nally, all of these issues must be constantly viewed through the lens of the linguistic percep-
tion of the transcriber. Certain distinctions particular to a certain language are sometimes per-
ceived quite differently by speakers of another language, even if they are fluent in both.295
In light of these principles, we may refine the approach of YUDITSKY, who generally
assumes that the presence or lack of gemination in the transcriptions reflects the same in the
phonology of the Hebrew. What is missing from his approach is an appreciation of how sig-
nificant a role linguistic perception can play in all of these cases. Because of misperception
across languages, a singleton in transcription may still represent a geminate in Hebrew, and
vice versa. The representation of gemination in Greek transcriptions of Latin and Akkadian
support this point. One may also consider modern examples, such as the transliteration of
Arabic names into English, in which gemination may go unrepresented (see 6.4.2.4).
6.3.9. Summary
The suggested realization of each consonantal phoneme in the Secunda is outlined below
(chart 19):
295. See, for example, HAN's study on Japanese geminates among native Japanese and fluent Americans. He
found that the Americans tended to pronounce the geminates with considerably less duration than did the
Japanese (1992). See also CELATA and CANCILA's study of the perception of geminates in the Lucchese
community in San Francisco. She found that the longer one had been in the United States the more difficulty
they had discerning the singleton-geminate distinction in Lucchese (2010).
- 248 -
Hebrew Letter Phoneme Phone Greek Grapheme Written Word Pronunciation
STOPS:
‫ב‬ [b] β βαναυ [bɔːnaw]

/b/
‫ֿב‬ [β] β αβδω [ʕaβdoː]
‫פ‬ [ph] φ φααλθα [phɔːʕalthɔː]
/p/
‫ֿפ‬ [ɸ] φ αφαχθ [hɔːɸaχth]
‫ד‬ [d] δ δερχ [dɪʀkh]
/d/
‫ֿד‬ [ð] δ ιαδω [jɔːðoː]
‫ת‬ [th] θ θαµιδ [thɔːmiːð]
/t/
ֿ‫ת‬ [θ] θ ασιθι [ʕɔːsiːθiː]
‫ג‬ [g] γ γαµ [gam]
/g/
ֿ‫ג‬ [ʁ] γ µαγεν [mɔːʁɪn]
‫כ‬ [kh] χ χι [khiː]
/k/
‫ֿכ‬ [χ] χ βαχ [bɔːχ]
SIBILANTS:
[s] σ σαµ [sɔːm]

‫ס‬ /s/
([z]) σ, ζ βεεζδαχ [bɪħɪzdɔːχ]
‫שׂ‬ /ś / [s] σ σαµου [sɔːmħuː]
‫שׁ‬ /š / [ʃ] σ σαλωµ [ʃɔːlõː(m)]
[z] ζ µαζµωρ [mazmoːʀ]
‫ז‬ /z/
([s]) ζ, σ νεγρεσθι [nɪʁʀæsthiː]
EMPHATICS:
‫ק‬ /q/ [kˀ] κ κωλ [kˀoːl]

‫ט‬ /t/̣ [tˀ] τ ταµνου [tˀɔːmnuː]
‫צ‬ /s ̣/ [ʦˀ] σ σεδκι [ʦˀɪðkˀiː]
SONORANTS:
[m] µ µαιµ [majm]

‫מ‬ /m/
[ṽ(m)] µ, ν θαµµιν [tɔːmıː(m)]
̃
[n] ν νααρ [nɔːhɔːʀ]
‫נ‬ /n/
[ṽ(n)] ν, µ296 ναθαν [nɔːθã(n)]
‫ל‬ /l/ [l] λ λανου [lɔːnuː]
296. One example of ν > µ is attested in external sources: βεδεµ ‫( ְבּעֵדֶ ן‬Gen. 2:8). In the Ambrosiana palimpsest,
initial /#n/ is once transcribed with µ before being corrected to ν: µνηερθ (Ps. 89:40).
- 249 -
‫ר‬ /r/ [ʀ] ρ ραµωθ [ʀɔːmoːθ]
GUTTURALS:
‫א‬ /ʔ/ [ʔ] Ø, α_α297 αηλ [hɔːʔeːl]

‫ה‬ /h/ [h] Ø, α_α µηηρα [meheːʀɔː]
‫ע‬ /ʕ/ [ʕ] Ø, α_α µαωγ [mɔːʕoːʁ]
‫ח‬ /h ̣/ [ħ] Ø, α_α αωσιµ [haħoːsıː̃ (m)]
SEMI-VOWELS:
ου γηουαθω [geːwɔːθoː]
[w]
‫ו‬ /w/ υ ( / α_ ) αυωναν [ʕawoːna(̃ n)]
[ʔ] α_α αων [ʕawõː(n)]/[ʕaʔõː(n)]
[j] ι, ϊ ωϊηβ [ʔoːjeːβ]
‫י‬ /y/
[j]/[ʔ] α_α αεα [hɔːjɔː]/[hɔːj̞ ɔː]
Chart 19: Consonantal Phonology, Phonetics, and Orthography in the Secunda
6.4. VOWELS
There are essentially two possible interpretations of the vowel system of Secunda Hebrew,
one which posits an eight-vowel system (/a/, /e/, /o/, /ā /, /ē /, /ō /, /ı̄ /, /ū /) (BRØNNO 1943, 12;
JANSSENS 1982, 51; YUDITSKY 2017, 71) and one which posits a ten-vowel system (/a/, /e/, /o/,
/ā /, /ē /, /ō /, /ı̄ /, /ū / + /ε/, /ɛ̄/) (BLAU 1984). Additionally, there is the question regarding whether
vocalic shewa is a real feature or merely the preservation of a short historical vowel in an
open unstressed syllable (6.5.1.2). Based on my analysis of the transcriptions, which will be
borne out in this section, the Hebrew tradition reflected in the Secunda has the following vo-
calic phonemic inventory:

Front Back
close ı̄ ū
e o
mid-close ē ō
mid-open (ɛ̄, ε) ɔ̄
open a
Chart 20: Vocalic Phonemic Inventory of the Secunda
297. The sequence α_α indicates that a hiatus between vowels may signify this consonant.
- 250 -
Oppositions between phonemes are indicated by the minimal pairs below. Due to the limited
corpus, minimal pairs are not always attested. Therefore, hypothetical, yet justifiable, forms
are reconstructed (marked with ***) on the basis of comparable forms (marked with →):298
/ı̄ / : /ē / (σιρι→) νιρι*** /nı̄ rı̄ / 'my fallow ground' /ɔ̄/ : /o/ (σαµ→) χαλ*** /kɔ̄l/ 'he enclosed'
νηρι /nē rı̄ / 'my light' χολ /kol(l)/ 'all'
/ē / : /e/ ηλ /ʔē l/ 'God of' (cstr.) /ɔ̄/ : /ō / βα /bɔ̄/ 'he is coming'
ελ /ʔel/ 'to' βω /bō / 'in him/it'
(/ē / : (ωση→) µαση*** /mah ̣sē / 'refuge of' /o/ : /ō / οζ /ʕoz(z)/ 'strength'
/ɛ̄/)299
µασε /mah ̣sɛ̄/ 'refuge' (µωτ→) ωζ*** /ʕō z/ '(to) seek refuge'
/a/ : /ɔ̄/ (ναθαν→) ζαχαρ*** /zɔ̄kar/ 'he remembered' /ō / : /ū / (δωρ→) σωρ*** /š ō r/ 'bull'
ζαχαρ /zɔ̄kɔ̄r/ 'male' σουρ /š ū r/ 'wall'

Chart 21: Vocalic Phonemic Oppositions in the Secunda
I will begin this section by analyzing the representation of /ē / and /ō / in the Secunda to ad-
dress the fundamental question as to whether or not the transcriber utilized the Greek vocalic
graphemes to represent length or quality in the Secunda. Following this, I will address each
Hebrew vocalic phoneme in turn, dealing with its graphemic representation, phonemic value,
and phonetic realization. The issue of vocalic shewa will be dealt with in 6.5.
6.4.1. Length and Quality: /ē / = η and /ō / = ω
In the Secunda, the transcription of long /ē /, /ō / and short /e/, /o/ corresponds with the histori-
cal-grammatical distinction between long and short vowels in Greek. Long /ē / is represented
by η, which represents a close-mid front vowel [e] in Palestinian Koine (4.5.3.1.7):
εκκης /ʕeqqē š / [ʕɪkˀːeːʃ] 'a perverse one' Ps. 18:27

ουην /w-ʔē n/ [(ʔ)uʔeːn] 'and there is no' Ps. 18:42
αµµααζερηνι /ham-mʔazzerē nı̄ / [hamːaʔazːɪʀeːniː] 'the one who girds me' Ps. 18:33
298. Unattested νιρι*** ‫* <( נ ִִירי‬qı̄ l) 'my fallow ground' has been reconstructed on the basis of attested σιρι ‫ִירי‬
ִ ‫שׁ‬
(< *qı̄ l) 'my song', the final vowel of a III-weak construct noun in unattested µαση*** (< *-vyv#) 'refuge of' on
the basis of the final vowel in attested ωση ‫* <( עֹשֵׂה‬-vyv#) 'doer of', the unattested 3ms strong suffix conjugation
form ζαχαρ*** ‫' זָכַר‬he remembered' on the basis of attested ναθαν ‫' נָתַ ן‬he gave', the unattested 3ms II-weak
suffix conjugation form χαλ*** ‫' כָּל‬he enclosed' on the basis of attested σαµ ‫' שָׂם‬he set', the unattested II-weak
qal infinitive ωζ ‫* <( עוֹז‬qō l) '(to) seek refuge' on the basis of attested µωτ ‫* <( מוֹט‬qō l) '(to) shake', and the
unattested II-weak noun σωρ ‫* <( שׁוֹר‬qō l) on the basis of attested δωρ ‫* <( דּוֹר‬qō l) 'generation'.
299. I argue that [ɛː], indicated by ε, was present in the Secunda, but it may be an allophone of /ē / (see 6.4.4).
- 251 -
Long /ō / is rendered by ω, which represents a mid back rounded vowel [o] in Palestinian
Koine at the time of the Secunda (4.5.3.1.11):
κωλ /qō l/ [kˀoːl] 'voice' Ps. 28:6

νωσηρ /nō s ̣ē r/ [noːʦˀeːʀ] 'keeping' Ps. 31:24
σαβαωθ /s ̣(a)bɔ̄ʔō t/ [ʦˀaβɔːʔoːθ] 'hosts' Ps. 46:8
For a discussion regarding the transcription of short /e/ with ε and short /o/ with ο, see 6.4.2.
Before analyzing other vocalic correspondences in the Secunda, we must begin by ad-
dressing a fundamental question: are both length and quality directly represented in the
transcription, or is only quality directly transcribed? Most scholars who have dealt with the
Secunda seem to assume that the Greek historically long vocalic graphemes η and ω must al-
ways represent long vowels and ε and ο must always represent short vowels (e.g., BRØNNO
1943; JANSSENS 1982; YUDITSKY 2017), but it has also been suggested that the Greek transcrip-
tions may correspond only with quality (BLAU 1984). In this section, I will argue that the Se-
cunda transcriptions represent primarily quality, with direct representation of quantity as a
possible secondary convention as long as it did not affect the perception of quality.
6.4.1.1. The Presence of Real Vowel Quantity in the Secunda
Although the terms "long vowels" and "short vowels" are often used to refer to qualitative
differences, especially in Biblical Hebrew, an important distinction must be made between
vowel quality and vowel quantity. Traditionally, vowel quality has been understood as the rel-
ative height and backness of the tongue when pronouncing particular vowels.300 Any vowel
may be described in terms of how high, low, back, or front it is.301 Vowel length refers to the
duration for which a particular vowel quality is pronounced. In many instances, length is
merely a phonetic feature, but many languages make use of length for phonemic contrasts.
300. There are also additional features such as rounding, ATR (width of the pharynx), rhotacization, and
nasalization (LADEFOGED 2001, 215).
301. While phoneticians have used such terms for a long time, "height" and "backness" actually correspond
more to acoustic frequencies than they do to the position of the tongue. The high-low distinction corresponds to
what is referred to as the first formant (F1) and the front-back distinction roughly corresponds to the difference
between the first formant (F1) and the second formant (F2) (LADEFOGED 2001, 14–15, 170–78, 232–33).
- 252 -
Unlike vowel quality, vowel length is a suprasegmental feature imposed on a particular vowel
segment. It is not a feature like height, backness, or roundness, but merely specifies the dura-
tion of the vowel segment (LADEFOGED 2001, 14–15, 170–78, 232–33).
There are several pieces of evidence that support the presence of real phonemic length
(i.e., duration) in Hebrew in the first few centuries CE. First, when discussing Christians who
mispronounce Hebrew names, which they only knew as they were presented in the Greek
transcriptions of the LXX (see HARVIAINEN 1977, 49–50; BRØNNO 1970, 205), Jerome writes:
And if we make a mistake in pronunciation, in lengthening or shortening of a
syllable, whether lengthening that which is short, or shortening that which is
long, they (the Jews) are accustomed to mock us for our ignorance ... 302
The Latin terms Jerome uses here, produco and brevio, are technical terms referring to the
lengthening and shortening of vowels and syllables (see MORENO 2008). Cicero (1st BCE), for
example, uses the term produco when referring to the compensatory lengthening of a vowel
before a nasal, and Quintilian (1st CE) uses the term brevio when referring to the shortening of
the long vowel in the name Amphı̄ on.303 Second, a number of ordered sound rules in the histo-
ry of Hebrew require real durational length to be present at the time of the Secunda.304 Third,
302. Commentary on Titus (3.9): Et si forte erraverimus in accentu, in extensione et brevitate syllabae, vel
brevia producentes, vel producta breviantes, solent irridere nos imperitiae, maxime in aspirationibus in
quibusdam cum rasura gulae litteris proferendis.
303. Orator (48.159): indoctus dicimus brevi prima littera, insanus producta, inhumanus brevi, infelix longa.
"We know that 'indoctus' is to be pronounced with the first letter short (brevi), 'insanus' long (producta),
'inhumanus' short (brevi), 'infelix' long (longa)."
Institutio Oratio (12.10.57): ...cum interrogasset rusticum testem, an Amphionem nosset, negante eo, detraxit
aspirationem breviavitque secundam eius nominis syllabam, et ille eum sic optime norat. huiusmodi casus
efficient, ut aliquando dicatur liter quam scribitur, cum dicere, quomodo scribendum est, non licet. "When he
asked a rustic witness whether he knew Amphı̄ on, and the witness replied that he did not, dropped the aspirate
and shortened (breviavit) the second syllable, whereupon the witness recognised him at once. Such situations,
when it is impossible to speak as we write, will sometimes make it necessary to speak in language other than
that which we use in writing" (translation from BUTLER 1922).
304. KHAN points out that at some point in the history of Hebrew, the pairs of long and short a/ā and e/ē vowels
became differentiated by quality in addition to length: a-ā > a-ᴐ̄ and e-ē > ε-ē . Another relevant sound change
was the lengthening of stressed syllables in a particular group of words, including final syllables in certain
verbal forms and originally monosyllabic nouns closed by gemination. This lengthening rule is demonstrated by
Tiberian Hebrew forms such as ‫* <( לֵב‬libb) and ‫* <( כּ ֹל‬kull). It is apparently problematic, however, that the
same sort of lengthening does not apply to Tiberian Hebrew ‫עַם‬. Being from geminate roots, all of these forms
exhibit short vowels in the Secunda (λεβ, χολ, αµ/εµ). In Tiberian Hebrew, the fact that all stressed syllables
were pronounced long—this was a separate, later lengthening than the one just mentioned—shows that the
difference between ‫כּ ֹל‬/‫ לֵב‬and ‫ עַם‬lies not in the fact that ‫ עַם‬is not lengthened, but in the fact that it did not
undergo the quality shift to /ᴐ/. This is best explained by assuming that the lengthening rule operated after the
- 253 -
certain minimal pairs in the Secunda, such as the distribution of Greek ο and ω, are best ex-
plained by assuming a real phonemic contrast in vowel duration.305

6.4.1.2. Contemporary vs. Historical Orthography
Previous scholars who have worked on the transcriptions concur that vowel quantity was
present and phonemic in the Hebrew of the Secunda (e.g., BRØNNO 1943, 12; JANSSENS 1982,
51; YUDITSKY 2017, 45–61). They also point out that Greek η and ω are used to represent the
Hebrew long vowels /ē / and /ō / and Greek ε and ο are used to represent the Hebrew short
vowels /e/ and /o/. While this is a correct description, incorrect assumptions, resulting from a
lack of precision and a lack of sensitivity to Greek orthography and phonology, have under-
girded the approach. For example, the Greek vocalic graphemes η and ω are considered to be
inherently long at the time of the Secunda (e.g., JANSSENS 1982, 20). Also, Greek ε and η are
portrayed as differing only in length, ε representing short /e/ and η representing long /ē / (e.g.,
YUDITSKY 2007a, 5; YUDITSKY 2017, 46).306
Neither of these assumptions is consistent with the Greek evidence. First, in Palestin-
ian Koine Greek of the Roman period—in fact, as early as the Koine Greek of the second
century BCE (HORROCKS 2014, 169)—vowel-length distinctions had been neutralized and the
Greek vocalic graphemes came to represent only quality (4.5.3.1.22). At the time of the Se-
cunda, one reading Greek would not have made phonemic length distinctions, just as
Jerome's contemporaries were unable to pronounce vowel length in Hebrew names correctly
ā > ᴐ̄ quality shift had ceased to operate (KHAN 1987, 45). We may summarize these changes as follows: (1) a-
ā > a-ᴐ̄, (2) stressed vowels in certain words subsequently lengthened, and (3) e-ē > ε-ē . While there is some
evidence for (1) in the Secunda, there is counter-evidence for (2), which suggests that phonemic length was still
been present in Secunda Hebrew. Note, however, that there may be evidence for (2) in Chrysostom's
transcriptions ωµ ‫' ח ֹם‬heat' (< */h ̣omm/) and ην ‫' חֵן‬favor' (< */h ̣enn/) (Fragmenta in Jeremiam, 64.969.50–51).
305. Because ο and ω both represent [o] in Palestinian Koine Greek during the Roman period, the best
interpretation of their complementary distribution in the Secunda, attested in such minimal patterns (minimal
pairs are not always attested) as the imperative ζχορ 'remember!' (Ps. 89:48) and the nominal βχωρ 'firstborn'
(Ps. 89:28), is that ω is utilized to represent long /ō /, despite the fact that it no longer represented a long vowel
in Greek at the time of the composition of the Secunda (4.5.3.1.22).
306. In his dissertation, YUDITSKY states that "in Greek, long e and short e are represented by different
[graphemes]: η for the long [vowel] and ε for the short [vowel]" (my translation) (2007a, 5). In his monograph,
YUDITSKY describes Greek ε as "a short front middle vowel, lower than /i/ and higher than /a/" and η as "a long
front middle vowel, lower than /i/ and higher than /a/" (my translation) (2017, 46).
- 254 -
because they only knew them from the Greek transcriptions of the LXX (6.4.1.1). Second,
not only was there no distinction in length between Greek η and ε at the time of the Secunda,
neither were they equivalent in quality. Greek η represented [e] and Greek ε represented [ε]
(or [e̞ ]) (4.5.3.1.7). YUDITSKY does operate under the assumption that the orthography of the
Secunda reflects the Greek pronunciation of at least a few centuries prior to Origen's time
(2017, 46), but at no stage in the history of Greek would a synchronic description result in η
and ε having no difference in quality (see PETROUNIAS 2007c, 602–605).307
Although the evidence demonstrates that in contemporary Greek spelling and pronun-
ciation the vocalic graphemes corresponded to vowel quality and not vowel quantity
(4.5.3.1.22),308 the historical pronunciations (with the vocalic graphemes corresponding to
both vowel quality and vowel quantity) would have been remembered by a portion of the lit-
erate population. Those educated in Greek grammar and literature would have been aware of
the fact that η/ω were associated with long vowels and ε/ο were associated with short vowels.
This knowledge is clearly preserved in grammatical works such as that of Dionysius Thrax
(170–90 BCE), who, when discussing the Greek alphabet, writes, "And of these, seven are
vowels ... and of the vowels, two are long, η and ω, two short, ε and ο, three of either length,
α ι υ ... "309 We must remember, however, that only a rudimentary knowledge of the alphabet
was needed both to compose and to use the Secunda transcriptions.
This leads to the following question: how did the author of the Secunda utilize a
graphemic system, which only represented vowel quality in his day, to transcribe a vocalic
system characterized by both quality and quantity? Did he follow the grapheme-phoneme
307. It is possible that during the third century BCE η and ε had similar qualities. However, if they did indeed
have similar qualities at that point, it was quite brief. Already by the second century BCE they were distinguished
in quality again (KNOBLOCH 1995, 124).
308. Note the various Greek renderings of the Hebrew name ‫יוסה‬: Ιωσε/Ιοσε/Ιωση (CIIP I/1, no. 46, 81, 573).
309. Τούτων φωνήεντα µέν ἐστιν ἑπτά ... Τῶν δὲ φωνηέντων µακρὰ µέν ἐστι δύο, η καὶ ω, βραχέα δύο, ε καὶ ο,
δίχρονα τρία, α ι υ.
- 255 -
correspondences of his day by representing only vowel quality or did he follow historical and
grammatical conventions310 to utilize the historically long Greek vocalic graphemes to repre-
sent the long vowels in Hebrew? In this section, I will advance the claim that, unless the au-
thor of the Secunda was working with two alternative Greek graphemes of the same quality
(e.g., o/ω = [o]), he prioritized quality, rather than quantity, in transcribing the Hebrew read-
ing tradition.
6.4.1.3. The Case of η and Vowel Harmony
The main piece of evidence in support of this theory concerns the use of the Greek grapheme
η to represent an assimilated vowel that is a result of vowel harmony. In the Secunda, when
preceding a guttural, "reduced"311 vowels may assimilate to the the vowel of the guttural:
βεεζδαχ312 ü ְ‫( ְבּ ַחסְדּ‬Ps. 31:8), µεεθθα ‫( ְמחִתָּ ה‬Ps. 89:41), χεεβλ ‫( ְכּ ֵאבֶל‬Ps. 35:14), λοοµ ‫( ְלחַם‬Ps.
35:1), βεειρ ‫( ְבּעִיר‬Ps. 31:22), αββωτεειµ ‫( הַבּ ֹ ְטחִים‬Ps. 49:7), ααλλελ ‫( ֲא ַחלֵּל‬Ps. 89:35), µηηρα
‫( ְמה ֵָרה‬Ps. 31:3), βηηκι ‫( ְבּחֵיקִי‬Ps. 89:51), and possibly also βεηναυ/βηηναυ313 ‫( ְבּעֵינָיו‬Ps.
36:3).314 In each instance, the reduced vowel assimilates in quality to the following vowel.
The assimilations to ε and ο are not especially interesting, but the assimilation to η in µηηρα
and βηηκι is significant for understanding the representation of length in the Secunda.
310. This seems to be the claim of YUDITSKY, who argues that even if the Secunda were composed during
Origen's lifetime, it would reflect pronunciation at least a couple hundred years earlier since writing is
conservative (2017, 45–46). However, speaking of a "conservative" writing system only makes sense when
there is an established historical tradition with historical spellings that may be preserved. Unless YUDITSKY
would argue that the Secunda continues a well-established tradition that is hundreds of years old, this argument
may be dismissed. The fact that we have variations in transcription of Hebrew proper names in Palestinian
epigraphy would argue against a well-established and well-known transcription tradition.
311. By "reduced vowels" I mean those vowels that are equivalent to phonological zero but not necessarily
phonetic zero. In the Secunda, as a general rule, reduced vowels may be defined as those which may be omitted
in transcriptions. This is certainly the case for the inseparable prepositions. For a fuller discussion, see 6.5.1.
312. It should be noted that the inseparable prepositions may be represented with or without a vowel; when they
are transcribed with a vowel, it is most commonly α (e.g., βα-, χα-, λα-).
313. The reading is uncertain.
314. There are also a couple instances in which vowel harmony seems to occur before non-gutturals: αµιµιµ
‫( ַע ָממִים‬or Aramaic ‫( ) ַע ְממִין‬Ps. 18:48) and εφικιδ ‫( אַ ְפקִיד‬Ps. 31:6) (see 6.5.1.3.2).
- 256 -
MARGOLIS compares this phenomenon to the rules outlined by the medieval Hebrew
grammarians Ben-Asher and Ḥayyū j.315 Though the system in the Secunda is not consistent,
the transcriptions µηηρα and βηηκι are "clearly [based] on the principle of assimilation"
(1909, 66). PRETZL argues that the first η in βηηκι and βηηναυ represents a short or ambigu-
ous e vowel assimilated to the following vowel, which indicates that the quantitative system
of transcription gave way to a qualitative one (1932, 9, 13). BRØNNO, while admitting that the
phenomenon of assimilation before a guttural exists in the Secunda, doubts that it applies in
the case of µηηρα and βηηκι for two reasons: (1) elsewhere, η reflects a long vowel and (2) in
another example, only one η appears in the same environment (βησαθ ‫[ ַבּ ֲעצַת‬Ps. 1:1]). The
forms with -ηη- could be a scribal error or represent an extra long /ē / after the elision of the
guttural (1943, 255–56, 340–41). JANSSENS acknowledges that shewa preceding a guttural
sometimes assimilates to the following vowel, including βηηκι in his examples (1982, 86).
YUDITSKY argues that the first vowel in the forms µηηρα and βηηκι has lengthened under the
influence of the following guttural (2017, 88–89).
A summary of previous explanations for µηηρα and βηηκι highlights the tension be-
tween the apparent assimilation of a reduced (or short) vowel and the representation of such a
vowel with a grapheme (η) used for long vowels everywhere else. There is no need to resort
to scribal error (contra BRØNNO) to explain these forms and the suggestion that the guttural
was not pronounced (see BRØNNO 1943, 256) has since been refuted by YUDITSKY (2008a;
2008b).
The idea that the vowel lengthened under the influence of the guttural (YUDITSKY
2017, 88–89) may be rejected for several reasons. First, this lengthening does not occur in
315. See, for example, Ḥayyū j's comments in Kitā b al-afˁā l ḏ awā t ḥurū f al-lı̄ n: ‫ﻣﻨﮭـﺎ ﻣـﺎ ـﯾﺤـﺮك ـﺑﻤﺜــﻞ ﺣـﺮﻛـﺔ ﻣـﺎ ـﺑﻌـﺪه‬ ‫وذﻟـﻚ أن ــ‬
‫ـــﺢ وإن ﻛـﺎن‬ ‫ـــﺢ ﺣـﺮﻛـﺖ الשבא ـــ‬
‫ﻗﺒﻠﮭـﺎ ﺑـﺎﻟﻔﺘ‬ ‫ـــﺎ ً أﻋﻨــﻰ إن ﻛـﺎن اﻟـﺬى ـﺑﻌـﺪ الשבא ﻣـﻦ ھـﺬه اﻻر ـﺑﻌـﺔ اﺣـﺮف ـﻣﺤـﺮﻛـﺎ ﺑـﺎﻟﻔﺘ‬
‫اذا ﻛـﺎن ﻣـﺎ ـﺑﻌـﺪه أﻟﻔــﺎ ً او ھـﺎء او ﺣـﺎء او ﻋﯿﻨ‬
‫ﻟﻜﺴـﺮ‬
‫ﻟﻜﺴـﺮ ﺣـﺮﻛـﺖ ﺑـﺎ ـ ـ‬ ‫ﻣﺤـﺮﻛـﺎ ﺑـﺎ ـﻟﻀـﻢ ﺣـﺮﻛـﺖ ﺑـﺎ ـﻟﻀـﻢ وإن ﻛـﺎن ﺑـﺎ ـ ـ‬‫ ـ‬... '[The shewa] is vocalized like the vowel that comes after it. If
an ˀalef, heh, ḥet, or ˁayn comes after it, that is, if one of these four letters comes after the shewa and is
vocalized with a fatḥa (i.e., an /a/ vowel), the shewa before it will be vocalized with a fatḥa, and if it has a
ḍamma (i.e., a /u/ vowel), the shewa will be vocalized with a /u/ vowel, and if it has a kasra (i.e., an /i/ vowel),
the shewa will be vocalized with a kasra'.
- 257 -
other forms with similar syllable structures (e.g., µεεθθα and βεεζδαχ).316 In order to explain
why the assimilated vowel is ε in µεεθθα yet η in µηηρα, one would need to posit that the
reduced (or short) vowel assimilated to the quantity of the following vowel, which is ex-
tremely rare cross-linguistically.317 Second, the fact that a vowel in this same phonological
environment may be omitted (e.g., βησαθ), indicates that it is not a long vowel. Third, unless
the guttural ceased to be pronounced,318 gutturals do not cause adjacent vowels to lengthen in
the other Hebrew reading traditions.
However, there may actually be a couple examples in Tiberian Hebrew of a compara-
ble phenomenon. Before certain weak consonants (‫)אהעחינל‬, vowels may be lengthened pho-
netically so that the weak consonants were not elided in pronunciation (KHAN 2013f, 983).
The fact that the prepositions ‫ל‬/‫כ‬/‫ ב‬are more likely to be transcribed with a vowel before a
guttural seems to support a similar phenomenon in the Secunda (6.5.1.6.2), but suggesting
that such a rule applies in the case of µηηρα and βηηκι is problematic in light of forms like
µεεθθα. The only other possible Hebrew parallel for the lengthening of the first vowel in
µηηρα and βηηκι is the musical shewa gaˁya in the Tiberian tradition which, when placed on
a shewa, lengthens the shewa to a long vowel.319 Positing that the initial traces of such a phe-
316. Further evidence that the initial vowel in a structure like this should be regarded as short is found in the
LXX's rendering of ‫ ְרחוֹב‬as Ροωβ (see KHAN 2013h, 551). At the time of the LXX, ο and ω were identical in
quality, but historically short ο is utilized to represent the short vowel and historically long ω to represent the
long vowel (see 6.4.1.5). Phonemic length may still have been applicable in Greek at that time.
317. The problem with "length harmony" is that real length (i.e., duration) does not indicate a feature of the
vowel but actually indicates that a particular vowel is maintained for two "moraic slots." The consensus among
phonologists is that the difference between the syllables Cv̆ and Cv̄ is not between Cv[-long] and Cv[+long] (an
erroneous representation), but between Cv and Cvv. These issues are presented and discussed by HYMAN and
UDOH (2007), who claim that "there is no known process by which a short vowel assimilates in length to a long
vowel in a neighboring syllable ... long vowels have been known to shorten in the context of another long
vowel" (2007, 75).
318. In Samaritan Hebrew, the vowel of the inseparable prepositions may "fuse" with the vowel of a word
which originally began with a guttural, as in bē š å r ‫ באשר‬and lū lå m ‫( לְעוֹלָם‬BEN-ḤAYYIM 2000, 316).
319. Tiberian gaˁya marks secondary stress. When musical shewa gaˁya, rare in the twenty-one books but
common in the three books (Psalms, Proverbs, Job), is placed on a shewa, it lengthens the shewa to a long
vowel. That shewa gaˁya is often found on a guttural may indicate that a phonetic impetus lies behind the shewa
gaˁya (KHAN 2013g, 8–9). An example of shewa gaˁya occurs on the preposition ‫ ְבּ‬in the word ‫' בְּ ֽ ֵ֘ע ָינ֤יו‬in his eyes'
(Ps. 15:4). Because the shewa precedes a guttural and is marked with gaˁya, it both assimilates to the quality of
the following vowel and lengthens. The resulting Tiberian realization, /b-ʕē nā w/ [beːʕeːˈnå ːv], would
correspond quite nicely with the uncertain Hexaplaric reading βηηναυ (Ps. 36:3), though its Tiberian counterpart
- 258 -
nomenon are attested in the Secunda, however tempting, is unwarranted. It is far simpler to
conclude that η did not represent a long vowel in these forms.
If we do not hold to the assumption that η must always represent a long vowel in the
Secunda, these forms are easily interpreted and our earlier question regarding whether the
transcriber followed contemporary or historical spelling is answered as well. If η is regarded
as primarily representing the vocalic quality [e] as it did in contemporary Greek, we may in-
terpret the transcriptions βηηκι and µηηρα as examples of a reduced vowel assimilating in
quality to the vowel of the following guttural. The resulting forms, /b-h ̣ē qı̄ / [beħeːkˀiː] and
/mhē rā / [meheːʀɔː], fit well with the other Biblical Hebrew reading traditions. At least in
these instances, the transcriber prioritized quality over quantity, following orthographic con-
ventions of his own day rather than historical or grammatical conventions.

6.4.1.4. Tense/Lax Distinction and Qualitative Transcription
If the transcriber operated according to the writing conventions of his own day, transcribing
quality rather than quantity, this means that the vowel quality [e] (= η) was regarded as a bet-
ter approximation of Hebrew long /ē / than the vowel quality [ε]/[e̞ ] (= ε) was. Apparently,
Hebrew long /ē / and short /e/ were not only distinguished by quantity but also by quality.
This presents two questions: First, if vowel duration is merely a suprasegmental feature, why
does there seem to be a qualitative difference between long /ē / and short /e/ in the Secunda?
Second, if only Hebrew long /ē / and not short /e/ was best approximated by the quality η in
the Secunda, why is it that the quality of the short vowel is nearest to the quality represented
by η only in a few transcriptions (e.g., µηηρα, βηηκι)?
Although a sharp distinction was made earlier between vowel quality as a segmental
feature and vowel quantity as a suprasegmental feature (6.4.1.1), there are actually a number
is without shewa gaˁya. The forms βηηκι and µηηρα would also correspond perfectly to their Tiberian
counterparts, if they were marked with a shewa gaˁya in the Tiberian tradition, which they are not. If one wanted
to maintain the interpretation that the first η in βηηκι and µηηρα represents a long vowel, one could argue that
these forms represent an isolated example of what would eventually develop into shewa gaˁya.
- 259 -
of features of quality that tend to be associated with length. These may be described in terms
of what is called "tenseness" and "laxness."320 The terms "tense" and "lax" usually correlate
with features of vowel length and vowel height. Tense vowels are associated with length and
lax vowels with shortness. Non-low tense vowels are typically higher and non-low lax vow-
els are typically lower. Tense vowels are generally more peripheral and lax vowels closer to
the acoustic center. All of these are trends rather than rules (HOCK 1991, 143–44). In this
work, I will use the terms "tense" and "lax" primarily to refer to the degree of peripherality or
closeness to the acoustic center. For example, [i] and [u] are tense whereas [ɪ] and [ʊ] are lax.
The correlation of length and tenseness has been borne out in a study conducted by
GENDROT and ADDA-DECKER (2007) on phonetic reduction and acoustic duration in eight of the
world's languages. They found that as the duration of a vowel decreases, the closer it is artic-
ulated to the acoustic center (i.e., more centralized or reduced). This is explained as a result
of articulatory "undershoot" and "overshoot" (GENDROT and ADDA-DECKER 2007, 1417, 1419).
With the acoustic center as the "default" articulatory point of departure, more time is required
to attain the target quality of vowels at a greater distance from that center. As the duration of
a vowel grows shorter and shorter, there is less time to reach the target quality and return to
the "default" articulatory position; thus, the vowel is "undershot," being articulated closer to
the acoustic center (PEARCE 2008, 137; CROSSWHITE 2001).
PEARCE has demonstrated that there is one exception to the correlation between vowel
duration and nearness to the acoustic center, namely, that it is nullified when spreading
processes such as vowel harmony are in effect. The idea behind this is that the acoustic center
acts as a sort of "neutral position" for pronunciation, with each individual vocalic segment
being treated in its immediate environment. The target quality must then be attained in what-
ever duration is allotted for a particular segment before returning to the neutral position. In
320. Although these terms are commonly used by linguists, phoneticians have not found any acoustic corollary
of what is called 'tenseness'. The term is only loosely defined as "greater muscular tension" (HOCK 1991, 143).
- 260 -
the case of vowel harmony, the shorter duration no longer impacts the quality of the vowel
because the articulators are set in position in anticipation of the following vowel. The antici-
pation nullifies the need to return to the neutral position and thus enables the articulators to
attain the target quality without being restricted by duration (PEARCE 2008; 2012).
These principles may be applied to the distribution of η and ε in the Secunda
transcriptions. Assuming that Secunda Hebrew is not a linguistic outlier with respect to the
relationship between vowel duration and proximity to the acoustic center,321 we may posit
that Hebrew long vowels were generally more tense and Hebrew short vowels were generally
more lax. Thus, Hebrew long /ē / would have been pronounced with a more tense-peripheral
quality [e] and Hebrew short /e/ (< */i/) with a more lax-centralized quality [ɪ]322 (or [ë]) (see
6.4.2). Accordingly, Hebrew long /ē / [eː] was transcribed with the more tense Greek η = [e]
and Hebrew short /e/ was transcribed with the more centralized Greek ε = [ε] (6.4.2). The
identification of long vowels with a tense quality and short vowels with a lax quality is also
supported by modern linguistic studies of cross-linguistic perception (see 6.4.2.4).
The only exception to this rule occurs in the words µηηρα and βηηκι, in which the
spreading process of vowel harmony was operative. Because the articulators did not need to
return to the neutral position and were assimilated in anticipation of the following vowel [eː],
the short vowel was realized with a quality normally only attained with a longer duration.
Thus, the distribution of ε/η reflects a transcription based on quality (not quantity):
Hebrew /ē / Hebrew /e/ Hebrew "shewa" (ə)

Realized as ... [eː] [ɪ] [e] ( / C_Geː)
Transcribed as ... η = [e] ε = [ε] η = [e]
321. There is actually positive evidence that the long and short vowels in ancient Hebrew also differed in
tenseness and laxness. This argument will be developed more fully below (see 6.4.2).
322. Note that Hebrew /e/ is the reflex of etymological short */i/.
- 261 -
Such an interpretation of the Secunda transcriptions should be preferred over one which re-
gards η as inherently long, because it offers greater explanatory power in the case of the ex-
ceptional spellings in the Secunda, while remaining consistent with internal-Hebrew and
cross-linguistic evidence. In sum, at least in the case of η and ε, the transcriber's modus
operandi in the Secunda was to transcribe according to quality, rather than quantity, thus
placing him within the conventions of Greek orthography and phonology of his day.
6.4.1.5. Length and Historical Orthographic Convention
At this point, a distinction must be made between the Hebrew long vowels whose quality cor-
responded with only one Greek grapheme and the Hebrew long vowels whose quality corre-
sponded with two Greek graphemes. In the case of /ē /, the transcriber chose Greek η ([e]) to
transcribe Hebrew long /ē / not because Greek η was inherently long (cf. µηηρα and βηηκι),
but because it best approximated the vowel quality of Hebrew /ē / [eː]. Since there was only
one Greek grapheme that represented the vowel quality [e], quality was the only factor in the
transcriber's choice. The same may be said about the transcriber's choice of Greek α to
transcribe Hebrew long /ɔ̄/ and Greek ου to transcribe Hebrew long /ū /. (For the transcription
of Greek long /ı̄ / [iː], which corresponded in quality to both the historically length-neutral ι
and the historically diphthongal/long ει, see the discussion in 6.4.5.)
The same principle does not apply, however, in the case of Hebrew long /ō / [oː], the
quality of which corresponded with two Greek graphemes: ο, ω = [o] (4.5.3.1.11). According-
ly, even after identifying [o] as the most appropriate vowel quality to represent Hebrew long
/ō /, the transcriber still had two options from which to choose: ο and ω. The transcriber opted
for ω to represent long /ō / and ο to represent short /o/, reflecting the historical-grammatical
distinction between Greek long ω and Greek short ο. We cannot be sure if the utilization of ω
to represent a long vowel reflects knowledge resulting from an education in Greek grammar
and literature or knowledge resulting from familiary with an earlier tradition of Greek
- 262 -
transcription of Hebrew such as the LXX. In either case, however, it would reflect a histori-
cal-grammatical convention no longer relevant for the phonology of his day.

6.4.1.6. Greek Transcription of Vowel Length in Other Languages
A primarily qualitative transcription practice—or at least one that did not treat the grapheme
η as inextricably linked to length—is also attested in Greek transcription of other languages.
In general, Greek η is used to represent a long /ē / vowel, though in certain phonetic environ-
ments it may also represent a short vowel. For example, in Greek transcription of Latin, while
Greek η almost always renders Latin ē , in the environment of r Latin short ĕ is transcribed
with η (5.3.1.1.2). In transcription of Arabic, while etymological short */i/ is usually rendered
by Greek ε, it may also be transcribed by η especially in the environment of liquids and
nasals (5.3.3.1.2). Greek η may also be used to transcribe a short vowel in Akkadian, though
it can also be interpreted as long (5.3.2.1.2). Finally, there are a few interchanges of ε and η
for e-vowels in both Phoenician and Aramaic (5.3.4.1; 5.3.5.1.5). In sum, while Greek η is
normally used to transcribe long /ē / in various languages, there are exceptional cases, such as
certain phonetic environments, in which it may also represent a short vowel. This too indi-
cates a qualitative rather than strictly quantitative transcription practice.
Greek ω for long /ō /, on the other hand, is more stable in transcription of both Latin
and Semitic (5.3.1.1.4; 5.3.2.1.4; 5.3.3.1.3; 5.3.5.1). The only real instance of length confu-
sion regarding ω and long /ō / occurs in Greek transcription of Latin long ō and short o, yet
this is more common after length distinctions are neutralized in both Greek and Latin.323 This
is not surprising, since there were two Greek graphemes which represented the quality [o].
It has been demonstrated that the transcriber operated in a manner consistent with the Greek
orthography and phonology of his day, prioritizing quality over historical-grammatical
quantity to best approximate the Hebrew vowels. This is demonstrated by the transcriber's
323. It should also be noted that paleogrpahic similarity between Greek ο and Latin o may have been a factor.
- 263 -
use of η, a historically long vowel, to transcribe the short (or reduced) vowel in the words
µηηρα and βηηκι. It was only when the transcriber had before him two Greek graphemes of
the same quality that he made use of historical-grammatical distinctions in his transcription
conventions, as demonstrated by the use of ω to represent Hebrew /ō /. A similar distribution
is attested in roughly contemporary Greek transcription of other languages.
Another implication of the present section is that the transcriptions of the Secunda are
indeed transcriptions and not transliterations. That is, contrary to the conclusions of previous
scholars, it is not necessarily true that one Greek vocalic grapheme must always represent
only one Hebrew vocalic grapheme. One Greek grapheme may represent multiple Hebrew
phonemes and, conversely, one Hebrew phoneme may be represented, in different circum-
stances, by more than one Greek grapheme.
All these findings will guide our interpretation of the Secunda transcriptions in the re-
mainder of this chapter. Methodologically, it will be assumed that the transcriber generally
chose the quality that best approximated that of the Hebrew vowel.324
6.4.2. Lax /e/ (< */i/) and /o/ (< */u/), Tense /ı̄ / and /ū /
In the Secunda, etymological short */i/ is usually transcribed by ε, which represents an open-
mid front vowel [ε] (or true mid [e̞ ]) in Roman Palestinian Koine (4.5.3.1.10) (for the phonet-
ic transcription of /e/ as [ɪ], see further below for discussion):
σεµα /ś emh ̣ɔ̄/ [sɪmħɔː] 'joy' Ps. 30:12

λεβ /leb(b)/ [lɪb] 'heart' Ps. 32:11
ελλελθ /h ̣ellelt/ [ħɪlːɪlt ]h
'you profaned' Ps. 89:40
Etymological long /ı̄ /, on the other hand, is usually transcribed in the Secunda by ι,
which represents a close front vowel [i] in Roman Palestinian Koine (4.5.3.1.1):
δερχι /derkı̄ / [dɪʀkhiː] 'my way' Ps. 18:33

324. BLAU (1984, 77) comes to a similar conclusion in his review of JANSSENS (1982).
- 264 -
Etymological short */u/ is usually transcribed in the Secunda by ο, which represents a
mid back rounded vowel [o] in Roman Palestinian Koine (4.5.3.1.11) (for the phonetic
transcription of /o/ as [ʊ], see further below for discussion):
ερδοφ /ʔerdop/ [ʔɪʀdʊɸ] 'I will pursue' Ps. 18:38

χολ /kol(l)/ [khʊl] 'all' Ps. 49:2
βεκοδσι /b-qodš ı̄ / [bɪkˀʊðʃiː] 'in my holiness' Ps. 89:36
Etymological long /ū /, on the other hand, is usually transcribed by ου,325 which repre-
sents a high back rounded vowel [u] in Roman Palestinian Koine (4.5.3.1.16–17):
βαρουχ /bɔ̄rū k/ [bɔːʀuːχ] 'blessed' Ps. 31:22

χισους /k-sū s/ [khisuːs] 'like a horse' Ps. 32:9
σασουου /š assū hū / [ʃasːuːhuː] 'they plundered it' Ps. 89:42
There are essentially two ways of interpreting the use of Greek ε and ο for transcribing ety-
mological */i/ and */u/. First, /i/ and /u/ had lowered to /e/ and /o/ in the pronunciation of He-
brew reflected in the Secunda. Second, Hebrew /i/ and /u/ maintained their close pronuncia-
tions, but because of some limitation in the Greek orthographic system (e.g., length, quality),
ε and ο were the nearest approximations of /i/ and /u/.326 After a brief review of scholarship, I
will argue that etymological */i/ and */u/ are represented with ε and ο in the Secunda because
they were phonetically realized as the more lax (i.e., centralized) vowels [ɪ] and [ʊ].
6.4.2.1. Review of Scholarship
KUTSCHER has argued that the use of ε and ο to transcribe Hebrew etymological */i/ and */u/
in the Secunda, in addition to other contemporary evidence,327 indicates that */i/ and */u/ had
shifted to /e/ and /o/ in the vocalic systems of contemporary Hebrew pronunciation by 200
BCE (1969). The problem with such an argument, however, is that short /i/ and /u/ are attested
325. There is one instance in which /ū / is transcribed by ευ: ιουχαλευ ‫( יֻכְלוּ‬Ps. 18:39). This is generally
corrected to ιουχαλου*, but there may be an orthographic basis for such a variant. In Egyptian Koine, the Latin
name Lucias, usually spelled in Greek as Λούκιος, also has a variant of Λεύκιος (GIGNAC 1976, 216).
326. One could also suggest that Greek ε and ο were realized with more close pronunciations. In fact, BRIXHE
found that ε and ο had more close pronunciations in Koine Greek of Asia Minor (2010, 233).
327. He cites evidence for /i/, /u/ > /e/, /o/ in Greek and Latin transcriptions of Biblical Hebrew, Jewish
Palestinian Aramaic (JPA) manuscripts, and Mishnaic Hebrew (MH) manuscripts (1969).
- 265 -
in the pronunciations of all of the main Hebrew reading traditions of the Middle Ages (Tiber-
ian, Babylonian, Palestinian). KUTSCHER deals with this problem by positing a "substandard"
colloquial pronunciation, in which /i/, /u/ > /e/, /o/ obtained, over against a "standard" pro-
nunciation used in prestigious biblical recitation, in which /i/ and /u/ maintained their close
pronunciations (1969, 226).
HARVIAINEN questions whether the transcriptions truly reflect a "substandard" reading
tradition, instead proposing that /i/ and /u/ were actually somewhat flexible, vacillating be-
tween [i e ɨ] and [u ᴐ ʉ] depending on their consonantal environment (1977, 37–47, 70–72,
75–76, 95–98). This explanation seems closer to the truth, given the fact that KUTSCHER's the-
ory would require Hebrew and Aramaic speakers to maintain two slightly distinct vocalic sys-
tems simultaneously. HARVIAINEN's explanation is also supported by MORAG's research on the
relationship between various vernaculars and Biblical Hebrew reading traditions, in which he
finds that most communities have the same vocalic phonemes in the reading tradition as in
the vernacular (MORAG 1958; 1963; HARVIAINEN 1977, 82).
The idea of a "substandard" tradition also goes against the distribution of the mater
lectionis ‫ י‬in Second Temple period Hebrew. At both Qumran and in the Judaean Desert texts,
while the mater lectionis ‫ ו‬is used for long /ū ō /, and short /o/ (< */u/), the mater lectionis ‫ י‬is
used only for long /ı̄ / (but not short /e/ [< */i/]). This is interpreted as indicating that short /i/
was slightly lower or more centralized in quality than long /ı̄ / (QIMRON 1986, 19; MOR 2015,
46–51). If this was a trait of a "substandard" tradition of Hebrew, then we might expect to
find some biblical texts exhibiting the "standard" tradition, in which the mater lectionis ‫ י‬was
used also for short /i/. However, at least at Qumran, no such texts have been found.
KHAN points out that when etymological short /i/ lengthens it results in s ̣ere and not
ḥiriq. Therefore, the realization of */i/ must be closer to [e]; words like ‫' עֵדָ ה‬congregation' and
‫' ֵענָב‬grape' should then be reconstructed as */ˁida/ [ˁɪda] and */ˁinab/ [ˁɪnab] with etymological
- 266 -
short */i/ being realized phonetically as [ɪ].328 Greek ε (parallel to Tiberian ḥiriq) in the
transcriptions is an attempt to reflect the Hebrew vowel quality [ɪ] (2013i, 329). Presumably,
a centralized realization [ʊ] of etymological short */u/ is also likely, since when */u/ length-
ens it results in ḥolem and not shureq/qibbus ̣: e.g., ‫ < *( א ְֶרדּ ֹף‬ʔirdup) 'I will pursue'.
The development of pretonic lengthening as outlined by KHAN actually helps mediate
between the theories of KUTSCHER and HARVIAINEN. If we assume that */i/ and */u/ were real-
ized with more lax pronunciations as [ɪ] and [ʊ], differences in speech production and percep-
tion (see below) may prove to be just as relevant for explaining their different representations
in the transcriptions and the reading traditions as differences in dialect or register.

6.4.2.2. Greek Transcription
An argument could be made that Hebrew */i/ (> /e/) and */u/ (> /o/) maintained their close
pronunciations and were only transcribed with ε and ο because of a limitation inherent in
Greek orthography. The evidence of Greek transcription of other languages, however,
demonstrates clearly that this is not the case. Moreover, the specific transcription choices
support a lax realization of [ɪ] and [ʊ] as opposed to a complete shift to [e] and [o].
There are three potential options for transcribing Hebrew /e/ (< */i/): ι-ει = [i], η = [e],
or ε-αι = [ε]. The fact that a short i-vowel could be transcribed by Greek ι is clearly demon-
strated by the regular transcription of short /i/ by ι in both Latin and Akkadian (5.3.1.1.3;
5.3.2.1.3). Arabic short /i/, on the other hand, is almost always transcribed by Greek ε. How-
ever, in the environment of liquids and nasals, Arabic short /i/ may also be transcribed by
Greek η (5.3.3.1.2). Latin and Semitic long /ı̄ /, on the other hand, are regularly transcribed by
ι (or ει). Taken together, all of this evidence would indicate that Hebrew long /ı̄ / and Hebrew
short /e/ (< */i/) were not of the same quality in the Secunda. If they were, Hebrew short /e/
(< */i/) should have been transcribed by ι just as it was in Akkadian.
328. KHAN points out that a slightly centralized [e], which would be very near in its articulation to [ɪ], is
transcribed as [ë] in IPA conventions (2013i, 329).
- 267 -
The precise realization of Hebrew etymological */i/, in light of its representation by ε,
is difficult to determine. Theoretically, if its quality was nearest to [e], it could be represented
by Greek η as it is on occasion in Arabic. One might argue that the association of η and
length might prevent such a transcription, but there are examples of ε and η interchanging in
Greek transcription of Hebrew names in Palestinian epigraphy: e.g., Ιωσε/Ιωση (for ‫)יוסה‬
(CIIP I/1, no. 46, 81, 573), Ελιας/Ηλιας (CIIP I/2, no. 1021; CIIP II, no. 1165.3a), and
Ιεσους/Ιησους (CIIP I/1, no. 247, 425). The realization of ε was rather open in Palestinian
Koine, as shown by the fact that ε in Greek loanwords is rendered by pataḥ in Mishnaic He-
brew (see 6.4.4.5), and it is unlikely that Hebrew /e/ (< */i/) was identical in quality to Greek
ε [ε] at the time of the Secunda. Rather, Hebrew /e/ (< */i/) was likely realized as a more cen-
tralized [ɪ] and was thus transcribed by Greek ε [ε] because of its greater proximity to the
acoustic center than ι or η (see 4.5.3.1.12 for Greek ε representing centralized vowels).
There are also three potential options for transcribing Hebrew short /o/ (< */u/): ου =
[u], υ = [y], or ο-ω = [o]. Because of its unique fronted quality, Greek υ [y] is not typically
used in transcription.329 Claims that ου could only transcribe long vowels because it was long
in Greek (see YUDITSKY 2017, 70) are not supported by contemporary transcriptional evi-
dence. The fact that short /u/ could be transcribed by ου is clearly demonstrated by the regu-
lar transcription of Latin ŭ with Greek ου, even in unaccented syllables. Latin ŭ is represented
by Greek ο only in certain phonetic environments (5.3.1.1.5). Semitic short /u/, on the other
hand, is regularly transcribed by Greek ο (5.3.6.), though there are exceptional cases in Ara-
bic in which /u/ is transcribed by ου (5.3.3.1.3). Latin and Semitic long /ū / are normally ren-
dered by ου in Greek. These data indicate that Hebrew /ū / and /o/ (< */u/), like /ı̄ / and /e/
(< */i/), were realized with different qualities in the Secunda. Had they both been realized
329. In the environment of λ, however, Latin /u/ seems to be fronted to [y] and transcribed as υ (instead of its
normal rendering ου): e.g., Μαρκυλλα.
- 268 -
with the quality of [u], Hebrew short */u/ would have also been rendered as Greek ου just
like Greek transcriptions of Latin ŭ and of Arabic /u/ in exceptional cases.
An argument could be made for either a centralized near-close [ʊ] realization or a
close-mid [o] realization of Hebrew etymological */u/ in the Secunda. Contemporary Greek
orthography and transcription conventions leave either possibility open. However, in light of
certain developments in the history of Hebrew, such as short */u/ becoming /ō / under pretonic
lengthening and the presence of short [u] in the medieval reading traditions, it seems best to
posit that Hebrew /o/ (< */u/) was realized as a more centralized [ʊ] and was thus transcribed
by Greek ο [o] because of its greater proximity to the acoustic center than ου. This claim may
also be supported by the fact that in Greek loanwords in the Mishnah, while Greek ο is usual-
ly rendered by ḥolem-waw ‫ וֹ‬in Hebrew, in a few instances it is also rendered by shurq ‫ וּ‬in He-
brew (5.4.1.1.5).
One final piece of evidence to be mentioned here concerns exceptional spelling in the
quotations of the Hexapla in external sources. When preceding geminate consonants,330 ι and
ου may replace expected ε and ο: e.g., γιββωρ ‫' גִּבּוֹר‬mighty warrior' (Isa. 9:5) (Chrysostom;
Procopius), ακοββαι/ακουββαει ‫ ֲע ֻקבַּי‬/‫' ֲע ֵקבַי‬those who cheat me' (Ps. 49:6), and ισοββουνι/
ισουββουνει ‫יְסֻבּוּנִי‬/‫' י ְסוּ ֵבּנִי‬they encircle me' (Ps. 49:6) (Ambrosiana; Chrysostom).331 While
more work needs to be done regarding the reliability and antiquity of such transcriptions, if
reliable, these spelling variants in specific phonetic environments attest at the very least to
two facts: First, the transcription of */i/ and */u/ with ε and ο were not merely a transliteration
convention, but were particularly chosen to more accurately reflect the quality of the vowel.
Second, at least in certain phonetic environments, */i/ and */u/ had not completely shifted to
330. KUTSCHER also finds variants in the Greek and Latin transcriptions in which ι and ου may precede
geminated consonants (1969, 219–226).
331. In each case, the transcription with Greek ο is from the Ambrosiana palimpsest and the example with
Greek ου is from a quotation of the Secunda in Chrysostom's commentary on Psalms.
- 269 -
[e] and [o]. It may be that /e/ [ɪ] and /o/ [ʊ] had allophonic realizations of [i] and [u] in un-
stressed syllables closed by a geminate consonant.

6.4.2.3. Comparative Evidence: Modern Semitic Dialects and the Development of Latin
Positing a tense realization of long /ı̄ / and /ū / as [iː] and [uː] and a lax realization of short /i/
and /u/ as [ɪ] and [ʊ] in ancient Hebrew has parallels in the vocalic systems of dialects of
modern Arabic, Neo-Aramaic, and in the diachronic develpoment of the Latin vocalic system.
In Cairene Arabic, the long high vowels /ı̄ / and /ū / are realized as tense [iː] and [uː],
and short /i/ and /u/ are realized as lax [ɪ] and [ʊ] (MCCARTHY 2005, 20). Syrian Arabic (SA),
which has five long vowel phonemes (/ā ı̄ ū ē ō /) and three short vowel phonemes (/a i u/),
exhibits a similar situation. ALMBARK and HELLMUTH's acoustic analysis of Syrian Arabic
demonstrated that short /a i u/ were more centralized than long /ā ı̄ ū /, so that long /a i u/ are
realized as [aː iː uː] and short /a i u/ are realized as [ɐ ɪ ʊ], with short /i u/ having allophonic
realizations of [e,ə o]. In fact, short /i u/ are nearer in quality to long /ē ō / than to long /ı̄ ū /
(ALMBARK and HELLMUTH 2015). Even an acoustic analysis of standard Quranic recitation, in
which are only short /a i u/ and long /ā ı̄ ū /, demonstrated that the short vowels were realized
with a more centralized pronunciation than the long vowels (NEWMAN and VERHOEVEN 2002).
The vocalic systems of various dialects of Neo-Aramaic provide similar parallels. In
the dialect of Sulemaniyya and Ḥalabja, /i/ and /u/ are realized as [iː] and [uː] when long and
as [ɪ]/[ə] and [ʊ] when short, respectively. There is, in fact, some overlap in the phonetic real-
izations of short /i/ and short /e/ this dialect (KHAN 2004, 48–50). A similar phonetic realiza-
tion of /i/ and /u/ is also found in the dialect of the Jews of Arbel (Khan 1999, 52–53).
Classical Latin was characterized by five long vowels /ı̄ ē ā ō ū / [iː eː aː oː uː] and five
corresponding, but more centralized (aside from /a/), short vowels /i e a o u/ [ɪ ε a ᴐ ʊ]
(ALLEN 1977, 47). When vowel-length distinctions were lost in the development from Classi-
cal Latin to Vulgar Latin, short /i u/ merged with long /ē ō / and not with long /ı̄ ū /. The results
of this merger are reflected in the modern Romance languages (MARTINEZ 1989, 106, 110). In
- 270 -
Latin loanwords in Koine Greek, Latin ı̆ is typically transcribed by Greek ι, but is sometimes
transcribed by Greek ε (e.g., Δοµετιανοῦ/Δοµιτιανοῦ). Also, although Greek ου typically
transcribes Latin ŭ , there are certain words in which ου interchanges with Greek ο (e.g.,
ταβολαρίου/ταβουλαρίου).
6.4.2.4. Cross-Language Perception and Production of [ɪ] and [ʊ]
That ε [ε] and ο [ο] would represent [ɪ] and [ʊ] and that ι [i] and ου [u] would represent /ı̄ /
and /ū / is also supported by studies on cross-language perception, for a number of reasons.
First, vowel length and vowel tenseness are often associated perceptually. In a study
of second language acquisition, it was found that Arab students of English tend to associate
the tense and lax vowels of English with the long and short vowels of Arabic. For example,
the tense [i] vowel in the word [tin] 'teen' was pronounced by Arab students with a long /ı̄ /
vowel [tiːn], whereas the lax [ε] vowel in the word [bεd] 'bed' was pronounced by Arab stu-
dents with following gemination [badd], thus indicating that it was interpreted as short
(BROSELOW 1988, 298–300). In another study of English vowel production by native Arabic
speakers, it was demonstrated that Arabic speakers exaggerated the durational difference in
their production of tense and lax vowels (MUNRO 1993, 44–45).
Second, when [ɪ] is absent from the vocalic inventory of a language that has /i/, /e/, /ε/
but not /ɪ/, it may be perceived as /ε/. The vocalic inventories of Italian (/i, e, ε, a, ᴐ, o, u/),
Catalan (/i, e, ε, a, ᴐ, o, u/), and Korean (/i, e, ε, y, ø, ʌ, ɑ, o, u, i/) are suitable examples for il-
lustration.332 In a number of vowel perception studies with Italian–English bilinguals, the
most common mistake the Italians made in the perception and production of English /ɪ/ was
to identify it with English /ε/ (MUNRO et al. 1996, 330; FLEGELAN and MEADOR 1999, 2977;
PISKE et al. 2002, 64). Two studies of Catalan students learning English found that inexperi-
enced students were most likely to identify English /ɪ/ with Catalan /e/, and less frequently
332. For the Italian vocalic inventory, see AGARD and DIPIETRO 1964 apud PISKE et al. 2002. For the Catalan
vocalic inventory, see RECASENS and ESPINOSA 2005. For the Korean vocalic inventory, see SONG 2005 apud JIN
and LIU 2014, 584.
- 271 -
identified it with Catalan /ε/ or /i/. An interesting trend found in one of the studies was that as
students became more experienced in English they grew to identify English /ɪ/ more with
Catalan /i/ and less with Catalan /ε/ or /e/ (CEBRIAN 2006, 378; FABRA and ROMERO 2012,
495).333 In studies on Korean–English bilinguals, Koreans generally identify English /ɪ/ with
Korean /i/. However, there was a tendency for Koreans less-experienced in English, when
they made a mistake in the production or perception of English /ɪ/, to either produce it as
English /ε/ or identify it with Korean /e/ or /ε/ (YANG 1996; FLEGE, BOHN, and JANG 1997, 443,
448; TROFIMOVICH et al. 2001, 175, 179; BAKER et al. 2002; BAKER and TROFIMOVICH 2005, 10–
19; TSUKADA 2005, 269, 278–80; JIN and LIU 2014, 587). These variations in perception may
be explained by the relatively lower more centralized articulation of Korean /i/ (YANG 1996,
257–58), somewhere between English /i/ and /ɪ/, and the relatively higher articulation of the
e-vowels in Catalan and Italian (FLEGELAN and MEADOR 1999, 2978; FABRA and ROMERO 2012,
494). Finally, German speakers tended to equate /ε/ and /ɪ/ in their production and perception
of English vowels (FLEGE BOHN, and YANG 1997, 449).
Third, when [ʊ] is absent from the vocalic inventory of a language that has /u/ and /o/
(but not /ʊ/) it may be perceived as /o/. For Italian–English bilinguals, the most common mis-
take made in the perception and production of English /ʊ/ was to identify it with English /o/,
though identifying /ʊ/ with /ʌ/ also occurred (MUNRO et al. 1996, 330; FLEGELAN and MEADOR
1999, 2977; PISKE et al. 2002, 64). For Catalan speakers, English /ʊ/ was most frequently
identified with Catalan /u/, less frequently with Catalan /o/. More experienced learners did
not identify English /ʊ/ with Catalan /o/ at all, instead opting for /u/ and occasionally for /ᴐ/
(FABRA and ROMERO 2012, 495). In studies on native Korean speakers, English /ʊ/ was identi-
fied most commonly with Korean /u/, less frequently with Korean /i/. In terms of mistakes of
production, English /ʊ/ was commonly misproduced as English /u/ by Koreans (TROFIMOVICH
333. It should be noted that in the part of the experiment that tested vowel production, the more experienced a
learner was the more centralized and lower they pronounced English /ɪ/ (FABRA and ROMERO 2012, 502).
- 272 -
et al. 2001, 175, 179; BAKER et al. 2002; BAKER and TROFIMOVICH 2005, 10–19; JIN and LIU
2014, 587).
Fourth, interchanges of i/e and u/o occur in representing the etymologically short Ara-
bic vowels */i/ and */u/ in the different contexts in which Arabic is written in Latin characters
in the modern world. In a study regarding Arabic students learning English as a second lan-
guage, one student by the name of Muhammad (‫ ُﻣـ َﺤـ ﱠﻤـﺪ‬/muh ̣ammad/) was found to spell the ini-
tial vowel of his name with a u on some occasions (M-u-h-a-m-a-d) and with an o on other
occasions (M-o-h-a-m-e-d, M-o-h-a-m-m-a-d). Moreover, the general variations in transliter-
ated Arabic words, such as Muslim/Moslem and Qur'an/Koran, also attest to i/e and o/u varia-
tion (THOMPSON-PANOS and THOMAS-RUŽIĆ 1983, 612). In another study on "ASCII-ized"
transcriptions of Gulf Arabic, short /i/ may be transcribed with either i or e, as in yimkin/
yemken/yemkin for /yimkin/ 'could be' and short /u/ is transcribed with either u or o, as in
shukran for /š ukran/ 'thanks' and sho for /š u/ 'what...?' (PALFREYMAN and AL-KHALIL 2003).
The reflexes of */ı̄ /, */ū /, */i/, and */u/ in Secunda Hebrew may be summarized as follows:
Hebrew /ı̄ / Hebrew /e/ (< */i/) Hebrew /ū / Hebrew /o/ (< */u/)
Realized as ... [iː] [ɪ] [uː] [ʊ]
Transcribed as ... ι = [i] ε = [ε] ου = [u] ο = [o]
KUSTCHER's 1969 article on the shift of */i/ and */u/ to /e/ and /o/ was indeed a seminal one,
yet not beyond the need for refinement. On the basis of Greek transcription conventions, a
comparison of modern and ancient vocalic systems, and modern linguistic studies on cross-
language perception and production, we have demonstrated that what might be attributed to a
difference in register or dialect, can also be attributed to the difference in linguistic percep-
tion from one language to another.
- 273 -
6.4.3. Long /ɔ̄/ [ɔː] and Short /a/ [a]/[æ]
̣ /ā / [aː] or /ɔ̄/ [ɔː]?
6.4.3.1. "Qamas":
In the Secunda, Hebrew etymological long */ā / is transcribed by Greek α, which probably
represents a back open unrounded vowel [a] (or [ɑ]) in Roman Palestinian Koine:334
αδαµ335 /ʔɔ̄dɔ̄m/ [ʔɔːðɔː̃ (m)] 'man' Ps. 31:20

εµµανου /ʕemmɔ̄nū / [ʕɪmːɔːnuː] 'with us' Ps. 46:8
σαµ /ś ɔ̄m/ [sɔː̃ (m)] 'is placing' Ps. 46:9
In Greek transcription of other languages, α is used regularly to transcribe both short /a/ and
long /ā /. In Greek loanwords in the Mishnah, α is typically rendered by a Hebrew /a/ vowel
(i.e., pataḥ/qamas ̣), but is sometimes rendered with an /o/ or /u/ vowel in the environment of
the sonorants λ, µ, ν, ρ or in the environment of κ (5.4.1.1.1).
The central issue regarding the transcription of etymological long */ā / by Greek α is
whether Hebrew */ā / had yet shifted to a half-open back rounded vowel /ɔ̄/, more commonly
referred to as "qamas ̣." After a brief review of scholarship, I will argue that there is tentative
evidence for positing /ɔ̄/ in the Hebrew of the Secunda and that Greek α could be a suitable
representation of it.
6.4.3.1.1. Review of Scholarship
There are different opinions as to when precisely Hebrew long /ā / shifted in quality to [ɔː].
HARVIAINEN argues that there is no positive consistent evidence for the ā > ᴐ̄ quality change
until the fifth century CE (HARVIAINEN 1977, 104–7). MORAG (1963) and MEYER (1958) argue
that forms such as ‫ כבושים‬in the Dead Sea Scrolls indicate that the shift had taken place by the
334. The relative infrequency of spelling interchanges with α in Palestinian epigraphy indicates that α remained
distinct from the other Greek vocalic phonemes. It was probably realized as α was realized in Classical and
Hellenistic Greek, as a back open unrounded vowel [a] (or perhaps near [ɑ]) similar to the a-vowels in the
Italian word amare [amare] (PETROUNIAS 2007b, 558; PETROUNIAS 2007c, 602–605).
335. In a word like ‫ אָדָ ם‬in Tiberian Hebrew, the second qamas ̣ is the result of tonic lengthening and the first
qamas ̣ is the result of pretonic lengthening. In the Secunda, there is evidence that tonic lengthening has taken
place from the form ααφης* /ha-h ̣ā pē s ̣/ ‫' ֶה ָחפֵץ‬who is pleased' (Ps. 35:27) and there is evidence that pretonic
lengthening has taken place from the form νηχαρ /nē kā r/ ‫' נֵכָר‬foreigner' (Ps. 18:46). Therefore, we may
reasonably conclude that, even though α can represent either /a/ or /ɔ̄/ in the Secunda, the vowels represented by
α in αδαµ and similar forms were long.
- 274 -
Second Temple period, but KUTSCHER (1974) and QIMRON (1986) argue that such spellings are
to be explained otherwise (for a full review of scholarship, see REYMOND 2014, 138–40).
6.4.3.1.2. Possible Evidence of Qamas ̣ in the Secunda
The are a few of pieces of evidence that may indicate that /ā / > /ɔ̄/ had already taken place by
the time of the Secunda, namely, the transcription ουαλεα found in Chrysostom, the etymolo-
gy of Ενωχ 'Enoch' in Philo, and data from Jerome and the Babylonian incantation bowls.
6.4.3.1.2.1. Chrysostom's ουαλεα
In Biblical Hebrew, when a pronominal suffix attaches to a noun with a plural base, the plural
construct ending is typically pointed with s ̣ere (e.g., ‫' סוּסֵינוּ‬our horses', ‫' סוּסֵיהֶם‬their horses',
‫' סוּסֵיכֶם‬your horses'). However, when the pronominal suffix is pointed with qamas ̣, the quality
of the preceding vowel is seghol (e.g., ü‫' סוּסֶי‬your horses', ‫' סוּסֶי ָה‬her horses'). The seghol is
typically explained as the result of the s ̣ere (< *ay) assimilating in quality to the following
qamas ̣, since seghol and qamas ̣ were of the same half-open vowel height (KHAN 2013k, 268).
Therefore, if the Secunda transcriptions reflect seghol in such an environment, it would
presumably indicate that the shift /ā / > /ɔ̄/ had already obtained.
While there are no relevant transcriptions in the Ambrosiana palimpsest, a quotation
of the Secunda in John Chrysostom's comments on Psalm 7:8 (Expositio in Psalmos, 55.90)
renders the parallel of MT ‫ ְו ָעלֶי ָה‬as ουαλεα (LXX: ὑπὲρ ταύτης 'for this'). We might expect the
3fs suffix on a preposition with a plural base such as ‫ על‬to be something like **αλα in the Se-
cunda on the basis of a comparison with the 3fs suffix on the plural (e.g., αµουδα ‫[ עַמּוּדֶ י ָה‬Ps.
75:4]) or the form in Aramaic (‫) ֲעלַהּ‬.336 It is possible, then, that the reading ουαλεα is not orig-
inal. It would not be unusual if an earlier transcription resembling **ουαλα (or ουαληα) was
later corrected in conformity with the MT, since this phenomenon is attested elsewhere.337
336. Note that in Origen's commentary on Psalms, he specifically says that αµουδα is the Hebrew for στυλους
(pl.) αυτης 'its pillars' (Selecta in Psalmos, 12.1060.11). For the possibility of the intrusion of a spoken Aramaic
form ‫ ֲעלַהּ‬, note that the preposition ‫ על‬with the 3fs suffix is written as ‫ עלה‬in the Judaean Desert texts (see
5/6Hev3 and XHev/Se13).
337. For example, in Origen's list of biblical books found in his commentary on Psalms and quoted again in
Eusebius's history, the original reading for the title of the Book of Chronicles, Δαβρηϊαµειν, is eventually
changed to Διβρὴ Ἀϊαµίµ (see MOMMSEN 1908, 574), suspiciously identical to the Tiberian vocalization and
- 275 -
There is currently no critical edition of Chrysostom's Expositio in Psalmos, but Henry Sav-
ile's 1612 text of Chrysostom's works, in which the reading ουαλεα is found, is the product of
the consultation of many manuscripts.338 Nevertheless, more text-critical work is necessary
before we can determine whether or not the reading ουαλεα is original. If ουαλεα is indeed a
good reading, however, it would likely indicate that long /ā / had shifted in quality to /ɔ̄/ al-
ready in the Secunda as a necessary precursor for the assimilation of *ē > ɛ̄ / _Cɔ̄́.
6.4.3.1.2.2. Philo's and Origen's Etymology of Enoch (á‫ = חֲנוֹ‬Ενωχ)
Both Philo (25 BCE–50 CE) and Origen explain the etymology of the name Ἐνώχ (MT °‫ )חֲנוֹ‬as
χάρις σου 'your grace' (Philo, De posteritate Caini, 36.1; Origen, Selecta in Genesim,
12.121.8), presumably based on metanalysis: εν 'grace' + -ωχ 'your' or ‫' חן‬grace' + ‫' וך‬your'. If
the etymology results from such a metanalysis, which is likely, then long /ō / was confused
with the 2ms suffix (-αχ in the Secunda, ‫ ָך‬in Mishnaic Hebrew). This etymology is probably
not original to Philo or Origen, but derived from another Jewish source. Interestingly, Jerome
provides the correct etymology of 'dedicatio' in his Liber de Nominibus Hebraicis (9).
6.4.3.1.2.3. Mid-First Millennium CE Evidence for Qamas ̣
There are two pieces of evidence from the mid-first millennium CE which seem to indicate /ɔ̄/
in Hebrew. First, in one of Jerome's letters to Damasus (d. 384), he transcribes Hebrew "qa-
mas ̣" as o in the word lochen (MT ‫( ) ָלכֵן‬Epistula XXXVI, 2).339 Second, "qamas ̣" is indicated
by ‫ ו‬in the Babylonian incantation bowls: e.g., ‫' בורוך‬blessed', ‫' הורוחות‬the spirits', and
‫' בירושולים‬in Jerusalem' (MISHOR 2007; ELITZUR 2013, 850). The fact that /ɔ̄/ is already attested
inconsistent with the phonology and orthography of the Secunda (cf. δαβρη [Ps. 35:20]).
338. According to SCHAFF, SAVILE invested no small effort in collating manuscripts of Chrysostom's work: "The
edition of Sir Henry Savile (Provost of Eton), Etonae, 1612, in 8 vols. for., is less complete than the Benedictine
edition, but gives a more correct Greek text (as was shown by F. Dübner from a collation of manuscripts) and
valuable notes. Savile personally examined the libraries of Europe and spent £8,000 on his edition. His wife was
so jealous of his devotion to Chrysostom that she threatened to burn his manuscripts" (1889, 3). Note that
ουαλεα is actually found in vol. 1 (1611).
339. There is, however, a variant in the apparatus of lachen, though lochen was judged by the editor as the more
faithful reading (see HILBERG 1910).
- 276 -
in the middle of the first millennium CE and present in both Tiberian and Babylonian strongly
suggests that it has more ancient roots than these mid-first millennium CE atttestations.
6.4.3.1.3. Greek α for [ɔ]?
The salient question that remains, then, is whether Greek α could represent a vowel of the
quality [ɔ(ː)], or if Greek ω/ο would be better suited to transcribe such a quality.
While there is insufficient ancient evidence to rule in favor of one possibility over the
other, some data may be interpreted as indicating that ω/ο would be preferred for transcribing
[ɔː]. In Greek transcriptions of Phoenician, the vowel resulting from the "Phoenician Shift"
(*/a/ > */ā / > */ɔ/ > /o/) is transcribed as ω/ο: e.g., λαβον 'white', ναδωρ 'he vowed', σαµω 'he
heard' (FRIEDRICH and RÖLLIG 1999, 40–41). It is not clear, though, whether ω/ο transcribes
the end result (/ó / [oː]) or the intermediate stage (/ɔ́/) of the shift (see also FOX 1996, 38–39).
In Greek transcriptions of Latin, short ŏ , which was realized phonetically as [ɔ], is usually
transcribed by Greek ο. However, we must remember that Latin o and Greek ο were very
similar or identical paleographically (5.3.1.1.4).
On the other hand, modern linguistic studies on cross-language perception suggest
that a vowel with the quality of /ɔ/ can be identified with either /o/ or /a/ depending on the re-
lationship of the vowel spaces of the respective languages. For example, in a study of Spanish
speakers' (Spanish vocalic inventory: /i/, /e/, /a/, /o/, /u/) identification of Southern Standard
British English (SSBE) and American English (AE) vowels, it was found that while SSBE /ɔ/
was almost always identified with Spanish /o/, AE /ɔ/ was identifed also with Spanish /a/
about 29% of the time. This is presumably because SSBE /ɔ/ is higher than AE /ɔ/ (ESCUDERO
and CHLÁDKOVÁ 2010, 256–57). In another study, native Catalan speakers (Catalan vocalic in-
ventory: /i/, /e/, /ɛ/, /a/, /ɔ/, /o/, /u/) were found to identify AE /ɔ/ most frequently with Cata-
lan /a/ (FABRA and ROMERO 2012, 494–95).
FABRA and ROMERO'S study of Catalan speakers' perception of AE vowels is especially
instructive, as well as cautionary, for our present analysis. Particularly noteworthy is the fact
- 277 -
that AE /ɔ/ was most frequently identified with Catalan /a/, even though Catalan has /ɔ/ (!) in
its own phonemic inventory. Moreover, a comparison of the normalized vowel spaces depict-
ed in FABRA and ROMERO's charts (2012, 494) reveals that AE /ɔ/ is between Catalan /o/ and
/ɔ/. The F1 and F2 frequencies illustrate this point (RECASSENS and ESPINOSA 2006, 655):
F1 F2
Catalan /o/ 489 1047
English /ɔ/ 570 840
Catalan /ɔ/ 608 1125
Catalan /a/ 730 1358
The fact that Catalan /a/ is more distant from English /ɔ/ than both Catalan /o/ and Catalan /ɔ/
demonstrates how non-intuitive perceptual assimilation can be across languages. This is espe-
cially relevant when we consider that, at least at face value, Greek ο/ω (a true mid [o̞ ]) was
likely nearer in the vowel space to Hebrew /ɔ̄/ [ɔː] than Greek α ([a] or [ɑ]) was.
These modern cross-linguistic studies provide justification for exploring the possibili-
ty that ancient Greek α could have represented "qamas ̣". If ancient Hebrew qamas ̣ was a half-
open /ɔ̄/ as in Tiberian, it probably would have been even lower than AE /ɔ/. Moreover, it is
probable that ancient Greek /a/ was a low back vowel as suggested by PETROUNIAS (2007b,
558; 2007c, 604). The vowel space of Palestinian Koine also supports this. Because ε was
particularly open (5.4.1.1.2), a more back realization of Greek /a/ (perhaps approaching [ɑ])
would have been more contrastive. The only difficulty with Greek α representing Hebrew /ɔ̄/
concerns the precise phonetic realization of ο/ω. If it was a true-mid vowel [o̞ ], it would be
closer to /ɔ̄/ and thus more likely to be the most apt for transcription (but note Catalan /a/ ≈
AE /ɔ/).
In sum, there is insufficient evidence to determine whether ο/ω or α would be the best
transcription choice for /ɔ̄/. Nevertheless, in light of modern cross-linguistic perceptual stud-
ies and ancient Greek phonology, it seems entirely possible that a low back Greek /a/ (≈ [ɑ])
- 278 -
vowel represented by α may transcribe a mid-open rounded /ɔ̄/. There are, of course, numer-
ous conceivable ways in which the various vocalic systems might relate to each other.
6.4.3.1.4. Summary
In light of the fact that it is entirely feasible that ancient Greek α may have represented /ɔ̄/,
the argument for the existence of /ɔ̄/ in the Secunda can be summed up as follows: The pres-
ence of qamas ̣ in both Tiberian and Babylonian indicates that "qamas ̣" probably has more an-
cient roots than its earliest attestations in Babylonian incantation bowls from the middle of
the first millennium CE. Greek ε (instead of η) in the transcription ουαλεα reflects assimilation
to a following /ɔ̄/. Therefore, if ουαλεα is an original reading and assimilation to qamas ̣ is the
only appropriate explanation for seghol in forms like ‫סוּסֶי ָה‬, the evidence seems to indicate
that /ɔ̄/ was present in the Secunda. While this issue requires further research and text-critical
work, we will operate on the very tentative supposition that "qamas ̣" /ɔ̄/ [ɔː] existed in the Se-
cunda and will transcribe it as such with the understanding that it may still have been realized
merely as /ā / [aː] (or [ɑː]).

6.4.3.2. Short /a/: [a] or [æ]?
In the Secunda, Hebrew short /a/ is normally transcribed by α in the Secunda:
αλ /ʕal/ [ʕal] 'on' Ps. 18:42

ραββιµ /rabbı̄ m/ [ʀabːıː̃ (m)] 'great' Ps. 32:10
σαµθ /ś amt/ [samth] 'you set' Ps. 89:41
There are also a number of instances in which an expected /a/ is transcribed by ε, which rep-
resents an open-mid front vowel [ε] (or true mid [e̞ ]) in Palestinian Koine (4.5.3.1.10).340
These can be divided into a number of categories, most (~16x) in the prefix vowel of hifˁil:341
θεριβ /terh ̣ı̄ b/ [thɪʀħiːβ] 'you widen' Ps. 18:37
340. The upshot of this section is as follows: Hebrew short /a/ is at least sometimes represented by Greek ε,
which indicates that its phonetic realization was a front vowel [a] and perhaps (at least in some instances) a
slightly raised vowel [æ]. While it lies beyond the scope of this dissertation to evaluate every possible instance
of Greek ε for Hebrew /a/, I have stated my reasonings for my transcriptions in the following footnotes. Full and
detailed argumentation for each of these transcription choices will have to be articulated in future works.
341. It is assumed that the prefix vowel of the hifˁil stem was /e/ [ɪ] as a result of analogy to the prefix vowel in
other verb stems such as qal. A similar analogy seems to occur in Palestinian Arabic (ELIHAY 2012, 760–61).
- 279 -
εττη /hetṭẹ̄ / [hɪtˀːeː] 'incline!' Ps. 31:3
There are also about five instances in which ε represents the initial vowel of the piˁel stem:342
χελλωθαµ /kallō tam(m)/ [khælːoːθa(̃ m)] 'annihilating them' Ps. 18:38

φελλετηνι /palletẹ̄ nı̄ / [phælːɪtˀeːniː] 'rescue me!' Ps. 31:2
εελλελεχ* /ʔhallelek(k)/ [ʔæhælːɪlɪkh] 'I will praise you' Ps. 35:18
Even before gutturals, the prefix vowel of the qal stem is represented with ε:343

ουϊερογου /w-yeh ̣r(o)gū / [(ʔ)ujɪħʀʊʁuː] 'and they will tremble' Ps. 18:46
There is one instance in which the initial vowel of the qal stem is represented with ε:344
σεωθι /š a(h ̣)h ̣ō tı̄ / [ʃæħoːθiː] 'I was bowed down' Ps. 35:14
In a few construct forms, expected /a/ is also transcribed with ε:345
βααδαρεθ /b-hadrat/ [bahaðaʀæθ] 'in raiment of' Ps. 29:2

βιεδ /b-yad/ [b(i)jæð] 'into the hand of' Ps. 31:9
µεϊεδ /mey-yad/ [mɪjːæð] 'from the hand of' Ps. 89:49
In guttural and geminate Qatl(-at)* nominal forms, expected /a/ is also transcribed with ε:346
θεθ /teh ̣t/ or /tah ̣t/ [thɪħth]/[thæħth] 'under' Ps. 18:39

ρεκ /raq(q)/ [ʀækˀ] 'only' Ps. 32:6
In waw consecutive forms, the vowel of the conjunction is sometimes represented with ε:347
342. On the basis of the forms ουβαρεχ °‫( וּב ֵָר‬Ps. 28:9), ζαµµερου* ‫( זַמְּרוּ‬Ps. 30:5), and φαλητ ‫( ַפלֵּט‬Ps. 32:7),
instances of ε in the stem of the piˁel are regarded as reflecting a raised [æ] realization of Hebrew short /a/.
343. There is a tendency for the pattern of the strong qal non-stative verb (qɔ̄tal, qō tē l, yeqtol) to be generalized
across the paradigm in the Secunda. For example, originally stative *ḥapis ̣ē (MT ‫ ) ֲח ֵפצֵי‬is realized as ωφση (Ps.
35:27) and original *tisʕadē nı̄ (MT ‫ )תִּ ְסעָדֵ נִי‬as θεσοδηνι (Ps. 18:36). It is assumed, then, that the /e/ prefix vowel
in I-guttural qal forms, represented by Greek ε, is the result of analogy to the qal strong verb: *kɔ̄tab : *yektob ::
*ʕɔ̄zab : ? ( > *yeʕzob).
344. On the basis of a comparison with Jerome's calloth ָ‫( קַלּוֹת‬SIEGFRIED 1884, 41), the ε in this form likely
reflects a raised realization [æ] of Hebrew short /a/, though raising could also be due to the sibilant (see 6.3.2).
345. Greek ε reflects a raised realization [æ] of Hebrew short /a/ in these instances (see 6.4.3.2.2.1), though the ε
in βιεδ and µεϊεδ may be due to assimilation to /y/ (see YUDITSKY 2017, 96–98).
346. The form ‫ תחת‬may have had an alternate pattern (i.e., *qitl). It is assumed that Greek ε in ρεκ reflects a
raised [æ] realization of Hebrew short /a/, though raising may also be due to the ‫( ק‬see YUDITSKY 2017, 96).
347. On the basis of a comparison with forms like ουαϊαλεζ ‫ז‬Äֲ‫( ַויַּע‬Ps. 28:7) and ουαθθεµας ‫( וַתִּ מְאָס‬Ps. 89:39)
and the realization of wayyiqtol throughout the various traditions of Hebrew, it is assumed that Greek ε reflects a
raised [æ] realization of Hebrew short /a/.
- 280 -
ουεθαζερηνι /wat-tʔazzerē nı̄ / [wæthː(a)ʔazːɪʀeːniː] 'and you girded me' Ps. 30:12
ε
ου ϊεριβου /way-yerh ̣ı̄ bū / [wæjːɪʀħiːβuː] 'and they made wide' Ps. 35:21
Finally, the interrogative pronoun ּ ‫מַה‬, the negative particle ‫אַל־‬, and the relative particle ‫ַאשֶׁר‬
all have a vowel transcribed by ε where we would expect /a/:348
µεββεσε /meb-bes ̣ʕ/ [mæbːɪʦˀɪʕ] 'what gain?' Ps. 30:10

ελθαρακ /ʔal terh ̣aq/ [ʔæl thaʀħakˀ] 'do not be far!' Ps. 35:22
εσερ /ʔš ɛr/ [ʔæʃɛʀ] 'which' Ps. 46:9
There are essentially two ways of interpreting the use of Greek ε to transcribe what we would
expect to be Hebrew /a/. First, Greek ε does indeed reflect Hebrew /a/ in these instances and,
for some phonetic reason, approximates Hebrew /a/ just as well as or better than α does. Sec-
ond, Greek ε here reflects the phoneme /e/ [ɪ] (or an allophone of another phoneme realized
as [ɪ]/[ə]/[ɛ]) in these forms just as it does regularly in the Secunda. In each case, there is ei-
ther a morphological or phonetic explanation for the presence of /e/ instead of /a/. It is also
possible, of course, that some but not all of the forms are given to one or the other explana-
tion. After a review of scholarship, I will argue that, though many of these forms can be ex-
plained morphologically or phonetically, the evidence suggests that at least in some of these
words, ε is used to represesnt Hebrew /a/, realized as a near-open front unrounded vowel [æ].
PRETZL argues that in a closed stressed syllable, in addition to representing */i/, ε can also rep-
resent an etymological */a/ vowel that had shifted to an open front vowel (perhaps [æ]?) in
the environment of certain consonants (liquids, gutturals, sibilants) (1932, 8, 13).
SPEISER argues that while "qamas ̣" is represented by α in the Secunda, "pataḥ" may be
represented by either α or ε, particularly in closed syllables far from the stress.349 At the time
348. Greek ε reflects a raised realization [æ] of Hebrew short /a/ in these instances (see 6.4.3.2.2.2–3), though
the ε in εσερ may be due to the sibilant (see 6.3.2).
349. His list includes the vowel of the prefix conjugation in the hifˁil stem (9 times), the prefix vowel of the
imperative in the hifˁil stem (5 times), the vowel between the first and second radical in various forms in the piel
stem (5 times), various verbal forms (6 times), segholate nouns (19 times), other nominal forms (7 times), and
pronouns and particles (15 times) (1943, 267–68).
- 281 -
of the Secunda, short /a/ inclined towards /e/ and long /ā / was more of a pure a-quality vowel.
After /ā / shifted in quality to /ɔ̄/, "pataḥ" then shifted in quality to occupy the space previous-
ly held by "qamas ̣" (1933, 35–44).350
BRØNNO is unwilling to accept that short /a/ can be signified by ε in the Secunda.
Rather, he argues for a pervasive /a/ > /e/ sound change occurring in closed unstressed sylla-
bles in the Secunda, similar to the rule known as "attenuation" (e.g., */magdā l/ > /migdā l/) in
Tiberian Hebrew. Many forms (e.g., the irregular /i/ in hifˁil) are the result of this change, but
others (e.g., θεθ) must ultimately derive from variant patterns or scribal errors (1943, 18, 26,
30–31, 203, 245–46, 267–68, 290–93, 301, 304, 307–309, 443, 439, 449).
JANSSENS continues in the line of PRETZL and SPEISER, arguing that etymological */a/, if
not lengthened to /ā /, was realized as a more close [æ] in the Secunda; thus, /e/ (< */i/) is
transcribed as ε, /a/ (< */a/) as α/ε, and /ā / (< */ā /) as α (1982, 67, 70–74).351
Finally, YUDITSKY argues that instances of ε in the Secunda are either the result of the
development of etymological */i/ or the result of contraction. To explain the unusual in-
stances of */a/ > ε cited above, he appeals to analogy, assimilation to adjacent consonants
(e.g., sibilants, /k ̣/, /y/), and derivation from different patterns. For example, he suggests that
the hifˁil and piˁel forms are the result of assimilation to the past tense and that θεθ derives
from a *qitl pattern. (2007b, 303 n13; 2017, 49–52, 150–51, 159–61, 222–23).
It is true that many of the forms discussed above can be explained on the basis of
analogy, assimilation, or derivation from variant patterns. However, there are some forms for
which these explanations are not sufficient: e.g., µεββεσε, ελ, and βααδαρεθ (see below). It is
difficult to determine whether such forms would indicate a general realization of /a/ as [æ] or
350. According to SPEISER, /ā / did not shift to /ɔ/ in Babylonian Hebrew and thus "pataḥ" remained as a more
front vowel. For this reason, pataḥ can signify both pataḥ and the equivalent of Tiberian seghol (1933, 35–44).
351. JANSSENS transliterates this hypothesized vowel as ä, but due to his use of the word "close" to describe the
pronunciation of the vowel he probably means something more like [æ] (1982, 67).
- 282 -
merely that a sporadic sound change (e.g., a > e / C_C[-stress]) was in operation. I concur with
YUDITSKY (2007a, 10–11) that a wide "attentuation" rule did not operate in the Secunda,
though we should keep in mind that reduction and centralization away from the stress is com-
mon in many languages.

6.4.3.2.2. Etymological */a/ > ε: Analysis of Forms
It would be unfruitful to discuss forms in which the presence of ε could likely be explained as
the result of analogy, such as the prefix vowel of the hifˁil stem. Only those forms with ε for
*/a/ which are unlikely to be explained by analogy, assimilation to nearby consonants, or de-
rivation from a variant pattern are valuable for argumentation. Additionally, a strong case for
ε = [æ] can be made if a particular transcribed word with ε has a biform with α elsewhere in
the Secunda and is attested in all other Hebrew reading traditions with /a/. If the most likely
interpretation of such words is that ε is representing a realization of /a/ [æ], the principle can
then be considered for other transcriptions. The three words in which it is most likely that ε
represents /a/ [æ] are βααδαρεθ, µεββεσε, ελ-, and -εννα They will be examined in turn.
6.4.3.2.2.1. βααδαρεθ
In the transcription βααδαρεθ ‫( ְבּהַדְ ַרת־‬Ps. 29:2), the feminine construct ending */-at/, which is
attested everywhere in Hebrew as /-at/,352 is transcribed as -εθ. Parallel Secunda forms show
that the feminine ending */-at/ is usually transcribed by -αθ: e.g., εµαραθ ‫' ִאמ ְַרת־‬word of' (Ps.
18:31), οννεχαθ ‫' ֲחנֻכַּת‬dedication of' (Ps. 30:1), and αρφαθ ‫' ח ְֶרפַּת‬reproach of' (Ps. 89:51).
BRØNNO explains the unusual ending on this form as either the result of dittography
(βααδαραθ > βααδαρθθ > βααδαρεθ) or a segholate pattern with an epenthetic (*hadart >
(βα)αδαρεθ) (1943, 152–53). YUDITSKY also argues that the transcription βααδαρεθ may de-
rive from the pattern *qatalt and the ε is an epenthetic vowel. However, because an epenthet-
ic is rare in the Secunda, he concludes that the transcription is corrupt (2017, 79, 192–93).
352. In Samaritan Hebrew it is realized as -å t as in yē š uwwå t 'salvation of', aš få t 'edge of', and wtirruwwå t 'and
the shout of' (FLORENTIN 2016b, 73, 75, 87).
- 283 -
It is unlikely that this form is the result of scribal error, since we have a parallel form
in which */-at/ is transcribed by -εθ in external sources. In John Chrysostom's comments on
Jeremiah 49:28 (30:23 in the LXX), he says that τῇ βασιλίσσῃ τῆς αὐλῆς 'to the queen of the
court' is realized in Hebrew as Μελχεθ Ασωρ (64.1029), presumably reflecting a variant ‫ַמ ְלכַּת‬
‫ חָצוֹר‬from the MT ‫וּ ְל ַמ ְמלְכוֹת חָצוֹר‬. It is unlikely that this is a scribal error in Chrysostom, since
he also renders βασίλισσα 'queen' with µελχεθ in his comments on Jeremiah 44:17 (51:17 in
the LXX) (64.1013).
The word βααδαρεθ may derive from a different pattern, but no such pattern is attest-
ed in Hebrew for this word. In Tiberian, the form is always vocalized as ‫הַדְ ַרת‬, Babylonian
only attests to tr'dh'b/] tr'd]h' (YEIVIN 1985, 927), and Palestinian only attests to hedrat (MURTO-
NEN 1988 I/Ba, 83). Targumic Aramaic has the forms ‫ הַדָּ ָרה‬and ‫הֲדַ ְרתָּ א‬, but even there the con-
struct form attested is ‫( הַדְּ ַרת‬JASTROW 1926, 335). YUDITSKY compares βααδαρεθ to forms such
as Ναζαρεθ ‫ נָצ ְַרת‬in the Gospels and µαελεθ ‫ ַמ ֲחלַת‬in the LXX (2017, 193). ELIZUR argues that
original *qatalt, as in Ναζαρεθ, tends to be realized as qatεlεt in an older layer of Hebrew, but
as qā tlat in a later layer of Hebrew: e.g., ‫' בּ ֶָרקֶת‬carbuncle' (type of stone) in the Pentateuch but
‫ בּ ֽ ְָרקַת‬in Ezekiel (2004, 227–28).353 However, the fact that the initial vowel in the various He-
brew vocalization traditions is short demonstrates that ‫ הדרת‬derives from a different pattern
than that of ‫ ברקת‬or ‫נצרת‬. Moreover, this pattern (*qatalt > qatεlεt/qatlat) seems to be as-
signed to special classes of nouns (e.g., place names, infirmities).
Even if we allow for a unique nominal pattern behind βααδαρεθ in the Secunda, both
the presence of epenthesis and the quality of the epenthetic vowel in βααδαρεθ are inconsis-
tent with the typology of segholation in the Secunda. Epenthesis in segholate nouns in the
Secunda is only found in the environment of gutturals or in the environment of resh when the
353. Note that ‫ בּ ֶָרקֶת‬is probably a loanword.
- 284 -
final consonant cluster follows the Sonority Sequencing Principle (SSP) (ιεθερ ‫)י ֶתֶ ר‬.354 In He-
brew and Semitic, epenthesis in a final consonant cluster is more likely with rising sonority
(e.g., ιεθερ), but less likely with decreasing sonority (e.g., ‫( ) ַויֵּבְךּ‬KIPARSKY 2003, 149, 160–61,
168–172). Also, except in the case of the pharyngeal /ʕ/, epenthetics in the Secunda have the
same quality as the adjacent vowel (YUDITSKY 2017, 79–80). In the case of βααδαρεθ, not
only does sonority fall significantly—C2 (‫ )ר‬is at the top of the sonority hierarchy in Hebrew
and C3 (‫ )ת‬is at the bottom (see ALVESTAD and EDZARD 2009, 49)—but the epenthetic is of a
different quality than the preceding vowel. Therefore, despite the comparative patterns in the
Gospels and the LXX such as Ναζαρεθ ‫ נָצ ְַרת‬and µαελεθ ‫( ַמ ֲחלַת‬YUDITSKY 2017, 193), it would
be inconsistent typologically with Secunda Hebrew to posit that ε in βααδαρεθ reflects a
helping vowel. A much simpler phonetic explanation is laid out in 6.5.1.5.1.
In sum, while it is possible that βααδαρεθ constitutes a unique pattern in the Secunda,
such a claim is inconsistent with the semantics of the noun, its attestations elsewhere in He-
brew, and the phonotactics of Secunda Hebrew. Alternatively, we may suggest that the ending
-εθ in the form βααδαρεθ, just as in the transcription µελχεθ in Chrysostom's commentary on
Jeremiah, reflects the Hebrew ending /-at/ realized as something like [æθ]. Whether the rais-
ing of the vowel was the general realization of short /a/ in the Secunda or due to a sporadic
change as a result of being far from the stress is unclear.

6.4.3.2.2.2. µεββεσε
In the word µεββεσε ‫( מַה־ ֶבּצַע‬Ps. 30:10), the interrogative pronoun ּ ‫ מַה‬is realized with ε for
an expected /a/. Parallel forms in the Secunda demonstrate that the word is regularly spelled
with α: e.g., χαµµα ‫' ַכּמָּה‬how long/much?' (Ps. 35:17), λαµα ‫' ָלמָּה‬why?' (Ps. 49:6), αδ·µα
‫" עַד־מָה‬how long?" (89:47), and αλ·µα·σαυ ‫שׁוְא‬

ָ ‫" עַל־מַה־‬on account of what vanity?" (89:48).
BRØNNO argues that the first ε in µεββεσε may be the result of an /a/ > /e/ shift in un-
stressed closed syllables. It corresponds etymologically with Tiberian Hebrew ‫ מֶה‬before non-
354. But cf. νεεµαναθ ‫נֶ ֱא ֶמנֶת‬. However, note the relevance of the OCP (see 6.5.2.2). For the SSP, see 6.5.2.1.2.
- 285 -
gutturals (1943, 211). YUDITSKY, though he admits that there is no clear solution for the pres-
ence of ε in µεββεσε, suggests that instead of being a transcription of the interrogative pro-
noun ‫' מה‬what?', µε may actually be a transcription of a negative particle which may be com-
pared with Mishnaic Hebrew ‫שׁמֵּא‬

ֶ and Samaritan Aramaic dimme (2017, 111).
In Tiberian Hebrew, the various syntactic states of the interrogative ‫ מה‬may be classi-
fied as proclitic (i.e., attached to following word [e.g., ‫)]מַה־שְּׁמוֹ‬, enclitic (i.e., after a preposi-
tion [e.g., ‫)] ַכּמָּה‬, or independent (i.e., with a disjunctive accent [e.g., ‫)]ו֣יהִי ָ֔מה‬.
ִ The vocaliza-
tion ‫ מֶה‬only occurs in proclitic ‫ מה‬before a pharyngeal with qamas ̣ (e.g., ‫)מֶה־ ָעשָׂה‬, in enclitic
non-pausal ‫( מה‬e.g., ‫) ַבּמֶּה‬, and in independent ‫( מה‬e.g., ‫)מֶ ֛ה ֥קוֹל ֶהה ָ֖מוֹן ַה ֶזּ֑ה‬.355 It is only the last of
these three, independent ‫ מה‬before non-gutturals, that might correspond with µεββεσε in the
Secunda. However, a diachronic analysis of ‫ מה‬in the various Hebrew reading traditions
demonstrates that the vocalization of independent ‫ מה‬as ‫ מֶה‬before non-gutturals in phrases
like ‫ מֶ ֛ה ֥קוֹל ֶהה ָ֖מוֹן ַה ֶזּ֑ה‬is most likely a later Tiberian innovation.356
355. When proclitic, it is vocalized as ּ ‫ מַה‬with gemination in the following consonant. When it is enclitic, it
occurs as ‫ מֶה‬in context (e.g., ‫ ) ַבּמֶּה‬and as ‫ מָה‬in pause (e.g., ‫ ) ַבּמָּה‬and in the word ‫ ָלמָה‬/‫ ָלמָּה‬. When it is independent,
it occurs as ‫ מֶה‬before a word and as ‫ מָה‬after a word. Before a guttural, it usually appears as ‫מָה‬. However, if it is
proclitic, it will appear as ‫ מַה‬when the guttural takes virtual gemination and as ‫ מֶה‬before a pharyngeal with
qamas ̣ (JOÜON and MURAOKA 2009, §37).
356. In Babylonian, the rules for ‫ מה‬are fairly similar, but in several instances Babylonian has qamas ̣ where
Tiberian has seghol (YEIVIN 1985, 1134–39). In Palestinian, ‫ מה‬is often left unpointed when it corresponds to
Tiberian pataḥ, but doubling may be marked (e.g., ylﬞ hﬞm). It may be pointed with qamas ̣ before a guttural (e.g.,
lydbh hmuw and l[ hmu) (REVELL 1970, 176), but with /e/ before ‫ ח‬with an /a/ vowel (e.g., ld$j] hm# [YAHALOM 2016b,
111–12]). In Samaritan Hebrew, ‫ מה‬is always vocalized as mā̊ . However, there is a distinction in vocalization
between ‫כמה‬/‫ במה‬bā̊ må /kā̊ må and ‫ למה‬lē må (BEN-ḤAYYIM 2000, 238–39, 320–21).
Assuming that ‫ מה‬pointed with a pataḥ and following gemination (ּ ‫ )מַה‬is the most original form, we may divide
the various changes in the vocalization of ‫ מה‬into two categories. First, there are sound rules that operate
relatively consistently based on the immediate phonological environment (e.g., proclitic ‫ מָה‬before gutturals and
‫ מֶה‬before pharyngeals with qamas ̣). Second, there are sound changes that operate based on the prosodic and
syntactic structure of the verse (e.g., enclitic and independent ‫מָה‬/‫)מֶה‬.
A few diachronic observations may be made regarding these distinctions as they relate to the Secunda. First,
there is evidence against the phonological rule a > e / _Cː[+laryngeal]ā , which applies to proclitic ‫מה‬, operating in
the Secunda (e.g., ααφης* ‫ ) ֶה ָחפֵץ‬and in Jerome's transcriptions. However, there is evidence for the a > e /
_Cː[+laryngeal]ā change in Tiberian (e.g., ָ‫)מֶה ָעשִׂית‬, Babylonian (e.g., htyç[; hm'), and Palestinian (e.g., ld$j] hm#).
Second, there is evidence against the prosodic change of a > e that applies to enclitic ‫ מה‬both in Jerome (bamma
|| ‫ ) ַבּמֶּה‬and Babylonian Hebrew (µym[p hm;k' || ‫ ַכּמֶּה ְפ ָעמִים‬,hm;b' || ‫ ַבּמֶּה‬, hm; ˜[y || ‫)יַעַן מֶה‬. Third, it is unclear if there is
evidence for the prosodic change of a > e that applies to independent ‫ מה‬in Babylonian Hebrew. Finally, when
Babylonian does not exhibit an a > e change in certain prosodic conditions, it often demonstrates an a > ā
change instead. These observation are summarized below (✓ = evidence for, ✕ = evidence against, — = not
enough evidence):
- 286 -
In light of the diachrony of ‫ מה‬in the various traditions, then, it is highly unlikely that
the Secunda form µεββεσε corresponds to prosodically-conditioned ‫ מֶה‬in Tiberian Hebrew as
BRØNNO suggests. Even if it were reasonable to reconstruct mĕ for the Hebrew of the Secun-
da, BRØNNO would have to assume that µεββεσε is a mixed form, since its vocalization would
correspond to an independent form even though the following gemination would indicate that
it is a proclitic form. While YUDITSKY's suggestion that µε in µεββεσε should be interpreted as
the negative particle cannot be disproven, it is highly unlikely for two reasons. First, all of the
ancient translations (Aquila, Symmachus, LXX, Theodotion, Jerome, Targum) understand the
word ‫ מה‬in the sequence ‫ מה בצע‬in Psalm 30:10 to be an interrogative. Second, there is no at-
testation anywhere else in Hebrew of a negative particle ‫ מה‬standing on its own with a vocal-
ization other than that of the definite article.357 This leaves no other reasonable option but to
interpret µεββεσε as the interrogative ‫ מה‬in the Secunda.
Phonologically-Motivated Changes: Secunda Jerome Tib. Bab. Pal.
➣ (1) Proclitic ‫מה‬: a > e / _Gːā ✕ ✕ ✓ ✓ ✓
➣ (2) Proclitic ‫מה‬: a > ā / _Gːv̄ ̆ [-open back long] — — ✓ ✓ ✓(?)
Prosodically-Motivated Changes: Secunda Jerome Tib. Bab. Pal.
➣ (3) Enclitic ‫)בכל(מה‬: a > ā — — ✓ ✓ —
➣ (4) Independent ‫מה‬: a > ā — — ✓ ✓ —
➣ (5) Enclitic ‫)בכל(מה‬: a > e — ✕ ✓ ✕ —
➣ (6) Independent ‫מה‬: a > e — — ✓ ? —
It seems clear from the chart that change (1) in proclitic ‫ מה‬occurred after the period of the transcriptions but
still relatively early. It is likely that change (5) in enclitic ‫ מה‬only applies in Tiberian Hebrew. Although the evi-
dence for change (6) is inconclusive, it makes the most sense to associate it with change (5) and assume that it
applies only in Tiberian. There is no evidence for a terminus ante quem for changes (3) and (4), but they are
probably as old as the reading tradition itself, since ‫מה‬, disjoined from the following word, would not have
brought about gemination of the following consonant.
357. In the examples YUDITSKY cites (‫שׁמֵּא‬ ֶ and dimme), the negative particle is enclitic with a preposed particle.
This is parallel to ‫שׁ ַלּמָּה‬
ַ in Song of Songs 1:7. When ‫ מה‬does function as a non-enclitic negative particle in
Biblical Hebrew, it is vocalized like the interrogative: e.g., ‫' וּמָה אֶתְ בּוֹנָן‬and I will not look' (Job 31:1).
Interestingly, Symmachus, the LXX, and Theodotion translate ‫ מה‬in this verse as a negative (FIELD 1875, 2:54).
- 287 -
Because gemination is indicated, µεβ(βεσε) should be regarded as a proclitic form.
The forms χαµµα, λαµα, and αδ·µα should be regarded as enclitic. The final attestation,
αλ·µα·σαυ, could be interpreted as either proclitic (i.e., 'on account of what vanity ... ?') or
enclitic (i.e., 'on account of what ... vainly?'). Both interpretations are found in the ancient
translations, most translating it as enclitic.358 The singleton σ supports an enclitic interpreta-
tion, but the fact that all three words are written on the same line with diacritics in between
them (|αλ·µα·σαυ|) and its proclitic status in the Tiberian vocalization support a proclitic
reading in the Secunda. If αλ·µα·σαυ does represent a proclitic form of ‫מה‬, it would support
reading the ε in µεβ(βεσε) as representing /a/. The evidence, however, is inconclusive.
In sum, the ε in µεβ(βεσε) could be interpreted as either a sporadic instance of /a/ > /e/
in a closed unstressed syllable, attested nowhere else in Hebrew for this word, or, and more
likely, the ε is merely an alternative representation of Hebrew /a/. YUDITSKY's claim that it is
to be interpreted as a negative particle lacks supporting evidence.

6.4.3.2.2.3. ελ
In the phrase ελθαρακ ‫( אַל־תִּ ְרחַק‬Ps. 35:22), the negative particle ‫ אַל־‬is rendered with an ε
where we would expect /a/. The fact that this word is attested eight times in the Secunda, al-
ways as ελ, precludes any possibility of this being a scribal error.
BRØNNO, based on an article by BLAKE (1911), claims that the negative particle was
originally just a vocalic /l̥ /. A prosthetic vowel of varying quality was added, resulting in ʔal
in Tiberian Hebrew but ʔel in Secunda Hebrew. He also suggests that /a/ might have shifted
to /e/ in the proclitic word (1943, 213–14). YUDITSKY, drawing on the interchanges of ‫ אֶל־‬and
358. Symmachus (proclitic interpretation): ἢ ἐπὶ τίνι µαταίωι ἔκτισας πάντας τοὺς υἱοὺς τῶν ἀνθρώπων 'On
account of what vanity did you create all the sons of men?' Aquila (enclitic interpretation): ἐπὶ τί εἰκῆ ἔκτισας
πάντας υἱοὺς ἀνθρώπων 'On account of what did you vainly create all the sons of men?' LXX (enclitic
interpretation): µὴ γὰρ µαταίως ἔκτισας πάντας τοὺς υἱοὺς τῶν ἀνθρώπων 'For have you vainly (lit. 'for not
vainly...?') created all the sons of men?' Theodotion (enclitic interpretation): ἐπὶ τί µαταίως ἔκτισας πάντας
τοὺς υἱοὺς τῶν ἀνθρώπων 'On account of what did you vainly create all the sons of men?' Jerome (enclitic
interpretation): numquid enim vane constituisti omnes filios hominum? 'For you have surely not vainly created
all the sons of men?' Targum (enclitic interpretation): ‫' מטול מה לבטלא בריתא כל בני־נשא‬On account of what did
you vainly create all the sons of men?'
- 288 -
‫ אַל־‬in Tiberian and Babylonian (see below), suggests that the prohibitive particle might have
had multiple allomorphs and that the Secunda reflects the /ʔel/ allomorph (2017, 216).
BRØNNO's claim that a preceding vowel is not original can be rejected on the basis of
comparative Semitic evidence. Negative ˀl is common in Semitic, attested in Geˁez ʔal, Saba-
ic/Qatabanic ˀl, Mehri əl (in the phrase əl ... lā ), Ugaritic ʔal, Phoenician ˀl, Old Aramaic ˀl,
and Hebrew ʔal (WENINGER 2011b, 170). According to SJÖRS, who argues that *ʔal was origi-
nally a "prohibitor" used to indicate negation and volition in Proto-Semitic, there is no per-
suasive reason to expect anything other than /a/ in Hebrew. He does, however, note that the
vowel of *ʔal is usually /a/ in Semitic, but may raise or lower depending on the particular re-
flex of */l/ in a given language (e.g., å l in Modern South Arabian due to the velarized /l/)
(2015, 86–89, 281–88, 303–309; pers. comm.).
The prohibitive particle is realized with an /a/ vowel (/ʔal/) in all of the main Hebrew
reading traditions.359 There are a small number of exceptions in both Tiberian and Babylonian
in which the negative particle ‫ אַל־‬is pointed like the preposition ‫אֶל־‬,360 but the fact that this
interchange goes in both directions indicates that the exceptional vocalizations are lexical
variants in the tradition—they substituted the word ʔεl 'to' for ʔal 'not'—and not variants in
pronunciation or vestiges of a tradition which realized *ʔal as ʔel. The regular vocalization of
‫ אל־‬as ελ in the Secunda can hardly be compared with these extremely rare interchanges of
‫אֶל־‬/‫ אַל־‬in Tiberian and Babylonian. In fact, there is no evidence that the negative particle ˀl
was regularly realized as anything other than /ʔal/ in any tradition of Hebrew or Northwest-
Semitic language.
359. The prohibitive particle is realized in Tiberian, Babylonian, and Palestinian as ʔal and in Samaritan as al
(YEIVIN 1985, 1117; YAHALOM 2016b, 115–16; FLORENTIN 2016b, 73, 83).
360. In Tiberian, there are a few instances in which the negative particle ˀal is pointed as ˀεl (Exod. 10:28, Deut.
2:9, Josh. 22:19, Jer 51:3 [2x]) and other instances in which the preposition ˀεl is pointed as ˀal (Judg. 19:23, Jer.
47:6, Prov. 12:28). In Babylonian, there is at least one occasion in which the manuscripts vary between ˀal and
the preposition ˀil (Jer. 47:6) (YEIVIN 1985, 1117).
- 289 -
In sum, there are a few options for interpreting the transcription ελ in the Secunda.
First, ελ could represent an innovation in Secunda Hebrew found nowhere else in the Hebrew
reading traditions or in Northwest Semitic. Second, ελ may reflect an instance of the sporadic
/a/ > /e/ shift in an unstressed closed syllable (BRØNNO 1943, 214–15). Third, and the most
likely, is that the ε in ελ may simply be an alternative representation of the phoneme /a/, con-
sistent with the rest of the reading traditions of Biblical Hebrew.

6.4.3.2.2.4. αννα/εννα
Although it comes from an external source, we may also mention the various renderings of
the long imperative ending with following ‫נָא‬. In Origen's Commentary on Matthew, Hebrew
‫ ַה ְצלִיחָה נָּא‬and ‫( הוֹשִׁיעָה נָּא‬Ps. 118:25) are rendered by ασιλαννα and ωσιεννα, respectively,
presumably reflecting either α or ε for short /a/ in precisely the same context.
6.4.3.2.2.5. Summary
For each of these irregular transcriptions, multiple interpretive possibilites have been put
forth, none of which is certain. In my view, the arguments for ε representing /a/ in the
transcriptions βααδαρεθ, µεββεσε, ελ, and εννα are more compelling than those against it,
which usually require the transcription to reflect a unique phenomenon particular to the Se-
cunda. In light of the evidence of these four words, we should be open to the possibility that,
at least on occasion, Greek ε may represent Hebrew /a/ in the Secunda.

6.4.3.2.3. Cross-Language Perception and Production of [æ], [a], and [ɑ]
That Greek ε [ε] might represent Hebrew /a/ is also supported by the linguistic typology of
open vowels and studies on cross-language perception of [æ] and [a].
First, cross-linguistically, in vowel systems with both long /ā / and short /a/, the short
/a/ tends to be realized with a more front articulation and the long /ā / tends to be realized with
a more back realization (HOCK 1991, 144). (In traditional pronunciations of Geˁez among
Amharic speakers, long /ā / is realized as a low central [a] or [ɑ] while short /a/ is realized as a
near open [æ] or [ä] (LAMBDIN 1978, 3).) In many modern Arabic dialects, short /a/ tends to
have a higher second formant (correlating with vowel frontness) than long /ā / (ROSENHOUSE,
- 290 -
AMIR, and AMIR 2014, 6). In Modern Persian, /a/ is realized as [æ] and /ā / is realized as [ɑː]. A
more front realization of short /a/ is also the case in numerous dialects of German and in con-
servative French (HOCK 1991, 144).
Second, when /æ/ is absent from the vocalic inventory of a language that has /ε/ and
/a/, it may be perceived as either /ε/ or /a/. We may take KIM's cross-linguistic study of Kore-
ans' perception of English vowels as an example.361 This study is particularly relevant because
the participants essentially engaged in a transcription exercise, writing out English words that
they heard in Korean orthography. English /æ/ was most frequently transcribed as Korean /ε/
(1972). In another perceptual study, it was found that when Spanish (Spanish vocalic invento-
ry: /i, e, a, o, u/) speakers attempted to produce English /æ/, they were prone to misproduce it
as /ɑ/ and less frequently as /ε/ (FLEGE BOHN, and YANG 1997, 448). Another study found that
English /æ/ was assimilated to Spanish /a/ 94% of the time and to Spanish /e/ only 6% of the
time. However, in a discrimination task, /æ/ and /ε/ were regarded to be the same vowel
about 80% of the time (JESKE 2012, 18). Finally, in a study of Salento Italian speakers' per-
ception of English vowels, it was found that English /æ/ was perceived as Salento /a/ 54% of
the time and as Salento /e/ 46% of the time (SISINNI, ESCUDERO, and GRIMALDI 2014, 716).
Third, an a/e interchange is also found in the Arabic-in-Latin-letters example cited
earlier: M-o-h-a-m-e-d, M-o-h-a-m-m-a-d (THOMPSON-PANOS and THOMAS-RUŽIĆ 1983, 612).

6.4.3.2.4. The Damascus Psalm Fragment: Arabic /a/ = ε
We find a transcriptional parallel in an eighth-century CE fragment of an Arabic translation of
the Psalms written in Greek script known as the Damascus Psalm Fragment. In this text, Ara-
bic short /a/ is transcribed by Greek ε: e.g., γεβελ /gabal/ 'mountain', γανεµ /ġ anam/ 'goats',
σεµα /samā (ʔ)/ 'sky', and µεσκεν /maskan/ 'dwelling'. In the environment of back consonants,
short /a/ is transcribed by α, though the distribution is not entirely clear. AL-JALLAD suggests
that perhaps Arabic short /a/ had been raised to [æ] in "non-backed environments" and thus
361. The Korean vocalic system has /i, e, ε, y, ø, ʌ, ɑ, o, u, i/ (see 6.4.2.4).
- 291 -
both α and ε were regarded as appropriate transcriptions of this phoneme (2017, 15–16). Be-
cause the text is so much later than our period, it was excluded from chapter 5. However,
since the pronunciations of Greek α and ε are quite similar from the end of the Koine period
until Modern Greek, it is applicable for this specific case.

6.4.3.2.5. Greek Loanwords in the Mishnah: Greek ε = Hebrew /a/
While cross-linguistic perceptual studies and Arabic transcription demonstrate that it is theo-
retically possible that Hebrew /a/ could have been realized with a more front and/or raised
pronunciation and thus equated with Greek ε on occasion, the vocalization of Greek loan-
words in Mishnaic Hebrew provides direct evidence of such an equivalency.
In Greek loanwords in the Mishnah, ε is typically rendered by a pataḥ in Hebrew,
usually in closed stressed syllables. HEIJMANS concludes on the basis of this fact that Greek ε
had a rather open realization in contemporary Palestinian Greek (see 5.4.1.1.2). Thus, He-
brew /a/ was regarded as a better approximation of Greek ε than Hebrew /e/ would have been.
The fact that we are dealing with cross-language equivalencies and not sound changes
is supported by a few points. First, in a minority of instances, Greek ε was realized in Hebrew
with an /e/ vowel (5.4.1.1.3). Second, Greek η is usually reflected in Hebrew by /e/ vowels
(usually s ̣ere) (5.4.1.1.3), thus demonstrating that Greek ε was more open than η and nearer
to Hebrew /a/. Third, in a few Latin loanwords in the Mishnah, Latin ĕ [ε] is also rendered by
Hebrew /a/ (e.g., catella ‫( ) ְק ַטלָּה‬HEIJMANS 2013, 262). In sum, the evidence from Greek loan-
words in the Mishnah seems to indicate that Hebrew short /a/ in Tannaitic Hebrew was of
such a quality that it was nearer to Greek ε than Hebrew /e/ was.
6.4.3.2.6. Concluding Remarks
A survey of the evidence in the Secunda, cross-linguistic perceptual studies, the Damascus
Psalm Fragment, and Greek loanwords in the Mishnah suggests that Hebrew /a/ may have
been realized as something like [æ] and thus transcribed by Greek ε, at least in the case of
βααδαρεθ, µεββεσε, ελ-, and -εννα. While previous scholars' presupposition that one
- 292 -
grapheme can only correspond with one phoneme may make an analysis of the transcriptions
less troublesome, it is unrealistic and inconsistent with the data.
Having established the principle in a small number of words, we may also point to
other potential instances in which Greek ε might correspond with Hebrew /a/ in the Secunda,
such as the piˁel imperative forms. While YUDITSKY argues that forms such as φελλετηνι and
ουνεσσηµ prove that the piˁel imperative had been formed in analogy to the past, the forms
ουβαρεχ (α due to compensatory lengthening before /r/) and ζαµµερου* may very well indi-
cate that the first vocalic phoneme in these forms was Hebrew /a/ and that α and ε are alterna-
tive representations of it. Thus, the transcription φελλετηνι may represent Hebrew /palletẹ̄ nı̄ /
just as Greek σέλλα is rendered as /sallā / in the Mishnah (HEIJMANS 2013, 262).
In sum, though we have made a strong case that Greek ε might represent Hebrew
short /a/ in the Secunda on occasion, this only facilitates phonemic transcription. Determining
the precise phonetic realization of Hebrew short /a/ is more difficult. However, a couple
points can be made. First, if the realization of short /a/ in Secunda Hebrew was similar to that
of Mishnaic Hebrew, then Greek ε was nearer to Hebrew /a/ than it was to Hebrew /e/. This
probably indicates that Hebrew short /a/ was realized as either a front open [a] (like pataḥ in
Tiberian Hebrew [KHAN 2013a, 95]) or a near-open front [æ]. Second, the phonetic quality of
Hebrew short /a/ must have generally been nearer to Greek α than it was to Greek ε. This sup-
ports an open realization of /a/ as [a].
Understanding that Hebrew was a real language, we should not be suprised if short /a/
vacillated in its phonetic realization, having allophones of [a] and [æ]. We may compare this
to the Neo-Aramaic dialect of the Assyrian Christians of Urmi, in which short /a/ vacillates in
its realization between [a], [æ], and [ɛ] (and [ɑ] in emphatic words) (KHAN 2016, 64), or the
oral production of modern Literary Arabic short /a/, which is typically realized between [a]
and [æ] (and [ɑ] near emphatics) (MITCHELL 1990, 72–82).
- 293 -
When it comes to phonetic transcription, then, we will adopt a conservative but in-
structive approach. Hebrew short /a/ will be phonetically transcribed as a front unrounded
open vowel [a], with the understanding that it may have had a slightly raised realization of
[æ]. In those particular cases in which Hebrew short /a/ is represented by Greek ε, it will be
represented in phonetic transcription by [æ], leaving the door open for understanding it as an
allophone in a particular environment. Given the distribution, it would not be surprising if
some raising/centralization occurred far from the stress362 or before geminate consonants.363
In sum, the transcription conventions for Hebrew /a/ and /ɔ̄/ may be depicted as follows:
Hebrew /a/ Hebrew /ɔ̄/

Realized as ... [a]/[æ] [ɔː]
Transcribed as ... α = [a]/[ɑ], ε = [ε] α = [a]/[ɑ]
6.4.4. Potential Phonemes: /ɛ̄/ and /ε/
Various reconstructions of the development of the Hebrew vowels would require positing that
ε represents the vocalic phonemes /ɛ̄/ and /ɛ/ in a number of transcriptions in the Secunda (see
6.4.4.1). First, what would be final ‫ ֶה‬in Tiberian Hebrew is rendered by ε, which represents
an open-mid front vowel [ε] (or true mid [e̞ ]) in Roman Palestinian Koine (4.5.3.1.10):

µασε /mah ̣sɛ̄/ [maħsɛː] 'a refuge' Ps. 46:2
ιερε /yerʔɛ̄/ [jɪʀʔɛː] 'he will see' Ps. 49:10; 89:49
ζε /zɛ̄/ [zɛː] 'this' Ps. 49:14
The parallel to pausal seghol, attested once in the Secunda, is also transcribed by Greek ε:
βεχι /bɛ̄́kı̄ / [ˈbɛːχiː] 'weeping' Ps. 30:6
However, non-pausal ‫ ְבּכִי‬, in which the first vowel would be short, is transcribed identically:
βεχι /bkı̄ /́ [b(ə)χiː] 'weeping' Ps. 30:6
362. There is evidence that when α becomes reduced/centralized far from the stress in Greek, it is represented
with ε (4.5.3.1.12; GIGNAC 1976, 278–93).
363. Note that Hebrew /e/ and /o/ may have raised allophones when they precede geminate consonants: e.g.,
γιββωρ ‫( גִּבּוֹר‬Isa. 9:5) and ακουββαει (Ps. 49:6).
- 294 -
In the context _́C1C2, original short *a and *i are also sometimes transcribed with Greek ε:
ελλελθ /h ̣ellelt/ [ħɪlːɪlth] 'you profaned' Ps. 89:40

δερχ /derk/ [dɪʀkh] 'path' Ps. 89:42
Finally, when epenthesis occurs in a non-II-guttural *qVtl pattern, the epenthetic is usually ε:
µεββεσε /meb-bes ̣ʕ/ [mæbːɪʦˀɪʕ] 'what gain?' Ps. 30:10

In section 6.4.1, we demonstrated that the transcriber prioritized quality over historical
quantity when transcribing Hebrew sounds into Greek. This opens the door to a couple possi-
bilities: First, Greek ε may have signified both short and long vowels. Second, if a Hebrew
phoneme with the quality [ε(ː)] existed in the Secunda, it would probably be transcribed by
Greek ε [ε]. After a brief review of scholarship, I will argue that the orthography of the Se-
cunda is consistent with the possibility, but not determinant, that both /ɛ̄/ and /ε/ were present.
BRØNNO (1943, 12), JANSSENS (1982, 51), and YUDITSKY (2017, 17) interpret the vocalic sys-
tem of the Secunda as having only two e-vowels: /e/, /ē /. This presumably reflects their view
of the history of Hebrew and, in the case of BRØNNO and YUDITSKY, the presupposition that
one letter should correspond with only one sound.364 BLAU, on the other hand, suggests that
we should expect four e-vowels in Secunda Hebrew: /e/, /ē /, /ε/, /ɛ̄/ (BLAU writes: ê ,̣ e ̣, ę ̂ , ę
[1984, 77]). While /ē / is represented by Greek η, the vowels /e/, /ɛ̄/, /ε/ are all represented by
Greek ε (1984, 77). SUCHARD's interpretation of the development of the Hebrew vowels also
requires four e-vowels (/e/, /ē /, /ε/, /ɛ̄/) at the time of the Secunda (2016, 276–79).365
The debate regarding the presence of a long /ɛ̄/ phoneme in the Secunda or lack there-
of centers primarily around the reflex of the final triphthong vyu#/vyi# in both nominal and
364. YUDITSKY writes: "according to the basic assumption (‫)הנחת היסוד‬, one letter reflects one sound" (my
translation) (2007a, 2n13). It is not entirely clear if he means that one Greek letter has one default phonetic
value or if he means that one Greek letter can only be used to represent one Hebrew sound. Regardless, the
former claim is demonstrated to be false by the hundreds of spelling interchanges documented in chapter 4; the
latter claim is unlikely on the basis of the various cross-language perception studies cited in this chapter.
365. It should be noted, however, that SUCHARD suggests that ɛ̄ may simply have been an allophone of ē .
- 295 -
verbal forms: e.g., µασε ‫ ַמ ֲחסֶה‬and ιερε ‫י ְִראֶה‬. BRØNNO, JANSSENS, and YUDITSKY interpret these
forms as ending in short /e/, but BLAU suggests that the final vowel in forms like µασε ‫ַמ ֲחסֶה‬
and ιερε ‫ י ְִראֶה‬should be interpreted as /ɛ̄/ instead of /e/, for two reasons: First, in the LXX, in
which quantitative transcription is the norm, these forms are transcribed with a final η: e.g.,
Μανασση ‫ ְמנַשֶּׁה‬and Ιεφοννη ‫י ְ ֻפנֶּה‬. Second, this final vowel is written as η both in the construct
(e.g., ωση ‫עֹשֵׂה‬, but cf. κασε ‫ ) ְקצֵה‬and in the imperative (e.g., εττη ‫( ) ַהטֵּה‬1984, 77). According
to SUCHARD's interpretation of the history of Hebrew, there are at least two other forms in the
Secunda in which final ε should be interpreted as reflecting long /ɛ̄/ instead of short /e/. First,
the grapheme ε in the demonstrative ζε ‫זֶה‬, the development of which is reconstructed as *ḏ ı̄ >
*zı̄ > *zē > *zɛ̄ > zε ‫זֶה‬, should reflect a long /ɛ̄/: /zɛ̄/. Second, since SUCHARD finds evidence
for what he terms "minor pausal lengthening" in the Secunda, ε in the form βεχι ‫ ֶבּכִי‬should
also reflect a long /ɛ̄/: /bɛ̄́kı̄ / (2016, 138, 249).
If a short /ε/ existed in the Secunda, depending on one's reconstruction of the history
of the Hebrew vowels, it might be found in the following forms: the non-pausal reflex of
*qaty patterns (a > ε / _Cy) such as βεχι /bεkı̄ /́ (?) ‫( ְבּכִי‬Mal. 2:13, not Ps. 30:6), in contexts
where Philippi's law operates (é > έ / _C1C2) such as ελλελθ /h ̣ellɛlt/(?) ָ‫( ִח ַלּלְתּ‬Ps. 89:40), in
originally *qatl/*qitl forms that correspond with qɛtɛl in Tiberian such as (β)δερχ /b-dɛrk/(?)
°‫( בְּדֶ ֶר‬Ps. 32:8), λανες /l-nεs ̣h ̣/(?) ‫( ָלנֶצַח‬Ps. 49:10), and αµµελχ366 /ham-mεlk/(?) °ֶ‫ ַה ֶמּל‬, and pos-
sibly also as the epenthetic vowel in forms like ιεθερ ‫( י ֶתֶ ר‬Ps. 31:24) and βεσε ‫( ֶבּצַע‬Ps. 31:2)
(see LAMBDIN 1985; GARR 1989; HUEHNERGARD 2013; SUCHARD 2016, 276–79). In addition to
these words, we may also add the second vowel in the reflex of ‫ ֲאשֶׁר‬, which was probably at
least phonetically realized as [ɛ] from an early period, perhaps as an allophone of /a/ (for the
origins of ‫ ֲאשֶׁר‬, see HUEHNERGARD 2006).
366. In Origen's commentary on Psalms, the form αµµελχ is part of the name he transcribes for I/II Kings:
Βασιλειῶν γ´ δ´ ἐν ἑνί Ουαµµελχδαυιδ ὅπερ ἐστὶν βασιλεία Δαυίδ 'and the third and fourth Book of Kingdoms
[in our tradition], in [the Jewish tradition is found] in one [book, whose title is]: "And King David ... ," which is,
"The Kingdom of David"'. It corresponds with the first words of I Kings.
- 296 -
6.4.4.2. Evidence for Long /ɛ̄/ in the Secunda
There is no direct evidence in the Secunda regarding whether or not the final ε vowel in
forms like µασε and ιερε is long or short. The interpretation of these forms depends on one's
interpretation of the development of final */ē #/367 in Hebrew. Did it lower and shorten under
the stress (ē > ε / _́#) or merely lower (ē > ɛ̄ / _́#)? Τhe interpretation of these specific forms
remains uncertain, but a number of points can be made. First, there may be clear evidence
elsewhere in the Secunda that the grapheme ε at least can represent long [εː]. Second, Greek ε
transcribes long vowels in transcription of other languages. Third, and finally, interchanges of
η and ε are attested in both the Secunda and contemporary Palestinian epigraphy.
6.4.4.2.1. Chrysostom's ουαλεα
In our discussion regarding "qamas ̣," we suggested that the transcription ουαλεα, if original,
was evidence of *ē > ɛ̄ / _Cɔ̄́ (6.4.3.1.2.1). Such a transcription is also relevant for our
present discussion, since it would prove that Greek ε could be used to indicate a long vowel
in the Secunda. Because other parts of the paradigm indicate that this vowel was long (e.g.,
ελωηνου ‫הֵינוּ‬Äֱ‫[ א‬Ps. 18:42], ουααρηεµ ‫[ וְאַח ֲֵריהֶם‬Ps. 49:14]), we can assume that the ε in
ουαλεα also indicates a long vocalic phoneme /ɛ̄/ or allophone [ɛː]. On another note, if origi-
nal, this form would also be significant because it would serve as the sole attestation of a 3fs
suffix on a preposition with a plural base in the Secunda, demonstrating that it matched the
Hebrew form rather than the Aramaic form. On this point, it should be noted that there are in-
stances of the preposition ‫ על‬with the 3fs suffix being written as ‫ עליהא‬at Qumran in both bib-
lical and non-biblical texts. In sum, then, confirming the validity of the reading ουαλεα would
also confirm that ε, at least in some instances, may signify [εː] in the Secunda.
367. Even though the final vowel in forms like ‫ ַמ ֲחסֶה‬develops from the triphthong vyu#/vyi# and the final vowel
in forms like ‫ זֶה‬probably develops from ı̄ yu#, both of these forms are thought to shift to /ē #/ before they shifted
to seghol in Tiberian Hebrew.
- 297 -
6.4.4.2.2. Greek ε for Long Vowels in Transcription of Other Languages
There does not seem to be any restriction on Greek ε representing long vowels in transcrip-
tion of other languages. It is used, though far less frequently than η, to transcribe Latin ē dur-
ing the Roman period (5.3.1.1.2). In the Damascus Psalm fragment, when long /ā / has shifted
to long /ē /, it is transcribed by ε: e.g., /kɛ̄n/ (* < /kā na/) (AL-JALLAD forthcoming, 17). Finally,
in transcription of Phoenician-Punic, the plural construct ending -ē is transcribed sometimes
by η and sometimes by ε: e.g., φενη and φανε for /panē / 'face of' (5.3.5.1).
6.4.4.2.3. Interchanges of η and ε for Final [eː]/[εː]
The interchange of final η/ε also occurs in both the Secunda and Palestinian epigraphy. In the
Secunda, the final vowel of two construct forms from *vyu#/*vyi# are transcribed by η/ε:368
ωση /ʕō ś ē / [ʕoːseː] 'those who do' Ps. 31:24

κασε /qs ̣ē / [kˀaʦˀeː] 'the edge of' Ps. 46:10
Based on the history of Hebrew, we would expect both of these transcriptions to end in η to
represent long /ē /. Interchanges of η and ε are attested in Palestinian and Egyptian Koine both
in word-final position and in the environment of σ (4.5.3.1.7; GIGNAC 1976, 242–49). There
are also examples of final η and ε interchanging in Greek transcriptions of Hebrew names in
Palestinian epigraphy from the Roman period. For example, the name ‫יוֹסֵה‬, which is vocal-
ized regularly with s ̣ere in the Kaufmann MS of the Mishnah, is transcribed with either ε or η
in Palestinian epigraphy: Ιωσε/Ιοσε/Ιωση (CIIP I/1, no. 46, 81, 573). Also, the final vowel in
the name ‫ ְמנַשֶּׁה‬is transcribed with η: Μενασση (CIIP III, no. 2222). The rendering of this
name in both the LXX (Μανασση) and Neo-Assyrian cuneiform texts (e.g., mi-na-si-i, me-
na-se-e, mi-in-se-e, mu-na-se-e, me-na-se-e) indicates that the final vowel was long at an ear-
lier stage of Hebrew (RÖLLIG 1960, 385–86; BAGG 2007, 342–45; MILLARD 2013, 839–40).
368. YUDITSKY discusses these forms as ωσε and κασε (2017, 145, 189–90), but the correct reading of the former
word in the palimpsest, verified recently by spectral imaging, is ωση.
- 298 -
6.4.4.3. Evidence for Short /ε/ in the Secunda
Two of the categories in which short /ε/ might occur in the Secunda, namely, ε where Philip-
pi's law operates and an initial ε in *qatl/*qitl > qεtɛl forms, are problematic. With respect to
Philippi's law, both short /e/ and short /ε/ would be transcribed by ε in a transcription such as
ελλελθ ָ‫( ִח ַלּלְתּ‬Ps. 89:40) or εκσερθ (Ps. 89:46). In the case of segholate forms from *qatl/
*qitl, there are no forms with ε in the Secunda that are unequivocally from *qatl.369 The ε in
originally *qitl forms (e.g., χεσλ ‫ ֵכּסֶל‬and σεθρ ‫)סֵתֶ ר‬, on the other hand, may simply reflect /e/
as the reflex of original */i/.
The precise interpretation of Greek ε as the initial vowel in the non-pausal reflex of
*qaty nouns and as the epenthetic vowel in non-II-guttural segholates is difficult in each case
for a similar reason. In the case of βεχι, it is not clear if ε represents a centralized shewa-type
vowel or the quality of the short vowel, since both ε and α seem to be used to represent a she-
wa-type vowel in the Secunda (see 6.4.6) and ε may be used to represent a centralized vowel
in Greek (4.5.3.1.12).370 In a transcription exhibiting a similar syllable structure, the initial
shewa-type vowel is represented with an unetymological α (λαβανι /l-bnı̄ / [laβaniː] 'to my
son' [Hos. 11:1]; cf. βεν [Ps. 9:1]) (but see 6.5.1.3.2).
In the case of ιεθερ and βεσε, one could make the case that the epenthetic is simply a
centralized [ə] vowel, though GARR argues that the seghol is the natural outcome of an origi-
nal epenthetic [ə] in the segholates: e.g., *ká lb > *ká ləb > *kɛ́ləb > [kɛ́lɛḇ ] [(1989, 112–15).
LAMBDIN (1985), on the other hand, who argues for the shift *kalb > *kɛlb before the inser-
tion of an epenthetic vowel, claims that the epenthetic would have been [ɛ] initially because it
matched the quality of the preceding vowel (see also HUEHNERGARD 2013). In sum, then, the
369. The only segholate forms with ε in the Secunda that seem to come from *qatl in Tiberian Hebrew are δερχ
and µελχ. In the case of δερχ, a good case can be made that it was originally a *qitl form. In the case of µελχ,
while it probably comes from *qatl originally as in Aramaic, Akkadian, and Ugaritic, there may be evidence that
in Phoenician and in some dialects of Hebrew it shifted to *qitl (note LXX, etc.).
370. Note the evidence for an early *qaty > *qity change in Hebrew (HUEHNERGARD 2015, 37). The following
forms could all be represented by βεχι in the Secunda: */bɛkı̄ / > [bɛχiː], */bikı̄ / > [bɪχiː], */bkı̄ / > [bəχiː].
- 299 -
second ε in ιεθερ and βεσε could represent (1) a centralized vowel [ə] > [ɛ] (/yetr/ [jɪθəʀ] or
[jɪθɛʀ]), (2) a vowel identical in quality to the preceding vowel /e/ [ɪ] (/yetr/ [jɪθɪʀ]), or (3) a
vowel identical in quality to the preceding vowel after segholization (/yɛtr/ [jɛθɛʀ]).
Finally, the second ε in εσερ ‫( ֲאשֶׁר‬Ps. 31:8) almost certainly reflects either a neutral
centralized [ə] or [ɛ]. Because of its realization in Tiberian (‫ ) ֲאשֶׁר‬and Babylonian (rç'a)'
(YEIVIN 1985, 112), it will be transcribed as [ɛ]. It is worth noting that, although unattested as
a regular noun, a construct form of Hebrew ‫' ָאשֵׁר‬Asher' (meaning: 'happy one') could poten-
tially result in a minimal pair contrasting with /ɛ/ in εσερ ‫ ֲאשֶׁר‬.

The evidence regarding the presence of the potential phonemes /ɛ̄/ and /ε/ in the Secunda is
quite scant. Positing these phonemes (or phones) in the Secunda largely depends on one's un-
derstanding of the development of the Hebrew vowels. What can be said is that the orthogra-
phy in µοσαυε, µασε, ιερε, βεχι, ελλελθ, δερχ, and ιεθερ is entirely consistent with the possi-
bility of the phonemes /ɛ̄/ and /ɛ/ (or phones [ɛː] and [ɛ]) existing in the Secunda. While /ɛ/ is
more ambiguous, a strong case can be made that long ɛ̄, either as its own phoneme or as an
allophone of /ē /, was present in forms like µασε and ιερε.
For the sake of phonetic transcription convention, we will posit the following: First,
because the final vowel was originally long in forms like nominal µοσαυε/µασε, verbal ιερε,
and demonstrative ζε, we will transcribe ε in such forms as /ɛ̄/ [εː]. Second, because
SUCHARD's reconstruction of the development of pausal forms is convincing, ε in pausal βεχι
will also be transcribed as long /ɛ̄/ [ɛː]: /bɛ̄́kı̄ / [ˈbɛːχiː].371 Third, because of the parallel exam-
ple λαβανι, in which "shewa" is represented with unetymological α, we will assume that ε in
non-pausal βεχι reflects a centralized schwa vowel: /bkı̄ /́ [bəˈχiː]. Fourth, because there is no
371. Note also that ‫' בכי‬weeping' and ‫' חצי‬half' develop differently in pause in both Tiberian and Babylonian:
‫ ַבּ ֶבּכִי‬, yIkb'b' with seghol/pataḥ but ‫ ָו ֵחצִי‬, yIxjew: with s ̣ere (YEIVIN 1985, 875). This may indicate that when (minor)
pausal lengthening occurred, there was a distinction in the vowel between ‫ בכי‬and ‫( חצי‬i.e., [ɛ] vs. [e]). We may
also contrast pausal ‫ ֶבּכִי‬with the imperative pausal form ‫' ֵלכִי‬go! (fs)', though the comparison with ‫ ֵלכִי‬may be
irrelevant since imperative forms tend to have different phonotactics from nouns.
- 300 -
relevant evidence by which to determine if the second ε in forms like ελλελθ reflects a lower
vowel quality than short /e/, these forms will be transcribed with short /e/ [ɪ] without any par-
ticular change: e.g., ελλελθ /h ̣ellelt/ [ħɪlːɪlth]. Fifth, because segholization is not general in the
Secunda, seems to be a phonetic phenomenon, and is prone to vowel harmony, the ε in forms
like ιεθερ and βεσε will be transcribed as [ɪ]: e.g., ιεθερ /yetr/ [jɪθɪʀ]. Sixth, and finally, εσερ
will be transcribed as /ʔš ɛr/ [ʔæʃɛʀ] for the reasons outlined above.
6.4.5. Etymological Long /ı̄ / = Greek ι, ει, and η
In the Secunda, etymological long /ı̄ / [iː] is usually transcribed by Greek ι (~240x) (6.4.2),
which represents a close front vowel [i] in Roman Palestinian Koine (4.5.3.1.1):
δερχι /derkı̄ / [dɪʀkhiː] 'my way' Ps. 18:33

Less frequently (25x), long /ı̄ / in a stressed syllable (YUDITSKY 2017, 60–61) is transcribed by
Greek ει, which also represents a close front vowel [i] in Roman Palestinian Koine
(4.5.3.1.1). It occurs in 11 distinct words, usually, but not always, following a guttural:
θαειρ /tɔ̄ʔı̄ r/ [tɔːʔiːʀ] 'you illuminate' Ps. 18:29

ϊεσει* /yesʕı̄ / [jɪʃʕiː] 'my salvation' Ps. 18:47
οζει /ʕozzı̄ / [ʕʊzːiː] 'my strength' Ps. 28:7 (+1x)
σελει /selʕı̄ / [sɪlʕiː] 'my rock' Ps. 31:4
βεειρ /b-ʕı̄ r/ [bɪʕiːʀ] 'in a city of' Ps. 31:22
βεσαυει /b-š aw(w)ʕı̄ / [bɪʃawʕiː] 'when I cried' Ps. 31:23
ραειθ* /rɔ̄ʔı̄ t/ [ʀɔːʔiːθ] 'you saw' Ps. 35:22
ελωειµ /ʔlō hı̄ m/ [ʔɪloːhıː̃ (m)] 'God' Ps. 36:2 (+6x)
εις /ʔı̄ š / [ʔiːʃ] 'man' Ps. 49:3 (+2x)
αββωτεειµ ̣ ̣ı̄ m/
/hab-bō t(ə)h [habːoːtˀɪħıː̃ (m)] 'those who trust' Ps. 49:7
λδαυειδ /l-dɔ̄wı̄ d/ [l(a)ðɔːwiːð] 'to David' Ps. 89:36 (+5x)
Finally, there are also four instances in which final /ı̄ #/ is transcribed by η, which represents a
close-mid front vowel [e] in Roman Palestinian Koine (4.5.3.1.6–7). Two of the instances are
after a nasal and two are after a pharyngeal (see YUDITSKY 2017, 84–85):
ουµαγεννη /w-mɔ̄gennı̄ / [(ʔ)umɔːʁɪnːeː] 'and my shield' Ps. 28:7
- 301 -
δελλιθανη /dellı̄ tɔ̄nı̄ /? [dɪlːiːθɔːniː] 'you drew me up' Ps. 30:2
ρουη /rū h ̣ı̄ / [ʀuːħiː] 'my spirit' Ps. 31:6
ουβσαλη /w-b-s ̣alʕı̄ / [(ʔ)uβʦˀalʕeː] 'and in my stumbling' Ps. 35:15
The question regarding the various representations of /ı̄ / in these forms is whether or not the
transcriptions with ει and/or η represent a different phonetic reality than the transcriptions
with ι. After a review of the Greek evidence and a review of scholarship, I will argue that the
data must be explained in subsets (see 6.4.5.3), with some subsets having phonological expla-
nations and others having orthographic explanations.

6.4.5.1. Greek Phonological, Orthographic, and Transcriptional Background
In Palestinian Koine of the Roman period, Greek ι represented the close front vowel [i]
(4.5.3.1.1). In transcription, Latin and Semitic long /ı̄ / is most commonly transcribed by
Greek ι. At the same time, Greek ι is also frequently used to transcribe short /i/ (5.3.1.1.3;
5.3.2.1.3; 5.3.3.1.2; 5.3.4.1; 5.3.5.1; 5.3.6). In Greek loanwords in the Mishnah, ι is usually
rendered with h ̣iriq in Hebrew (5.4.1.1.4).372 Therefore, the use of the grapheme ι (= [i]) to
represent Hebrew long /ı̄ / indicates a close front realization [iː] (see 6.4.2).
Greek ει also represented the close front vowel [i] in all phonetic environments in
Palestinian Koine (4.5.3.1.1). The fact that ει represented [i] and not [e] during the Roman
period is supported by the relative frequency of various spelling interchanges. While there are
only four total instances of interchanges of ει with either η or ε datable to the Roman period
(4.5.3.1.6),373 interchanges of ει and ι are found in abundance, with ει usually substituting for
ι (4.5.3.1.1). Moreover, it is more common for η to interchange with ε than with ει during the
Roman period (4.5.3.1.7). Therefore, it is clear that ει was identified with [i] and not [e].374
Only in the Byzantine period, when η (= [e]) shifts to [i], is ει identified with η (4.5.3.1.5–6).
372. There is reduction or centralization in closed unstressed syllables (5.4.1.1.4).
373. These interchanges occur in environments prone to vowel raising.
374. One could make the case that interchanges such as Γειωργιου (for Γεωργιου) support a lower realization of
ει before a vowel (variation 20.3–4). However, it is more likely that ε had a raised allophone of [i] in this
environment (see 6.3.7.2).
- 302 -
In Koine Greek of Asia Minor, ει was particularly associated with historically long vowels
(BRIXHE 2010, 232). In transcription, Greek ει is usually associated with length, most fre-
quently representing long /ı̄ / in both Latin and Semitic. Overall, however, Greek ι for long /ı̄ /
occurs more frequently. On occasion, ει may also be used to represent short /i/ (5.3.1.1.3;
5.3.2.1.3; 5.3.3.1.2; 5.3.4.1; 5.3.5.1; 5.3.6). Only in Akkadian is long /ē / transcribed by ει
(5.3.2.1.2). In Greek loanwords in the Mishnah, ει is usually rendered with h ̣iriq in Hebrew,
though Yemenite manuscripts attest one example of s ̣ere for ει (5.4.1.1.4).
Greek η, on the other hand, represented the close-mid front vowel [e] in Palestinian
Koine. The chronological distribution of the η/ε, η/ι, and η/ει interchanges indicate that η
maintained its close-mid [e] realization during the Roman period and did not shift to a close
[i] until the Byzantine period (4.5.3.1.6–7). In transcription, Greek η most frequently repre-
sents a long /ē / vowel, though it may also represent a short vowel of similar quality. In Greek
loanwords in the Mishnah, Greek η is usually rendered by a s ̣ere.

With respect to η for /ı̄ /, SPEISER (1925, 354) and BRØNNO (1943, 253–54) tend to assume
scribal error. JANSSENS, on the other hand, argues that the 1cs suffix /ı̄ / may be realized as [eː]
after a guttural or sonorant (1982, 125). YUDITSKY likewise affirms that η represents lowering,
explaining ρουη and ουβσαλη as the result of lowering after a guttural and ουµαγεννη and
δελλιθανη as the result of a general rule of final /ı̄ #/ lowering to [eː] similar to the phenome-
non in XHev/S13 in which final /ı̄ / is written with a ‫ה‬- (e.g., ‫ עינגדה‬for ‫( )עֵין גְּדִ י‬84–85, 2017).
Contemporary Greek evidence also supports interpreting η as indicating a lower [eː],
but two points are still in order. First, the lowering in ρουη and ουβσαλη is probably due to
adjacent pharyngeals (/h ̣/, /ʕ/) specifically, as is common in Semitic (MCCARTHY 1994, 207–
13), and not "gutturals" more generally. Second, while lowering in ουµαγεννη and δελλιθανη
may reflect a general /ı̄ #/ > [eː] rule, it is worth noting that it is common cross-linguistically
for nasalization to result in the "centralization" of vowel height. That is, high vowels are low-
- 303 -
ered and low vowels are raised (GAHL 2015, 99). It is also possible that the use of η in these
transcriptions is a Greek orthographic variant for /ı̄ /. It is not uncommon for ι to interchange
with η in the environment of a nasal in Greek (4.5.3.1.5). Even in transcription, we find
examples such as Σηµων and Βενιαµην (variation 26.1–2) already in the first century in
Jerusalem. Nevertheless, because of the consistent correspondence of η and vowels of the [e]
quality in the Secunda, it is likely that η reflects a lowered allophone [eː] in these words.
The variant forms with ει are more difficult to explain. BRØNNO suggests several ex-
planations for the digraph ει in these forms: ε is a helping vowel in the environment of the
guttural, ε represents the guttural, or Hebrew /ı̄ / was realized as [ei] after a guttural (1943,
274–77). YUDITSKY claims that ει reflects the lowering of /ı̄ / due to the influence of the guttur-
al (or a general /ı̄ #/ > [eː] rule, see above). In support of his theory, he cites those forms in the
Secunda in which /ı̄ / is represented by η after a guttural (see above) (2017, 46, 60–61, 84–85,
103). However, it is not generally "gutturals" that effect lowering in Semitic but specifically
pharyngeals (BUTCHER and AHMAD 1987; MCCARTHY 1994, 208–13; AL-ANI 2006; BROSELOW
2006), and there are multiple instances in which ει occurs after non-pharyngeals (e.g., εις,
ελωειµ). YUDITSKY's suggestion that ει reflects the lowering of /ı̄ / > [eː] after a guttural is also
problematic because not all of the examples occur in such an environment (e.g., λδαυειδ).
For further support, YUDITSKY also appeals to ALLEN (1974, 69) to suggest that Greek
ει did not totally merge with ι until the second century CE (2017, 46, 60–61, 84–85, 103).
What ALLEN actually says, however, is that even though the interchange of ι and ει is common
already in the third century BCE, ει seems to be pronounced with a mid quality up until the
second century CE only when preceding a vowel, as shown by Latin renderings of Greek
words (1974, 66–69). However, ALLEN is addressing Greek in general. In Palestinian Koine
during the Roman period, the interchange of ει and ι before a vowel is far more frequent (e.g.,
variation 1.50, 66, 71, 83, 89–96, 142, 150, 151, 169; 2.2, 11, 13, 22, 27, 36–28, 40, 46, 78)
than the interchange of ει and η before a vowel, which only occurs twice (variation 24.1, 10).
- 304 -
This indicates that even before a vowel, ει was identified with ι and not η. Moreover, in no
transcription above does the digraph ει occur before another vocalic grapheme, thus negating
the relevance of such an allophone even if it had existed in Roman Palestinian Koine. There-
fore, unless the Secunda reflects a Greek pronunciation from centuries before, we can be fair-
ly confident that ει, as a digraph, represented the vocalic quality [i] (see 6.4.5.1). It is always
possible, however, for ει to be read as ε [ε] + ι [i] (= εϊ), rather than as ει [i].
6.4.5.3. Orthographic and Phonological Explanations
The data actually requires division into further subsets, with each subset having its respective
explanation. Those transcriptions with η for /ı̄ / were already dealt with in the review of schol-
arship, and it was concluded that they indicate a lowered allophone of word-final /ı̄ #/. The re-
maining occurrences of ει for /ı̄ / amount to eleven distinct words, which may be divided into
the following categories: /ı̄ / after a non-guttural consonant (I), /ı̄ / after a guttural phonologi-
cally but after a vowel orthographically (II), /ı̄ / after a guttural phonologically but word-ini-
tial orthographically (III), and /ı̄ / after a guttural (/ʕ/) phonologically but after a consonant or-
thographically (IV) (G = guttural consonant, C = non-guttural consonant):

Phonology Orthography Transcriptions
C_ (I) C_ οζει /ʕozzı̄ /, λδαυειδ /l-dɔ̄wı̄ d/
(II) V_ θαειρ /tɔ̄ʔı̄ r/, βεειρ /b-ʕı̄ r/, ραειθ* /rɔ̄ʔı̄ t/, ελωειµ /ʔlō hı̄ m/, αββωτεειµ /hab-bō tḥ ̣ı̄ m/
G_ (III) #_ εις /ʔı̄ š /
(IV) ʕ_ ϊεσει* /yeš ʕı̄ /, σελει /selʕı̄ /, βεσαυει /b-š aw(w)ʕı̄ /
Chart 22: Representation of Long /ı̄ / as ει in the Secunda
In the following sections, I will argue that in group (I), the digraph ει is merely an alternative
spelling for long /ı̄ / as in contemporary Greek, in groups (II) and (III), ει is as an orthographic
variant of ϊ (note trema ¨), and in group (IV), ει is to be read as ε + ι (i.e., εϊ), ε being a per-
ceptual approximation of the phonetic reality of the transition to /ʕ/ at the C-V boundary.
6.4.5.4. (I): Greek ει as Common Orthographic Variant for /ı̄ /
In group (I), ει for /ı̄ / is simply an orthographic variant of ει for ι. It only occurs in two words:
οζει and λδαυειδ. In the case of οζει, a phonetic argument could be made that the pharyngeal
- 305 -
/ʕ/ spreads across the whole word and causes the lowering of the suffix: [ʕʊˁzːˁeː]. In the case
of λδαυειδ, one could argue that /ı̄ / lowers to [eː] in assimilation to the semivowel /w/. How-
ever, both of these explanations are rather unlikely. It seems more reasonable to suggest that
ει simply represented a long vowel [iː] as it does often in Greek transcription. In fact, there
are numerous instances of long /ı̄ / being represented by ει in external sources: e.g., Epipha-
nius has σαλωειµ (for σαλωσιµ/σαλωσειµ*) ‫שִׁים‬Äְ‫( שׁ‬Gen. 5:5), εµµουνειµ ‫( אֱמוּנִים‬Isa. 26:2),
σααρειµ ‫שׁע ִָרים‬

ְ (Isa. 26:2), and ωλεµειµ ‫( עוֹ ָלמִים‬Isa. 26:4); the Syro-Hexapla indicates σειειµ/
σιειν ‫( ִציּ ִים‬Jer. 50:39);375 Chrysostom has σωσανειµ ‫שׁנִּים‬

ַ ‫( שׁוֹ‬Ps. 45:1), Θαρσεις ‫( תַּ ְרשִׁישׁ‬Ps.
48:8), ιδαθει ‫( חִידָ תִ י‬Ps. 49:5), and ισουββουνει ‫( י ְסוּ ֵבּנִי‬Ps. 49:6). The same phenomenon is also
attested in Palestinian epigraphic transcriptions of Hebrew: e.g., Σειλωνει ‫שִׁילוֹנִי‬. In sum, pho-
netic explanations for lowering are somewhat ad hoc when contemporary Greek orthography
routinely uses ει to signify [i] and there are numerous examples of ει for long /ı̄ / attested in
the Secunda in external sources.

6.4.5.5. (II) and (III): Greek ει as an Orthographic Variant of ϊ = [i] / V_
The use of ει for /ı̄ / in group (II), like group (I), also constitutes a spelling variant, albeit a
more sophisticated one. While the spelling variant described above constitutes ει for ι, the
spelling variant exhibited in group (II) should be described as ει for ϊ (note trema ¨). When a
guttural is followed by a long /ı̄ / and preceded by a non-/i/ vowel, its presence is indicated by
means of a hiatus between vowels. However, because the following long /ı̄ / is normally repre-
sented by ι, the sequence of vowel (α, ε, ω) + ι could be mistaken as a digraph. Greek αι
could be mistakenly read as [ε],376 ει as [i], and ωι as [o]. What the transcriber intended as two
distinct vowels could be mistaken for one vowel, perhaps even of an inappropriate quality.
Two orthographic conventions resolved this problem. First, one could write ι with trema (¨),
indicating that ϊ was to be pronounced separately from the preceding vocalic grapheme:
375. Note that the two consecutive digraphs ει + ει in σειειµ could reflect the following change: -iyyı̄ - > -ı̄ ʔı̄ -.
376. Note the dubious transcription βσαιµ ‫( ְבּשֵׁם‬Ps. 118:26) found in external sources.
- 306 -
ραϊθ /rɔ̄ʔı̄ t/ [ʀɔːʔiːθ] 'you saw' Ps. 31:8
αϊρα /hɔ̄ʕı̄ rɔ̄/ [hɔːʕiːʀɔː] 'awaken!' Ps. 35:23
ουβανγαϊµ /w-b-ngɔ̄ʕı̄ m/ [(ʔ)uβanʁɔːʕıː̃ (m)] 'and with smitings' Ps. 89:33
Alternatively, if one were not going to use trema, one could replace ϊ with ει. For
example, in the Greek texts from the Judaean Desert, we find τροπαιεικον for τροπαϊκον
(4.5.3.1.10). In Palestinian epigraphy, a similar convention for transcribing VC[+guttural]ı̄ is at-
ttested when the Hebrew proper name ‫ יאיר‬is rendered in Greek as Ιαειρος (CIIP I/1, no. 164,
400a–b, 401b). That αϊ and αει—more generally, vowel + ϊ and vowel + ει—were equivalent
is also indicated by the alternative representations of ָ‫ראִית‬:

ָ ραϊθ and ραειθ*.377 Moreover, the
word ‫הִים‬Äֱ‫א‬, usually written as ελωειµ, is found in a Secunda quotation in Eusebius as ἐλωῒµ.
Other examples of this phenomenon are also attested in external references to the Secunda:
e.g., ουαεει* ‫( ַויְחִי‬Gen. 5:5), αειρ ‫( ָהעִיר‬Gen. 28:9), αρβαειµ ‫אַר ָבּעִים‬
ְ (I Sam. 4:18) and ραφαειµ
‫( ְר ָפאִים‬Isa. 14:9). The distribution of the various transcriptions of the Hebrew noun ‫' עִיר‬city'
may be particularly instructive. It occurs twice in the Ambrosiana palimpsest and once in ex-
ternal sources. After a vowel, long /ı̄ / is transcribed with ει: βεειρ (Ps. 31:22) and αειρ (Gen.
28:19). When it occurs word-initially, /ı̄ / is transcribed by ι alone: ιρ (Ps. 46:5).
The pattern outlined for the word ιρ ‫ עִיר‬does not fit as nicely for group (III), which
only contains the word εις ‫אִישׁ‬. It occurs three times in the Secunda, always as εις. It should
be noted, though, that the previous word always ends in a vowel, and in two out of the three
instances, the preceding word is written on the same line. In external sources, the word ‫אִישׁ‬
'man' is attested twice, once by itself and once after the definite article. By itself, it is
transcribed as ις (Ps. 92:4). After the definite article, it is transcribed as ἀῒς with trema on ι
(Ps. 1:1). A variant in the Vatican MS has αεις (Ps. 1:1). Like ραϊθ/ραειθ*, the variant
spellings ἀῒς/αεις support the idea that αϊ was a transcription convention equivalent to αει.
377. Though representing the diphthong /ay/ instead of the sequence VC[+guttural]ı̄ , the same equivalency of αει =
αϊ is evidenced in the pair ακουββαει (Chrysostom) vs. ακοββαϊ (Ambrosiana) (Ps. 49:6).
- 307 -
Although the orthographic conventions for transcribing ‫ אִישׁ‬are not as neat as ιρ/αειρ,
they do highlight the importance of taking into account orthographic tendencies and transmis-
sion history as a factor in explaining variant spellings. It should also be kept in mind that
trema (¨) is easily dropped in transmission and there are likely places in the transmission of
the Secunda where it should be restored. The alternative explanation, namely, that /ʔ/ caused
the lowering of the high vowel /ı̄ / to [eː] in εις seems phonetically implausible and contrary to
contemporary Greek orthography. In sum, then, although the data is not perfectly conforma-
tive, it seems preferable to posit an orthographic explanation for groups (II) and (III) over a
phonetic explanation. That is, vowel + ει serves as an orthographic alternative to vowel + ϊ,
both of which represent /ı̄ / [iː] after a guttural in groups (II) and (III).
That this phenomenon was merely an orthographic convention and not reflective of a
phonetic reality is further supported by the fact that when word-medial long /ı̄ / follows a gut-
tural consonant phonologically but follows a non-guttural consonant transcriptionally, long /ı̄ /
is always represented with simple ι: θεριβ ‫( תַּ ְרחִיב‬Ps. 18:37), εριµ ‫( ה ְִרעִים‬Ps. 29:3), ουεϊεριβου
‫( ַויּ ְַרחִיבוּ‬Ps. 35:21). Were the digraph ει indicative of lowering following a pharyngeal, we
would expect it to be present in these transcriptions as well.

6.4.5.6. (IV): Greek ει as an Orthographic Variant of εϊ = [ɛi]
Group (IV), on the other hand, which contains the words ϊεσει*, σελει, and βεσαυει, has a dif-
ferent explanation, partially orthographic and partially phonetic. In each of these cases, ει fol-
lows a consonant in transcription. Assuming that trema could have been lost in transmission,
or never written but conventionally assumed, we can read -ει in these forms as ε + ι. Good
evidence for reading ει as ε + ι in βεσαυει is actually found in the Hexapla palimpsest of
Psalm 22 from the Cairo Genizah published in 1900. Because the second column is so frag-
mentary in this palimpsest, often containing only a few letters per line, it has generally been
ignored in treatments of Origen's Secunda. However, on the seventh line down of the second
folio, the final letters of the transcription of ‫שׁוְּעוֹ‬

ַ ‫' וּ ְב‬and in his crying' appear as ****]αυεω,
- 308 -
probably to be restored as [ουβσ]αυεω. Apart from the restoration, however, the fact that this
transcription concludes with εω indicates that the ε is either part of the syllable structure of
the word (i.e., /w-b-š awweʕō /) or an approximation of the guttural sound in this context.
This same method is used in a couple words in the LXX: e.g., Συµεων ‫שׁמְעוֹן‬
ִ , Γεδεων
‫גִּדְ עוֹן‬, and ɸαραω ‫פּ ְַרע ֹה‬. KNOBLOCH argues that the lack of direct representation of gutturals in
transcription can cause the reader to syllabify the word incorrectly. The addition of a vowel
helps ensure that the word is pronounced according to the original syllable structure. It is also
possible, especially in the case of pharyngeals like /ʕ/, that the additional vowel actually ap-
proximates the guttural sound in a language that did not have it (1995, 219–224).
In light of these principles, then, there are three ways to read ε in βεσαυει: first, it may
actually reflect the syllable structure (i.e., /b-š awweʕı̄ / [bɪʃawwɪʕiː]), second, it may act as a
placeholder to preserve the syllable structure though syncope had occurred (i.e., /b-š aw(w).ʕı̄ /
[bɪ.ʃaw.ˈʕiː]), or third, it may actually approximate the guttural sound perceptually (i.e., /b-
š aw(w)ʕı̄ / [bɪʃawʕiː]). The first interpretation is unlikely for two reasons. First, the ει ending
also occurs on the nouns ϊεσει* and σελει, in which ε is unlikely to be part of the syllable
structure, since segholates with a suffix have no need of a helping vowel.378 Second, syncope
occurs in the same syllable structure elsewhere: ϊεσαυου /yš aw(w)ʕū / [jɪʃawʕuː] (Ps. 18:42).
It seems preferable, then, to posit that the ε in these forms either served to preserve the sylla-
ble structure or to approximate the sound of the guttural. The latter of these two explanations
may be preferred for the very reason that the transition at the V-C boundary of /ʕ/, which hap-
pens to be the guttural present in all of these forms, also seems to be approximated by ε in at
least a couple other transcriptions of Origen (see discussion in 6.3.6.2).
In sum, then, in light of the transcription [ουβσ]αυεω, it seems best to posit that final
ει originally had a trema (εϊ), or at least was intended to be read as ε + ι, and that the ε before
378. Note also how the name ‫שׁעִי‬ ְ ִ ‫' י‬Ishi' is rendered in the LXX as Ισεϊ (I Chr. 4:20, 5:24) or Ιεσι (I Chr. 4:42).
But cf. the transcription ιεσερι ‫( יִצ ְִרי‬Num. 26:49) and the effect of the SCL outlined in 6.5.1.5.1.
- 309 -
the final ι is an approximation of the relatively centralized vocalic quality of the transition at
the V-C boundary of /ʕ/ and /ı̄ / (see 6.3.6.2). This is not without precedent in contemporary
transcription, since Latin ĕ ı̄ is transcribed as ει in a second-century CE papyrus (5.3.1.1.2).

6.4.6. Summary
The suggested typical realizations of the vocalic phonemes in the Secunda are charted below:
Front Back
close ı̄ [iː] ū [uː]
e [ɪ] o [ʊ]
mid-close ē [eː] ō [oː]
mid-open (ɛ̄ [ɛː], ε [ɛ]) ɔ̄ [ɔː]
open a [a]/[æ]
Chart 23: Phonetic Realization of the Vocalic Phonemes in the Secunda
Phoneme Phone Greek Grapheme Written Word Pronunciation

/ı̄ / [iː] ι = [i] σιρ [ʃiːʀ]
/ē / [eː] η = [e] νηχαρ [neːχɔːʀ]
/e/ (< */i/) [ɪ] ε = [ɛ] δερχ [dɪʀkh]
(/ɛ̄/) [ɛː] ε = [ɛ] µασε [maħsɛː]
(/ɛ/) [ɛ] ε = [ɛ] εσερ [ʔæʃɛʀ]
[a] α = [a]/[ɑ] σαµθ [samth]
/a/
([æ]) (ε = [ɛ]) (µεββεσε) ([mæbːɪʦˀɪʕ])
/ɔ̄/ [ɔː] α = [a]/[ɑ] ιαδω [jɔːðoː]
/o/ (< */u/) [ʊ] ο = [o̞ ] χολ [khʊl]
/ō / [oː] ω = [o̞ ] κωλ [kˀoːl]
/ū / [uː] ου = [u] σουρ [ʦˀuːʀ]
Chart 24: Vocalic Phonology, Phonetics, and Orthography in the Secunda
6.5. SHEWA AND SYLLABLE STRUCTURE
Unlike our interpretation of Tiberian Hebrew, for which medieval sources provide additional
information, our interpretation of syllable structure in the Secunda depends almost entirely on
the transcriptions themselves. As with any reading tradition of Biblical Hebrew, a description
of syllable structure in the Secunda depends to a large extent on the nature of shewa in the
tradition reflected therein. This section, which addresses both shewa and syllable structure in
- 310 -
the Secunda, is divided into two main parts. The first part (6.5.1) addresses the phonetic and
phonemic status of shewa and the nature of word-initial and word-medial consonant clusters.
The second part (6.5.2) addresses final consonant clusters mainly through the lens of segho-
late nouns. My analysis of both shewa and syllable structure in the Secunda follows, to a
large degree, KHAN's work on shewa and syllable structure in Tiberian and Babylonian (1987;
2013a, 98–107; 2013b; 2013h) and KIPARSKY's work on syllables and moras in Arabic (2003).
6.5.1. Shewa
In the Secunda, the parallel of Tiberian vocalic shewa is usually left unrepresented:
βδαµι /b-dɔ̄mı̄ / [b(a)ðɔːmiː] 'in my blood' Ps. 30:10

φλαγαυ /plɔ̄gaw/ h
[p (a)lɔːʁaw] 'its streams' Ps. 46:5
βνη /bnē / [b(a)neː] 'the sons of' Ps. 89:48
Less frequently, it is represented with Greek α:
νακαµωθ /n(a)qɔ̄mō t/ [nakˀɔːmoːθ] 'vengeances' Ps. 18:48

βαφιεµ /b-pı̄ hem(m)/ [baɸiːhɪ(̃ m)] 'with their mouth' Ps. 49:14
χαµω /k(a)mō / [khamoː] 'like' Ps. 89:47
Still less frequently, it is represented with Greek ε:

σεµω /š mō / [ʃəmoː] 'his name' Ps. 29:2
λεβουσι /lbū š ı̄ / [ləβuːʃiː] 'my clothing' Ps. 35:13
It may also assimilate to the vowel of a following guttural:
µηηρα /mhē rɔ̄/ [meheːʀɔː] 'speedily' Ps. 31:3

βεεζδαχ /b-h ̣esdɔ̄k/ [bɪħɪzdɔːχ] 'in your mercy' Ps. 31:8
µεεθθα /mh ̣ettɔ̄/ [mɪħɪthːɔː] 'a terror' Ps. 89:41
The issues regarding shewa in the Secunda range from the question of its very existence to its
phonetic realization and phonemic status. In this section, we will begin with a general review
of the concept of schwa in modern linguistics and shewa in Biblical Hebrew. Here I should
note that I follow the convention of the Encyclopedia of Hebrew Language and Linguistics
by using the term "schwa" to refer either to the vowel of neutral quality (represented in the
IPA by [ə]) or to the concept of schwa in modern linguistics and the term "shewa" to refer
- 311 -
specifically to the concept of shewa in the Biblical Hebrew reading traditions. Following our
review of the concept of schwa/shewa, we will continue with a review of scholarship specifi-
cally on shewa in the Secunda. Subsequently, we will deal with the phonetic realization of
vocalic shewa, the nature of complex onsets, the vowel syncope rule with its resulting clus-
ters, and the behavior of the conjunction waw and the inseparable prepositions in the
Secunda.
6.5.1.1. The Concept of Schwa/Shewa: Linguistics and the Hebrew Reading
Traditions379
Because Hebrew shewa is viewed through the lens of niqqud and often misunderstood, it is
necessary to define what exactly we mean when we refer to the existence of shewa in the Se-
cunda. In modern linguistics, the term schwa refers to either a vowel of "neutral" quality (IPA
[ə])380 or a vowel that interchanges with zero as a result of the historical processes of epenthe-
sis or deletion. Because it is often the "neutral" vowel [ə] that is deleted or epenthesized,
these two meanings tend to overlap (VAN OOSTENDORP 2013). It should also be noted that it has
become increasingly common for linguists to describe the phonetic quality of schwa as vari-
able, changing according to its immediate phonological context (see 6.5.1.6).
In the Hebrew reading traditions, vocalic shewa is similarly the product of deletion
and subsequent epenthesis. In its earliest stages, the Hebrew vowel system was made up of
short and long vocalic phonemes (/a/, /i/, /u/, /ā /, /ı̄ /, /ē /, /ō /, /ū /). At some point in the history
of the language, etymological */i/ and */u/ underwent reduction in a number of environments.
Eventually, all short vowels in open syllables at least two places from the stress underwent
reduction or deletion. As a result, consonant clusters at the onset of a syllable, at least at an
underlying phonological level, were created (e.g., *yiktubū > *yiktbū ; *dabā rı̄ m > *dbā rı̄ m).
379. In this section, I make a terminological distinction between schwa, which refers to the general concept of
schwa as it is discussed in modern linguistics, and shewa, which refers specifically to shewa in the Hebrew
reading traditions. There is a high degree of overlap between these two concepts, but they are not identical.
380. It should be noted, however, that [ə] is actually used for a variety of "non-peripheral" vowels that could
potentially be signified by other IPA symbols (VAN OOSTENDORP 2013).
- 312 -
The various Hebrew reading traditions deal with these clusters in different ways.
Tiberian tends to introduce an epenthetic short vowel after the first consonant to break up the
cluster, usually of the quality [a] (e.g., *yiktbū > [jiχtavuː]). This vowel, which was phoneti-
cally no shorter than a pataḥ in an unstressed closed syllable (e.g., ‫' תְּ דַ בֵּר‬you will speak' is re-
alized as [taðabːeːʀ]), is what is referred to as "vocalic shewa." Babylonian, on the other
hand, does not always have a vocalic segment where Tiberian has vocalic shewa (e.g., *yikt-
bū > [jiχtvuː]), thus exhibiting a different degree of tolerance for clusters. Regardless of its
phonetic realization ([a] or Ø), however, this "shewa slot" is equivalent to phonological zero.
This is the reason that both vocalic and quiescent shewa are indicated by the same sign (ְ ) in
Tiberian Hebrew (KHAN 2013h, 543–48; 2013b, 674; 2013f, 981).
With respect to the Secunda, then, we must examine trends of reduction, deletion, and
epenthesis in the transcriptions in order to understand the nature of vocalic shewa, namely, a
short vowel that functions as an allophone of zero. After a brief review of scholarship, these
and other issues will be addressed.

MARGOLIS, who was among the first to conduct research on the hexaplaric transcriptions,
made a number of observations and hypotheses about the behavior of shewa in the Secunda.
Before a guttural, the vocalic quality of the shewa often assimilates to the following vowel as
in the rules outlined by the medieval Hebrew grammarians. Elsewhere, shewa tends to be
represented by either ε or α. Nevertheless, it is more often left unrepresented in the transcrip-
tions. The inconsistency in the representation of shewa indicates that its phonetic quality was
unstable. If shewa was not represented in transcription, it was not pronounced (1909).
BRØNNO, who compares the Tiberian forms with those of the Secunda, finds that vocal
shewa is rendered by Ø 126 times, by α 43 times, by ε 33 times, and by ο 5 times. On the ba-
sis of pairs like βανη || βνη ‫ ְבּנֵי‬, he suggests that shewa may still have been pronounced even
when it was not transcribed. The phonetic value of vocal shewa was /e/. Etymological corre-
- 313 -
spondences such as ιεφφολου ‫* <( יִפְּלוּ‬yippulū ) are attributed to interference from other forms
(i.e., **ιεφφολ) (1943, 327, 329, 333). JANSSENS affirms the presence of shewa in the Secun-
da, transcribing it as a short /e/. Though often unrepresented, it was still pronounced (e.g.,
βκωλω /beqō lō /) (1982, 89–110). BLAU suggests that α/ε for shewa in the Secunda indicates a
phonetic quality of [æ] and affirms the phenomenon of assimilation before a guttural (1984).
YUDITSKY denies that there is any evidence in the Secunda for a so-called (vocalic)
"shewa" vowel, which he refers to as "a short vocalic entity whose quality is unclear."381
Every instance of apparent shewa is actually a short vowel corresponding in quality to that of
the etymological vowel. When the vowel quality differs from the historical vowel, it may be
explained on the basis of phonetically conditioned changes due to the immediate consonantal
environment. The reason vocalic "shewa" is left unrepresented in transcription so frequently
is because it was especially short, presumably even shorter than a short vowel. These points
may be illustrated with the preposition ‫' ב‬in' in the Secunda: the historical Hebrew vowel for
the preposition ‫ ב‬is /a/ (e.g., βαµεθγε ‫) ְבּמֶתֶ ג‬, but it may be raised in the environment of a sibi-
lant (e.g., βεσοχχα ‫ ) ְבּ ֻסכָּה‬and is usually omitted (e.g., βκωλω ‫( )בְּקוֹלוֹ‬2005).
Finally, the most accurate description of shewa in the Secunda, though brief, is that of
KHAN. He regards the representation of shewa with ε in the Secunda as indicative of a quality
resembling that of Palestinian shewa, and most examples of α for shewa as preserving a his-
torical */a/ vowel (KHAN 2013h, 550–51).
Though YUDITSKY has made a significant contribution for how we understand "shewa"
in the Secunda and its relationship to the diachronic development of Hebrew, two of his con-
clusions are in need of refinement. First, at least in Tiberian, it is not entirely accurate to re-
gard shewa as "a short vocalic entity whose quality is unclear." Phonetically, in terms of vow-
el quality, vocalic shewa is realized as [a] generally, [i] before yod, and as the quality of the
381. ‫( מהות תנועית קצרה שאיכותה אינה ברורה‬YUDITSKY 2005, 138).
- 314 -
following vowel when preceding a guttural.382 Phonologically, vocalic shewa is equivalent to
phonological zero (see above). Second, while there seems to be truth in the claim that the vo-
calic representation parallel to Tiberian shewa in the Secunda often coincides with the quality
of the etymological vowel, this does not apply in all instances. While etymological */a/ is of-
ten preserved at a distance from the stress and thus represented with α, it seems that many of
the instances of ε, which YUDITSKY interprets as representing etymological */i/, are actually
better explained as signifying a reduced schwa vowel. These points will be elaborated in the
following sections.
6.5.1.3. The Phonetic Realization of Vocalic Shewa
We noted earlier that in modern linguistics the term schwa can refer either to a vowel that in-
terchanges with zero (i.e., phonological zero) or to a "neutral" mid-central vowel (i.e., pho-
netic [ə]) (6.5.1.1). However, linguists are prone to suggest even finer nuances with regard to
the phonetic quality of schwa.
According to FLEMMING, in addition to a "neutral" mid-central vowel [ə], there is also
a variable schwa, whose quality changes according to context. While both types of schwa can
be the result of reduction, mid-central schwa ([ə]) constitutes a "moderate reduction" whereas
variable schwa constitutes an "extreme reduction" and strong assimilation to the immediate
phonological context (FLEMMING 2007, 2).
Another approach is outlined by VAN BERGEM, who makes a distinction between the
phonetic description of vowel reduction, which he terms "acoustic reduction," and the phono-
logical description of vowel reduction, which he terms "lexical reduction." Acoustic reduc-
tion occurs as a result of speakers relaxing the articulators when pronouncing vowels in "less
informative parts of an utterance"; as a result, vowels are shortened and (usually) pronounced
382. YUDITSKY himself acknowledges these rules at the beginning of his article (2005, 121n4).
- 315 -
closer to the acoustic center [ə]. Lexical reduction occurs when a full vowel is phonologically
replaced by a schwa vowel (or another vowel easier to pronounce) (1991).
In this section, I will argue that some of the confusion in previous scholarship regard-
ing the phonetic value of shewa in the Secunda can actually be cleared up by making a dis-
tinction between mid-central and variable schwa, on one hand, and between acoustic reduc-
tion and lexical reduction, on the other.383 I will demonstrate that (1) a reduced centralized
vowel (i.e., vocalic shewa) was generally realized as [ə] or [ε] in the Secunda, (2) assimilato-
ry tendencies in vocalic shewa also point towards a variable realization in certain contexts,
and (3) the preservation of historical /a/ in "shewa-vowel" slots is best understood as a
"snapshot" during the transition process from mere acoustic reduction to lexical reduction.
6.5.1.3.1. Vowel Reduction and Centralization in the Secunda: Greek ε for [ə] or [ε]
When the parallel of Tiberian vocalic shewa is represented in transcription, if not a preserva-
tion of a historical /a/ (see below 6.5.1.3.3), it is usually represented by ε in the Secunda:
γεδουδ /gdū d/ [gəðuːð] 'a troop' Ps. 18:30

ζερουωθαϊ /zrō ʕō tay(y)/ [zəʀuːʕoːθaj] 'my arms' Ps. 18:35
σεµαχ /š mɔ̄k/ [ʃəmɔːχ] 'your name' Ps. 31:4
In Palestinian Koine Greek, centralized reduced vowels may be represented by ε: e.g.,
Πτελεµ[αικ]ου (for Πτολεµαϊκοῦ), ηποµενοντες (for ὑποµένοντος), and possibly
δεσδεκαλλου/δεσκαλου (for διδασκάλου) (variations 88, 90). At the same time, there may be
instances in which α is used to reflect centralization: e.g., αδαλφου (for ἀδελφοῦ) and
ενανηκοντα (for ἐνενήκοντα) (4.5.3.1.12). While the same is true of Egyptian Koine regard-
ing α and ε, the fact that centralized Greek ο tends to interchange with Greek ε may indicate
that ε was ultimately the most favorable grapheme for transcribing a centralized reduced
vowel (GIGNAC 1976, 278–93).
383. It should be noted that FLEMMING's mid-central schwa and the result of VAN BERGEM's "acoustic reduction"
are more or less the same entity. However, acoustic reduction does often but not always result in a vowel near
the value of [ə] (VAN BERGEM 1991, 3).
- 316 -
In Greek transcription of Phoenician-Punic, a reduced /a/ may be represented by
Greek ε, as can be seen from the transcription ɸενη Βαλ as opposed to the regular ɸανε Βαλ/
ɸανηβαλος /panē baʕl/ 'face of Baal'. In Greek transcription of Aramaic, vocalic shewa is
usually transcribed by α, though the only attestations are in the inseparable prepositions, the
relative ‫ד‬, and before a guttural: βανισαν ‫' בניסן‬in Nisan', λαµαν ‫' למן‬to whom', δαελαα ‫דאלהא‬
'that God', and δααβ ‫' דהב‬gold'.
In Greek transcriptions of Biblical Hebrew from Nikolaos of Otranto in the 12th/13th
centuries CE, vocalic shewa is usually transcribed by Greek ε: βερεσιθ ‫( בּ ְֵראשִׁית‬Gen. 1:1),
βεεθ ‫( ְואֶת־‬Gen. 1:1), νεχικοθ ‫( נְשִׁיקוֹת‬Song 1:2) (Disputatio contra Judaeos, 5.11, 245.18).384
Although it is far removed from the Secunda both geographically and chronologically, its
data regarding shewa are applicable here, provided they are understood within their linguistic
context. At the time of Nikolaos, Greek ε represented a true-mid front vowel [e̞ ] (PETROUNIAS
2007c, 604–605; HORROCKS 2014, 167), hardly different from the realization of ε during the
time of the Secunda. However, Greek η had since become a high front [i], merging with ι and
ει. Medieval Hebrew manuscripts from Italy indicate that vocalic shewa was usually pro-
nounced as /e/ (= s ̣ere/seghol) in the vocalization tradition there (RYZHIK 2008, 61–64; 2013,
365), which Greek ε [e̞ ] adequately approximated. If vocalic shewa in the Secunda was pho-
netically similar to the Sephardic shewa, then, it would probably be represented with ε.
In sum, the use of Greek ε to represent vocalic shewa in the Secunda indicates that
vocalic shewa was realized either as a centralized schwa vowel [ə] or a front vowel more in
the region of [e̞ ] or [ε]. It is unlikely that it reflects [e], for which the grapheme η would have
been utilized (see the discussion regarding βηηκι in 6.4.1.3). Such a realization of shewa
aligns Secunda Hebrew more with Palestinian than Tiberian (see KHAN 2013h, 549–51). It is
also worth noting that centralized vowels in the region of [ə] or [ɪ] tend to move to peripheral
384. Note also the curious transcription of Greek χ for Hebrew /š / in νεχικοθ ‫נְשִׁיקוֹת‬, which reflects the medieval
Greek pronunciation of χ as [ç] / _i,e (BRIXHE 2010, 235; for χ = /š / in Arabic, see AL-JALLAD [forthcoming, 50]).
- 317 -
qualities, especially [e̞ ] or [e] when they undergo lengthening.385 This may help explain the
diachronic relationship between Secunda shewa and Palestinian shewa (see 6.5.1.3.4).
The description of shewa here is not exhaustive of all reduced vowels in the Secunda.
Rather, it corresponds to what FLEMMING calls "mid-central schwa" ([ə]), arising due to what
VAN BERGEM calls "acoustic reduction." At the same time, if the reduced vowel represented by
ε actually constituted its own phoneme and not merely a reduction of other historical vow-
els—this would be the case if the frequent, but not unfailing, preservation of historical /a/ in
the same contexts was regarded as phonemic—then by contrasting with /a/ in these environ-
ments it would also constitute an example of what VAN BERGEM regards as "lexical reduction."
6.5.1.3.2. Variable Schwa in the Secunda: Assimilatory Tendencies of Vocalic Shewa
In the Secunda, there are a number of instances in which a vowel that may interchange with
zero exhibits assimilation to its immediate context. These are best attributed to the category
of variable schwa. Assimilatory tendencies of shewa are found (assimilation bolded):
1) before gutturals (6.4.1.3) (see also YUDITSKY 2017, 86–89):
µηηρα /mhē rɔ̄/ [meheːʀɔː] 'speedily' Ps. 31:3

λοοµ /lh ̣am/ or /lh ̣om/ [lʊħʊ̃(m)] 'do battle!' Ps. 35:1
µεεθθα /mh ̣ettɔ̄/ [mɪħɪthːɔː] 'a terror' Ps. 89:41
That the first vowel in a pattern like µεεθθα (< *CaCiCCā ) may interchange with zero is
demonstrated by the fact that such a vowel is not transcribed in the form θελαθαχ ü ֶ‫תְּ ִהלָּת‬
(< *CaCiCCā + suffix) (Ps. 35:28). The form θελαθαχ is elucidated, in terms of both the un-
represented shewa and the lack of gemination, by alternations in Greek transcription of the
Hebrew name ‫ תְּ ִחנָּה‬in Palestinian epigraphy, in which the initial shewa vowel alternates with
zero: e.g., Θενας, Θεννας, and Θεεννας (CIIP I/1, no. 22, 323, 427).
2) when originally silent shewa becomes vocalic:
εµαραθ /ʔemrat/ [ʔɪmaʀæθ] 'the word of' Ps. 18:31
385. Note also the discussion of etymological */i/ [ɪ] lengthening to [eː] in the history of Hebrew (see 6.4.2.1).
In the NENA dialects, /ə/ also tends to shift to /e/ when lengthened (e.g., see KHAN 2008b, 66, 77).
- 318 -
εφικιδ /ʔepqı̄ d/ [ʔɪɸikˀiːð] 'I entrust' Ps. 31:6
Alternation with zero in these patterns is demonstrated by parallels such as αρφαθ ‫ח ְֶרפַּת‬
(< *qVtlat) (Ps. 89:51) and εσχιλεχ üְ‫שׂכִּיל‬

ְ ‫( ַא‬Ps. 32:8).
3) when, synchronically, a vocalic shewa is inserted to block syncope due to the Syl-
lable Contact Law (SCL) (for an explanation of the SCL, see 5.3.2.1.5; 6.5.1.5.1):
ουαναυαθαχ /w-ʕanwɔ̄tɔ̄k/ [(ʔ)uʕanawɔːθɔːχ] 'and your humility' Ps. 18:36

Alternation with zero in these patterns is shown by the parallel βσεδκαθαχ ü ְ‫( ְבּצִדְ קָת‬Ps. 31:2).
4) when, synchronically, a vocalic shewa is inserted to block syncope due to the OCP:
αµιµιµ /ʕam(ə)mı̄ m/ [ʕamimıː(m)]

̃ 'peoples' Ps. 18:48
ερωµεµεχ /ʔrō m(ə)mek(k)/ [ʔɪʀoːmɪmɪkh] 'I will exalt you' Ps. 30:2
λααραρι /l-har(ə)rı̄ / [lahaʀaʀiː] 'for my mountain' Ps. 30:8
The form αµιµιµ is either from original *qalalı̄ m having undergone syncope (*qalalı̄ m >
*qal(ə)lı̄ m) or from influence of the Aramaic form ‫ ַע ְממִין‬. Alternation with zero may be shown
by the near-parallels αλµωθ and αρβωθ, for which the same development is posited (see
6.5.1.4.1), though it should be noted that αλµωθ and αρβωθ are not from geminate roots.
The only exceptions to (4), in which assimilation of shewa does not seem to occur, are
the forms θσωβαβηνι ‫( תְּ סוֹ ְב ֵבנִי‬Ps. 32:7) and ισωβαβεννου ‫( י ְסוֹ ְבבֶנּוּ‬Ps. 32:10), in which shewa
is realized as a non-historical /a/ vowel (see KHAN 2013h, 550; YUDITSKY 2017, 156). One
possible explanation of how variable schwa might be realized with a lower pronunciation in
such an environment lies in what is called the "trough effect" in modern phonetics. When a
labial stop is adjacent to high vowels, the tongue body actually lowers during the production
of the labials; thus, schwa might assimilate to the lower position of the tongue body in the ar-
ticulation of the labials rather than the high vowels. The degree to which the first formant
- 319 -
(F1) of schwa is higher (i.e., the vowel is lower) than the surrounding high vowels also in-
creases when the schwa is between two labials instead of just one (FLEMMING 2007, 14–15).386
5) in the theme vowel of qal prefix verbal forms that have undergone reduction:
ιασαβου /yah ̣š bū / [jaħʃaβuː] 'they think' Ps. 35:20

Alternation with zero in these patterns is shown by the parallel ουϊφρου ‫( ְוי ַ ְחפְּרוּ‬Ps. 35:26) and
ιεµρου ‫( י ֶ ְחמְרוּ‬Ps. 46:4)—but note that ουϊφρου is III-/r/ and ιεµρου is II-/m/ and III-/r/.387
However, because two consonants intervene between the vowel to which shewa is supposed
to assimilate, it is also possible that ε and α here simply reflect reduction and centralization.
Finally, it should be noted that all the instances in the Secunda in which a shewa vow-
el changes quality in assimilation to an adjacent consonant could also potentially reflect an
assimilatory tendency of variable schwa (for the various effect of consonants on vowels, see
section 6.3 and YUDITSKY [2017, 86–98]).
From external sources, in this case Cod. 86 of the LXX, we may also add the
transcription λαβανι ‫ ִל ְבנִי‬, in which shewa is realized as a non-historical /a/. Though the lack
of syncope is a separate issue—we would expect **λβνι, **λαβνι, or **λεβνι—it is possible
that shewa assimilated to the previous /a/ vowel of the preposition /l-/.
Postulating the existence of a variable schwa in the Secunda is also supported by oth-
er contemporary Hebrew evidence. Assimilation of shewa to a following vowel, even across
non-guttural consonants, is attested in the LXX (e.g., Σοδοµα ‫ סְדוֹם‬and Γοδολιας ‫)גְּדַ ְלי ָהוּ‬, the
Dead Sea Scrolls (e.g., ‫ סְדוֹם || סודום‬and ‫ֲמוֹרה || עומרה‬

ָ ‫)ע‬, and Mishnaic Hebrew (e.g., || ‫ְבּסִיד‬
386. But cf. methnosasoth ‫( מִתְ נוֹסְסוֹת‬Zech. 9:16) in Jerome, in which case this explanation would not work.
387. See 6.5.1.3.2. YUDITSKY argues that these forms should be vocalized with an epenthetic between the first
and second radicals: i.e., ουϊφρου = wyiḥiprū and ιεµρου = yiḥemrū /yiḥemrū (2017, 47, 121–22). However, there
are at least two arguments against YUDITSKY's interpretation. First, the prothetic epenthesis to which he appeals,
characteristic of the Babylonian tradition, does not occur for the root ‫ חפ"ר‬in Babylonian Hebrew (YEIVIN 1985,
458). Second, the only other instance of a I-/h ̣/ and II-sonorant verb in the Secunda does not exhibit prothetic
epenthesis: ‫ ְויַח ְְרגוּ‬ουϊερογου (Ps. 18:46). It seems more conservative with the data, then, to assume that ουϊφρου
and ιεµρου do not exhibit any irregular syllabification patterns.
- 320 -
‫( )ביסיד‬KHAN 2013h, 550–51). Aquila and Theodotion, roughly contemporaries of Origen,
transcribe ‫( ְכּסִילֵיהֶם‬Isa. 13:10) as χισιλεεµ. Since ‫ ְכּסִיל‬is an Aramaic loan and thus inherited
with an initial shewa vowel—it was originally *qatı̄ l—this form is especially instructive; the
unusual quality ι, as opposed to more regular ε, can only be explained by assimilation. Final-
ly, Jerome's transcription nifilim ‫* <( נְ ִפלִים‬napı̄ lı̄ m) (Gen. 6:4), like χισιλεεµ, can only be ex-
plained by assimilation of the shewa across the non-guttural consonant to the following /ı̄ /.
It is also worth noting that although the medieval Hebrew grammarians discuss the as-
similation of shewa only before gutturals and yod, medieval Judaeo-Arabic texts with Tiber-
ian vocalization reflect the assimilation of shewa to a following vowel also across non-guttur-
al consonants (at least in the Arabic fuʕū l pattern): e.g., Classical Arabic ‫ اﻟــﻤــﻠــﻮك‬al-mulū k 'the
kings' is rendered in Hebrew script as ‫ אלמְלוּך‬and ‫ اﻟـــﺸـــﺮور‬al-š urū r 'the evils' as ‫אלשְרוּר‬. Note,
however, that Classical Arabic short /u/ is represented with qibbus ̣ in other words: e.g., ‫ﻣـــﺤـــﺐ‬
muḥibb 'loving' is rendered as ‫ ֻמחִב‬. Spellings such as ‫ אלמְלוּך‬and ‫ אלשְרוּר‬were likely formed by
analogy with the pronunciation of shewa in forms like ‫[ בְּעוּלָה‬buʕuːlɔː] (Isa. 54:1) and ‫כְּהוּט‬
[kuhuːtˁ] (Song 4:3), in which assimilation does occur due to the guttural (KHAN 1992, 110).
6.5.1.3.3. Acoustic and Lexical Reduction in the Secunda: Historical Vowels and Shewa
Several scholars (e.g., YUDITSKY 2005; KHAN 2013h, 550) have pointed out that there is a ten-
dency for a historical /a/ vowel that is parallel to vocalic shewa in Tiberian to be preserved
(transcribed as α) in the Secunda. Preservation of historical /a/ is most commonly found:388
1) in the inseparable prepositions (see YUDITSKY 2017, 224–29):
λαχολ /l-kol(l)/ [laχʊl] 'to all' Ps. 18:31

βακααλ /b-qɔ̄hɔ̄l/ [baqɔːhɔːl] 'in a congregation' Ps. 35:18
χασων /k-s ̣ō (ʔ)n/ [khaʦˀoːn] 'like flocks' Ps. 49:15
Alternation with zero is shown by the fact that /b-/, /k-/, and /l-/ are often transcribed without
a vowel: e.g., λµαωλ ‫( ְלמָחוֹל‬Ps. 30:12), βκωλω ‫( בְּקוֹלוֹ‬Ps. 46:7), and χσεδκαχ* üְ‫( ְכצִדְ ק‬Ps.
388. The transcription ιουχαλου ‫( יוּכְלוּ‬Ps. 18:39) was not included in any category since it has no parallels.
- 321 -
35:24). Acoustic reduction may occur in the form λεριβι* ‫( ל ְִריבִי‬Ps. 35:23), and, in external
sources, βεγεβουροθαυ ‫( ִבגְבוּר ֹתָ יו‬Ps. 150:2) and λεβνη ‫( ִל ְבנֵי‬Ps. 11:9).389
2) in the qatā lı̄ m/qatā lō t (and more generally, qatv̄ lı̄ m/qatv̄ lō t) plural pattern:390
νακαµωθ /n(a)qɔ̄mō t/ [nakˀɔːmoːθ] 'vengeances' Ps. 18:48

σαβαωθ /s ̣(a)bɔ̄ʔō t/ [ʦˀaβɔːʔoːθ] 'hosts' Ps. 46:8, 12
λσαχηναυ /l-š (a)kē naw/ [l(ɪ)ʃaχeːnaw] 'to his neighbors' Ps. 89:42
The attestations of qatā lı̄ m/qatā lō t in external sources also preserve the initial /a/ vowel: e.g.,
ραφαειµ ‫( ְר ָפאִים‬Isa. 14:9) and σααρειµ ‫שׁע ִָרים‬

ְ (Isa. 26:2). The preservation of /a/ in these pat-
terns is also well-attested in other contemporary Hebrew transcriptional evidence: e.g., the
LXX has Ναβαιωθ ‫( נְבָיוֹת‬Isa. 60:7), Καδηµωθ ‫( קְדֵ מוֹת‬I Chr. 6:64), and ναθινιµ ‫( נְתִ ינִים‬Ezra
2:70); Theodotion has σαβαχωθ ‫שׂבָכוֹת‬

ְ (II Chr. 4:12) and καδησιµ ‫( קְדֵ שִׁים‬Judg. 5:21); Jerome
has cadesoth ‫( קְדֵ שׁוֹת‬Hos. 4:14), sababim ‫שׁ ָבבִים‬

ְ (Hos. 8:6), and saridim ‫( שׂ ְִרידִ ים‬Joel 3:5).
Although the /a/ vowel is usually preserved in transcription, there is evidence of both
reduction and alternation with zero. Reduction is demonstrated in the following transcription:
αδδεβαρειµ /had-dbɔ̄rı̄ m/ [hadːəβɔːʀıː̃ (m)] 'the words' Deut. 1:1
Greek ε likely reflects centralization and thus an example of acoustic reduction. Alternation
with zero in qatā lı̄ m/qatā lō t is demonstrated by parallels such as αρσαειµ ‫שׁעִים‬
ָ ‫( ְר‬Ps. 1:1) and
φλαγαυ ‫( ְפּ ָלגָיו‬Ps. 46:5), in which no vowel at all is transcribed in the initial vowel slot—but
note the variant φαλαγαυ.
3) in the initial vowel of the piˁel participle:391
389. The ε in λεριβι* may indicate assimilation to /ı̄ / and thus variable schwa (see also YUDITSKY 2017, 228).
390. We could perhaps also add the suffixed form ιασουαθι ‫( י ְשׁוּעָתִ י‬Ps. 89:27) to this category, for which an
alternation with zero may be demonstrated by ισουωθ ‫( י ְשׁוּעוֹת‬Ps. 28:8) and θβουνωθ ‫( תְּ בוּנוֹת‬Ps. 49:4).
391. YUDITSKY argues that the prefix vowel of the piˁel participle was /a/ (i.e., *maqattel) in ancient Hebrew on
the basis of transcriptions in the Secunda, the LXX (e.g., Μανασση ‫) ְמנַשֶּׁה‬, Jerome (manaem ‫) ְמנַחֵם‬, and a number
of peculiar vocalizations in the reading traditions (YUDITSKY 2005, 128–29; 2017, 150). While the evidence he
cites probably indicates that during the Second Temple period *maqattel was a common vocalization of the
participle, it is unlikely that *maqattel was the original form. In cuneiform transcription of Hebrew, proper
names formed from the piˁel participle usually exhibit a prefix vowel of /e/ or /i/ into the Late Babylonian
period: e.g., me-ni-ḫ i-im-me and mi-na-aḫ -ḫ e-e-mu for ‫ ; ְמנַחֵם‬me-na-se-e and mi-na-si-i for ‫( ְמנַשֶּׁה‬MILLARD 2013,
840, 843). There is, however, one instance in which the prefix vowel is /a/ (ma-na-si-iˀ ‫) ְמנַשֶּׁה‬, which KHAN
- 322 -
αµµααζερηνι /ham-mʔazzerē nı̄ / [hamːaʔazːɪʀeːniː] 'the one who girds me' Ps. 18:33
µαλαµµεδ /m(a)lammed/ [malamːɪð] 'training' Ps. 18:35
ουµασαννεαϊ /w-m(a)ś anneʔay(y)/ [(ʔ)umasanːɪʔaj] 'and those who hate me' Ps. 18:41
λαµανασση /l(am)-m(a)nas ̣s ̣ē h ̣/ [lam(ː)anaʦˀeːħ] 'to the choirmaster' Ps. 31:1 (+3x)
While there are no clear instances in which the initial vowel of the piˁel participle is omitted,
it is possible that alternation with zero occurs in the transcription αµµιαλιµ ‫( ַה ְמי ַ ֲחלִים‬Ps.
31:25). Alternatively, the ι could be interpreted as representing the sequence [ij], in which
case the initial vowel would have assimilated to the following yod. Assimilation of the initial
vowel (to the preceding labial) is also attested in the transcription µοσαυε ‫שׁוֶּה‬
ַ ‫( ְמ‬Ps. 18:34).
The preservation of /a/ in these forms may be due to the fact that the vowel occurs in a prefix.
Cross-linguistically, schwa tends to be longer in a prefix; thus, the degree of shortening and
reduction is curtailed (HANIQUE, SCHUPPLER, ERNESTUS 2010).
4) in the construct pattern of *qatē /*qalē :
φανη /pnē / [phaneː] 'the face of' Ps. 18:43

βανη /bnē / [baneː] 'sons of' Ps. 18:46
κασε /qs ̣ē / [kˀaʦˀeː] 'the edge of' Ps. 46:10
Alternation with zero is demonstrated by βνη ‫( ְבּנֵי‬Ps. 29:1 + 4x) and ιµη ‫( יְמֵי‬Ps. 89:46).
5) in the construct pattern of *qatō l (including the infinitive construct):
χαβωδ /k(a)bō d/ [khaβoːð] 'the honor of' Ps. 29:2

καρωβ* /q(a)rō b/ [kˀaʀoːβ] 'coming near' Ps. 32:9
σαλωµ /š (a)lō m/ [ʃaloːm] 'the peace of' Ps. 35:27
Alternation with zero is demonstrated by the form λφνωθ ‫( ִלפְנוֹת‬Ps. 46:6).
Two other miscellaneous forms in the Secunda also demonstrate that historical short
/a/ was not always preserved propretonically, but could undergo acoustic reduction:
ιελεδεθεχ* /yledtek(k)/ (< *yaladtı̄ ka) [jəlɪðəθɪkh] 'I have begotten you' Ps. 110:3
βελιαλ /blı̄ yaˁl/ (< *balı̄ yaʕl) [bəliːjaʕl] 'worthlessness' Prov. 16:27
suggests may reflect assimilation to the following vowel (2013h, 552).
- 323 -
In each case, original */a/ is represented by Greek ε, which likely indicates reduction.
Before we proceed to interpret what these data mean for the realization of shewa, it is
also worth noting that there are a few instances in which historical */u/ is preserved where we
might expect reduction in the Secunda:
θεσοδηνι /tesʕ(o)dē nı̄ / [thɪsʕoðeːniː] '(it) supports me' Ps. 18:36

ιεφφολου /yepp(o)lū / [jɪphːʊluː] 'they will fall' Ps. 18:39
ουϊερογου /w-yeh ̣r(o)gū / [(ʔ)ujɪħʀʊʁuː] 'and they will tremble' Ps. 18:46
To this list we may also add the form ακκοδασιµ (II Kgs. 23:7) from external sources.392 Both
alternation with zero and acoustic reduction for the theme vowel in the verbal forms
ιεφφολου and ουϊερογου is demonstrated by category (5) in 6.5.1.3.2.
We have demonstrated that historical /a/ (rarely /u/) is often preserved in the Secunda
in environments where we would expect vocalic shewa. At the same time, there is also evi-
dence that it reduces and interchanges with zero in those very same environments.
This inconsistency can be explained in light of VAN BERGEM's work on vowel reduc-
tion. According to VAN BERGEM, the phonetic process of acoustic reduction and the phonologi-
cal process of lexical reduction should be seen as a linear diachronic development. In the first
place, acoustic reduction occurs when speakers are aiming for the quality of a particular vo-
calic phoneme, but due to factors like distance from stress, they undershoot and centralize it
to some degree, sometimes even pronouncing it as [ə]. Nevertheless, regardless of its phonet-
ic realization at this stage, its phonological status is still that of a full vowel. In the second
place, after acoustic reduction has rendered the vowel void of any clear quality—note FLEM-
MING's concept of variable schwa—it is then replaced permanently and lexically with a schwa
vowel or another vowel whose quality facilitates articulation with little effort. Because it is a
diachronic process, instances of acoustic reduction may increase and affect more and more
speakers until the reduced vowel is finally made a "permanent part of the lexical system."
392. Note also χωµαρειµ ‫( ְכּמ ִָרים‬Zeph. 1:4) in Theodotion.
- 324 -
However, during such a period of linguistic change, there would be two lexical variants for
the same word, one with the original historical full vowel and one with schwa (1991).
The data from the transcriptions indicate that, at the time of the Secunda, historical /a/
in certain contexts was passing through this period of linguistic change from acoustic reduc-
tion to lexical reduction. While some speakers probably pronounced ‫' פלגיו‬its streams' as
[phalɔːʁaw] with historical /a/ [a], others pronounced it as something more like [phəlɔːʁaw] or
even [phlɔːʁaw]. Variation could have even been an element of speech speed or register. For
example, HERCUS has shown that in a language where vowels reduce and are regularly real-
ized as [ə], the distinct quality of the vowel might still be pronounced when the word is said
carefully (HERCUS 1986; CROSSWHITE 2001, 4). For this reason, a sacred reading tradition
might be more prone to maintain historical vowels. The fact that there was a lag for the
reduction of historical /a/, which is more sonorous than /i/ and /u/, has parallels in other Se-
mitic languages. In Arabic consonantal orthography, for example, the accusative case (with
/a/) is represented while the nominative (with /u/) and genitive (with /i/) are not: e.g., ‫ﻛــــﺘــــﺎﺑــــﺎ‬
'book (acc.)', but cf. ‫' ﻛــﺘــﺎب‬book (nom. or gen.)'. Moreover, in Geˁez, the Proto-Semitic vowels
*/i/ and */u/, but not */a/, have become /ə/ (WENINGER 2011c, 1128).
6.5.1.3.4. Concluding Remarks: Variation in Shewa at the Time of the Secunda
In sum, we can make the following generalizations about the results of vowel reduction evi-
denced in the Secunda: (1) in some contexts, a reduced vowel behaved like variable schwa,
assimilating in quality to its immediate context, (2) in other contexts, acoustic reduction led
to a more centralized realization of [ə] (or perhaps [ɛ]), and finally, (3) a historical /a/ vowel
was preserved in many of the same environments in which reduction occurred in the previous
two groups. The precise realization of all three of these principles probably varied by speaker
and other factors. It is out of such a variegated linguistic situation that the more regular lexi-
cal realizations of vocalic shewa, such as [a] in Tiberian and [e] in Palestinan, developed.
- 325 -
At this point, VAN BERGEM's work on lexical reduction is especially helpful when con-
sidered in light of the history of the development of shewa in the various Hebrew traditions.
It is important to note that when VAN BERGEM discusses lexical reduction, he does not stipulate
that the acoustically reduced vowel must be replaced only by the /ə/ phoneme. Rather, the
reduced vowel can also be lexically replaced by another vowel in the system that entails a
particularly low-effort articulation. In the case of Tiberian, we may suggest that the historical
[a] vowel preserved in "shewa-vowel slots" became the lexically substituted vowel for the
reduced [ə]/[ɛ] and variable schwa elsewhere, whereas in Palestinian, the vowel /e/, some-
what near centralized [ə] or [ɛ], became the lexically substituted vowel for both the reduced
vowels and the historical /a/ vowels, which presumably had first undergone centralization.
Because lexical reduction entails replacement by another vowel of low articulatory effort, the
different phonetic realizations of shewa in Tiberian and Palestinian may be caused by differ-
ent bases of articulation in each tradition. Finally, remnants of variable schwa are still pre-
served in the behavior of vocalic shewa before gutturals and yod in both traditions.
6.5.1.4. Complex Onsets
6.5.1.4.1. Frequency of Sonorants and Sibilants in Complex Onsets
Complex onsets ((C).CCv) are often broken up by the insertion of an epenthetic in the Secun-
da (see 6.5.1.3.1).393 There are, however, numerous instances in which a complex onset is rep-
resented in transcription without an intervening vowel. These are listed below; onsets with
gutturals, onsets with yod, and complex onsets beginning with the prepositions /b-/, /k-/, and
/l-/ have been excluded, since each of these categories seems to have special conventions:394
σµα, σµαε* /š maʕ/ [ʃmaʕ] 'listen!' Ps. 28:6, 30:11

βνη /bnē / [b(a)neː]395 'sons of' Ps. 29:1 (+4x)
393. Τhe tendency of Hebrew/Aramaic phonotactics to resolve an initial consonant cluster may be demonstrated
by the fact that the normal initial cluster χθ in the name Χθουσιων is resolved by an epenthetic in the form
Χαθουσιωνος (for Χθουσιωνος) in 5/6 Hev 15.
394. Instances of Cv1C2C2vC3v > C1vC2(C2)C3v are interpreted as reflecting degemination on the post-lexical
level (see 6.3.8.6).
395. In some cases, I have enclosed a vowel in parentheses in my phonetic transcription (e.g., [(a)]). This is
because there is evidence from parallel forms that a vowel may have been pronounced in such an environment,
- 326 -
µσουδωθ /ms ̣ū dō t/ [mʦˀuːðoːθ] 'fortresses' Ps. 31:3
εργλαϊ /reglay(y)/ > /rglay(y)/ [əʀglaj] 'my feet' Ps. 31:9
λσωνωθ /lš ō nō t/ [lʃoːnoːθ] 'tongues' Ps. 31:21
θσωβαβηνι /tsō b(ə)bē nı̄ / [th(ɪ)soːβaβeːniː] 'you surround me' Ps. 32:7
ισµου* /ś mh ̣ū ?/ [ismħuː] 'rejoice!' Ps. 32:11
σφτηνι* /š ptẹ̄ nı̄ ?/ [ʃəɸtˀeːniː] 'judge me!' Ps. 35:24
ουϊφρου /w-yeh ̣prū / [(ʔ)ujiħphʀuː] 'let them be ashamed' Ps. 35:26
σµηη /ś mē h ̣ē / [smeːħeː] 'those who rejoice in' Ps. 35:26
ιεµρου /yeh ̣mrū / [jɪħmʀuː] 'they will foam' Ps. 46:4
φλαγαυ /plɔ̄gaw/ [ph(a)lɔːʁaw] 'its streams' Ps. 46:5
µσ῾χνη /maš knē / [m(i)ʃkhneː] 'the dwellings of' Ps. 46:5
ουεζρα /w-yeʕzrɔ̄(h)/ [(ʔ)ujɪʕzʀɔː(h)] 'and he will help her' Ps. 46:6
λχου /lkū / [lχuː] 'come!' Ps. 46:9
θβουνωθ /tbū nō t/ [thβuːnoːθ] 'understandings' Ps. 49:4
χσιλ* /ksı̄ l/ [(ə)khsiːl] 'a fool' Ps. 49:11
βχωρ /bkō r/ [bχoːʀ] 'firstborn' Ps. 89:28
σφωθαϊ /ś pō tay(y)/ [sɸoːθaj] 'my lips' Ps. 89:35
βριθ(ι) /brı̄ t/; /brı̄ tı̄ / [bʀiːθ]; [bʀiːθiː] 'covenant of'; 'my ... ' Ps. 89:35, 40
µσιαχ /mš ı̄ h ̣ɔ̄k/ [m(ɪ)ʃiːħɔːχ] 'your anointed' Ps. 89:39, 52
ζχορ, ηζχορ /zkor/ [zχoʀ] 'remember!' Ps. 89:48, 51
Even a brief glance at this list is sufficient to underscore the fact that almost all of the com-
plex onsets involve at least one sonorant or sibilant consonant:
(C).C1C2 C1[+sonorant] C1[+sibilant] C1[-sonorant, -sibilant]
C2[+sonorant] 1 4 6
C2[+sibilant] 3 0 1
C2[-sonorant, -sibilant] 2 3 2
Chart 25: Sonorants and Sibilants in Complex Onsets in the Secunda
Statistically, a sonorant (blue) or a sibilant (yellow) is present as either C1 or C2 in 91% of
transcribed complex onsets, with either a sonorant or a sibilant present in both C1 and C2
(depicted in green) in 36% of complex onsets. Transcribed complex onsets without either a
but may also have undergone reduction and deletion.
- 327 -
sonorant or a sibilant (depicted in gray) account for only 9% of all transcribed complex on-
sets. It is possible, then, that complex onsets may have been more permissible when one of
the consonants was either a sonorant or a sibilant.
To the list above we may add the examples of complex onsets in quotations of the Se-
cunda from external sources:
βρησιθ /b-rē (ˀ)š ı̄ t/ [bʀeːʃiːθ] 'in the beginning (of)' Gen. 1:1
σµωθ /š mō t/ [ʃmoːθ] 'names of' Ex. 1:1
αρσαειµ /rš ɔ̄ʕı̄ m/ [ʔaʀʃɔːʕıː̃ (m)] 'wicked ones' Ps. 1:1
ουθασρηου /w-tah ̣srē hū / [(ʔ)uθaħsʀeːhuː] 'and you made him lower' Ps. 8:6
ιερχθη /yerktē / [jɪʀkhteː] 'uttermost parts of' Ps. 48:3
φθοου /ptoh ̣ū / [pthʊħuː] 'open!' Isa. 26:2
σµοωχ(!) /smū k/? [smuːχ]? 'steadfast' Isa. 26:3
In five out of seven transcriptions listed above, the complex onset contains either a sonorant
or a sibilant, and in four out of these five instances the complex onset contains both a sono-
rant and a sibilant.396 We may also add the quotation of the Secunda's rendering of ‫שּׁמִינִית‬
ְ ‫ַה‬
(Ps. 12:1) in Chrysostom. Though it is transcribed as ασεµινιθ, a variant reading has ασµενιθ.
Moreover, in one of the two transcriptions without a sonorant or a sibilant in the complex on-
set, ιερχθη, the complex onset is directly preceded by /r/, exhibiting a similar syllable struc-
ture to εργλᾱι above. If the word-medial -C1C2C3- sequence in ιερχθη is permitted because of
the presence of the sonorant /r/ as C1, we might compare such a phenomenon to instances in
Mishnaic Hebrew in which ‫ בג"ד כפ"ת‬consonants after ‫ ר‬with silent shewa are rafeh as in
‫' ַבמּ ְַר ֿפֵק‬in the elbow' (BAR-ASHER 2015, 100–102; see also the weakening of /r/ in Modern He-
brew [BOLOZKY 2013, 390]). In each case (ιερχθη and ‫) ַבמּ ְַר ֿפֵק‬, syllable-final /r/ would have
weakened so that ‫ =( כ‬χ) and ‫ פ‬would have been virtually post-vocalic, exhibiting features
otherwise restricted to post-vocalic environments. If, alternatively, it is not the sonorant that
makes the cluster permissible, then ιερχθη may be taken as evidence that in the Secunda, se-
396. Note also Jerome's amsuchan ‫ ַה ְמּ ֻסכָּן‬, where even degemination of /m/ seems to occur.
- 328 -
quences of three consecutive moras were tolerated, with the second consonant being extra-
syllabic:
ω
σ σ
µ µ µ µ µ
y e r k t e
ιερχθη = /(yer).k.(tee)/ [jɪʀkhθeː]

Figure 23: Moraic Representation of ιερχθη
Regardless of our interpretion of ιερχθη, however, the external attestations of complex onsets
in the Secunda also support a correlation between complex onsets and sibilants/sonorants.
The retention of complex onsets when one or both of the consonants are sibilants or
sonorants may be compared to the Neo-Aramaic dialects of Qaraqosh and Sulemaniyya/Ḥal-
abja, in which complex onsets are not broken up by an epenthetic if a sibilant is the first con-
sonant and/or a sonorant continuant is the second consonant (KHAN 2002, 64; 2004, 58).
At the same time, there are a number of instances in which potential clusters with
sonorants and sibilants are broken up by the insertion of an epenthetic as in other contexts in
the Secunda, even in patterns and words identical with some of those above:

ζερουωθαϊ /zrō ʕō tay(y)/ [zəʀuːʕoːθaj] 'my arms' Ps. 18:35
βανη (cf. βνη) /bnē / [baneː] 'sons of' Ps. 18:46
µεσιω (cf. µσιαχ) /mš ı̄ h ̣ō / [mʃiːħoː] 'his anointed' Ps. 28:8
σεµω (cf. σµωθ) /š mō / [ʃəmoː] 'his name' Ps. 29:2
φαλαγαυ (cf. φλαγαυ) /plɔ̄gaw/ [phalɔːʁaw] 'its streams' Ps. 46:5 (Chrys.)
There are essentially two ways of interpreting such transcription-doublets as βνη | βανη,
µσιαχ | µεσιω, σµωθ | σεµω, and φλαγαυ | φαλαγαυ. First, the inconsistency in the Greek
transcription reflects inconsistency in the Hebrew vocalization tradition. Second, the varia-
tion in the Greek transcription reflects diverse conventions of transcribing the same Hebrew
sounds and structure; thus, the Hebrew vocalization is not inconsistent but rather the Greek
- 329 -
conventions for representing it are. Both explanations are possible, and we should allow for
language-internal variation and inconsistency, yet the Greek evidence may support the latter.
Inconsistent transcription of the same phonetic reality, resulting from the transcriber
perceiving the Hebrew sounds differently as they were mapped onto the Greek graphemic-
phonemic system, could arise in a couple specific ways. First, it is possible that the short cen-
tralized epenthetic vowel was perceived as part of the sibilant or the sonorant, partly due to
the transcriber's Greek accent. Cross-linguistically, a sequence of vC[+sibilant] or C[+sibilant]v may
be perceived merely as C[+sibilant], a phenomenon also attested elsewhere in the Secunda
(6.3.2). According to SILVERMAN, due to its short duration and high coarticulatory nature,
"schwa may be confused with its absence" (2011, 629; 6.5.1.6). In Palestinian epigraphy,
Greek ἰσχίων is attested as σχίων, spelled with a complex onset (4.5.3.1.20). On the other
hand, it is also common for a prosthetic vowel to develop before σ + consonant in Egyptian
Koine (GIGNAC 1976, 312). Finally, vowel deletion in both Palestinian and Egyptian Koine is
especially common in the environment of sonorants (4.5.3.1.20).
Second, it is also possible that the transcriber was inconsistent in his representation of
Hebrew, occasionally conceiving of the Hebrew words phonemically instead of phonetically.
Native speakers do tend to conceive of their own language in terms of its "phonemic struc-
ture" rather than its "phonetic reality" (DIRVEN and VERSPOOR 2004, 115). This is especially
true for those trained in literacy and spelling. These doublets, then, can be explained by posit-
ing that vocalic shewa was phonemically zero but phonetically realized as a short vowel.
When the Hebrew phones were mapped onto the Greek graphemic-phonemic system, unsur-
prisingly, the transcriber vacillated between representing the phonemic structure and the pho-
netic reality. A transcription like βνη, then, reflets the phonemic structure /bnē / from the per-
spective of the transcriber's Hebrew accent, whereas a transcription like βανη reflects the
phonetic reality [baneː] from the perspective of his Greek accent:
- 330 -
ω
µ µ
b n e
βανη, βνη = /(bnee)/ [baneː]

Figure 24: Moraic Representation of βνη
The inconsistency in transcribing phonological consonant clusters that are realized phoneti-
cally with an epenthetic vowel may be compared to doublets in Tiberian, in which a non-gut-
̣
tural consonant is pointed with the regular shewa sign (ְ ) in one form and with a ḥateph vow-
el in an identical form elsewhere: e.g., ‫[ נֵ ֽ ְלכָה־נָּא‬neːlaˈχɔːnnɔː] 'let us go!' (II Kgs. 6:2), but
‫[ נֵ ֽ ֲלכָה־נָּא‬neːlaˈχɔːnnɔː] (Ex. 3:18) (see KHAN 2013a, 101).
It should be noted, however, that aside from these doublets, there is evidence that in
particular instances sonorants actually occasioned a distinct syllable structure in a word and
were not merely perceived differently. This phenomenon will be examined below.
6.5.1.4.2. The Effect of Sonorants and Sibilants on Syllable Structure
When complex onsets are resolved by the insertion of an epenthetic in the Secunda, the
epenthetic is usually inserted between the consonants of the initial cluster (.CC > .CvC).
However, several scholars have pointed out that if the first consonant in the cluster is a sono-
rant, especially /r/, the epenthetic may be inserted before the complex onset (.CC > v.CC)
(KHAN 2013b, 674–75; YUDITSKY 2017, 75–76). This occurs especially in verbal forms:
ϊκερσου /yeqrs ̣ū / [jikˀəʀʦˀuː] 'they will wink' Ps. 35:19

ισουµου* /yeś mh ̣ū / [jisumħuː] 'let them rejoice' Ps. 35:19
ιεσεµου /yeś mh ̣ū / [jɪsɪmħuː] 'let them rejoice' Ps. 35:24
ουειεσοµου /w-yeś mh ̣ū / [wɪjɪsʊmħuː] 'and let them rejoice' Ps. 35:27
These forms probably indicate that, at an underlying phonological level, the sonorant of the
second radical was extra-syllabic (see KHAN 2013b, 675):
- 331 -
ω ω
σ σ σ σ
µ µ µ µ µ µ µ µ µ µ
y e q r s. u y e ś m h. u
ϊκερσου = /(yeq).r.(s ̣ú u)/ [jɪkˀəʀʦˀuː] ιεσεµου = /(yes).m.(h ̣ú u)/ [jɪsəmħuː]
Figure 25: Moraic Representation of ϊκερσου and ιεσεµου
It may be that an alternative spelling, ιεσµου (‫שׂמְחוּ‬

ְ ִ ‫?י‬, cf. MT ‫שׂמְּחוּ‬
ַ ְ ‫[ י‬Ps. 46:5]), reflects either
a misperception based on the transcriber's Greek accent or a phonemic representation. There
are also a couple nominal forms in which an initial /r/ in a complex cluster seems to engender
the insertion of an epenthetic before the consonant cluster:
αρσαειµ /rš ɔ̄ʕı̄ m/ [ʔaʀʃɔːʕıː̃ (m)] 'wicked ones' Ps. 1:1

εργλαϊ /reglay(y) > rglay(y)/ [əʀglaj] 'my feet' Ps. 31:9
The epenthetic being inserted before the consonant cluster in αρσαειµ probably indicates that
the initial /r/ is regarded as extra-syllabic at an underlying phonological level:

ω
σ σ
µ µ µ µ
r š O Q i m
αρσαειµ = /r.(š ɔɔ).(ʕı́i).m/ [(ʔ)aʀʃɔːʕıː̃ (m)]
Figure 26: Moraic Representation of αρσαειµ
The transcription αρσαειµ is especially significant because the pattern *qatalı̄ m is almost al-
ways transcribed with the preservation of the historical */a/ vowel after the first consonant.
The only clear example of a sibilant occasioning a different syllable structure is in the
imperative form ηζχορ ‫( זְכָר־‬Ps. 89:48) listed above. Although the quality of η is unexpected
for a prosthetic vowel, which may reflect that an epenthetic was at least occasionally realized
more like s ̣ere [e], it has been argued that there is at least one instance in Judaean Hebrew in
- 332 -
which a prosthetic ‫ א‬is added before a qal imperative (MOR 2015, 148–49).397 If ηζχορ is an
original reading, it would indicate that the initial sibilant /z/ of the cluster is extra-syllabic:
ω
µ µ
z k o r
ηζχορ = /z.(kó r)/ [(ʔ)ezχʊʀ]
Figure 27: Moraic Representation of ηζχορ
Additionally, the transcription ισµου cited above, if not a scribal error (**σιµου > ισµου),
would also indicate prothesis. Finally, the Aramaic loan χσιλ ‫* <( ְכּסִיל‬qatı̄ l) (Ps. 49:11) may
also be evidence of a different syllable structure. In the previous word, ιααδε ‫יַחַד‬, an inexplic-
able ε is added word-finally, despite no apparent phonetic cause. The final ε may be a helping
vowel to facilitate the pronunciation of the initial cluster: i.e., ιααδε χσιλ [jaħað əkhsiːl].398 As
a final note, it is worth mentioning that prosthetic vowels may exhibit assimilatory features
like variable schwa (see 6.5.1.3.2): the prosthetic is ε before /k/ (e.g., ιααδε χσιλ), but raised
to η or ι before a sibilant (e.g., ισµου and ηζχορ) and lowered to α before /r/ (e.g., αρσαειµ).
The phenomenon of prothetic epenthesis in the Secunda has been compared to the
epenthetic inserted in Babylonian verbal forms with a sonorant second radical such as wubrqiti
(KHAN 2013b, 674–75). Although less frequently cited, it should also be mentioned that along
with the sonorants ‫ מ‬,‫ נ‬,‫ ל‬,‫ר‬, YEIVIN also cites ‫ ז‬,‫ צ‬,‫ ש‬as consonants for which an epenthetic is
inserted prothetically (1985, 388–89).399 These Secunda transcriptions are significant because
397. On this point, it is worth considering whether the initial vowel transcribed in qal imperatives with initial
gutturals such as εζακ ‫( ֲחזַק‬Ps. 31:25), εζου* ‫( חֲזוּ‬Ps. 46:9), and αϊη ‫( ֲהי ֵה‬Ps. 30:11) may actually reflect prothetic
epenthesis rather than anaptyctic epenthesis. Babylonian also behaves differently than Tiberian on this point,
tending to introduce epenthetics before the guttural rather than after when it is preceded by the conjunction
waw: e.g., µk;jwi, ˚dohw', lwOmjw' (YEIVIN 1985, 482–84). It is also worth noting that the transcription εζου* actually
reads εεζου in the manuscript, perhaps indicating /h ̣zū / [(ʔ)əħzuː] (see 6.3.6.2).
398. Compare the Mishnaic Hebrew rendering ‫ אכסניה‬of the Greek loanword ξενία (ROSÉN 1963, 69).
399. Compare the fact that Greek loanwords with initial clusters beginning with σ are often resolved by
prothetic epenthesis when rendered in Hebrew: e.g., ‫ איזמל‬σµίλη, ‫ איסטסית‬στάσις, ‫ איספלנית‬σπλήνιον, and ‫איצטרובל‬
- 333 -
they demonstrate that, when the first consonant of a complex onset was highly sonorous (see
also 6.5.1.3.3) or perhaps a sibilant, the first consonant of the cluster was not always syllabi-
fied with the following Cv sequence as in Tiberian, but sometimes regarded as extra-syllabic
as in Babylonian, thus prompting an epenthetic to be inserted before the complex onset.

6.5.1.4.3. The Effect of /y/ on Syllable Structure
Like complex onsets whose first radical is either /r/ or /m/ (.rC or .mC), there may be evi-
dence that complex onsets beginning with yod (.yC) also prompt the insertion of a prosthetic
rather than anaptyctic vowel. However, because ι is used in the Secunda to signify [iː], [i],
[ji(ː)], and [j], it is not always clear how to interpret complex onsets beginning with ι. Some
of the possible instances of #yC > #vy.C are listed below:
Nouns:
ισουωθ /yš ū ʕō t/ [(ʔ)i(ː)ʃuːʕoːθ] 'salvations' Ps. 28:8

ιριβαϊ400 /yrı̄ bay(y)/ [(ʔ)i(ː)ʀiːβaj] 'those who contend with me' Ps. 35:1
ϊϊδαθι /yh ̣ı̄ dɔ̄tı̄ / [jiħiːðɔːθiː] 'my only one' Ps. 35:17
ιµινω /ymı̄ nō / [(ʔ)i(ː)miːnoː] 'his right hand' Ps. 89:26
ιµιν /ymı̄ n/ [(ʔ)i(ː)miːn] 'the right hand of' Ps. 89:43
ιµη /ymē / [(ʔ)i(ː)meː] 'the days of' Ps. 89:46
Verbs:
ιζαµµερεχ /yzammerek(k)/ [(ʔ)i(ː)zammɪʀɪkh] 'I will sing to you' Ps. 30:13

ιδαββερ /ydabber/ [(ʔ)i(ː)ðabːɪʀ] '(it) will speak' Ps. 49:4
ιαλληχουν /yhallē kū n/ [(ʔ)i(ː)halːeːχuːn] 'they will walk' Ps. 89:31
ιαλληλου /yh ̣allē lū / [(ʔ)i(ː)ħalːeːluː] 'they will profane' Ps. 89:32
ιµαλλετ /ymallet/̣ [(ʔ)i(ː)malːɪtˀ] 'he will rescue' Ps. 89:49
A transcription like ιµη can be interpreted in essentially three different ways, all determined
based on the reading of the initial ι. First, if the ι represents only /y/ [j], then the word should
be read with an initial cluster beginning with /y/ [j]: ιµη = /ymē / [jmeː]. Second, if the ι repre-
sents only the vowel /i/ [i], then the word should be read as beginning with a vowel: ιµη =
στρόβιλος (HEIJMANS 2013, 26–31). Note also ‫ אצטדיון‬στάδιον cited in ROSÉN (1963, 69).
400. Note that spectral imaging of the manuscript has revealed that there is probably no ι after the β (i.e.,
ιριβιαϊ) as others have read it. The correct reading is ιριβαϊ.
- 334 -
/ymē / [(ʔ)i(ː)meː]. KHAN suggests that such is the case at least for the transcription ιµη, posit-
ing the following development: ymē > iymē > ı̄ mē (2013h, 551). Third, and finally, if the ι
represents both /y/ and a following short high vowel (i.e., [jɪ] or [ji]), then the word should be
read with an anaptyctic vowel: /ymē / [jimeː] or [jɪmeː] (see YUDITSKY 2017, 32–33, 96–98).
There is insufficient evidence to determine with certainty how the initial ι in these
transcriptions should be read, but a case can be made that in many of these instances, word-
initial ι should be read as a simple vowel ([i], [iː], or [ʔi(ː)]). If the transcriber intended to rep-
resent [jɪ] (or [ji]), he would have been more likely to write ι with a vowel (ιε/ϊε) or perhaps ι
with trema (ϊ). This point can be illustrated by comparing the rendering of the 3m prefix
/y(e)-/ in qal and piˁel verbs (forms with a prefixed waw or an /a/ prefix vowel are excluded):
ι ι / _σ ϊ / _κ ιε ϊε Total
Piˁel/Polel: 6 (75%) 1 (12.5%) 0 0 1 (12.5%) 7
Qal: 0 2 (9%) 2 (9%) 13 (59%) 5 (23%) 22

Chart 26: Greek Representation of the Piˁel and Qal 3ms Verbal Prefix in the Secunda
The distribution of transcription conventions between the piˁel and qal prefixes clearly cuts
against the idea that they both were phonologically realized as /ye-/ and phonetically realized
as [jɪ]/[ji]. In the qal prefix, transcription conventions favoring a consonant + vowel realiza-
tion (ιε, ϊε, ϊ) are implemented, whereas in the piˁel prefix, conventions favoring a vowel real-
ization (ι) are implemented. Only before σ or κ, environments in which vowels tend to raise
in the Secunda (see 6.3.2; 6.3.3.1), does the representation of the qal prefix resemble that of
the piˁel. YUDITSKY attributes the use of ι in the piˁel—instead of ιε/ϊε as in the qal—to the
raising of /e/ in the environment of yod: i.e., (ye >) yi (> ˀi) (2017, 96–97; 150–53), but this
does not explain the large disparity in the representation of the /y(e)-/ prefix between the piˁel
and qal stems, both of which begin with yod.
Because YUDITSKY does not acknowledge the presence of shewa in the Secunda, the
implication of his theory must be that e > i / y_ occurred far more often in the piˁel than in the
- 335 -
qal for some reason. However, if we affirm that shewa did exist in the phonology of the Se-
cunda, we may explain the disparity between the piˁel and qal prefix forms by assuming a
complex onset in the piˁel stem. Just as the highly sonorous /r/ and /m/ engendered the inser-
tion of an epenthetic vowel before the cluster, so did the even more sonorous /y/ at least on
some occasions. Accordingly, we may suggest that word-initial yod in these forms was an ex-
tra-syllabic semisyllable, assuming the same sort of development that KHAN posits for Secun-
da ιµη (see above) and Babylonian dwOsyI (ysō d > iysō d > ı̄ sō d) (KHAN 2013h, 551; see also
YEIVIN 1985, 269–82): i.e., ιδαββερ = ydabber > iydabber > ı̄ dabber/idabber:
ω ω
σ σ σ
µ µ µ µ µ µ
y d a b e r y m i n
ιδαββερ = /y.(dab).(bé r)/ [(ʔ)i(ː)ðabːɪʀ] ιµιν = /y.(mii).n/ [(ʔ)i(ː)miːn]

Figure 28: Moraic Representation of ιδαββερ and ιµιν
On the other hand, it should not be ruled out that the epenthetic may have been inserted in be-
tween the consonants of the complex onset and was realized as [i] due the preceding /y/,
which subsequently elided: ydabber > yidabber > (ʔ)idabber. This same phenomenon, before
elision, occurs in the Neo-Aramaic dialect of the Jews of Urmi (e.g., ytəwli [jiˈtɪβli̞ ] 'I sat
down') (KHAN 2008a, 44), and, after elision, in the Neo-Aramaic dialect of the Jews of Arbel
(e.g., *yliple > [ʔiːˈliˑpleː] 'he learnt') (KHAN 1999, 65). Note also the Syriac forms ‫ ܺܝـــ ܷܬܒ‬ı̄ ṯ eḇ
ܳ ‫ ܻܐ ـ‬ı̄ qā rā 'honor' (NÖLDEKE 1904, 27). In modern Arabic, it is

'sat' and ‫ ܺܝـ ܰܕܥ‬ı̄ daʕ 'knew', and ‫ܝܩـ ܴܪܐ‬
common for VC- dialects to realize the reflex of *yukallimū as iká llmu with an initial vowel,
presumably deriving from an initial cluster *ykallmu (KIPARSKY 2003, 148). However, it
should be noted that there is not always such a sharp distinction between #yC and #iC pho-
netically. For example, ERWIN notes that in Iraqi Arabic the quality of an initial /y/ in a com-
plex onset is more "vowel-like" (i.e., [iː]) than "consonant-like" (2004, 31).
- 336 -
There is, in fact, evidence that such variant forms (i.e., yidabber) may have existed at
least side-by-side with those suggested above (i.e., idabber/ı̄ dabber). For example, at least in
some instances and/or in some forms, a word-initial phonological cluster beginning with yod
was realized phonetically with a vowel (epenthetic or historical) after the yod:
ϊεσαυου /yš aw(w)ʕū / [jɪʃawʕuː] 'they will cry out' Ps. 18:42
ϊασουαθι /yš ū ʕɔ̄tı̄ / [jaʃuːʕɔːθiː] 'my salvation' Ps. 89:27
YUDITSKY argues that the ε in ϊεσαυου represents the etymological prefix for piˁel (2017, 150–
52), but it is better interpreted as an epenthetic even here. In the case of ιασουαθι, on the oth-
er hand, the α does seem to reflect the preservation of a historical vowel. It should also be
noted that the trema on the initial ι in ϊϊδαθι (see above) probably indicates a consonantal yod.
Consonantal yod is also preserved after the conjunction ‫ ו‬and the preposition ‫ל‬:
ουεµιναχ /w-ymı̄ nɔ̄k/ [(ʔ)u(j̞ )ɪmiːnɔːχ] 'and your right hand' Ps. 18:36
ουϊεδαββερ /w-ydabber/ [(ʔ)ujɪðabːɪʀ] 'and he will speak/subdue' Ps. 18:48
λειρηαχ /l-yrē ʔɔ̄k/ [lɪj(i)ʀeːʔɔːχ]/[ləj(i)ʀeːʔɔːχ] 'for those who fear you' Ps. 31:20
In the case of ουεµιναχ, the lack of ι in transcription may reflect the weakening of the
semivowel, in which case the ε would represent a vowel following the weakened reflex of
consonantal yod [wəʔɪmiːnɔːχ]. Alternatively, the transition from ου to ε may be an uncon-
ventional attempt to signify /y/ [j] (6.3.7.2). In the case of ουϊεδαββερ, assuming it is a w +
yiqtol form,401 it provides another example in which consonantal yod is preserved. In the case
of λειρηαχ, the digraph ει could be an example of ει for long /ı̄ /, but this is unlikely since ει
for long /ı̄ / in the Secunda normally occurs only on the main stress of the word. Therefore,
λειρηαχ should be read as λε + ιρηαχ, in which case it would also indicate a consonantal yod.
Finally, there is one transcription from external sources in which consonantal geminated yod
401. That ουϊεδαββερ is a w + yiqtol form is supported by Aquila, Symmachus, Theodotion, the Vulgate, and
Targum Onkelos. If the form ουϊεδαββερ is actually a wayyiqtol form, then it bears a similar syllable structure
to the form µειδηχεµ (i.e., CVyyVCV) and indicates that geminated yod was typically preserved before shewa.
- 337 -
with shewa seems to be preserved after the preposition ‫מן‬, contrary to most reading traditions
of Hebrew (see YUDITSKY 2017, 96):
µειδηχεµ /mey-ydē kem(m)/ [mɪjːiðeːχɪm] 'from your hands' Mal. 2:13
At the same time, however, the consonantal pronunciation of yod does not seem to be
maintained in the plural construct of the word ‫ יום‬after the inseparable prepositions (cf. YUDIT-
SKY 2017, 63, 181–82, 228):
βιµη /b-ymē / [biːmeː] 'in the days of' Ps. 49:6

χιµη /k-ymē / [khiːmeː] 'like the days of' Ps. 89:30
The transcriptions βιµη and χιµη are more consistent with the general behavior of yod with
shewa in such an environment in other Hebrew traditions. In both Tiberian and Babylonian,
yod in these circumstances is normally elided: e.g., ü‫ִיראֶי‬

ֵ ‫ ל‬,‫ וִידַ בֵּר‬,‫ מִידֵ יכֶם‬,üְ‫ ;וִימִינ‬rpek'ywI, ydeybi, ymeymi,
wOnymiyli (YEIVIN 1985, 525, 773, 891, 1157). There are, however, a number of rare vocalizations,
mostly in Middle and Late Babylonian, in which the yod is pronounced: e.g., ˚;n}ymiyw', qjeç'yIw},
µk'ymey}mi (normal for MB and LB) (525, 1154, 1157). Even in Old Babylonian, yod at the begin-
ning of a word-initial complex onset was not always pointed with ḥiriq, sometimes being left
̣ (i.e., vocalic shewa). One example of

without niqqud and sometimes being pointed with ḥitfa
inconsistency is that there is a greater tendency for word-initial yod to be pointed with a ḥiriq
when there is only one syllable that follows it in the word (e.g., dwOsyI) (YEIVIN 1985, 269–71).
In sum, just as in the case of the sonorants /r/ and /m/, highly sonorous /y/ at the be-
ginning of a complex onset in the Secunda seems to bring about a different sort of syllable
structure, namely, the insertion of an epenthetic vowel prior to the consonant cluster. This ap-
proach to epenthesis is shared with the Babylonian tradition. At the same time, however, a
number of counterexamples in the Secunda seem to indicate that this might not have been the
only way that these forms developed in the tradition(s) and pronunciation(s) upon which the
transcriber drew. After all, even in Old Babylonian, there is a degree of inconsistency with re-
spect to the vocalization of these forms (YEIVIN 1985, 271–75).
- 338 -
6.5.1.5. Vowel Syncope and Consonant Clusters: CvCvCv(C) > CvCCv(C)
6.5.1.5.1. Short + Short: CvCvCv̆̄ (C) > CvCCv̆̄ (C)
In a series of two consecutive short (or shewa) vowels in open syllables, there is a tendency
for the second vowel to undergo syncope: v > Ø / Cv̆ C_Cv̆ (C). Aside from a small number of
exceptions (see below), this phenomenon is general in the Secunda:402
Nouns:
λαβλωµ /l-blō m/ (< *la-balō m)403 [laβloːm] 'to curb' Ps. 32:9
λαµσω /l-ms ̣ō (ʔ)/ (< *la-mas ̣ō ʔ) [lamʦˀoː] 'to find' Ps. 36:3
βαρσωνω /b-rs ̣ō nō / (< *ba-ras ̣ō nō ) [baʀʦˀoːnoː] 'in his favor' Ps. 30:6
ουαλσωνι /w-lš ō nı̄ / (< *wa-laš ō nı̄ ) [walʃoːniː] 'and my tongue' Ps. 35:28
δαβρη /dabrē / (< *dabarē ) [daβʀeː] 'words of' Ps. 35:20
λαµαλχη /l-malkē / (< *la-malakē ) [lamalχeː] 'to the kings of' Ps. 89:28
βαλβαβαµ /b-lbɔ̄am(m)/ (< *ba-libā bamm) [balβɔːβa(̃ m)] 'in their heart' Ps. 35:25
χαβηµωθ /k-bhē mō t/ (< *ka-bahimō t) h
[k aβheːmoːθ] 'like cattle' Ps. 49:13
ουβανγαϊµ /w-b-ngɔ̄ʕı̄ m/ (< *wa-ba-nigā ʕı̄ m) [(ʔ)uβanʁɔːʕıː̃ (m)] 'and by smitings' Ps. 89:33
βσεδκαθαχ /b-s ̣edqɔ̄tɔ̄k/ (< *ba-s ̣adaqā tā k) [b(ɪ)ʦˀɪðkˀɔːθɔːχ] 'in your righteousness' Ps. 31:2
ουθφελλαθι /w-tpellɔ̄tı̄ / (< *wa-tapillā tı̄ ) [(ʔ)uθphɪlːɔːθiː] 'and my prayer' Ps. 35:13
Verbs:
ουαρηµ /w-rʕē m/ [waʀʕeːm] 'and shepherd them!' Ps. 28:9

ουαδου /w-dʕū / [waðʕuː] 'and know!' Ps. 46:11
There are, however, a number of exceptional cases in which this syncope rule does not apply.
First, when syncope would result in a consonant cluster of identical or homorganic conso-
nants, they are broken up by an epenthetic shewa vowel:
αµιµιµ /ʕam(ə)mı̄ m/ [ʕamimıː̃ (m)] 'peoples' Ps. 18:48

The lack of syncope in these forms may be attributed to the OCP, which states that consonant
clusters are not allowed for identical or homorganic consonants (see MCCARTHY 1981; 1986;
402. Note that KHAN states that in the Secunda "there is sometimes no vowel where the shewa is silent in the
Tiberian reading tradition, e.g., in the middle of a word after a short vowel" (2013h, 551).
403. We cannot determine with certainty whether the infinitive construct should be reconstructed as *qutul or
*qatō l. However, the infinitive construct seems to be attested as qatō l elsewhere in the Secunda (e.g., καρωβ ‫קְר ֹב‬
[Ps. 32:9]). At least synchronically, then, the forms λαβλωµ and λαµσω reflect syncope of *la-qatō l > laqtō l.
- 339 -
SCHWARZWALD 2013) and are consequently broken up by an epenthetic. This differs from
Tiberian, in which such sequences are geminated unless the first vowel is long: e.g., ‫ה ְַר ִרי‬
[haʀːiː] but ‫[ ָלקְקוּ‬lɔːqaquː] (KHAN 2013h, 545). In Middle Babylonian, the first resh has a vo-
calic segment in similar forms: e.g., yErr]h' (YEIVIN 1985, 798). In two examples, one in the
palimpsest and one in external sources, the OCP may block expected syncope (see 6.3.1.3):
βρεδεθι /b-redtı̄ / [bəʀɪðɪθiː] 'when I go down' Ps. 30:10
ιελεδεθεχ* /yledtek(k)/ [jəlɪðəθɪkh] 'I have begotten you' Ps. 110:3
Diachronically, these forms derive from *ba-ridtı̄ and *yaladtı̄ kā ̆ . Synchronically, the forms
presumably derive from *redt + ı̄ and *yaladt + *ek. At the post-lexical level, the final conso-
nant cluster in the non-suffixed forms *redt and *yaladt would have been resolved by an
epenthetic and likely realized as something like [ʀɪðɪθ] and [jɔːləðəθ]. When the pronominal
object suffix was added to the verb, the series of CvCvCv at the end of the word should have
resulted in syncope: *ρεδεθ + *ι > **ρεδθι; *ιελεδεθ + *εκ > **ιελεδθεχ. It seems that syn-
cope was blocked by the homorganic articulation of /d/ and /t/ according to the OCP.404
Second, in two construct forms from original *qatalat, in which we would expect syn-
cope in light of the principles outlined above, a medial /a/ is present:
ουαναυαθαχ /w-ʕanwɔ̄tɔ̄k/ [(ʔ)uʕanawɔːθɔːχ] 'and your humility' Ps. 18:36

The only other construct form from original *qatalat in the Secunda (βσεδκαθαχ) exhibits
syncope. Previous scholars have explained the lack of syncope in ουαναυαθαχ and βααδαρεθ
as indicative of either the preservation of the original vowel or the derivation of the form
from a variant pattern (see YUDITSKY 2017, 191–93). There is, however, a more likely expla-
nation. According to the Syllable Contact Law (SCL), which has been found to be valid for
Hebrew in two studies of ‫ פ"ח‬verbs (DECAEN 2003; ALVESTAD and EDZARD 2009, 51), a fall in
sonority is preferred in the transition from the end of one syllable to the beginning of another.
404. Compare also the Modern Hebrew form lamá deti ‫( ָלמַדְ תִּ י‬cf. patá xti ‫( )פָּתַ חְתִּ י‬SCHWARZWALD 2013, 573).
- 340 -
When there is an increase in sonority, the cluster may be resolved by means of contact
anapytxis, as in English thatway > thataway (DECAEN 2003, 38–39). The rise in sonority in
the syllables [ʕan.wɔː] and [hað.ʀæθ], then, might have been resolved by means of contact
anaptyxis. This explanation is further supported by the fact that anaptyxis seems to occur in a
pattern without an original medial vowel (i.e., *qitlat)405 but with rising sonority at the onset
of the second syllable:406
εµαραθ /ʔemrat/ [ʔɪmaʀæθ] 'the word of' Ps. 18:31
The same sort of phenomenon is attested in Palestinian Greek epigraphy, probably also due to
the SCL: e.g., Απερος (for Ἄπρος), Δυστερου (for Δύστρου) (variations 93–94), and
Ιαναουαριαις (for Ιανουαριαις) (5/6Hev 11). In Egyptian Koine and Modern Greek, epenthet-
ics also often arise in similar environments: e.g., Egyptian Koine has δραχαµας (for
δραχµάς), πραγαµατος (for πράγµατος), and µενήµ[ης] (for µνήµης); Modern Greek has
καπινός (for καπνός) (GIGNAC 1976, 311–12).
These forms may also be compared to such Tiberian forms as ‫[ ִאמֲרוֹת‬ʔiːmaˈr ̣oːθ] (for
*‫[ ִאמְרוֹת‬ʔimˈr ̣oːθ]; Ps. 12:7) and °Äְ‫[ הֲתִ ֽמ‬haθiːmaˈloːχ] (for °Äְ‫[ הֲתִ מ‬haθimˈloːχ]; Jer. 22:15) (see
KHAN 2013a, 102). In multiple sonority scales suggested for Tiberian, ‫ ר‬and ‫ ל‬are both more
sonorous than ‫ מ‬and ‫( נ‬see DECAEN 2003, 38; ALVESTAD and EDZARD 2009, 49). We may also
compare the transcription αβανηθ ‫ אַ ְבנֵט‬found in Josephus (Antiq. 3:156), απαδανω ‫ַאפַּדְ נוֹ‬
(Dan. 11:45) and σατανηζ ‫שׁ ַע ְטנֵז‬

ַ (Lev. 19:19) in Theodotion, ιεσερι ‫( יִצ ְִרי‬Num. 26:49) and
Νινευη ‫( נִינְוֵה‬Gen. 10:11) in the LXX (cf. Νινυα ‫ נִינְוֵה‬in Josephus and Ninua in Akkadian),
and asamath ‫שׁמַת‬

ְ ‫( ַא‬Am. 8:14) in Jerome.
Third, in a few instances after ‫ ל‬,‫ כ‬,‫ ב‬and the conjunction ‫ו‬, syncope does not occur:
ουµασαννεαϊ /w-m(a)ś anneʔay(y)/ [(ʔ)umasanːɪʔaj] 'and those who hate me' Ps. 18:41
405. The form ‫ ִאמ ְָרה‬does not derive from ‫ ֲאמ ָָרה‬but is a *qitl biform of ‫*( אֹמֶר‬qutl) attested only in the construct
state. A *qitl biform of *qutl is also found in ‫ ִבּסְרוֹ‬for typical ‫ בֹּסֶר‬with a suffix (FOX 2003, 109, 153).
406. The OCP may also be relevant for interpreting the transcription αχαµωθ ‫' ָחכְמוֹת‬wisdom' (Ps. 49:4).
- 341 -
βσεβωθαµ /b-š mō tam(m)/ [b(ɪ)ʃəmoːθa(̃ m)] 'by their names' Ps. 49:12
λσαχηναυ /l-š (a)kē naw/ [l(ɪ)ʃaχeːnaw] 'to his neighbors' Ps. 89:42
According to the rules outlined above, we would expect these forms to be transcribed as
ου(α)µσαννεαϊ**, β(α)σµωθαµ**, and λ(α)σχηναυ**. The explanation for these forms is
unclear, but it is noteworthy (1) that in two out of three instances the second consonant is /š /
and (2) that the syllabic sequence normally prompting syncope occurs across a morpheme
boundary. It may be that the prefixed conjunction or preposition and the subsequent lexeme
were conceived apart from their relationship to one another. Thus, both the affixed word (i.e.,
the conjunction waw or the preposition) and the following word were pronounced as they
were generally. This may be compared to how speakers of Israeli Hebrew generalize the pro-
nunciation of waw as ve- in all contexts, even before initial clusters: e.g., ‫ ֶעשׂ ְִרים וּשְׁתַּ י ִם‬esrı́ m
ve-shtá yim (COFFIN and BOLOZKY 2005, 179–180). Note that similar variation also occurs in
both the Leningrad Codex (L) and the Aleppo Codex (A), reflecting non-standard Tiberian
features: e.g., ü‫ ְבּשְׂפֿתֿ וֹתֶ י‬in L (cf. ü‫שׂ ֿפְתֿ וֹתֶ י‬

ִ ‫ ְבּ‬in A) and ‫ ְכ ְמטִּיל‬in A (cf. ‫ ִכּ ְמטִיל‬in L) (BLAPP 2016).407
Finally, when an inseparable preposition is attached to a word beginning with a gut-
tural (e.g., βααλωθαµ), syncope does not often occur.
Before concluding the section, we must also mention that there are a number of words
in which syncope (CvCvCv(C) > CvCCv(C)) occurs in the Secunda but not in Tiberian:
αλµωθ /ʕalmō t/ [ʕalmoːθ] 'Alamoth' Ps. 46:1

εσδαχ /h ̣esdɔ̄k/ [ħɪsðɔːχ] 'your mercies' Ps. 89:50
αβδαχ /ʕabdā k/ [ʕaβðɔːχ] 'your servants' Ps. 89:51
γαδρωθαυ* /gadrō taw/ [gaðʀoːθaw] 'his walls' Ps. 89:41
407. It is likely, however, that the examples ü‫ ְבּשְׂפֿתֿ וֹתֶ י‬and ‫ ְכ ְמטִּיל‬do not represent a phonetic phenomenon, but
merely a graphical one. They may be compared to instances in L and other Tiberian manuscripts in which a
shewa represents a short vowel in a closed syllable: e.g., ‫[ ַוה ְֲרגְנֻהוּ‬vahaʀaʁˈnuːhuː] (Judg. 16:2) (DOTAN 1985;
KHAN 2013h, 548). However, instances of shewa substituting for pataḥ in forms like ‫ ַוה ְֲרגְנֻהוּ‬should be regarded
as an extension of the use of the shewa sign (ְ ) to indicate short [a], which was the phonetic value of vocalic
shewa in the Tiberian tradition. It is not clear, however, if such a comparison is relevant for ü‫ ְבּשְׂפֿתֿ וֹתֶ י‬, in which
the shewa sign is actually substituting for ḥiriq.
- 342 -
To this list we may add the quotation of the Secunda αρβωθ ‫( ח ֳָרבוֹת‬Ps. 9:7) found in Chrysos-
tom's commentary on the Psalms. YUDITSKY explains these forms by positing two plural
allomorphs, *qVtalı̄ m/*qVtalō t and *qVtlı̄ m/*qVtlō t, as original to Hebrew (2017, 193–94).
In the case of αλµωθ ‫ ֲעלָמוֹת‬, εσδαχ ָ‫ ֲחסָדֶ יך‬, and αβδαχ ü‫ ֲעבָדֶ י‬, however, /a/ insertion in the
plural forms is a feature inherited from Proto-Semitic and attested everywhere in Northwest
Semitic (HUEHNERGARD 1991, 284; 2013); thus, the absence of it can only be explained as the
result of syncope.408 Moreover, the regular plural pattern with /a/ insertion is attested in non-
suffixed non-construct forms more frequently in the Secunda: e.g., νακαµωθ ‫נְקָמוֹת‬, σαβαωθ
‫ ְצבָאוֹת‬. The syllable structure of these words in the Secunda is more similar to that of construct
forms (e.g., ‫ ַעבְדֵ י‬,‫ ַחסְדֵ י‬,‫) ַעלְמוֹת‬.409 We might explain this similarity by positing that vowel syn-
cope occurred in these forms before pretonic lengthening was operative, thus aligning them
with the construct forms, in which pretonic lengthening of the second vowel never occurred
because the word was unstressed. However, it is unlikely that a different pattern of syncope
would occur only in these words. Rather, these forms may be explained as the result of the in-
fluence of Aramaic, in which the regular plural base of *qVtl and *qatil is *qVtl-. Aramaic
forms might also have been more prone to be used when pronominal suffixes were attached
to a noun. Note that two of the forms under discussion (εσδαχ, αβδαχ) have the Aramaic form
of the 2ms suffix for a plural noun. Finally, it is worth noting that αλµωθ may not necessarily
correspond with ‫ ֲעלָמוֹת‬.410
The Secunda form γαδρωθαυ is actually more consistent with the general phonotac-
tics of Hebrew than Tiberian ‫גְּדֵ ר ֹתָ יו‬. Synchronically, a s ̣ere that is not the result of compen-
408. Note the spirantized kaf in the Aramaic plural ‫ ַמ ְלכִין‬, which indicates an originally post-vocalic consonant.
409. YUDITSKY cites the Babylonian construct form trodg' for comparison (YUDITSKY 2017, 193), but aside from
the initial /a/ vowel, which would be expected in Babylonian, there is no difference from the Tiberian construct
form ‫גִּדְ ר ֹת‬. In both traditions, the plural form with an addition exhibits a similar syllable structure: e.g., ‫גְּדֵ ר ֹתָ יו‬
and µyt;rodegwi (place name in Josh. 15:36) (YEIVIN 1985, 921).
410. Note how in the superscription to Psalm 9, the Masoretic phrase ‫ עַלמוּת ַלבֵּן‬is rendered in the Secunda as
Αλµαυθ βεν or Αλµωθ βεν and interpreted as ‫' עַל‬concerning' + ‫' ָמוֶת‬death' in Eusebius (Generalis elementaria
introductio, 75.19–22).
- 343 -
satory lengthening or the collapse of a diphthong is normally reduced to shewa in propretonic
position: e.g., ‫ זְ ֵקנִים‬but ‫זִ ְקנֵיהֶם‬. Diachronically, the s ̣ere in the absolute form ‫ גְּדֵ רוֹת‬is the result
of pretonic lengthening operating on etymological */i/. Thus, instead of the expected form
**‫גִּדְ רוֹתָ יו‬, the Tiberian form ‫ גְּדֵ ר ֹתָ יו‬has been formed based on analogical restoration of the pat-
tern (HUEHNERGARD 2015, 43–44). The Secunda form γαδρωθαυ, on the other hand, exhibits
the expected development of such forms. As YUDITSKY points out, similar variants exist in
Tiberian and Babylonian as well, such as ‫( ְל ַחצְרוֹתָ יו‬Ps. 96:8) over against ‫( ֲחצֵר ֹתָ יו‬Ps. 100:4)
(2017, 193). The fact that both of these developments are attested in multiple traditions indi-
cates that they have ancient roots.

6.5.1.5.2. Long + Short: CvvCvCvv > CvvCCvv
Vowel syncope also occurs frequently in a short open syllable between long vowels:411
µεϊωρδη /mey-yō rdē / (< *yō ridē ) [mɪjːoːʀðeː] 'from those who go down' Ps. 30:4
ταµνου /tɔ̣̄ mnū / (< *tạ ̄ manū ) [tˀɔːmnuː] 'they hid' Ps. 31:5
ασσωµριµ /haš -š ō mrı̄ m/ (< *haš -š ō mirı̄ m) [haʃːoːmʀıː̃ (m)] 'those who keep' Ps. 31:7
ϊωµρου /yō (ʔ)mrū / (< *yō mirū ) [joːmʀuː] 'they will say' Ps. 35:25
ηρφου /h ̣ē rpū / (< *hē ripū ) [ħeːʀɸuː] 'reproached' Ps. 89:52
These transcriptions are important because they demonstrate that CVVC syllables were toler-
ated at least on the post-lexical level. In light of the variation in this pattern (see below), how-
ever, the second consonant was probably licensed as a semisyllable (see KHAN 1987; 2 013b,
who bases his semisyllable analysis on KIPARSKY 2003):

ω
σ σ σ
µ µ µ µ µ µ µ µ
h a š o m r i m
ασσωµριµ = /(haš ).(š oo).m.(rı́i).m/ [haʃːoːmʀiːm]

Figure 29: Moraic Representation of ασσωµριµ
411. For the reduction of vowels in this context, see SUCHARD 2016 (112, 115, 137).
- 344 -
There are a number of exceptions to this rule. First, the plural forms of ‫' אוֹי ֵב‬enemy'
with suffixes do not necessarily exhibit syncope: (ου)οϊεβαϊ ‫( אֹיְבַי‬Ps. 18:38, 41), οϊεββαϊ ‫אֹיְבַי‬
(Ps. 30:2), ωεβη ‫( אֹיְבַי‬Ps. 35:19), and οϊβαχ ü‫( אֹיְבֶי‬Ps. 89:52) (see YUDITSKY 2017, 126,
128).412 It is actually possible, however, that a Hebrew form with syncope lies behind these
transcriptions, yet the transition from /y/ [j] to [β] was perceived by the Greek accent as a
short vowel. On the other hand, a sophisticated linguistic explanation may not be necessary. It
is possible that two patters of the qal plural participle existed side-by-side. The Sephardic
communities, for example, differentiate between the biblical pronunciation š omerı́ m and the
Mishnaic pronunciation š ó mrim (KHAN 2013h, 549). Morever, though a word like ‫שׁמְרוּ‬
ָ was
normally realized as š ɔ̄mrū in Tiberian, early masoretic sources indicate that there may have
been variation from reader to reader, some pronouncing it as š ɔ̄mrū and others as š ɔ̄marū
(KHAN 1987, 56–57).
Second, as above, when syncope would result in a cluster of identical or homorganic
consonants, they are broken up by an epenthetic shewa vowel according to the OCP:
ερωµεµεχ /ʔrō m(ə)mek(k)/ [ʔɪʀoːmɪmɪkh] 'I will exalt you' Ps. 30:2
θσωβαβηνι /tsō b(ə)bē nı̄ / [th(ɪ)soːβaβeːniː] 'you surround me' Ps. 32:7
ισωβαβεννου /ysō b(ə)bennū / [(ʔ)i(ː)soːβaβɪnːuː] '(it) will surround him' Ps. 32:10
A similar phenomenon occurs in Tiberian as well, in which the first of two consecutive she-
was on two identical consonants after a long vowel is realized as vocalic shewa: e.g., ‫ֽוֹר ִרים‬
ְ ‫ס‬
[soːʀaʀiːm] (Isa. 30:1) with shewa, but ‫סוֹר ִרים‬

ֲ [soːʀaʀiːm] (Ps. 68:7) with ḥatef-pataḥ
̣ .
Third, syncope is blocked if the following onset marks a significant rise in sonority:
ιουχαλου* /yū k(a)lū / [juːχaluː] 'they will be able' Ps. 18:39
YUDITSKY interprets the α in this form as indicating the preservation of etymological short */a/
in an open unstressed syllable (2017, 70, 134–35). This is indeed probably the case, since it is
the only instance of etymological */a/ in this environment in the Secunda, yet there may be
412. In the last case, ϊ may be interpreted as representing [ji] or [jɪ].
- 345 -
another factor that facilitated its preservation. As in the case of αναυαθαχ, βααδαρεθ, and
εµαραθ (see above), there is a rise in sonority from the end of the first syllable to the onset of
the second. Therefore, syncope may have been blocked according to the SCL (see above).
This may be compared to certain Tiberian forms: e.g., ‫[ תּ ֹא ֲכלֶנּוּ‬toːχaˈlεːnːuː] 'you will eat it'
(Num. 18:10), but cf. ‫[ תּ ֹא ְכלֶנּוּ‬toːχaˈlεːnːuː] (Deut. 12:24); ‫[ הֽוּטֲלוּ‬huːtaˈluː]

̣ (Jer. 22:28).
Finally, if the vowel expected to undergo syncope is next to a pharyngeal, syncope
does not occur: e.g., λωαµαϊ ‫ ֲחמַי‬Ä [Ps. 35:1]; αββωτεειµ ‫[ הַבּ ֹ ְטחִים‬Ps. 49:7]).
6.5.1.6. The Conjunction waw and the Inseparable Prepositions
Because a vowel is not typically transcribed after the conjunction waw or the inseparable
prepositions, they should be interpreted as shewa and represented phonemically as /w-/, /b-/,
/k-/, and /l-/. Their precise phonetic realizations and relationship to syllable structure will be
examined below.
6.5.1.6.1. Conjunction waw: ου = /u/
In the Secunda, the conjunction waw /w-/ is usually represented only by ου (99x):
ουλω /w-lō (ʔ)/ [(ʔ)uloː] 'and not' Ps. 18:38

ουγιλου /w-gı̄ lū / [(ʔ)uʁiːluː] 'and rejoice!' Ps. 32:11
ουβαρουχ /w-bɔ̄rū k/ [(ʔ)uβɔːʀuːχ] 'and blessed' Ps. 18:47
Aside from those instances in which it is prefixed to a vayyiqtol past-tense form (e.g.,
ουαϊαλεζ ‫ז‬Äֲ‫[ ַויַּע‬Ps. 28:7]; ουαθθεµας ‫[ וַתִּ מְאַס‬Ps. 89:39]), the conjunction ου /w-/ is only fol-
lowed by a vowel in eight instances, which can essentially be categorized into two groups
(see YUDITSKY 2017, 230–32).413 First, ου is transcribed with a vowel when /w-/ is followed
by a word with an initial consonant cluster:
ουαρηµ /w-rʕē m/ [waʀʕeːm] 'and shepherd them!' Ps. 28:9

ουαλσωνι /w-lš ō nı̄ / [walʃoːniː] 'and my tongue' Ps. 35:28
ουαδου /w-dʕū / [waðʕuː] 'and know!' Ps. 46:11
ουεβροβ /w-b-rob(b)/ [wɪβʀʊb]/[wəβʀʊb] 'and in the abundance of' Ps. 49:7
413. YUDITSKY concludes that the vowel in wa-, while typically not transcribed, is indicated in pre-tonic position
in pause and before a consonant cluster (2017, 231). My analysis of the data leads to similar conclusions.
- 346 -
This reflects the syncope rule v > Ø / Cv̆ C_Cv̆ (C) outlined in 6.5.1.5.1. While it is normal for
the vowel preceding the cluster to be /a/ (e.g., λαµσω [< *lamas ̣ō (ˀ), δαβρη [< *dabarē ]), the
ε in ουεβροβ may reflect a resolution of the initial cluster similar to Babylonian Hebrew, in
which waw before a cluster is usually pointed with ḥiriq (e.g., µwOybwI). There is variation, how-
ever, even in Babylonian, with examples of waw + pataḥ before a cluster in the Safra (e.g.,
rwOçyImbw') (YEIVIN 1985, 1152). It is also possible that the ε reflects a centralized epenthetic [ə].
Second, ου has a vowel when it precedes a stressed syllable in monosyllabic nouns:
ουαρεσν /wɔ̄-resn/ [wɔːʀɪsn] 'and a bridle' Ps. 32:9

ουαδωρ /wɔ̄-dō r/ [wɔːðoːʀ] 'and generation' Ps. 49:12
In each case, the noun is the second item in a fixed phrase and would have a disjunctive ac-
cent in Tiberian (βαµεθγε ουαρεσν ‫' ְבּמֶתֶ ג ו ֶָרסֶן‬with bit and bridle'; λδωρ* ουαδωρ ‫' לְד ֹר וָד ֹר‬to
generation and generation'). In these sorts of contexts, the conjunction waw exhibits pretonic
lengthening in both Tiberian (‫ ) ֶלחֶם ָויַי ִן‬and Babylonian Hebrew (e.g., ˜yy'w: µjl) (YEIVIN 1985,
1154–55; for pretonic lengthening of waw, see KHAN 2013j, 228). Accordingly, the vowel
should be transcribed as long (on /w-/ + vowel, see YUDITSKY 2017, 230–32).
The only other two potential examples of the conjunction waw with a vowel that are
not in a wayyiqtol context are ουειεσοµου ‫שׂמְחוּ‬

ְ ִ ‫( ְוי‬Ps. 35:27) and ουααρηεµ* ‫' וְאַח ֲֵריהֶם‬and af-
ter them' (Ps. 49:14). In the first example (ουειεσοµου), the preceding word ends in a long
unstressed /ū / (ιαροννου ουειεσοµου ‫שׂמְחוּ‬

ְ ִ ‫ )י ָר ֹנּוּ ְוי‬and thus the conjunction /w-/ might not have
been realized phonetically if it was not consonantal (cf. the distribution of /w-/ as [u-] and
[w-] in Syrian Arabic below). In the second example (ουααρηεµ*), it is not clear if the
transcription should be read as ου + ααρηεµ or ουα + αρηεµ (see YUDITSKY 2017, 215, 230).
Because /w-/ is nowhere else transcribed with a vowel when preceding a guttural (e.g.,
ουαδαµ ‫[ וְאָדָ ם‬Ps. 49:13]; ουαθ ‫[ ְואַתָּ ה‬Ps. 89:39]), the first interpretation should be favored.
In sum, the conjunction waw /w-/ (excluding instances when it is prefixed to a verbal
form in wayyiqtol) is transcribed as ου unless it precedes a consonant cluster or is lengthened
- 347 -
historically before the stress in fixed phrases. Attestations of the conjunction waw both in ex-
ternal quotations of the Secunda and in Jerome also support this distribution. In external
sources, it is transcribed without a vowel generally (e.g., ουοµρ ‫[ וְחֹמֶר‬Hos. 3:2]; ουαγιθ ‫ְו ָהגִית‬
[Ps. 49:4]), with a vowel in some instances of wayyiqtol (e.g., ουαεει ‫[ ַויְחִי‬Gen. 5:3]; ουαθετ
‫[ וַתֵּ ט‬Ps. 44:19]), and with a vowel once before a monosyllabic noun potentially in a fixed
phrase (ουαθεσα ‫[ וּתְ שַׁע‬Gen. 5]).414 A seldom-quoted Latin transcription of Genesis 14:18–20
in Jerome also exhibits a similar, and fairly consistent, distribution:

umelchisedech melech salem hosi lehem uaiain, uhu cohen lehel helion:
uaibarcheu uaiomer baruch abram lehel helion: cone samaim uares: ubaruch
hel helion eser maggen sarach biadach uaiethen lo maaser mecchol.415
In the passage above, the conjunction /w-/ is realized as u in general (umelchisedech, uhu,
uares, ubaruch), as ua regularly in wayyiqtol forms (uaibarcheu, uaiomer, uaiethen), and as
ua once before the second part of a pair in a fixed phrase (lehem uaiain). While it is always
possible that such transcriptions are the result of later emendation, a strong case for the au-
thenticity of this passage can be made on the basis of a number of features that are character-
istic of both contemporary Hebrew transcriptions and specifically Jerome.416
Aside from special cases in which /w-/ precedes a consonant cluster or a stressed syl-
lable, then, the conjunction is regularly transcribed as ου with no additional vocalic grapheme
in the Secunda. While ου- could represent a number of different phonetic realizations (e.g.,
[w-], [wa-], [wə-], [wu-], [(ʔ)u(ː)]), the essential question is whether or not ου- represents the
414. The transcription ουαθεσα is found in construct in the following phrase in Epiphanius (Mensuris et
Ponderibus, 22–23): ουαεει Αδαµ σαλωειµ σανα ουαθεσα µηωθ σανα, presumably parallel to ‫שׁנָה‬ ָ ‫שִׁים‬Äְ‫ַויְחִי אָדָ ם שׁ‬
‫שׁנָה‬
ָ ‫וּתְ שַׁע מֵאוֹת‬. The problem, however, is that Epiphanius's text seems to have conflated Gen. 5:3 and Gen. 5:8.
Moreover, there are variant readings of ουαθεσα that do not have a vowel. Nevertheless, even though it is in
construct, because it is the second half of a number pair (e.g., ‫שִׁים ָושֵׁשׁ‬Äְ‫)שׁ‬, it could be conceived of as a fixed
phrase and thus the preceding waw could undergo lengthening.
415. Gen. 14:18–20: ‫ש ַמי ִם‬ ׁ ָ ‫ אַב ְָרם ְלאֵל ֶעלְיוֹן קֹנֵה‬°‫שׁלֵם הוֹצִיא ֶלחֶם ָויָי ִן וְהוּא כֹהֵן ְלאֵל ֶעלְיוֹן׃ ַויְב ְָרכֵהוּ וַי ֹּאמַר ָבּרוּ‬
ָ °ֶ‫וּ ַמ ְל ִכּי־צֶדֶ ק ֶמל‬
‫שׂר מִכ ֹּל‬
ֵ ‫ ַויִּתֶ ּן־לוֹ ַמ ֲע‬ü ֶ‫ ְבּי ָד‬ü‫שׁר־ ִמגֵּן צ ֶָרי‬
ֶ ‫ אֵל ֶעלְיוֹן ֲא‬°‫ָאָרץ׃ וּבָרוּ‬
ֶ ‫ו‬
416. First, the gutturals are often transcribed by h (e.g., lehel, helion, hel). Second, the 2ms suffix on both plural
and singular nouns is -ach (sarach, biadach). Third, the relative particle is eser (cf. εσερ). Fourth, the piˁel form
of ‫מגן‬, contra the MT, resembles an Aramaic paˁel form (maggen). Fifth, gemination is not always represented
(e.g., uaiethen; cf. ϊεθεν). Sixth, the r in the piˁel stem is degeminated, resulting in syncope of the following
vowel (uaibarchehu; cf. ηρφου).
- 348 -
consonant [w] or the vowel [u]. The evidence of Greek transcription of other languages fa-
vors the latter interpretation, which also finds parallels in modern Semitic languages.
In other Greek transcription traditions, a short vowel following word-initial [w] is
almost always represented in transcription. In transcription of Latin, words beginning with
consonantal v followed by a short vowel in an unstressed open syllable are represented by ου
+ vowel: e.g., ουοκ[ατιω] vŏ cā tı̆ ō (1st CE, P.Berol. 21246) and ουοκαβουλωρουµ vŏ cā bŭ lō rŭ m
(2nd CE, P.Oxy. XLIX.3452).417 In Greek transcription of Arabic, the conjunction waw is repre-
sented with ου/ω/ο + vowel both in the third-century CE inscription from north-eastern Jordan
(e.g., αουα ειραυ /wa-yirʕaw/, ωα βαναα /wa-Bannā ʔa/) (AL-JALLAD 2015b, 52–53) and in the
Damascus Psalm fragment (e.g., οα ρυγζ /wa-rugz/) (AL-JALLAD forthcoming, 32, 118). This
evidence from transcription of Arabic, in which the conjunction was presumably pronounced
as [wa-], seems to indicate that it was actually the transition from a back rounded vowel ([u],
[o]) to another vowel (or vice versa) that approximated [w] and not the digraph ου itself (see
6.3.7.1). This is especially clear from the presence of an initial α in the transcription αουα
[wa]. Had ου by itself been sufficient to indicate [w], there would have been no need for the
preceding α. In lieu of an adjacent vowel, Greek ου would merely represent [u] as in contem-
porary Greek orthography (4.5.3.1.17). There is one possible exception in Greek transcription
of Phoenician-Punic, in which the singular occurrence of the conjunction waw is transcribed
by ου: ου λυ ρυβαθων /w-l-rabbaton/ 'and to our lady'. However, since Latin transcriptions of
Punic often render the conjunction /w-/ with only u (FRIEDRICH and RÖLLIG 1999, 185), this
may reflect an internal-Phoenician sound change of wa- > w(u)- > u-. We may assume, then,
that /w-/ in the Secunda was generally realized vocalically as [u], but, before a cluster, was
417. Note, however, two exceptions: Latin Vespasianus and veteranus are usually transcribed as Οὐεσπασιανός
and οὐετρανός, but there are variants of each in the Egyptian papyri without a vowel following ου: e.g.,
Οὐσπασανός (3rd CE) and οὐτρανοῦ (261 CE) (GIGNAC 1976, 305–6, 356). However, in light of the abundance of
examples of ου + vowel (with various vowels), these exceptions may actually reflect an isolated development in
substandard Egyptian Latin ([we-] > [wu-] > [u]) rather than a common variant orthography.
- 349 -
prone to be realized consonantally as [wa-]. Instances of pre-tonic lengthening (e.g.,
ουαρεσν) are distinct in that the following long vowel was phonemic: /wɔ̄resn/.
Such a hypothesis finds parallels in both modern Arabic dialects and Neo-Aramaic di-
alects. In Palestinian Arabic, for example, roughly the same distribution obtains for the con-
junction /w-/ (< */wa-/), which is realized as [u-] before #V and #CV, but as [wə-]/[wɛ-] be-
fore #CCV: e.g., u-ana qā ˁed 'and I (was) sitting', il-yō m u-bukrɑ 'today and tomorrow', but
bē t ͜ mrattab w-enḍı̄ f 'a clean and tidy house' (ELIHAY 2012, 611, 614, 772). In Syrian Arabic,
e
/w-/ is normally realized as [u-] between a word ending in a consonant and one beginning
with a single consonant, but as [w-] everywhere else: e.g., təffā ḥ u-mō z 'apples and bananas',
but w-mō z kamā n 'and bananas too' and ʔalam w-əktā b 'a pencil and a book'. COWELL does
note, however, that there is a good deal of variation between u- and w-/w-ə among speakers
due to the phonetic similarity between the two. Moreover, this distribution only applies in
"close phrasing" (2005, 21, 392). Finally, in Iraqi Arabic, the quality of initial /w-/ is more
like that of a vowel (i.e., [uː]) than a consonant (ERWIN 2004, 31, 307).
In the Neo-Aramaic dialect of the Jews of Arbel, the conjunction w- typically shifts to
ʔu- before a word beginning with a consonant, reflecting assimilation to an epenthetic (i.e.,
wi > wu > ʔu), though wı̆ - is occasionally maintained: e.g., ʔu-gozè 'and nuts' but wı̆ -bā bè u
'and his father' (KHAN 1999, 65). The same realization of the conjunction /w-/ is attested in
the Neo-Aramaic dialect of Qaraqosh (e.g., ˀu-mə́nhə 'and from them') and in the dialect of
Sulemaniyya and Ḥalabja (e.g., ˀu-xadrı́ wa 'and they became') (KHAN 2002, 65; 2004, 59).
6.5.1.6.2. Inseparable Prepositions: β, χ, λ
As a default convention, the prepositions /b-/, /k-/, and /l-/ are transcribed without a vowel:
βδαµι /b-dɔ̄mı̄ / [b(a)ðɔːmiː] 'in my blood' Ps. 30:10

χφαρδ* /k-pard/ [kh(a)ɸaʀd] 'like a mule' Ps. 32:9
λδωρ* /l-dō r/ [l(a)ðoːʀ] 'for generation' Ps. 49:12
- 350 -
The clearest indication that the default form of the inseparable prepositions was vowelless is
the fact that, when prefixed to the tetragrammaton, no vowel is represented (e.g., λ‫[ יהוה‬Ps.
30:5]). There are, however, a number of environments in which a vowel may be represented.
First, as a rule, with a few explicable exceptions, a vowel is transcribed preceding a
guttural. The vowel of the preposition typically assimilates to that of the following guttural:418
βααφφω /b-ʔappō / [baʔaphːoː] 'in his anger' Ps. 30:6

χεεβλ /k-ʔebl/ [khɪʔɪβl] 'like the mourning of' Ps. 35:14
βααµ /b-ʕam(m)/ [baʕa(̃ m)] 'in/among a people' Ps. 35:18
βηηκι /b-h ̣ē qı̄ / [beħeːkˀiː] 'in my breast' Ps. 89:51
Aside from the fact that vocalic shewa often assimilates to the vowel following a guttural
(6.5.1.3.2), this may be merely an orthographic phenomenon. Since gutturals are only explic-
itly indicated by hiatus, failing to transcribe a vowel after the preposition could lead to in-
correct syllabification of the word (see KNOBLOCH 1995, 219–224). For example, while a
transcription like βααµ is clearly to be parsed as βα + αµ 'in/among' + 'a people', a spelling
like βαµ would actually be identical to the transcription βαµ ‫' בָם‬in them' (Ps. 49:15).419At the
same time, however, it is entirely possible that the gutturals actually occasioned the insertion
of an epenthetic (cf. Neo-Aramaic below). As YUDITSKY points out, exceptions occur when /ʔ/
has elided (e.g., χαϊαλωθ [khajːɔːloːθ] ‫[ ְכּ ַאיּ ָלוֹת‬Ps. 18:34], λαηριµ [laːħeːʀiːm] ‫[ ַל ֲאח ִֵרים‬Ps.
49:11], βαµουναθι [baːmuːnɔːθiː] ‫[ ֶבּאֱמוּנָתִ י‬Ps. 89:34]), when the guttural, being between back
rounded vowels, is not conspicuous (e.g., λωλαµ [loʕoːlɔːm] ‫[ לְעוֹלָם‬Ps. 30:7]), and due to the
syncope rule (see 6.5.1.5.2) (e.g., ουβελωαϊ [ʔuːβʔɪloːhaj] ‫הַי‬Ä‫[ וּבֵא‬Ps. 18:30]) (2017, 225).420
418. YUDITSKY also notes the presence of a vowel before gutturals and numerous instances of assimilation
before gutturals (2017, 228, 231). For his treatment of /b-/, /k-/, /l-/, see YUDITSKY (2017, 224–29).
419. We could also cite a number of other transcriptional minimal pairs: e.g., χααφαρ ‫' ְכּ ָעפָר‬like dust' vs.
hypothetical **χαφαρ ‫' ָכּפַר‬he covered', βααλωθαµ ‫' ַבּחֲלוֹתָ ם‬in their sickness' vs. hypothetical **βαλωθαµ ‫בַּלּוֹתָ ם‬
'wearing them out', λααβδ ‫' ְל ֶעבֶד‬to the servant of' vs. λαβδ 'Lamed' (see 6.5.2.3), and χεεβλ ‫' ְכּ ֵאבֶל‬like mounring
of' vs. hypothetical **χεβλ ‫' ֶכּבֶל‬fetter'.
420. Note, however, that YUDITSKY does not cite βαµουναθι as an example, since he reads it as βαεµουναθι. The
correct reading of the mansucript, however, is βαµουναθι. In light of the transcription βαεµουναθαχ ü ֶ‫( ֶבּאֱמוּנָת‬Ps.
- 351 -
Second, a vowel is indicated when the preposition precedes an initial clsuter:
λαβλωµ /l-blō m/ [laβloːm] 'to curb' Ps. 32:9

βαλβαβαµ /b-lbɔ̄am(m)/ [balβɔːβa(̃ m)] 'in their heart' Ps. 35:25
χαβηµωθ /k-bhē mō t/ [khaβheːmoːθ] 'like cattle' Ps. 49:13
This reflects the syncope rule v > Ø / Cv̆ C_Cv̆ (C) outlined in 6.5.1.5.1. This phenomenon is
characteristic of both Tiberian and Babylonian, in which the epenthetic is usually ḥiriq. Be-
fore /h ̣/ in Babylonian, however, the preposition has pataḥ: e.g., twbr;j]l' (YEIVIN 1985, 1151).
Third, when the initial consonant of the following word is homorganic or has a similar
articulation to that of the preposition, a vowel is often transcribed:

λανεγδ /l-negd/ [lanɪʁd] 'in front of' Ps. 36:2
ουβοµωτ /w-b-mō t/̣ [(ʔ)uβomoːtˀ] 'and in the shaking of' Ps. 46:3
βαφιεµ /b-pı̄ hem(m)/ [baɸiːhɪ(̃ m)] 'with their mouth' Ps. 49:14
λαµαλχη /l-malkē / [lamalχeː] 'to the kings of' Ps. 89:28
This phenomenon accounts for a substantial number of transcriptions with a vowel after /b-/,
/k-/, /l-/ before a non-guttural consonant. It is best explained according to the OCP and sup-
ports the notion that vocalic shewa was an epenthetic vowel inserted on the post-lexical level.
Such an interpretation is especially favored by the third transcription in the list (ουβοµωτ),
since a preposition after the conjunction /w-/ typically closes a syllable in the Secunda in
such environments421 according to the syncope rule (6.5.1.5.2). It is not clear how many
transcriptions can be explained in this way, but we might also add initial sequences such as
λασωλ ‫( ִלשְׁאוֹל‬alveodental nasal + alveodental fricative in Greek) (Ps. 49:15) and λεριβι*
‫( ל ְִריבִי‬alveodental liquid lateral + alveodental liquid trill in Greek) (Ps. 35:23).
89:50), this demonstrates internal variation.
421. See also ουβελωαϊ ‫הַי‬Ä‫( וּבֵא‬Ps. 18:30), ουλµαν ‫( וּ ְל ַמעַן‬Ps. 31:4), ουβσαλη ‫( וּ ְב ַצ ְלעִי‬Ps. 35:15), ουεβροβ ‫וּבְר ֹב‬
(Ps. 49:7), and ουβµεσφατι ‫שׁ ָפּטִי‬ ְ ‫( וּ ְב ִמ‬Ps. 89:31). However, when at an underlying phonological level a word
begins with CCC, the epenthetic is inserted between the last two consonants: ουβαναρωθ ‫( וּ ַבנְּהָרוֹת‬Ps. 89:26) and
ουβανγαϊµ ‫( וּ ִבנְגָעִים‬Ps. 89:33).
- 352 -
While these three categories (before gutturals, syncope rule, homorganic consonants)
account for most of the data, the remaining instances of a vowel indicated after a preposition
can be grouped roughly into four categories: when it precedes a monosyllabic word (e.g.,
λαχολ ‫[ לְכ ֹל‬Ps. 18:31]), when it precedes yod (e.g., βιαδαχ ü ְ‫[ ְבּי ָד‬Ps. 31:6]), when its quality is
other than /a/ due to assimilation to the following consonant (e.g., βεσοχχα ‫[ ְבּ ֻסכָּה‬Ps. 31:21]),
and when the vowel of the following consonant is /a/422 (e.g., βακααλ ‫[ ְבּ ָקהָל‬Ps. 35:18]).
In sum, what initially seems to be an inconsistent transcription system can actually be
explained to a large degree by two simple principles. As the default realization, the insepara-
ble prepositions are represented without a vowel. Before gutturals, a vowel is indicated to
prevent mispronunciation and preserve syllable structure. Other than these two principles,
most of the data can be explained by the same sort of principles that operate elsewhere in the
Secunda (syncope rule, the OCP, assimilation, etc.). The fact that when a vowel is indicated,
it is typically represented by α, probably points to the preservation of historical */a/ as the
phonetic realization ([a]) of vocalic shewa in these instances (6.5.1.3.3).
As a final observation, it is worth noting how the behavior of the inseparable preposi-
tions has parallels in living Semitic languages. For example, in the Neo-Aramaic dialect of
the Jews of Arbel, the preposition /b-/ is normally, but not always, prefixed to the noun with-
out an epenthetic: e.g., b-lelè 'in the night'. Before pharyngeals, /b-/ is followed by a short
epenthetic [a]: e.g., ba-ʕIrā ̀ q 'in Iraq'. Before laryngeals, /b-/ is followed by a short epenthetic
assimilated in quality to the following vowel: e.g., bĕ -ʔelà 'during the festival'. Sometimes an
initial /ʔ/ may be elided after /b-/: e.g., b-immá 'in a hundred' (< *ʔimma). Finally, before a
consonant cluster, an epenthetic intervenes between /b-/ and the cluster: e.g., badé niwa bi-
xlulá 'they began the wedding' (KHAN 1999, 186–87).
422. This final categorization (i.e., when the following vowel is /a/) is quite possibly merely a statistical
coincidence, but it is worth considering that a following /a/ vowel may be more prone to encourage the
preservation of a historical /a/ vowel in light of the fact that vocalic shewa is prone to assimilate.
- 353 -
6.5.2. Final Consonant Clusters
Up until this point, our analysis has only dealt with various features of word-initial and word-
medial consonant clusters. This is because consonant clusters behave similarly at the begin-
ning of a word and in the middle of a word. In the present section, we will analyze the phono-
tactics of final consonant clusters by primarily focusing on the segholate nominal pattern.
6.5.2.1. Segholates (*qVtl)
In the Secunda, the nominal pattern *qVtl is usually transcribed without an epenthetic vowel:
κοδς /qodš / [kˀʊðʃ] 'holiness' Ps. 29:2

σεθρ /setr/ [sɪθʀ] 'a hiding place' Ps. 32:7
αρς /ʔars ̣/ [ʔaʀʦˀ] 'land' Ps. 35:20
The main question regarding these forms is diachronic. It is not entirely clear why the typo-
logically more archaic forms without epenthesis (i.e., qVtl) are characteristic of the Secunda,
while the more innovative forms with epenthesis (i.e., qVtvl) are characteristic of the LXX,
which was composed a few centuries prior to the Secunda: e.g., Τοφολ ‫טֹפֶל‬, Σερεδ ‫ס ֶֶרד‬, ɸαρες
‫( פּ ֶֶרץ‬see KNOBLOCH 1995, 193). After a brief review of scholarship, it will be argued that the
lack of representation of an epenthetic in the Secunda indicates a phonetic reality and thus a
higher toleration for final consonant clusters than in other traditions of Hebrew.
The disparity in the representation of segholates between the Secunda and the LXX is gener-
ally explained in one of two ways. The first explanation argues that the forms in the Secunda
and the forms in the LXX reflect different orthographic conventions for representing the
same phonetic reality. According to BLAU, when final short vowels were elided in Hebrew,
epenthesis immediately arose in a portion of the segholates due to phonetic constraint. Never-
theless, although epenthesis resulted in a bisyllabic phonetic realization of (certain) segho-
lates, they remained monosyllabic morphophonemically. The LXX transcriptions, then, re-
flect a phonetic realization, whereas the Secunda transcriptions reflect a morphophonemic
realization (1998b, 102–103). That is, both Κορε (LXX) and κορ (Secunda) (MT ‫[ ק ַֹרח‬Ps.
- 354 -
45:1]) represent the same phonetic reality, albeit by means of a different transcription con-
vention. The second explanation contends that the transcriptions in the Secunda reflect a dif-
ferent dialect (or register) in which final consonant clusters were tolerated; thus, the differ-
ence in transcription does indeed reflect a different phonetic reality (YUDITSKY 2017, 178).
There are two problems with BLAU's argument. First, although it is almost certainly
correct that the segholates were monosyllabic at an underlying phonological level (see also
MALONE 1971; 1993; GREENSTEIN 1992; COETZEE 1999; KHAN 2013l), it is not necessarily true
that a portion of *qVtl forms would require epenthesis as soon as final short vowels were
elided (see BLAU 1998b, 102). In modern Arabic dialects, for example, in which final short
vowels have also elided, the phonotactics of final consonant clusters vary from dialect to di-
alect. Some dialects tolerate final -CC# clusters without restriction (e.g., Moroccan), others
insert an epenthetic vowel according to the Sonority Sequencing Principle (SSP) (e.g., most
Levantine dialects), and still others resolve all final -CC# clusters with an epenthetic (e.g.,
Baghdad Christian Arabic) (KIPARSKY 2003, 149; WATSON 2007, 339). Moreover, in the Neo-
Aramaic dialect spoken by the Jews of Arbel, the same words are sometimes realized with an
epenthetic and sometimes without: e.g., ḥà wiš || ḥà wš 'enclosure' and waxit || wá xt 'time'
(KHAN 1999, 67–68). It would not only be possible, then, but actually expected, to find simi-
lar variation in the ancient dialects of Hebrew.423
The second problem with BLAU's argument is that it ignores the fact that an epenthetic
vowel is inserted in not a small number of forms in the Secunda, most of which can be ex-
plained phonetically (see below). According to YUDITSKY, the fact that epenthetics tend to
423. Such variation may actually be attested in cuneiform transcriptions of Hebrew names from the first
millennium BCE. On one hand, -CC./-CC# consonant clusters seem to be tolerated (note that a final -u in
Akkadian is merely an orthographic necessity and does not represent a final short vowel in Hebrew): e.g., ab-di-
mil-ku for °ֶ‫ ֶעבֶד־ ֶמל‬and pa-la-as-tú for ‫ ְפּ ֶלשֶׁת‬. On the other hand, sometimes -CC. is broken up by an epenthetic:
e.g., s ̣i-di-iq-yá -ma ‫( צִדְ ִקיּ ָה‬MILLARD 2013, 839, 842, 844).
- 355 -
arise in the environment of gutturals in *qVtl forms indicates that the lack of a transcribed
epenthetic in other forms must represent a phonetic reality (2013, 817).

6.5.2.1.2. Conditions for Epenthesis in *qVtl in the Secunda
Α number of trends can be identified regarding the insertion of an epenthetic in *qVtl forms
in the Secunda. Epenthetics tend to be present in III-/ʕ/ roots, absent in II-/ʕ/ and III-/h ̣/ roots,
and intermittent in II-/h ̣/ roots (cf. YUDITSKY 2017, 30–31, 77–79, 180–81, 185, 213–14):424
C_Cʕ# C_ḥC# C_ʕC# C_Cḥ#
ø > v: ρεγε ‫ֶרגַע‬ φααδ ‫ַפּחַד‬ ουβααρ ‫ָו ַבעַר‬
(see below)
µεββεσε ‫מַה־ ֶבּצַע‬ ασσααθ/ ‫שּׁחַת‬
ַ (ַ‫)ה‬
σααθ
φεσα ‫ֶפּשַׁע‬ ιααδ/ε ‫יַחַד‬
µεσσααρ ‫שּׁחַר‬
ַ ‫ִמ‬
ø > ø: βσακ ‫שּׁחַק‬

ַ ‫ַבּ‬ βαρ ‫ַבּעַר‬ κορ ‫ק ַֹרח‬
µηρεµ ‫מ ֵֶרחֶם‬ βαχας ‫ְבּ ַכעַס‬ λανες ‫ָלנֶצַח‬
θεθ ‫תַּ חַת‬ λαµαν/ ‫)וּ( ְל ַמעַן‬
ουλµαν
ουλακεθ ‫ְו ָל ַקחַת‬ νερ ‫נַעַר‬
βαβαλ ‫ַבּ ַבּעַל‬
Chart 27: Epenthesis in Segholate (*qVtl) Forms in the Secunda
The distribution of these forms may indicate that in the Hebrew tradition reflected in the Se-
cunda, /ʕ/ was a more sonorous consonant with weaker consonantality and /h ̣/ was a less
sonorous consonant with stronger consonantality. This hypothesis is consistent with the fact
that /h ̣/ is susceptible to (virtual) doubling in the Secunda (e.g., σεωθι ‫[ שַׁחוֹתִ י‬see KHAN 2013d,
502–503]), whereas /ʕ/ is not (e.g., µηεµµω* ‫) ֵמעִמּוֹ‬. Although these are only trends and not
rules, it would be difficult to argue that phonetic conditioning was not a significant factor in
determining whether an epenthetic was transcribed.
Other instances of epenthesis in *qVtl forms in non-guttural environments further sup-
port the significance of sonority in determining epenthesis. For example, according to the
424. While YUDITSKY recognizes the difference in behavior between III-/ʕ/ and III-/h ̣/ roots, he does not point
out the fact that there is also a marked difference in the distribution of epentehtics between II-/ʕ/ and II-/h ̣/ roots.
- 356 -
SSP, a significant rise in sonority from C2 to C3 seems to occasion the insertion of an
epenthetic vowel in the following transcription (6.3.5.2) (note perhaps also ουβααρ above):425
Finally, there is one case of paragogic epenthesis in the *qVtl pattern in the Secunda:
The nature of epenthesis in βαµεθγε cannot be accounted for on the basis of BLAU's theory,
which assumes a regular untranscribed epenthetic in the segholates in the Secunda. Rather,
the insertion of a word-final epenthetic can only be explained by assuming a /-tg#/ cluster.
6.5.2.2. Verbal and Participial Forms
A survey of the phonotactics of final -CC# clusters in verbal and participial forms also
demonstrates the significance of phonetic factors in determining epenthesis. In general, final
/-Ct#/ in the suffix conjugation is preserved without an epenthetic:
αφαχθ /hɔ̄pakt/ [hɔːɸaχth] 'you turned' Ps. 30:12

ελλελθ /h ̣ellelt/ [ħɪlːɪlth] 'you profaned' Ps. 89:40
εκσερθ /heqs ̣ert/ [hɪkˀʦˀɪʀth] 'you cut short' Ps. 89:46
Because /t/ is lowest on the Hebrew sonority scale and paradigmatic pressure would encour-
age a consistent shape throughout the verbal suffix conjugation, it is not surprising that there
is generally no epenthesis in these forms.
However, the OCP seems to explain a couple instances of epenthesis in the system. In
III-/d/ roots, for example, epenthesis may occur between the final radical and the suffix /-t/:426
εεµεδεθ /heʕmedt/ [hɪʕɪmɪðɪθ] 'you established' Ps. 30:8
This principle is also present in a transcription of the feminine nifˁal participle:
νεεµαναθ /neʔmant/ [nɪʔɪmanaθ] 'is faithful' Ps. 89:29
Though the final cluster /-nt#/ is permitted according to the SSP, the alveolar realization of
both /n/ and /t/ might have occasioned the epenthetic. That epenthesis was not regular in par-
425. But cf. σεθρ ‫( סֵתֶ ר‬Ps. 32:7).
426. But cf. ουφα<κα>δθι ‫( וּ ָפקַדְ תִּ י‬Ps. 89:33). Note also the discussion in 6.3.1.3.
- 357 -
ticipial forms with similar sonority sequencing is demonstrated by the transcription Κωελθ
‫( ק ֹ ֶהלֶת‬Qoh. 1:1) found in Origen's list of biblical books. Finally, it should be noted that in fi-
nal CC# forms resolved by an epenthetic, the quality of the vowel is typically identical to that
of the preceding vowel (see YUDITSKY 2017, 79–80), as is especially clear from νεεµαναθ.
6.5.2.3. Concluding Remarks: Dialectal Variation
The fact that most instances of epenthesis in the Secunda can be explained on the basis of
phonetic principles further supports our hypothesis that what distinguishes the segholates in
the Secunda from the segholates in the LXX is not transcription convention but dialect. Ac-
cordingly, we must assume that both qVtl# and qVtvl# existed side-by-side at least from the
time of the LXX onwards, differing from dialect to dialect and from phonetic environment to
phonetic environment within a particular dialect.
While it is difficult to isolate pairs of cognate transcriptions of the segholates across
various transcription traditions, a potential illustration of ancient dialectal variation is found
in the transcription of the Hebrew letters, five of which derive from *qVtl patterns: *ʔalp,
*giml, *dalt/*dilt, *lamd, and *samk. These letters are found transcribed in lists of the names
of the letters among the church fathers (Eusebius and Epiphanius) as well as in the sectional
headings of the acrostic Psalm 119 in the LXX (Catenae and P. Codex Oriental 5000) (clus-
ters in red and epenthetics in green):

Eusebius (4th c.) Epiphanius (4th c.) Catenae (5th c.) LXX MS: P. Codex Oriental 5000
(6th/7th c.)
*ʔalp αλφ αλεφ αλφ αλφ
*giml γιµελ γιµηλ γιµλ γιµαλ
*dalt δελθ δελεδ δαλεθ δελθ
*lamd λαβδ λαµεδ λαµδ λαβδ
*samk σαµχ σαµεχ σαµχ σαµχ

Chart 28: Various Greek Transcriptions of the Segholate Letter Names
Transmission history no doubt plays a significant role in bringing about variants, but the sort
of differences exemplified above probably have origins in real linguistic variation. For exam-
- 358 -
ple, epenthesis most commonly occurs in *giml, the only final cluster with rising sonority. In
light of the comparison with the Arabic dialects above, we might suggest that the dialect tra-
dition underlying Catenae generally tolerated -CC# (assuming δαλεθ is from a different
source), that of Epiphanius did not tolerate -CC#, and those of Eusebius and P. Codex Orien-
tal 5000 tolerated -CC# according to the SSP. Such a description is obviously an oversimplifi-
cation of a complicated web of linguistic development, textual transmission, and correction,
but the principle is likely valid. In ancient times, there were probably a number of different
dialects of Hebrew, each with its own particular phonotactics for dealing with final -CC#. It is
also worth noting that the analogy to modern Arabic and Neo-Aramaic dialects suggests vari-
ation even within the same phonetic environments in the same dialect.
- 359 -
7. CONCLUSION
This dissertation has demonstrated that, in order to best understand the phonology and or-
thography of the second column of Origen's Hexapla, it is necessary to approach it contextu-
ally within the Hellenistic/Roman Near East and analyze it linguistically in light of Roman
Palestinian Koine Greek phonology and orthography. In chapter 2, I argued on the basis of
Origen's own writings that he lacked the requisite Hebrew proficiency to compose the Secun-
da himself. In chapter 3, on the basis of comparative texts in the Hellenistic Near East and the
social and linguistic setting in Palestine, I argued that the Greek transcriptions underlying the
Secunda text were originally composed with a didactic or scholarly purpose for the Jewish
community in Caesarea sometime in the second or third century CE. In chapter 4, on the basis
of a thorough analysis of spelling interchanges in Palestinian Greek epigraphy, the phonology
of Roman Palestinian Koine Greek was reconstructed. In chapter 5, roughly contemporary
examples of Greek transcription of other languages were summarized and analyzed.
In the final substantial chapter (6), the findings of the preceding chapters (4 and 5)
were utilized to elucidate the phonemic and phonetic values of the consonants, vowels, and
shewa as well as the syllable structure of the Hebrew reflected in the Secunda. While this
analysis resembled previous analyses of the Secunda in that it approached the transcriptions
from the perspective of historical Hebrew linguistics and a comparison with other reading tra-
- 360 -
ditions, it differed from previous treatments of the phonology of the Secunda primarily in
three respects. First, the orthography of the Secunda transcriptions was approached in light of
an in-depth study of contemporary Palestinian Greek pronunciation and orthography and
roughly contemporary Greek transcription conventions. Second, the process of transcription
was viewed in light of theoretical models of cross-language perception, namely, the PAM, the
PAM-L2, and the SLM. Third, the phonology of the Secunda was analyzed from the perspec-
tive of moraic theory, with a clear distinction being made between the underlying phonologi-
cal representation of a word and the post-lexical surface realization of a word.
While the main contribution of this dissertation is producing both a phonemic and
phonetic transcription of Secunda Hebrew that is grounded in contemporary Greek pronunci-
ation and orthography, applying these new approaches to the Secunda in chapter 6 has also
yielded a number of other helpful insights. For example, I argued that the transcriber priori-
tized quality over historical quantity when transcribing the Hebrew vowels. This led to the
hypothesis that the Hebrew short vowels were realized phonetically with a more lax or cen-
tralized realization than the long vowels. Another product of this hypothesis was that Greek ε
can represent a long [ɛː] in the Secunda. One potential attestation of this, ουαλεα for ‫ ְו ָעלֶי ָה‬,
may prove the existence of "qamas ̣" /ɔ̄/ [ɔː] in the vocalic system of Secunda Hebrew. A sen-
sitivity to Greek phonology and orthography also proved helpful in disentangling the various
representations of Hebrew etymological long /ı̄ / (as ι, ει, η), explaining some instances as re-
flecting phonological change and others as mere orthographic variants. Finally, I have argued
that a "shewa" vowel should be posited for the Secunda, yet understood as somewhat variable
in its phonetic realization and, in numerous contexts, undergoing the transition from being the
result of acoustic reduction to being the result of lexical reduction.
It should also be noted that, as a recurring theme throughout chapter 6, many of the
linguistic phenomena reflected in the Secunda were found to be paralleled by similar phe-
nomena in contemporary Hebrew/Aramaic and by modern living languages such as Arabic
- 361 -
and Neo-Aramaic. This likely indicates that the Hebrew reflected in the Secunda cannot be
totally divorced from its connection to the living language(s) of second-century CE Palestine.
The scope of this dissertation has been limited to addressing the phonology of the He-
brew reflected in the Secunda in light of contemporary Greek pronunciation and orthography.
However, there are a number of issues that this dissertation has left unexamined as well as a
number of promising areas for further study for which the present work has laid a foundation.
First, not all of the words of the Secunda or issues of grammar such as morphology
have been addressed in the present work. The application of the methodology utilized here
may be expanded to analyze the Secunda comprehensively. Second, it is not entirely clear
how exactly the process of transcription was carried out, whether it involved one bilingual
transcriber or a dictator and a transcriber. I expect that a more in-depth application of the
principles of the PAM, the PAM-L2, and the SLM may help elucidate this issue. Third, while it
was lightly touched on in this dissertation, it is not entirely clear to what degree contempo-
rary spoken Hebrew or Aramaic might have influenced the reading tradition. Advancements
in the modern linguistic discipline of language contact may be applied to the Secunda to fur-
ther address this point. Fourth, it is unclear how the Hebrew tradition reflected in the Secunda
relates to other Hebrew reading traditions such as Tiberian, Babylonian, and Palestinian. The
innovative features of the Hebrew tradition reflected in the Secunda may be systematically
compared with other Hebrew traditions to better understand this relationship. Fifth, and final-
ly, it should be noted that similar methodologies might be applied to other instances of Greek
and Latin transcription of Hebrew. For example, in analyzing Jerome, one might conduct a
thorough analysis of contemporary Latin pronunciation in 4th CE Palestine. Additionally, one
might explore how the principles of the PAM(-L2) and the SLM might be reflected in Jerome's
transcriptions, which actually reflect the product of L2 learning. While I plan to address these
issues in future publications, the present work has laid an adequate foundation.
- 362 -
8. APPENDIX A: PALESTINAN GREEK SPELLING INTERCHANGES
Spelling interchanges are tabulated and presented in four columns. The first column contains
the word,427 the second column contains the reference (either inscription number in the CIIP
or the name of the manuscript from the Judaean Desert), the third column contains the date,428
and the fourth column contains the location.

1 ει > ι
1 επ<ε>ι Mur115 124 CE Judaean Desert
2 θαρσι 1515 2nd CE-3rd CE Caesarea
3 θαρσι 749 3rd CE-4th CE Jerusalem
4 ουδις 2437 4th CE South Coast
5 β(οηθ)ι 804 Late Roman Jerusalem
6 θαρσι 924 2nd CE-5th CE Jerusalem
7 ουδις 924 2nd CE-5th CE Jerusalem
8 βο(η)θι 2080 410 CE Caesarea
9 θαρσι 2564 411 CE South Coast
10 ουδις 2564 411 CE South Coast
11 ταξι 1197 465-467 CE Caesarea
12 εκλαβιν 1197 465-467 CE Caesarea
13 υπερτεθισαν 1197 465-467 CE Caesarea
14 διαφερι 1197 465-467 CE Caesarea
427. It should be noted that, especially in formulae, various abbreviations are quite common in inscriptional
Greek. Therefore, it is not always clear whether a spelling interchange might also be explained as an
abbreviation. This means that sometimes whether a spelling interchange is present is dependent on the editors'
judgment and/or my judgment. In the CIIP, the following diacritical marks are used: ( ) for resolution of an
abbreviation; [ ] for restoration of missing text; < > for correction of a mistake or omission; { } for superfluous
text; ⟦ ⟧ for text erased in antiquity; [...] for missing text with relatively certain numbers of letters; [--] for
missing with uncertain number of letters; ˹ ˺ correction of a letter by the editor, . a dot beneath indicates an
uncertain reading (CIIP 1/1, xxv). In inscriptions from the Judaean Desert, the following diacritical marks are
used: ⟦ ⟧ for empty space with no writing; . a dot beneath indicates a probably reading; < > indicates a modern
editor's correction; {{ }} indicates various deletions such as erasure and scribal marks; [ ] indicates a
reconstructed letter; ( ) indicates solutions for symbols and abbreviations (ABEGG, BOWLEY, and COOK 2016, xvi).
428. If the inscription or text is undated, it is marked with a question mark (?). However, it should be noted that
undated texts from the Judaean Desert are most likely from the first or second century CE, undated texts from
Jerusalem numbered 1–704 are most likely from the first century CE or earlier, undated texts from Jerusalem
numbered 705–783 from the Roman period, undated texts from Jerusalem numbered 784–1087 from the
Byzantine period, and undated texts from Caesarea from the Byzantine period.
- 363 -
15 δια[φ]ερι 1197 465-467 CE Caesarea
16 βασιλια 2477 474 CE South Coast
17 ποιµενι 2327 493/494 CE South Coast
18 µνηµιον 983 5th CE Jerusalem
19 µνηµιον 2190 3rd CE-6th CE South Coast
20 κιτε 2190 3rd CE-6th CE South Coast
21 Αλεξανδριας 2196 3rd CE-6th CE South Coast
23 Ισας 2193 3rd CE-6th CE South Coast
24 διαφερι 2228 3rd CE-6th CE South Coast
25 ιρηνη 2231 3rd CE-6th CE South Coast
26 µνηµιον{ν} 1123 4th CE-6th CE Caesarea
27 υγια 1419 5th CE-6th CE Caesarea
28 κιτε 2477 541 CE South Coast
29 ευτυχ<ε>ι 802 6th CE Jerusalem
30 νοσοκοµιου 967 6th CE Jerusalem
31 βοηθι 1178 4th CE-7th CE Caesarea
32 κιτε 1462 4th CE-7th CE Caesarea
33 βοηθι 2113B 4th CE-7th CE Caesarea
35 ορφαν<ο>τροφιω 1168 5th CE-7th CE Caesarea
36 υγια 796 6th CE-7th CE Jerusalem
37 απολαυσ<ε>ιας 796 6th CE-7th CE Jerusalem
38 κατακιτε 978 6th CE-7th CE Jerusalem
39 φοβισθαι 1335 6th CE-7th CE Caesarea
41 υγια 2513 6th CE-7th CE South Coast
42 κιτ(αι) 875 7th CE Jerusalem
43 βοηθια 2457 732 CE South Coast
44 βοηθι 1339 Byz Caesarea
45 ευτυχι 1343 Byz Caesarea
46 υγια 1125 Byz? Caesarea
47 αυξι 2334 Late Antique South Coast
48 αυξι 2334 Late Antique South Coast
49 βοηθι 795 ? Jerusalem
50 [β]ασιλια 809 ? Jerusalem
51 βασιλι 810 ? Jerusalem
52 πυεισι 810 ? Jerusalem
53 εγ<ε>ιρας 842.1 ? Jerusalem
54 [βοηθ]ι 842.8 ? Jerusalem
55 βοηθι 842.27 ? Jerusalem
56 γεροκοµιον 859 ? Jerusalem
57 ταπινων 859 ? Jerusalem
58 βασιλευι 862 ? Jerusalem
59 µνηµιον 896 ? Jerusalem
60 κιτε 949 ? Jerusalem
61 θαρσι 950 ? Jerusalem
62 β(οη)θι 960 ? Jerusalem
63 κιται 1004 ? Jerusalem
64 κοιµηθισα 1004 ? Jerusalem
65 ιρηνη 1004 ? Jerusalem
66 νοσοκοµιου 1008 ? Jerusalem
67 διακιµενου 1008 ? Jerusalem
68 πεινας 1021 ? Jerusalem
69 γινοσκι 1084 ? Jerusalem
70 βοηθι 1177 ? Caesarea
71 [µ]νηµιον 1569 ? Caesarea
72 εξουσιαζιν 1613 ? Caesarea
75 βοηθια 1689 ? Caesarea
76 ευψυχ(ε)ιτωσαν 2255 ? South Coast
77 αυξιτω 2310 ? South Coast
78 θαρσι 2343 ? South Coast
79 γινοσκις 2356 ? South Coast
80 αυξι 2395 ? South Coast
81 αιπαρθ[ενου] 2451 ? South Coast
82 κιται 2489 ? South Coast
83 πρεσβιες 2531 ? South Coast
84 κιτη 2575 ? South Coast
85 χ<ε>ιρος Mur122 ? Judaean Desert
86 χ[ι]ρογραφων 5/6Hev5 ? Judaean Desert
87 οφ[ι]ληµατος 5/6Hev5 ? Judaean Desert
- 364 -
88 δ[α]νιου 5/6Hev5 ? Judaean Desert
89 τροφια 5/6Hev13 ? Judaean Desert
90 τρ]οφια 5/6Hev13 ? Judaean Desert
91 τροφια 5/6Hev15 ? Judaean Desert
92 ασφαλιας 5/6Hev15 ? Judaean Desert
93 τροφιων 5/6Hev15 ? Judaean Desert
94 υπατιας 5/6Hev16 ? Judaean Desert
95 γυναικιαν 5/6Hev18 ? Judaean Desert
96 γυναικιαν 5/6Hev18 ? Judaean Desert
97 χαιριν 5/6Hev20 ? Judaean Desert
98 ις 5/6Hev20 ? Judaean Desert
100 γιτνευουσιν 5/6Hev20 ? Judaean Desert
101 γιτονων 5/6Hev20 ? Judaean Desert
103 γιτονες 5/6Hev20 ? Judaean Desert
104 κατεχις 5/6Hev21 ? Judaean Desert
105 λεγις 5/6Hev21 ? Judaean Desert
106 <ο>φιλης 5/6Hev21 ? Judaean Desert
107 οφιλης 5/6Hev22 ? Judaean Desert
108 δωσις 5/6Hev22 ? Judaean Desert
109 σταθισα 5/6Hev22 ? Judaean Desert
110 οφιλουσα 5/6Hev22 ? Judaean Desert
111 π[α]ρην[γ]ιλεν 5/6Hev23 ? Judaean Desert
112 διακρατις 5/6Hev23 ? Judaean Desert
113 υ]πατιας 5/6Hev23 ? Judaean Desert
114 παρηνγιλεν 5/6Hev23 ? Judaean Desert
116 παρεδρευιν 5/6Hev23 ? Judaean Desert
118 επιδη 5/6Hev24 ? Judaean Desert
119 αποδιξε 5/6Hev24 ? Judaean Desert
120 διακ[ρ]ατις 5/6Hev24 ? Judaean Desert
121 απι[θις 5/6Hev24 ? Judaean Desert
123 δ[ια]κρατις 5/6Hev24 ? Judaean Desert
124 απ[ι]θις 5/6Hev24 ? Judaean Desert
126 συνεξελθιν 5/6Hev25 ? Judaean Desert
128 παρε[δ]ρευιν 5/6Hev25 ? Judaean Desert
129 υφε]ξις 5/6Hev25 ? Judaean Desert
130 παρηνγιλες 5/6Hev25 ? Judaean Desert
131 αποκριθουσα 5/6Hev25 ? Judaean Desert
132 [π]αρε[δ]ρευιν 5/6Hev25 ? Judaean Desert
133 παρανγιλε 5/6Hev25 ? Judaean Desert
134 συνεξελθιν 5/6Hev25 ? Judaean Desert
135 παρεδρ]ευι[ν 5/6Hev25 ? Judaean Desert
136 υφεξις 5/6Hev25 ? Judaean Desert
137 πα[ρ]ηνγιλες 5/6Hev25 ? Judaean Desert
138 [απο]κριθουσα 5/6Hev25 ? Judaean Desert
139 εχις 5/6Hev25 ? Judaean Desert
141 εχις 5/6Hev25 ? Judaean Desert
145 παρηνγιλα 5/6Hev26 ? Judaean Desert
146 εχιν 5/6Hev26 ? Judaean Desert
147 χαιρι[ν] 5/6Hev27 ? Judaean Desert
149 τρ]οφιων 5/6Hev27 ? Judaean Desert
150 τελιων 5/6Hev27 ? Judaean Desert
151 ερµηνια{ς} 5/6Hev27 ? Judaean Desert
153 τ[ρο]φιων 5/6Hev27 ? Judaean Desert
154 τρις 5/6Hev27 ? Judaean Desert
155 [τρ]ις 5/6Hev27 ? Judaean Desert
156 επι 5/6Hev30 ? Judaean Desert
157 [ε]χις 5/6Hev32a ? Judaean Desert
158 προκιµενου 5/6Hev34 ? Judaean Desert
- 365 -
161 χαιρι[ν] XHev/Se60 ? Judaean Desert
162 υπατιαν XHev/Se60 ? Judaean Desert
163 προκιµενης Xhev/Se62 ? Judaean Desert
164 προκιµενης Xhev/Se62 ? Judaean Desert
165 γυναικιας Xhev/Se65 = 5/6Hev37 ? Judaean Desert
166 γυ[ναι]κιοις Xhev/Se65 = 5/6Hev37 ? Judaean Desert
167 δ]εδανισµενοι Xhev/Se66 ? Judaean Desert
168 δεδανικοτι[--] Xhev/Se66 ? Judaean Desert
169 ξυλιας Xhev/Se67 ? Judaean Desert
2 ι > ει
1 Νεικανορος 98 1st BCE-1st CE Jerusalem
2 Ειφιγενειας 105 1st BCE-1st CE Jerusalem
3 Iαειρος 164 1st BCE-1st CE Jerusalem
4 Σαφειρα 398 1st BCE-1st CE Jerusalem
5 Ιαειρος 400 1st BCE-1st CE Jerusalem
6 Ιαειρος 400 1st BCE-1st CE Jerusalem
7 Ιαειρου 401 1st BCE-1st CE Jerusalem
8 Σκυθοπολειτης 411 1st BCE-1st CE Jerusalem
9 Σκυθοπολειται 412 1st BCE-1st CE Jerusalem
10 Αφρεικανος 416 1st BCE-1st CE Jerusalem
11 Φουλειος 416 1st BCE-1st CE Jerusalem
12 Αφρεικανος 421 1st BCE-1st CE Jerusalem
13 Φο<υ>λεια 423 1st BCE-1st CE Jerusalem
14 Αφρεικανα 423 1st BCE-1st CE Jerusalem
15 Φουλεια 424 1st BCE-1st CE Jerusalem
16 Αφρεικανα 424 1st BCE-1st CE Jerusalem
17 [Τειµει]σειων[ος] 497 1st BCE-1st CE Jerusalem
18 Τει[µει]σειωνος 497 1st BCE-1st CE Jerusalem
19 Τει[µει]σειωνος 497 1st BCE-1st CE Jerusalem
20 [Τει]µεισειωνο[ς] 497 1st BCE-1st CE Jerusalem
21 Εισµαηλ 526 1st BCE-1st CE Jerusalem
22 Σαλλαµσειων 588 1st BCE-1st CE Jerusalem
23 Ιουδειθ 590 1st BCE-1st CE Jerusalem
24 Λευεις 354 1st CE Jerusalem
25 πολειτην 2336 1st CE South Coast
26 Εισιδοτη 2162 1st CE-2nd CE South Coast
27 τοπαρχειας Mur115 124 CE Judaean Desert
28 νυνει Mur115 124 CE Judaean Desert
29 γεινοµεν[ης] Mur115 124 CE Judaean Desert
30 Ανεινας Mur89 100-135 CE Judaean Desert
31 κρε(ιθῆς) Mur91 100-135 CE Judaean Desert
36 Μαλχείων Mur91 100-135 CE Judaean Desert
37 Πανδειων Mur91 100-135 CE Judaean Desert
38 Μαλ]χειων Mur91 100-135 CE Judaean Desert
39 Σειλωνει Mur92 100-135 CE Judaean Desert
40 Φελειου Mur94 100-135 CE Judaean Desert
41 Ανναβ Mur94 100-135 CE Judaean Desert
42 Ελειεζρο[ς] Mur95 100-135 CE Judaean Desert
43 Ααβαει Mur103 100-135 CE Judaean Desert
44 πεινοµε[ -- ] Mur112 100-150 CE Judaean Desert
45 δισχειλια Mur116 100-150 CE Judaean Desert
46 Στατειλιου Mur114 171 CE Judaean Desert
47 επεικτη[σωµαι] Mur114 171 CE Judaean Desert
48 γεινεσθω Mur117 185 CE Judaean Desert
49 Α]ντωνει{νει}ν[ου 2337 187/188 CE South Coast
50 Πρεισκε 1531 1st CE-3rd CE Caesarea
51 Οσειρις 1531 1st CE-3rd CE Caesarea
52 Οσειρις 1531 1st CE-3rd CE Caesarea
53 Λειο[υιος 1195 2nd CE-3rd CE Caesarea
54 Εισιω[ν]ος 2621 2nd CE-3rd CE South Coast
55 ηµιλειτριν 2616 3rd CE South Coast
56 Εισιδωρου 934 3rd CE-4th CE Jerusalem
57 Λευειτης 2182 3rd CE-6th CE South Coast
58 Ειλασιου 2185 3rd CE-6th CE South Coast
59 Εισακ 2186 3rd CE-6th CE South Coast
60 Ειακω 2192 3rd CE-6th CE South Coast
61 Λευει 2194 3rd CE-6th CE South Coast
62 Νεικοµηδου 2225 3rd CE-6th CE South Coast
63 Ειοπιτων 2227 3rd CE-6th CE South Coast
- 366 -
64 Εισακιου 2236 3rd CE-6th CE South Coast
65 Ζοειλος 2245 3rd CE-6th CE South Coast
66 Απολλωνειδου 1015 4th CE-6th CE Jerusalem
67 δουκει 784 492 or 507 CE Jerusalem
68 βειου 2490 541 CE South Coast
69 ειερηω[ν] 1504 3rd CE-7th CE Caesarea
70 Βαεισηου 1460 4th CE-7th CE Caesarea
71 Δειονυσει. 1343 Byz Caesarea
72 Δειονυσει. 1343 Byz Caesarea
73 πασειν 2355 ? South Coast
74 Σαβει[νου] 2423 ? South Coast
75 Εισα 2537 ? South Coast
76 Νεικανορ[ -- ] Μur120 ? Judaean Desert
77 Σα̥λουειδιη̥[νοῦ 5/6Hev5 ? Judaean Desert
78 επαρχειας 5/6Hev5 ? Judaean Desert
79 χειλια 5/6Hev5 ? Judaean Desert
80 τειµης 5/6Hev5 ? Judaean Desert
83 τε[ι]µης 5/6Hev5 ? Judaean Desert
84 µεικρου 5/6Hev5 ? Judaean Desert
85 πατρει 5/6Hev5 ? Judaean Desert
86 χειλ[ι]ον 5/6Hev5 ? Judaean Desert
87 Μανειου 5/6Hev11 ? Judaean Desert
88 Ακειλιου 5/6Hev11 ? Judaean Desert
89 µειλιαριας 5/6Hev11 ? Judaean Desert
90 Μανειου 5/6Hev11 ? Judaean Desert
92 µειλιαριας 5/6Hev11 ? Judaean Desert
93 Ιουστεινου 5/6Hev11 ? Judaean Desert
94 Αφροδεισιω 5/6Hev12 ? Judaean Desert
95 Αφροδεισιω 5/6Hev12 ? Judaean Desert
97 Ακυλεινου 5/6Hev14 ? Judaean Desert
98 Ακυλεινου 5/6Hev15 ? Judaean Desert
100 οµειλιαν 5/6Hev15 ? Judaean Desert
101 υµειν 5/6Hev15 ? Judaean Desert
102 ἀπειθαρ]χ̥ε̥[ί]α̥ς 5/6Hev15 ? Judaean Desert
103 τροφιων 5/6Hev15 ? Judaean Desert
104 υµειν 5/6Hev15 ? Judaean Desert
105 ἡγεµωνε[ίας] 5/6Hev15 ? Judaean Desert
106 απειθαρχειας 5/6Hev15 ? Judaean Desert
107 ενει 5/6Hev15 ? Judaean Desert
108 Ατειλιου 5/6Hev16 ? Judaean Desert
110 Φλωρεντεινου 5/6Hev16 ? Judaean Desert
111 κρειθης 5/6Hev16 ? Judaean Desert
112 µειγµατος 5/6Hev16 ? Judaean Desert
114 γεινοµενων 5/6Hev16 ? Judaean Desert
117 Πρεισκος 5/6Hev16 ? Judaean Desert
118 Πουπλειου 5/6Hev17 ? Judaean Desert
119 Μετειλιου 5/6Hev17 ? Judaean Desert
121 επαρ]χειας 5/6Hev17 ? Judaean Desert
122 γ]εινεσθαι 5/6Hev17 ? Judaean Desert
123 διατετειµηµενην 5/6Hev18 ? Judaean Desert
124 τειµογραφιαν 5/6Hev18 ? Judaean Desert
125 γεινοµενης 5/6Hev18 ? Judaean Desert
126 γεινεσθαι 5/6Hev18 ? Judaean Desert
127 Που[π]λει[ο]υ 5/6Hev18 ? Judaean Desert
128 Μετειλι[ου] 5/6Hev18 ? Judaean Desert
130 διατετει<µη>µενην 5/6Hev18 ? Judaean Desert
131 τειµογ[ρ]αφιαν 5/6Hev18 ? Judaean Desert
132 γεινοµενης 5/6Hev18 ? Judaean Desert
133 γεινεσθαι 5/6Hev18 ? Judaean Desert
134 Με]τ̥ε̥[ιλίου] 5/6Hev19 ? Judaean Desert
135 ειδιαις 5/6Hev20 ? Judaean Desert
136 ειδιαις 5/6Hev20 ? Judaean Desert
- 367 -
137 Κρισπεινα 5/6Hev20 ? Judaean Desert
138 Κοειντου 5/6Hev21 ? Judaean Desert
139 Κοειντου 5/6Hev22 ? Judaean Desert
140 νυνει 5/6Hev25 ? Judaean Desert
141 υπ]ατιας 5/6Hev26 ? Judaean Desert
142 Αυγορεινου 5/6Hev27 ? Judaean Desert
143 κατακρειν[α]τωσαν 5/6Hev28 ? Judaean Desert
144 µεχρει 5/6Hev29 ? Judaean Desert
145 κατακρεινατωσαν 5/6Hev29 ? Judaean Desert
146 τειµην XHev/Se60 ? Judaean Desert
147 [Π]ερειτιου XHev/Se60 ? Judaean Desert
148 Πρεισκος XHev/Se61 ? Judaean Desert
149 επαρχειας Xhev/Se62 ? Judaean Desert
150 αποτειµησεως Xhev/Se62 ? Judaean Desert
151 Φλωρεντεινο[υ] Xhev/Se62 ? Judaean Desert
152 Σιµω[ο]ς Xhev/Se62 ? Judaean Desert
153 κρειθης Xhev/Se62 ? Judaean Desert
154 κρε[ι]θη[ς] Xhev/Se62 ? Judaean Desert
155 κ]ρειθης Xhev/Se62 ? Judaean Desert
156 τ[ε]λουν Xhev/Se62 ? Judaean Desert
157 --]χεελανης Xhev/Se62 ? Judaean Desert
158 Σεδαλλου Xhev/Se62 ? Judaean Desert
161 Ληουει Xhev/Se63 ? Judaean Desert
162 Λ]ηουει Xhev/Se63 ? Judaean Desert
163 Ληουει Xhev/Se63 ? Judaean Desert
164 Δειου Xhev/Se64 ? Judaean Desert
165 Ληουειου Xhev/Se64 ? Judaean Desert
166 φοινεικωνος Xhev/Se64 ? Judaean Desert
167 Ληουειου Xhev/Se64 ? Judaean Desert
168 φοινεικωνων Xhev/Se64 ? Judaean Desert
169 φοινεικος Xhev/Se64 ? Judaean Desert
170 Ρουφεινου Xhev/Se65 = 5/6Hev37 ? Judaean Desert
171 επαρχειας Xhev/Se65 = 5/6Hev37 ? Judaean Desert
172 τειµογ[ρ]αφιαν Xhev/Se65 = 5/6Hev37 ? Judaean Desert
173 γεινεσθαι Xhev/Se65 = 5/6Hev37 ? Judaean Desert
174 φ[οι]νεικωνω[ν] Xhev/Se68 ? Judaean Desert
175 Ζειφηνης Xhev/Se69 ? Judaean Desert
176 Φειλωνος 1Mish2 ? Judaean Desert
177 Ανεινας 34Se4 ? Judaean Desert
178 Αλεξειµ[αχος--] 34Se4 ? Judaean Desert
179 ηµιλιτρε[ιον] Mas741 ? Judaean Desert
180 Λειτρον Mas802 ? Judaean Desert
181 Λειτρο(ν) Mas817 ? Judaean Desert
182 Επαφρ]ωδειτω 1389 ? Caesarea
183 Αντειχουον 2122 ? Caesarea
184 επεισκ(οπου) 920 ? Jerusalem
185 Ειλαριου 921 ? Jerusalem
186 Βειθυνικη 944 ? Jerusalem
3 ιει > ει
1 π(ι)ειν 395 1st BCE Jerusalem
4 ι > ιε
1 Ιαιερε 452 1st BCE-1st CE Jerusalem
5 ι > ιι
1 Ιουλι{ι}α 5/6Hev20 ? Judaean Desert
6 ιο > ε
1 θεσιδεν 1650 4th CE-7th CE Caesarea
7 ιο > ει
1 Ερωταρειν 322 1st BCE-1st CE Jerusalem
2 Ερωταρειν 322 1st BCE-1st CE Jerusalem
8 ιο > ι
1 Ιουδιν 550 1st BCE-1st CE Jerusalem
2 ηµιλιτριν 2597 2nd CE-3rd CE South Coast
3 ηµιλιτριν 2623 2nd CE-3rd CE South Coast
4 [ηµιλ]ιτρ[ι]ν 2624 2nd CE-3rd CE South Coast
5 ηµιλειτριν 2616 3rd CE South Coast
6 ηµιλιτριν 2642 2nd CE-4th CE South Coast
7 Αρπαγι(ο)ς 936 4th CE Jerusalem
8 ενγονιν 2240 4th CE South Coast
9 Σωσεβις 2492 587 CE South Coast
10 Ευσεβις 987 599 CE or 614 CE Jerusalem
11 πλακι(ο)ν 997 6th CE-7th CE Jerusalem
- 368 -
12 [Αν]αστασι<ο>ς 1005 6th CE-7th CE Jerusalem
13 εγ]γονιν 2321 605-606 CE South Coast
14 Λεοντακις 2528 6th CE-8th CE South Coast
15 δηναριν 5/6Hev15 ? Judaean Desert
16 Θαδαις 5/6Hev15 ? Judaean Desert
17 Θα[δ]αδαις 5/6Hev23 ? Judaean Desert
18 πιττακιν 5/6Hev25 ? Judaean Desert
19 πιττακιν 5/6Hev25 ? Judaean Desert
20 ΥΙΣ KhQ_Arch25 ? Judaean Desert
21 Ευσεβι<ο>ς 842.12 ? Jerusalem
22 Ποπι<ο>ς 842.19 ? Jerusalem
23 Ιανο[υα]ρι<ο>ς 842.26 ? Jerusalem
24 Μακαρις 842.43 ? Jerusalem
25 Ευσταθι<ο>ν 842.58 ? Jerusalem
26 Ευφρονι<ο>ν 842.61 ? Jerusalem
27 κλιβαναρι<ο>ς 949 ? Jerusalem
28 Βαρωχι<ο>ς 976 ? Jerusalem
29 µιµορι<ο>ν 979 ? Jerusalem
30 Ζηνοβις 2571 ? South Coast
9 ιο > η
1 Παρηγορης 2226 ? South Coast
10 ια > ι
1 διφεροντων 848 ? Judaean Desert
2 Ιουλινου 2210 3rd CE-6th CE South Coast
11 ιο > ο
1 υ(ι)ος 2193 3rd CE-6th CE South Coast
2 υος 1140 5th CE-6th CE Caesarea
3 δηνάρ̥ο̥[ν 5/6Hev11 ? Judaean Desert
4 δ[η]ναρον 5/6Hev11 ? Judaean Desert
5 Βελλικ<ι>ου 5/6Hev12 ? Judaean Desert
6 Σαλου(ι)ος 1723 ? Caesarea
12 ιω > ω
1 αγ(ι)ω 2321 605-606 CE South Coast
2 δηνά[ρ]ων 5/6Hev11 ? Judaean Desert
3 δη[να]ρων 5/6Hev11 ? Judaean Desert
4 κυρ(ι)ω 2264 ? South Coast
13 εω > ω
1 αντιληµψως 848 ? Jerusalem
14 ια > α
1 η[µ]ιωρ<ι>αν Xhev/Se64 ? Judaean Desert
2 ηµιωρ<ι>αν Xhev/Se64 ? Judaean Desert
15 ια > ιεα
1 Μαριεαµη 451 1st BCE-1st CE Jerusalem
16 ι > αε
1 Μαρκαεος 23 1st BCE-1st CE Jerusalem
17 ε>ι
1 Βινιαµιν 419 1st BCE-1st CE Jerusalem
2 λεγιωνος 2335 65 CE South Coast
3 ετελι[ωθη 1871 2nd CE-3rd CE Caesarea
4 Εζικιαν 2186 3rd CE-6th CE South Coast
5 Ειζικια 2194 3rd CE-6th CE South Coast
6 Ηζικια 2196 3rd CE-6th CE South Coast
7 ικ 784 492 or 507 CE Jerusalem
8 ετελιωθη 2313 512 CE South Coast
9 θεωφιαιστατου 2448 530 CE South Coast
10 Εµισ(ης) 1760 6 Caesarea
11 επιτροπευ[σ]α 5/6Hev16 ? Judaean Desert
12 τελισοις 5/6Hev22 ? Judaean Desert
13 ετελιωθη 2427 ? South Coast
18 ι>ε
1 δεσδεκαλλου 212 1st BCE-1st CE Jerusalem
2 δεσκαλου 214 1st BCE-1st CE Jerusalem
3 Μαρκεος 23 1st BCE-1st CE Jerusalem
4 Δοµετιλλ 924 2nd CE-5th CE Jerusalem
5 Δοµετ[ι]ανος 2471 5th CE South Coast
6 Ελεου 1504 3rd CE-7th CE Caesarea
7 Εληνεστι 5/6Hev52 ? Judaean Desert
8 Εβραεστι 5/6Hev52 ? Judaean Desert
9 Σισεννιο[υ] 931 ? Jerusalem
10 παππεας 959 ? Jerusalem
11 Δοµετ(ιανου) 1414 ? Caesarea
12 Μελτιαδου 2310 ? South Coast
13 Γε[ων] 2449 ? South Coast
- 369 -
14 Δοµετιου 825 ? Jerusalem
19 ει > ε
1 παταξε 451 1st BCE-1st CE Jerusalem
2 ες 2227 3rd CE-6th CE South Coast
3 χερος XHev/Se60 ? Judaean Desert
4 χερος Xhev/Se69 ? Judaean Desert
5 ες 1021 ? Jerusalem
6 ες 1036 ? Jerusalem
7 θαρσε<ι> 2094 ? Caesarea
8 ερινη 2562 ? South Coast
20 ε > ει
1 παρεισχηµ[ενων 2267 163 BCE South Coast
2 Ειζικια 2194 3rd CE-6th CE South Coast
3 Γειωργιου 2143 ? Caesarea
4 Γειωργιου 2143 ? Caesarea
5 ειπι 2458 ? South Coast
21 η>ε
1 Ιωσε 46 1st BCE-1st CE Jerusalem
2 Μαρες 48 1st BCE-1st CE Jerusalem
3 Ιοσε 81 1st BCE-1st CE Jerusalem
4 Ιεσουα 295 1st BCE-1st CE Jerusalem
5 Ιεσους 247 1st CE Jerusalem
6 Ιεσους 247 1st CE Jerusalem
7 Τελε(µαχος) 768 2nd CE Jerusalem
8 ναικε 1701 3rd CE-4th CE Caesarea
9 Γηθσεµανις 986 4th CE-5th CE Jerusalem
10 µενι 2487 467 CE South Coast
11 εµων 2477 474 CE South Coast
12 σωτε[ριας] 803 5th CE Jerusalem
13 µεµοριον 1456 3rd CE-6th CE Caesarea
14 Ιβερω[ν] 1000 5th CE-6th CE Jerusalem
15 σωτεριας 1143 5th CE-6th CE Caesarea
16 προσενηγκαµεν 2460 509 CE South Coast
17 µενι 2444 528 CE South Coast
18 ανεπαε 2492 587 CE South Coast
19 σωτερ(ιας) 2321 605-606 CE South Coast
20 σωτερ(ιας) 2321 605-606 CE South Coast
21 Ελεου 1504 3rd CE-7th CE Caesarea
22 µεµο[ριον 1653 4th CE-7th CE Caesarea
23 επερωτηµενης 5/6Hev20 ? Judaean Desert
24 επερωτηµε[ν]ης 5/6Hev20 ? Judaean Desert
27 ενενοχ[ε]ναι Xhev/Se64 ? Judaean Desert
28 σωτεριας 864 ? Jerusalem
29 θετε 933 ? Jerusalem
30 µνεµειον 959 ? Jerusalem
31 Ιβερον 977 ? Jerusalem
32 Ελιας 1021 ? Jerusalem
33 Ιβεριω[ν] 1078 ? Jerusalem
34 µεµοριον 1488 ? Caesarea
35 µεµ[οριον]? 1978 ? Caesarea
36 Ιεσου 2424 ? South Coast
37 αµεν 2486 ? South Coast
38 αποθεµενε 2489 ? South Coast
39 αµεν 2501 ? South Coast
40 ανεπαε 2507 ? South Coast
41 θεκε 2575 ? South Coast
22 ε>η
1 Ευγηνια 59 1st BCE-1st CE Jerusalem
2 δοτη 1091 1st CE Jerusalem
3 µησου Jer4 132-135 CE Judaean Desert
4 [τρ]αφησηται Mur116 100-150 CE Judaean Desert
6 Ηαρι[νη] 1332 post 450 CE Caesarea
7 Νε]ηµιας 1479 3rd CE-6th CE Caesarea
8 Ηζικια 2196 3rd CE-6th CE South Coast
9 προσενηγκαµεν 2460 509 CE South Coast
10 κατετηθη 2498 563 CE South Coast
11 εγενητω 2542 576 CE South Coast
12 εθηµελιοθη 2432 6th CE South Coast
13 ειερηω[ν] 1504 3rd CE-7th CE Caesarea
14 Γηοργιας 1460 4th CE-7th CE Caesarea
- 370 -
15 κατετηθη 2495 602 and 609 CE South Coast
16 κατετηθη 2495 602 and 609 CE South Coast
17 Ηνγαδη[νο]ς 5/6Hev19 ? Judaean Desert
18 Ηνγαδης 5/6Hev19 ? Judaean Desert
19 Ηνγαδηνος 5/6Hev20 ? Judaean Desert
20 Ηνγαδηνη 5/6Hev20 ? Judaean Desert
21 Ηνγαδοις 5/6Hev20 ? Judaean Desert
24 Ην[γ]αδηνου 5/6Hev20 ? Judaean Desert
26 Ην[γαδηνο]ς 5/6Hev23 ? Judaean Desert
27 Ηνγαδην[ο]ς 5/6Hev23 ? Judaean Desert
28 Ηνγαδηνη<ν> 5/6Hev26 ? Judaean Desert
30 Ηληονα 857 ? Jerusalem
31 ευσηβηστατος 959 ? Jerusalem
32 ευσηβηστατος 959 ? Jerusalem
33 ηµοι 960 ? Jerusalem
34 Μαρηαβδηνου 2479 ? South Coast
23 η > ει
1 [Τει]µεισειωνο[ς] 497 1st BCE-1st CE Jerusalem
2 Σειλωνει Mur92 100-135 CE Judaean Desert
3 Σαµουειλου 1543 3rd CE-6th CE Caesarea
4 ει 2478 548 CE South Coast
6 συν]ειργη[σε 1012 ? Jerusalem
24 ει > η
1 Ηδηα 243 1st CE Jerusalem
2 ηρηνη 2223 3rd CE-6th CE South Coast
3 η(ς) 2295 4th CE-6th CE South Coast
4 η[ρ]ηνη 2167 6th CE-7th CE South Coast
5 φυλαξη 801 ? Jerusalem
6 ησοδ[ον] 810 ? Jerusalem
7 ηρινης 856 ? Jerusalem
8 ταπηνος 1082 ? Jerusalem
9 ηρινη 1189 ? Caesarea
10 ηερα[τ]ηου 2318 ? South Coast
11 βοηθη 2472 ? South Coast
12 τηχος 2476 ? South Coast
25 η>ι
1 Νατανιλου 255 1st CE Jerusalem
4 κοµιτος 2151 4th CE-5th CE Caesarea
5 φροντιστι 2196 3rd CE-6th CE South Coast
6 Παριγοριο[υ 2221 3rd CE-6th CE South Coast
7 Παριγοριου 2227 3rd CE-6th CE South Coast
8 ανεθικαµεν 2234 3rd CE-6th CE South Coast
9 µιµοριων 2236 3rd CE-6th CE South Coast
10 αδε]λφις 2246 3rd CE-6th CE South Coast
11 µ]νισθι 2295 4th CE-6th CE South Coast
12 µ]νισθι 2295 4th CE-6th CE South Coast
13 κωνχις 2468 4th CE-6th CE South Coast
14 Φαρβεθιτις 2290 5th CE-6th CE South Coast
15 σπουδι 800 549/550 CE or 564/565 CE Jerusalem
16 Βερνικι 1176 4th CE-7th CE Caesarea
17 µνιµιν 1453 4th CE-7th CE Caesarea
18 µνιµιν 1453 4th CE-7th CE Caesarea
19 θηκι 1465 4th CE-7th CE Caesarea
20 Εζινωβιας 1473 4th CE-7th CE Caesarea
21 Σεβιρου 1556 4th CE-7th CE Caesarea
22 πεδαρικι 1565 4th CE-7th CE Caesarea
23 µιµορ[ιον 1656 4th CE-7th CE Caesarea
24 θικι 997 6th CE-7th CE Jerusalem
25 θικι 997 6th CE-7th CE Jerusalem
26 αυτις 997 6th CE-7th CE Jerusalem
27 µνιστιτη 904 7th CE-8th CE Jerusalem
29 τις 2108 Byz? Caesarea
30 κωµις 2108 Byz? Caesarea
31 ταυτις 2108 Byz? Caesarea
32 ιγουµ[ενου] 2394 Late Antique South Coast
- 371 -
33 βοηθισον 798 ? Jerusalem
34 προσκυνισοµεν 810 ? Jerusalem
35 ιµων 810 ? Jerusalem
36 βοιθος 810 ? Jerusalem
37 φοβιθισοµε 810 ? Jerusalem
39 βοιθος 810 ? Jerusalem
40 τιν 810 ? Jerusalem
41 [µν]µις 819 ? Jerusalem
42 εψιφωθ[η] 854 ? Jerusalem
43 ηρινης 856 ? Jerusalem
44 κοµιτος 856 ? Jerusalem
45 µνησθιτι 867 ? Jerusalem
46 θηκι 883 ? Jerusalem
47 Ονισ(ιµου) 888 ? Jerusalem
48 oσιωτατι 909 ? Jerusalem
49 χαρατι 909 ? Jerusalem
50 θικη 912 ? Jerusalem
51 τι 966 ? Jerusalem
52 (πεντη)κοστι 966 ? Jerusalem
53 µιµορι<ο>ν 979 ? Jerusalem
54 Προβατι{σ}κις 980 ? Jerusalem
55 τις 980 ? Jerusalem
56 θηκι 998 ? Jerusalem
57 Γεβαλινου 1021 ? Jerusalem
58 κοµι[τος] 1024 ? Jerusalem
59 ηρινη 1189 ? Caesarea
60 µιµ[οριον] 1553 ? Caesarea
61 µν]ισθητι 1870 ? Caesarea
62 [θη]κι? 1918 ? Caesarea
63 βοηθισον 2143 ? Caesarea
64 Σωτιριχου 2143 ? Caesarea
65 καταργισον 2356 ? South Coast
66 βοη[θ]ισον 2424 ? South Coast
67 ερινη 2562 ? South Coast
26 ι>η
1 Σηµων 210 1st BCE-1st CE Jerusalem
2 Βενιαµην 523 1st BCE-1st CE Jerusalem
3 Γησχα·δ̣αν̣[ -- ] Mur92 100-135 CE Judaean Desert
4 αρχησυν[αγωγος] 1001 3rd CE-6th CE Jerusalem
5 Λεοντηου 1499 3rd CE-6th CE Caesarea
6 Σησηνιου 1499 3rd CE-6th CE Caesarea
7 Σησηνιου 1499 3rd CE-6th CE Caesarea
8 Ησηδωρου 2204 3rd CE-6th CE South Coast
9 Ησηδωρου 2204 3rd CE-6th CE South Coast
10 Ησσης 2230 3rd CE-6th CE South Coast
11 δηαφ[ερ(ουσα)] 993 5th CE-6th CE Jerusalem
12 Κλεοντηου 993 5th CE-6th CE Jerusalem
13 κτηστη 2148 5th CE-6th CE Caesarea
14 Ασκα]λουνητης 2292 516-526 CE South Coast
15 Αρτεµησιου 2478 548 CE South Coast
16 κτησµα 861 6th CE Jerusalem
17 δηαφεροντα 967 6th CE Jerusalem
18 ινδηκτ(ιωνος) 2432 6th CE South Coast
19 Βαεισηου 1460 4th CE-7th CE Caesarea
20 δηαφεροσα 1533 4th CE-7th CE Caesarea
21 επη 1141 6th CE-7th CE Caesarea
22 µηνη 875 7th CE Jerusalem
23 ψηφοσης 2109 7 Caesarea
25 Σηλβανου 1150 Byz Caesarea
26 Χρηστιανων 842.42 ? Jerusalem
27 Συρης 842.47 ? Jerusalem
28 αµαρτηων 903 ? Jerusalem
29 Αγηου 1785 ? Caesarea
30 ηερα[τ]ηου 2318 ? South Coast
31 ηερ[εως 2319 ? South Coast
32 φιλοχρηστων 2428 ? South Coast
33 επη 2476 ? South Coast
34 πιστης 2562 ? South Coast
27 η > ηι
1 βουλη{ι} 2335 65 CE South Coast
2 Ιωανη{ι}ς 5/6Hev31 ? Judaean Desert
- 372 -
28 ιη > η
1 Ησους 5/6Hev15 ? Judaean Desert
29 ιη > ι
1 Ισουος 2460 509 CE South Coast
30 αι > α
1 θαµα 508 1st BCE-1st CE Jerusalem
2 κα Jer5a 132-135 CE Judaean Desert
3 κα Jer5b 132-135 CE Judaean Desert
4 κα(ι) 1548 3rd CE-6th CE Caesarea
5 µελαναν 5/6Hev21 ? Judaean Desert
6 µελαναν 5/6Hev22 ? Judaean Desert
7 κρινα<ι> 933 ? Jerusalem
8 κα<ι> 2172 ? South Coast
31 αι > η
1 Ωρηα 326 1st CE Jerusalem
6 Ηληονα 857 ? Jerusalem
7 κιτη 2575 ? South Coast
32 η > αι
1 Ωσαιας 588 1st BCE-1st CE Jerusalem
2 σωταιριας 2451 ? South Coast
33 αι > ε
1 ηριθµηµε Mur114 171 CE Judaean Desert
2 κε 1134 1st CE-3rd CE Caesarea
3 κε 1457 3 Caesarea
4 περετερω 1197 465-467 CE Caesarea
5 περεωθισης 1197 465-467 CE Caesarea
6 πε]ρετερ[ω] 1197 465-467 CE Caesarea
7 Πουσεου 1197 465-467 CE Caesarea
8 ποιµενι 2327 493/494 CE South Coast
9 Κεσαριου 932 5th CE Jerusalem
10 αποκερδεν[εσθαι] 784 492 or 507 CE Jerusalem
11 πρεποσιτους 784 492 or 507 CE Jerusalem
12 κε 2179 3rd CE-6th CE South Coast
17 κ<ι>τε 2203 3rd CE-6th CE South Coast
20 κε 1485 4th CE-6th CE Caesarea
22 κε 999 5th CE-6th CE Jerusalem
24 Φαρβεθιτις 2290 5th CE-6th CE South Coast
25 Υβερβερετεου 2499 505 CE South Coast
26 προς]δεξσ 2447 528-530 CE South Coast
27 προσδεξε 2445 529-530 CE South Coast
28 Δεσιω 2493 539 CE South Coast
29 κιτε 2477 541 CE South Coast
30 µηνιεα 1196 6th CE Caesarea
31 Καλοκερια 1347 6th CE Caesarea
32 πεδαρικη 1514 6th CE Caesarea
34 κιτε 1462 4th CE-7th CE Caesarea
42 πεδαρικι 1565 4th CE-7th CE Caesarea
43 κ(ε) 1574 4th CE-7th CE Caesarea
44 κ(ε) 1574 4th CE-7th CE Caesarea
47 κατακιτε 978 6th CE-7th CE Jerusalem
48 ευξετε 978 6th CE-7th CE Jerusalem
- 373 -
49 Δικεοσυνη 2153 6th CE-7th CE Caesarea
50 Δεσιου 2350 600 CE South Coast
51 εὐ]φρενοµ[ένων Mur156 1000-1100 CE Judaean Desert
52 τες 5/6Hev15 ? Judaean Desert
53 προγεγραµµενες 5/6Hev15 ? Judaean Desert
54 ερεσασιν 5/6Hev15 ? Judaean Desert
55 παλεαν 5/6Hev19 ? Judaean Desert
56 δικεωµα 5/6Hev20 ? Judaean Desert
57 συνκεχωρηκενε 5/6Hev20 ? Judaean Desert
58 ληµψωµε 5/6Hev21 ? Judaean Desert
59 προγεγραπτε 5/6Hev22 ? Judaean Desert
60 προγεγραπτε 5/6Hev22 ? Judaean Desert
61 δικεον 5/6Hev24 ? Judaean Desert
65 ενγισε 5/6Hev26 ? Judaean Desert
66 ξεν[ο]κριτε 5/6Hev30 ? Judaean Desert
67 πεµσε 5/6Hev52 ? Judaean Desert
68 χερειν Xhev/Se64 ? Judaean Desert
70 εποψοµε 810 ? Jerusalem
71 κε 842 ? Jerusalem
72 κε 842.42 ? Jerusalem
73 γυνεκων 859 ? Jerusalem
75 Ρωµεας 900 ? Jerusalem
76 γυνεκων 901 ? Jerusalem
78 κιτε 949 ? Jerusalem
79 Βατανεας 952 ? Jerusalem
81 [κει]τε 984 ? Jerusalem
83 κερον 1021 ? Jerusalem
84 δεχοµε 1080 ? Jerusalem
85 προσδεξε 1084 ? Jerusalem
86 κε 1349 ? Caesarea
87 Κ]εσαρεω[--] 1586 ? Caesarea
88 Χεροµενην 1682 ? Caesarea
89 κε ? 1972 ? Caesarea
90 κε ? 1972 ? Caesarea
91 κ(ε) 2082 ? Caesarea
92 ευχες 2531 ? South Coast
93 πρεσβιες 2531 ? South Coast
94 Φεδρα 2568 ? South Coast
95 Φεδρα 2568 ? South Coast
34 ε > αι
1 υποµενονταις 1142 6th CE-7th CE Caesarea
2 Αινγαδδων 5/6Hev16 ? Judaean Desert
3 Αινγαδηνος 5/6Hev17 ? Judaean Desert
4 Α[ινγα]δ[ηνος] 5/6Hev17 ? Judaean Desert
5 Αινγαδων 5/6Hev18 ? Judaean Desert
6 θεαι 842.42 ? Jerusalem
7 εθος 1019 ? Jerusalem
8 αναπαυσαι(ως) 2454 ? South Coast
35 αι > ι
1 Ιοδιου 552 1st BCE-1st CE Jerusalem
2 Δισιου 987 599 CE or 614 CE Jerusalem
36 ι > αι
1 Αναινας 99 1st BCE-1st CE Jerusalem
2 ναικε 1701 3rd CE-4th CE Caesarea
37 αι > αιει
1 τροπαιεικον 5/6Hev15 ? Judaean Desert
2 τροπαι]εικὸν̥ 5/6Hev15 ? Judaean Desert
38 αι > αη
1 Ιηνναη 2208 3rd CE-6th CE South Coast
39 υ>ι
1 αυριχωρον 5/6Hev19 ? Judaean Desert
2 Αιδα 947 ? Jerusalem
3 λιχν(απτης) 966 ? Jerusalem
40 ι>υ
1 κηρυκου 977 ? Jerusalem
- 374 -
41 υ>η
1 προσσυλητου 1456 3rd CE-6th CE Caesarea
2 <σ>ηβιου 2322 6th CE-7th CE South Coast
3 ηποµενοντες 903 ? Jerusalem
4 πρεσβητ(ερου) 959 ? Jerusalem
42 η>υ
1 κυµυσεως 2477 474 CE South Coast
2 υµων 2487 467 CE South Coast
43 υι > υ
1 υ(ι)ου 452 1st BCE-1st CE Jerusalem
2 υ(ι)ος 548 1st BCE-1st CE Jerusalem
44 υ > υι
1 ενγυιου 5/6Hev21 ? Judaean Desert
45 υι > ι
1 ιου 1481 2nd CE-5th CE Caesarea
46 υι > υει
1 υειος 2224 3rd CE-6th CE South Coast
2 υειου 2243 3rd CE-6th CE South Coast
47 υι > οιι
1 οιιος 2178 3rd CE-4th CE South Coast
2 οιιου 2229 3rd CE-6th CE South Coast
48 οι > ο
1 [π]ο(ι)ησουσιν 833 ? Jerusalem
49 οι > ου
1 αδαλφου 452 1st BCE-1st CE Jerusalem
50 ου > οι
1 εγµισθοιν 5/6Hev11 ? Judaean Desert
51 οι > ω
1 εξωκιζω(ν)τον 440 1st BCE-1st CE Jerusalem
52 ω > οι
1 αποδοι 5/6Hev17 ? Judaean Desert
53 οι > υ
1υ 451 1st BCE-1st CE Jerusalem
2 τυς 452 1st BCE-1st CE Jerusalem
3 ανυγηναι 986 4th CE-5th CE Jerusalem
4 κυµυσεως 2477 474 CE South Coast
5 υκονοµων 2469 580 CE South Coast
6 τεκνυς 1476 4th CE-7th CE Caesarea
7 τεκνυς 1560 4th CE-7th CE Caesarea
8 τε]κνυς 1581 4th CE-7th CE Caesarea
9 τεκ]νυς 1598 4th CE-7th CE Caesarea
10 τεκν]υς 1599 4th CE-7th CE Caesarea
11 ηνυγµενα 5/6Hev20 ? Judaean Desert
12 ηνυγµενον 5/6Hev20 ? Judaean Desert
14 µυ 810 ? Jerusalem
15 συ 889 ? Jerusalem
16 τυς 892 ? Jerusalem
17 τεκνυς 1487 ? Caesarea
18 τεκν]υς 2069 ? Caesarea
54 υ > οι
1 Σοιµωνος 2224 3rd CE-6th CE South Coast
55 οι > αι
1 πανταιοις 5/6Hev19 ? Judaean Desert
56 οι > ε
1 κυρε 452 1st BCE-1st CE Jerusalem
57 η > οι
1 οι Mur156 1000-1100 CE Judaean Desert
2 οικοιµατων 5/6Hev19 ? Judaean Desert
3 οικοιµατων 5/6Hev19 ? Judaean Desert
4 οικοιµα[των 5/6Hev19 ? Judaean Desert
5 τελισοις 5/6Hev22 ? Judaean Desert
6 οικοιµατα Xhev/Se64 ? Judaean Desert
7 οικοιµατα Xhev/Se64 ? Judaean Desert
58 οι > ει
1 ει 1142 6th CE-7th CE Caesarea
59 οι > ι
1 ινοπρατου 1563 4th CE-7th CE Caesarea
2 ζωοπ<ο>ιω Mur156 1000-1100 CE Judaean Desert
3 δοµησοιων 5/6Hev20 ? Judaean Desert
60 ι > οι
- 375 -
1 Φλαυοιου 5/6Hev20 ? Judaean Desert
2 Φλαυοιου 5/6Hev20 ? Judaean Desert
61 οι > οοι
1 προοικος 5/6Hev21 ? Judaean Desert
2 προοικος 5/6Hev22 ? Judaean Desert
62 ευ > ε
1 θε(υ)µν<α>τος 214 1st BCE-1st CE Jerusalem
2 Ζεξιπ(που) 1530 4th CE-7th CE Caesarea
63 υ > ευ
1 πρεσβευτηο 2151 4th CE-5th CE Caesarea
64 ευ > εου
1 ιερεους 2178 3rd CE-4th CE South Coast
65 ευ > εθ
1 εθλογια 2355 ? South Coast
66 εο > ευ
1 Κλευπα<τ>ρους 594 1st BCE-1st CE Jerusalem
67 αυ > α
1 εατων 2336 1st CE South Coast
2 Α<υ>τοκρατορ(ος) 2639 117-138 CE South Coast
3 αναπα<υ>σαµενη 985 ? Jerusalem
68 αυ > αου
1 αουτου 1554 3rd CE-6th CE Caesarea
69 αυ > αυου
1 αυουτης 1548 3rd CE-6th CE Caesarea
70 αου > αυου
1 Φλαυουιου 5/6Hev21 ? Judaean Desert
2 Φλ[α]υουιου 5/6Hev22 ? Judaean Desert
71 αυ > ω
1 Ωλον 2335 65 CE South Coast
72 αυ > ο
1 οτου 2491 ? South Coast
73 ου > υ
1 θεοδωρ<ο>υ 436 1st BCE-1st CE Jerusalem
2 Μαρυλλα 486 1st BCE-1st CE Jerusalem
3 Ανεµ<ο>υ 550 1st BCE-1st CE Jerusalem
4 µ<ο>υλος 765 Late Roman Jerusalem
5 Συλλα 2193 3rd CE-6th CE South Coast
6 Ιυδας 2208 3rd CE-6th CE South Coast
7 Μιµµυλλα 1015 4th CE-6th CE Jerusalem
8 Μα]χ̥υ̥θ̥[α 5/6Hev5 ? Judaean Desert
9 Ελαζαρ<ο>υ 5/6Hev11 ? Judaean Desert
10 [α]κολ<ο>υθως Xhev/Se69 ? Judaean Desert
11 λυτρου 825 ? Jerusalem
74 υ > ου
1 βουργος 1261 500 CE Caesarea
2 µουλον 1044 7th CE Jerusalem
3 µουλον 1045 7th CE Jerusalem
75 ο>υ
1 Πτυλεµαικη 331 1st CE Jerusalem
76 υ>ο
1 Ορκανος 236 1st BCE-1st CE Jerusalem
77 ου > ει
1 υψειν 5/6Hev19 ? Judaean Desert
78 ου > ο
1 Σαολος 269 1st BCE-1st CE Jerusalem
3 Ιοδιου 552 1st BCE-1st CE Jerusalem
4 Μεταβορο<υ> 562 1st BCE-1st CE Jerusalem
5 Ορσιλας 577 1st BCE-1st CE Jerusalem
6 το<υ> 1728 44th CE-5th CE4 CE Caesarea
7 Ινστοληιον 2335 65 CE South Coast
8 Αρριστοβολα 325 1st CE Jerusalem
9 Ιοδα 2200 3rd CE-5th CE South Coast
11 δηαφεροσα 1533 4th CE-7th CE Caesarea
12 δολω 1181 Byz Caesarea
13 Αυγορεινου 5/6Hev27 ? Judaean Desert
14 Ποπι<ο>ς 842.19 ? Jerusalem
15 Ποπιηνο(ς?) 842.33 ? Jerusalem
16 δο<υ>[λ]ους 842.56 ? Jerusalem
17 ωλο<υ> 1021 ? Jerusalem
18 Κυρικο 2575 ? South Coast
79 ο > ου
- 376 -
1 τουπου 452 1st BCE-1st CE Jerusalem
2 Εντουλιου 1466 4th CE-7th CE Caesarea
3 Πουπλιου 5/6Hev27 ? Judaean Desert
4 δο<υ>[λ]ους 842.56 ? Jerusalem
80 oυ > οε
1 θεωδωροε 1143 5th CE-6th CE Caesarea
81 ω>ο
1 Βερνικεος 20 1st BCE-1st CE Jerusalem
2 Ιορ 24 1st BCE-1st CE Jerusalem
3 Ιοσε 81 1st BCE-1st CE Jerusalem
4 Ιοσ(ηφ) 318 1st BCE-1st CE Jerusalem
5 Σιµον 391 1st BCE-1st CE Jerusalem
6 Νισον 431 1st BCE-1st CE Jerusalem
7 Νισον 431 1st BCE-1st CE Jerusalem
9 πολεος 579 1st BCE-1st CE Jerusalem
10 Πολεος 1732 1st BCE - 1st CE Caesarea
11 Ἐξοµο̣λ̣[ογ]ήσα̣[το Mur115 124 CE Judaean Desert
12 σ[υ]νβιωσεος Mur115 124 CE Judaean Desert
13 ανανεωσεος Mur115 124 CE Judaean Desert
14 πραξεος Mur115 124 CE Judaean Desert
15 πολεος 2584 148/149 CE South Coast
16 πολεος 2585 2nd CE-4th CE South Coast
17 πολεος 2587 2nd CE-4th CE South Coast
18 σκηνοµα 986 4th CE-5th CE Jerusalem
19 Αλεξανδρεος 2180 3rd CE-6th CE South Coast
20 ιερεος 2202 3rd CE-6th CE South Coast
21 Ιοση 2206 3rd CE-6th CE South Coast
22 γναφεος 2212 3rd CE-6th CE South Coast
23 Ιακοβ 2229 3rd CE-6th CE South Coast
24 εγο 2234 3rd CE-6th CE South Coast
25 Συµονος 2236 3rd CE-6th CE South Coast
26 Αλεξανδρεος 2243 3rd CE-6th CE South Coast
27 Ζοειλος 2245 3rd CE-6th CE South Coast
28 ζοης 1015 4th CE-6th CE Jerusalem
29 ευχαριστον 2468 4th CE-6th CE South Coast
30 Ιονα 991 5th CE-6th CE Jerusalem
31 µακαριοτατου 2499 505 CE South Coast
32 Ζηνονος 2499 505 CE South Coast
33 µακαριο(τατου) 2497 529 CE South Coast
34 µοναζον 2445 529-530 CE South Coast
35 ηµον 2448 530 CE South Coast
36 οσιοτατου 2450 549 CE South Coast
37 αγιοτατου 2450 549 CE South Coast
38 Σαλαονος 2455 594 CE South Coast
39 οδε 2455 594 CE South Coast
40 τον 858 6th CE Jerusalem
41 αρτοκοπον 880 6th CE Jerusalem
42 εος 1034 6th CE Jerusalem
43 εθηµελιοθη 2432 6th CE South Coast
44 εψηφοθη 2432 6th CE South Coast
45 Γηοργιας 1460 4th CE-7th CE Caesarea
46 Συµον[ος] 1556 4th CE-7th CE Caesarea
47 ετο[ν?] 1447 5th CE-7th CE Caesarea
48 τον 796 6th CE-7th CE Jerusalem
49 σον 796 6th CE-7th CE Jerusalem
50 τον 796 6th CE-7th CE Jerusalem
51 Σολοµον 1005 6th CE-7th CE Jerusalem
52 ανο 1022 6th CE-7th CE Jerusalem
53 Βικτορος 2369 6th CE-7th CE South Coast
54 Φροντονος 2376 6th CE-7th CE South Coast
55 ζοην 2321 605-606 CE South Coast
56 σοµατ(ος) 2480 616 CE South Coast
57 Λοο 2480 616 CE South Coast
58 Λοο 2480 616 CE South Coast
59 Οκτοβριω 875 7th CE Jerusalem
60 οσιοτ(ατ)ω 879 7th CE Jerusalem
61 ψηφοσης 2109 7th CE Caesarea
62 χαµωψηφοσεος 2456 732 CE South Coast
64 Νονιας 1150 Byz Caesarea
65 το Mur156 1000-1100 CE Judaean Desert
66 Μα]ριο KhQ_Arch40 ? Judaean Desert
- 377 -
67 Ισιονι Jer19 ? Judaean Desert
68 αποδοσο EinGedi1 ? Judaean Desert
69 αποδοσο EinGedi1 ? Judaean Desert
70 δυσεος 5/6Hev11 ? Judaean Desert
71 δυσεος 5/6Hev11 ? Judaean Desert
72 Θεωδορος 5/6Hev11 ? Judaean Desert
73 ανθοµολογηµενης 5/6Hev20 ? Judaean Desert
75 ανθοµολογηµενη[ς] 5/6Hev21 ? Judaean Desert
77 οκτοκαιδεκα[τη] Xhev/Se62 ? Judaean Desert
78 γειτωνες Xhev/Se64 ? Judaean Desert
79 γει]τωνες Xhev/Se64 ? Judaean Desert
82 γειτωνες 5/6Hev19 ? Judaean Desert
83 Αντονιου 795 ? Jerusalem
84 προσκυνισοµεν 810 ? Jerusalem
85 προσπεσοµεν 810 ? Jerusalem
86 αυτό 810 ? Jerusalem
87 το 810 ? Jerusalem
88 εγο 810 ? Jerusalem
89 εος 810 ? Jerusalem
90 τον 810 ? Jerusalem
92 θεοδορας 856 ? Jerusalem
94 Δορο[θεου] 872 ? Jerusalem
96 Ιοα(ννου) 959 ? Jerusalem
97 Γεοργιου 962 ? Jerusalem
98 Παχοµιος 966 ? Jerusalem
99 θυρορ[ων] 968 ? Jerusalem
100 Ιβερον 977 ? Jerusalem
101 Βενετον 1026 ? Jerusalem
102 τριον 1076 ? Jerusalem
104 θεοδο[--] 1785 ? Caesarea
105 Αρτεµονος 2143 ? Caesarea
106 Αριστον 2262 ? South Coast
107 οσιοτατου 2320 ? South Coast
109 ανα[παυ]σεος 2451 ? South Coast
110 ανενεοθη 2476 ? South Coast
111 κα]λλοπισεν 2561 ? South Coast
82 ο>ω
2 δεκ[α]δυω Mur114 171 CE Judaean Desert
3 δυω Mur114 171 CE Judaean Desert
4 υιω(ς) 2464 3rd CE-5th CE South Coast
5 µηνοριων 1548 3rd CE-6th CE Caesarea
6 διαφερων 2218 3rd CE-6th CE South Coast
7 µιµοριων 2236 3rd CE-6th CE South Coast
9 θεωδωροε 1143 5th CE-6th CE Caesarea
10 θεωδοτης 2497 529 CE South Coast
11 µωναζων 2446 528-530 CE South Coast
12 θεωφιαιστατου 2448 530 CE South Coast
13 ωσιωτατου 2542 576 CE South Coast
14 θεωφιλ(ε)στατου 2542 576 CE South Coast
15 παραµωναριου 2542 576 CE South Coast
16 εγενητω 2542 576 CE South Coast
17 καρποφωριας 794 6th CE Jerusalem
18 Εζινωβιας 1473 4th CE-7th CE Caesarea
19 Γρηγω[ριου 1590 4th CE-7th CE Caesarea
20 θεωδ[--] 1658 4th CE-7th CE Caesarea
21 [--]αµως(?) 1447 5th CE-7th CE Caesarea
22 κωµερκιαριος 978 6th CE-7th CE Jerusalem
23 ορθωδοξους 1182 6th CE-7th CE Caesarea
24 ενδοξωτ(ατου) 1263 6th CE-7th CE Caesarea
25 ενδοξω(τατου) 1331 6th CE-7th CE? Caesarea
27 τωπος 1490 Byz? Caesarea
- 378 -
28 θεοδωρου 1490 Byz? Caesarea
29 δυω 5/6Hev5 ? Judaean Desert
31 Θεωδορος 5/6Hev11 ? Judaean Desert
33 ηγεµωνος 5/6Hev14 ? Judaean Desert
34 ἡγ̥[ε]µ̥ων[εί]ας̥ 5/6Hev15 ? Judaean Desert
35 ἡγεµωνε[ίας] 5/6Hev15 ? Judaean Desert
36 ηγεµωνος 5/6Hev15 ? Judaean Desert
37 επιτρωπων 5/6Hev15 ? Judaean Desert
46 αποδω[σεως] Xhev/Se69 ? Judaean Desert
47 τω 848 ? Jerusalem
48 θεωδοσια 878 ? Jerusalem
49 διαφερων 892 ? Jerusalem
50 Πωντου 931 ? Jerusalem
51 ω 1021 ? Jerusalem
52 ωλο<υ> 1021 ? Jerusalem
53 [θε]ωδωρ[ου] 1061 ? Jerusalem
54 θεοδωσιος 1082 ? Jerusalem
55 ελπιζωντων 1348 ? Caesarea
56 Επαφρ]ωδειτω 1389 ? Caesarea
57 Βωτρυς 1396 ? Caesarea
58 τοπως 1567 ? Caesarea
59 ωλης 2318 ? South Coast
60 ω 2507 ? South Coast
61 ωσιου 2546 ? South Coast
83 ω(ι)ο > οο
1 υπερων Xhev/Se64 ? Judaean Desert
2 υπερων Xhev/Se64 ? Judaean Desert
84 ω > ου
1 Ασκα]λουνητης 2292 516-526 CE South Coast
2 τοπου 2167 6th CE-7th CE South Coast
3 σουτηριαν 2422 ? South Coast
85 ου > ω
1 Κυρω 2291 5th CE-6th CE South Coast
2 Κυρω 2291 5th CE-6th CE South Coast
3 υιω 2422 ? South Coast
4 Νωµ[--]ς 842.42 ? Jerusalem
5 Βαρωχι<ο>ς 976 ? Jerusalem
6 υιω 2422 ? South Coast
86 ω>υ
1 Ιυσης 2211 3rd CE-6th CE South Coast
87 ε>α
1 αδαλφου 452 1st BCE-1st CE Jerusalem
2 παραγενετ[ο] 2490 541 CE South Coast
3 ερεσασιν 5/6Hev15 ? Judaean Desert
4 αυριχωρον 5/6Hev19 ? Judaean Desert
5 µεθαξει 5/6Hev19 ? Judaean Desert
6 ενανηκοντα Xhev/Se65 = 5/6Hev37 ? Judaean Desert
7 Αγλα Xhev/Se69 ? Judaean Desert
88 α>ε
2 δεσκαλου (cited elsewhere) 214 1st BCE-1st CE Jerusalem
3 Σελασιων 500 1st BCE-1st CE Jerusalem
4 Ανεστασια 2477 474 CE South Coast
5 Ανεστασια 2477 474 CE South Coast
6 τεσερεσκαιδεκατου Xhev/Se69 ? Judaean Desert
7 Aβιδελλα 842.15 ? Jerusalem
8 βρεκαριω 898 ? Jerusalem
89 α>η
1 Ιηνναη 2208 3rd CE-6th CE South Coast
90 ο>ε
1 Πτελεµ[αικ]ου 2246 3rd CE-6th CE South Coast
2 ηποµενοντες 903 ? Jerusalem
91 α>ο
- 379 -
1 Ζοορων 5/6Hev20 ? Judaean Desert
6 Γροπτη Xhev/Se64 ? Judaean Desert
92 ο>α
1 αγαρα 5/6Hev20 ? Judaean Desert
2 καθαραποιουντος 5/6Hev22 ? Judaean Desert
93 ø>α
1 Xαθουσιωνος 5/6Hev15 ? Judaean Desert
94 ø>ε
1 Θεενας 5/6Hev15 ? Judaean Desert
2 Δυστ{ε}ρου 5/6Hev17 ? Judaean Desert
3 Δυστ{ε}ρου 5/6Hev17 ? Judaean Desert
6 Απερος 5/6Hev20 ? Judaean Desert
7 Απερος 5/6Hev20 ? Judaean Desert
95 α>ø
1 θε(υ)µν<α>τος 214 1st BCE-1st CE Jerusalem
2 αιων<α>ς 2487 467 CE South Coast
3 Ραββαθµωβοις 5/6Hev16 ? Judaean Desert
4 Βαβθα 5/6Hev16 ? Judaean Desert
96 ε>ø
1 [ -- ο]ὐετρανός Mur113 100-150 CE Judaean Desert
2 (ε)κηδευσεν 2576 3rd CE-5th CE South Coast
3 θεωφιλ(ε)στατου 2542 576 CE South Coast
4 <ε>το(υς) 2536 586 CE South Coast
5 <ε>ξοδ[ον] 826 ? Jerusalem
6 Ευθηρι<ε> 950 ? Jerusalem
97 η>ø
1 µνµα 2186 3rd CE-6th CE South Coast
2 Κορν<η>λιας 1464 4th CE-7th CE Caesarea
3 σωττ(αι)ριας 2458 ? South Coast
98 ο>ø
1 ορφαν<ο>τροφιω 1168 5th CE-7th CE Caesarea
2 <ο>φιλης 5/6Hev21 ? Judaean Desert
99 ι>ø
1 (ι)σχιων 1702 2nd CE-6th CE Caesarea
2 (ι)σχιων 1703 3rd CE-5th CE Caesarea
3 <ι>νδι(κτιωνος) 2497 529 CE South Coast
4 (ι)σχιων 2156 ? Caesarea
100 υ>ø
1 ευδαιµοσ<υ>νης 2482 3rd BCE South Coast
101 π>φ
1 ενφ 1163 Byz Caesarea
102 φ>π
1 Σαπιρα 208 1st BCE-1st CE Jerusalem
2 Ἰωση̣π Mur90 100-135 CE Judaean Desert
3 Ιωσηπ Mur94 100-135 CE Judaean Desert
103 φθ > πθ
1 πθονε 1420 ? Caesarea
104 ππ > π
1 παπος 117 1st BCE-1st CE Jerusalem
2 Καπαδοκος 2192 3rd CE-6th CE South Coast
3 Καπαδοκων 2203 3rd CE-6th CE South Coast
4 Ειοπιτων 2227 3rd CE-6th CE South Coast
5 παπου 5/6Hev20 ? Judaean Desert
6 π]απου 5/6Hev20 ? Judaean Desert
105 π > ππ
1 αππαν[τα] Xhev/Se64 ? Judaean Desert
106 β>π
1 πολ<υ>ποτων 977 ? Jerusalem
107 π>β
1 βουργος 1261 500 CE Caesarea
2 Υβερβερετεου 2499 505 CE South Coast
108 ββ > β
1 Σαβατις 330 1st CE Jerusalem
2 Ραβι 2200 3rd CE-5th CE South Coast
3 αβα 1564 4th CE-7th CE Caesarea
4 Σαβατιω 2113B 4th CE-7th CE Caesarea
109 µβ > β
- 380 -
1 <σ>ηβιου 2322 6th CE-7th CE South Coast
110 µπ > π
1 Λαπαδιου 896 ? Jerusalem
111 µσ > σ
1 Σελασιων 500 1st BCE-1st CE Jerusalem
112 µ > µµ
1 α]ν[ο]ιωγµµενον Xhev/Se64 ? Judaean Desert
2 ανοιωγµµ[ενο]ν Xhev/Se64 ? Judaean Desert
113 µµ > µ
1 Αµια 361 1st BCE-1st CE Jerusalem
2 κωµερκιαριος 978 6th CE-7th CE Jerusalem
114 µ>ν
1 Σαλων 591 1st BCE-1st CE Jerusalem
2 µηνοριων 1548 3rd CE-6th CE Caesarea
3 Ναρκελ[λα] 2424 ? South Coast
115 µβ > νβ
1 Νοενβριων Mur115 124 CE Judaean Desert
2 συνβιον Mur115 124 CE Judaean Desert
3 σ[υ]νβιωσεος Mur115 124 CE Judaean Desert
4 συνβιου 1554 3rd CE-6th CE Caesarea
5 συνβιου 2192 3rd CE-6th CE South Coast
6 ενβολου 2543 578 CE South Coast
7 Κινβερ 5/6Hev20 ? Judaean Desert
8 Κινβερ 5/6Hev20 ? Judaean Desert
9 Δ]εκενβρι[ων 5/6Hev23 ? Judaean Desert
10 Δεκενβριων 5/6Hev23 ? Judaean Desert
11 συνβιου 1586 ? Caesarea
116 µβ > νµβ
1 Παρε{ν}µβολης 2240 4th CE South Coast
117 µπ > νπ
1 Πονπηια 709 2nd CE Jerusalem
2 συνπαροντος 5/6Hev16 ? Judaean Desert
3 συνπαροντ[ος] 5/6Hev17 ? Judaean Desert
4 συνπαρ[ο]ντος 5/6Hev17 ? Judaean Desert
5 συνπαν 5/6Hev20 ? Judaean Desert
6 συνπαροντος Xhev/Se64 ? Judaean Desert
118 µσ > µψ
1 Σ[ελα]µψιωνην 5/6Hev18 ? Judaean Desert
2 Σελαµψιωνη<ν> 5/6Hev18 ? Judaean Desert
3 Σελαµψιους 5/6Hev18 ? Judaean Desert
4 Numerous other instances
of Σελαµψιων
119 µπτ > µτ
1 πεµτου 5/6Hev20 ? Judaean Desert
4 πε]µ[τ]ου 5/6Hev22 ? Judaean Desert
120 µ>ø
1 Μαναηου 5/6Hev21 ? Judaean Desert
121 ντ > τ
2 Α<ν>τιγονα 513 1st BCE-1st CE Jerusalem
3 διαφερο(ν)τα 1548 3rd CE-6th CE Caesarea
4 προ[σ]ηνεγκο(ν)τος 2463 6th CE South Coast
5 διαφερο(ν) την 968 ? Jerusalem
122 νθ > θ
1 ε(ν)θα[δε] 958 5th CE-6th CE Jerusalem
123 ν > νν
1 Βεννιαµιν 2193 3rd CE-6th CE South Coast
2 Ιωανην 2445 529-530 CE South Coast
3 ερµην{ν}εια XHev/Se61 ? Judaean Desert
4 εν{ν} 5/6Hev20 ? Judaean Desert
124 νν > ν
1 θενου 21 1st BCE-1st CE Jerusalem
2 θενας 22 1st BCE-1st CE Jerusalem
3 Ιωανας 64 1st BCE-1st CE Jerusalem
4 [Ιω]ανης 179 1st BCE-1st CE Jerusalem
5 Ιωανηου 267 1st BCE-1st CE Jerusalem
6 Ιωανης 333 1st BCE-1st CE Jerusalem
- 381 -
11 Ιωανου 400 1st BCE-1st CE Jerusalem
12 Ανις 417 1st BCE-1st CE Jerusalem
13 Ιωανας 447 1st BCE-1st CE Jerusalem
14 Ιωανου Mur115 124 CE Judaean Desert
15 Ἰω]άν̣[ο]υ Mur115 124 CE Judaean Desert
16 Ἰωάν̣[ο]υ̣ Mur115 124 CE Judaean Desert
17 [Ἰωά]ν̣ου Mur115 124 CE Judaean Desert
18 Ιωανης Jer16 ? Judaean Desert
19 Ιωανης Jer16 ? Judaean Desert
20 Ιωανη[ς] Jer19 ? Judaean Desert
21 Numerous other instances
of Ιωανης
25 Ανιανον 842.52 ? Jerusalem
26 δεκαενεα 964 ? Jerusalem
27 Ιωανης 989 ? Jerusalem
125 ν>µ
1 διακοµου 2458 ? South Coast
126 ν# > ø
1 αυτό(ν) 451 1st BCE-1st CE Jerusalem
2 εριο(ν?) 673 1st BCE-1st CE Jerusalem
3 αυτό(ν) 986 4th CE-5th CE Jerusalem
4 Μενιαµι(ν) 2223 3rd CE-6th CE South Coast
5 [Β]ΟΗΘΨΣΟ<Ν> Fesh_Arch4 ? Judaean Desert
6 προκειµενω<ν> 5/6Hev12 ? Judaean Desert
7 προκειµενω<ν> 5/6Hev12 ? Judaean Desert
8 ἐµαρτυροποιησάµη<ν> 5/6Hev15 ? Judaean Desert
9 Σελαµψιωνη<ν> 5/6Hev18 ? Judaean Desert
10 συνγρα[φ]η<ν> 5/6Hev18 ? Judaean Desert
11 το<ν> 5/6Hev20 ? Judaean Desert
12 ε<ν> 5/6Hev20 ? Judaean Desert
15 Μαωζηνη<ν> 5/6Hev23 ? Judaean Desert
16 Μαωζηνη<ν> 5/6Hev23 ? Judaean Desert
17 αυτη<ν> 5/6Hev25 ? Judaean Desert
18 αυτη<ν> 5/6Hev25 ? Judaean Desert
20 Λειτρο(ν) Mas817 ? Judaean Desert
21 ηµω(ν) 825 ? Jerusalem
22 βοηθησο(ν) 842.55 ? Jerusalem
23 βοηθω(ν) 2355 ? South Coast
24 µοχθω(ν) 2504 ? South Coast
127 τ>θ
1 θαφος 959 ? Jerusalem
2 [γ]ρυθο[πωλου] 1552 ? Caesarea
128 θ>τ
1 Βερουτος 293 1st BCE-1st CE Jerusalem
2 Νατανιλου 255 1st CE Jerusalem
129 τ > ττ
1 Απολ]λοδοτ{τ}ου 2337 187/188 CE South Coast
2 σωττ(αι)ριας 2458 ? South Coast
130 ττ > τ
1 πιτακιου 5/6Hev16 ? Judaean Desert
2 πιτακιου 5/6Hev16 ? Judaean Desert
131 τρ > ρ
1 Κλευπα<τ>ρους 594 1st BCE-1st CE Jerusalem
132 θθ > θ
1 Μαθεθ<ος> 5/6Hev20 ? Judaean Desert
133 σθ > στ
1 ευφραινεστε 395 1st BCE Jerusalem
134 νθ > ντ
1 επληθυντησαν 1173 5 Caesarea
135 δδ > δ
1 Θαδαιος 5/6Hev14 ? Judaean Desert
2 Θαδαις 5/6Hev15 ? Judaean Desert
3 Θαδαιου 5/6Hev15 ? Judaean Desert
4 Θαδαι[ο]ς 5/6Hev20 ? Judaean Desert
5 θαδαιου 5/6Hev20 ? Judaean Desert
6 Θα[δ]αδαις 5/6Hev23 ? Judaean Desert
- 382 -
7 Θ[αδ]αιου 5/6Hev23 ? Judaean Desert
136 δ > δδ
1 Αινγαδδων 5/6Hev16 ? Judaean Desert
137 δ>τ
1 τρυφακτου 2 23 BCE-70 CE Jerusalem
138 σσ > σ
1 ιερισης 297 1st BCE-1st CE Jerusalem
2 τεσερεσκαιδεκατου Xhev/Se69 ? Judaean Desert
3 Αππιοσς Mas788 ? Judaean Desert
4 Βεσα 962 ? Jerusalem
139 σ > σσ
2 εξεσστω Xhev/Se66 ? Judaean Desert
140 σ > σ(σ)τ
1 Εµεσ{τ}ηνου 2176 3rd CE-6th CE South Coast
2 Ιστραηλ 2231 3rd CE-6th CE South Coast
3 Ιστρα[ε]λ 2167 6th CE-7th CE South Coast
4 Ισστραηλ 2509 ? South Coast
141 στ > τ
1 Αρι<σ>των(ος) 308 1st BCE-1st CE Jerusalem
2 Ιουτου 1140 5th CE-6th CE Caesarea
3 ωτε 1156 ? Caesarea
142 σθ > θ
1 [µν]η<σ>θητ[ι] 842.60 ? Jerusalem
143 στ > σ
1 Χρισ<τ>ος 976 ? Jerusalem
144 σ# > ø
1 προ<ς> ( / _σ) 3 18-17 BCE Jerusalem
2 αυτή<ς> 566 1st BCE-1st CE Jerusalem
3 υιω(ς) 2464 3rd CE-5th CE South Coast
4 η(ς) 2295 4th CE-6th CE South Coast
5 τοι<ς> 5/6Hev12 ? Judaean Desert
6 Αβδοοβδα<ς> 5/6Hev13 ? Judaean Desert
7 ενο<ς> 5/6Hev16 ? Judaean Desert
8 πρε<σ>βευτου 5/6Hev26 ? Judaean Desert
9 πρωτη<ς> 5/6Hev27 ? Judaean Desert
10 ετου<ς> 5/6Hev27 ? Judaean Desert
11 αδερφο(ς) 2575 ? South Coast
145 σ>ø
1 [ε]λεη<σ>ον 842.3 ? Jerusalem
146 σ>ζ
1 πρεζβευτου 2177 3rd CE-5th CE South Coast
2 αγοραζµατος 5/6Hev22 ? Judaean Desert
3 αµφιαζµου 5/6Hev27 ? Judaean Desert
4 αµφιαζµου 5/6Hev27 ? Judaean Desert
147 ζ>σ
1 Εσκιας 389 1st BCE-1st CE Jerusalem
148 ζ > ζζ
1 τευχιζζει 5/6Hev19 ? Judaean Desert
149 ζζ > θζ
1 Α{θ}ζαν 2235 3rd CE-6th CE South Coast
150 κ > κκ
1 Ιακκωβου 1481 2nd CE-5th CE Caesarea
2 Ιακκωβ[ου--] 34Se4 ? Judaean Desert
3 [περδι]κ{κ}ες 1019 ? Jerusalem
151 κκ > κ
1 εκλ[(ησιαις)] 785 533-565 CE Jerusalem
2 εκλ[ησι?] 905 ? Jerusalem
3 εκλησιας 2318 ? South Coast
4 εκλησια 2427 ? South Coast
152 κχ > χχ
1 Βαχχιου Xhev/Se66 ? Judaean Desert
153 χχ > χ
1 Ζαχαι 2209 3rd CE-6th CE South Coast
154 χ>κ
1 Μικαηλ 2154 ? Caesarea
155 κ>χ
1 Αντειχουον 2122 ? Caesarea
156 χ > κκ
1 Ζακκαριας 959 ? Jerusalem
157 κ > σκ
1 αντισκινησ(ας?) 451 1st BCE-1st CE Jerusalem
2 Προβατι{σ}κις 980 ? Jerusalem
- 383 -
158 κ>γ
1 εγ 2482 3rd BCE South Coast
2 εγδω Mur116 100-150 CE Judaean Desert
3 εγ Mur114 171 CE Judaean Desert
4 εγ Jer4 ? Judaean Desert
5 εγδικα[ -- ] Mur120 ? Judaean Desert
6 εγ Mur120 ? Judaean Desert
7 εγµισθοιν 5/6Hev11 ? Judaean Desert
8 εγδικησωµεν 5/6Hev20 ? Judaean Desert
9 εγ 2172 ? South Coast
159 γ > γγ
1 ζυγγ(ων) 1196 6 Caesarea
160 γγ > γ
1 εγ<γ>εγραµµενον 5/6Hev12 ? Judaean Desert
2 Λογινος 1577 ? Caesarea
161 γ>δ
1 ηδορακα Mur122 ? Judaean Desert
2 ηδορακα Mur122 ? Judaean Desert
162 γγ > νγ
1 παρανγει… Mur115 124 CE Judaean Desert
2 ενγιστα Mur114 171 CE Judaean Desert
4 συνγενικης 2227 3rd CE-6th CE South Coast
5 Ε]νγαδα EinGedi2 ? Judaean Desert
6 ενγυ[η] 5/6Hev5 ? Judaean Desert
7 Ενγαδοις 5/6Hev11 ? Judaean Desert
8 Ενγαδηνος 5/6Hev11 ? Judaean Desert
11 ενγιστα 5/6Hev14 ? Judaean Desert
12 παρηνγει[λεν 5/6Hev14 ? Judaean Desert
13 παρανγελλω 5/6Hev14 ? Judaean Desert
14 παρήν̥[γ]ε̥ι̥λ̥έ 5/6Hev15 ? Judaean Desert
15 παρηνγειλα 5/6Hev15 ? Judaean Desert
16 συνγραφην 5/6Hev17 ? Judaean Desert
17 συνγρα[φ]η<ν> 5/6Hev18 ? Judaean Desert
18 Αινγαδων 5/6Hev18 ? Judaean Desert
19 συνγραφην 5/6Hev18 ? Judaean Desert
20 παρ]αν[γ]ει<λ>ει 5/6Hev19 ? Judaean Desert
21 Ηνγαδη[νο]ς 5/6Hev19 ? Judaean Desert
22 Ηνγαδης 5/6Hev19 ? Judaean Desert
23 ενγυς 5/6Hev19 ? Judaean Desert
24 παρανγειλει 5/6Hev19 ? Judaean Desert
30 Ην[γ]αδηνου 5/6Hev20 ? Judaean Desert
34 π[α]ρην[γ]ιλεν 5/6Hev23 ? Judaean Desert
35 Ην[γαδηνο]ς 5/6Hev23 ? Judaean Desert
37 Ηνγαδην[ο]ς 5/6Hev23 ? Judaean Desert
39 [π]αρανγελλω 5/6Hev25 ? Judaean Desert
40 παρηνγιλες 5/6Hev25 ? Judaean Desert
44 πα[ρ]ηνγιλες 5/6Hev25 ? Judaean Desert
47 παρηνγιλα 5/6Hev26 ? Judaean Desert
48 ενγισε 5/6Hev26 ? Judaean Desert
50 ενγυ[ς] 1613 ? Caesarea
163 γκ > νκ
1 µετενενκη 385 1st CE (before 70) Jerusalem
2 ασυνκριτε 1515 2nd CE-3rd CE Caesarea
3 ουνκιαι 2644 3rd CE-5th CE South Coast
- 384 -
4 ενκληµατικων 1197 465-467 CE Caesarea
5 Συνκλητικην 2234 3rd CE-6th CE South Coast
6 ενκληµατι 5/6Hev17 ? Judaean Desert
7 συνκεχωρηκεναι 5/6Hev20 ? Judaean Desert
8 {{παρα}}συνκεχωρηκεναι 5/6Hev20 ? Judaean Desert
9 συνκεχωρηκεναι 5/6Hev20 ? Judaean Desert
10 ενκαλουν[τος 5/6Hev28 ? Judaean Desert
11 ενκαλουµε[νου 5/6Hev28 ? Judaean Desert
12 ενκαλουντος 5/6Hev29 ? Judaean Desert
13 ε]ν[κ]αλ[ο-]υν[τος] 5/6Hev30 ? Judaean Desert
14 ενκαλουµεν[ου] 5/6Hev30 ? Judaean Desert
164 γχ > νχ
2 εντυνχανω 5/6Hev34 ? Judaean Desert
3 εντυνχ[ανω] 5/6Hev34 ? Judaean Desert
165 γ>ø
1 γιν[ωσκει] 869 7th CE Jerusalem
2 γινωσκει 1152 1150 CE Caesarea
3 [γι]νωσκε 5/6Hev24 ? Judaean Desert
4 γινωσκε 5/6Hev25 ? Judaean Desert
166 λλ > λ
1 Κυριλη 296 1st BCE-1st CE Jerusalem
2 Καλ(λ)ωνος 372 1st BCE-1st CE Jerusalem
3 Σαλλαµσειων 588 1st BCE-1st CE Jerusalem
4 αλως 5/6Hev52 ? Judaean Desert
5 Εληνεστι 5/6Hev52 ? Judaean Desert
6 Απολιναριων 901 ? Jerusalem
167 λ > λλ
168 ρ>λ
1 Φουλειος 416 1st BCE-1st CE Jerusalem
3 Φουλεια 424 1st BCE-1st CE Jerusalem
4 Γληγοριας 2210 3rd CE-6th CE South Coast
5 λιβλαριος 5/6Hev15 ? Judaean Desert
9 [λ]ιβλαριου 5/6Hev21 ? Judaean Desert
10 λιβλαριου 5/6Hev22 ? Judaean Desert
169 λ>ρ
1 θεοφιρος 2575 ? South Coast
2 αδερφο 2575 ? South Coast
170 λ>ø
1 αδεφου 1547 ? Caesarea
2 α<λ>υπε 2577 ? South Coast
171 ρ > ρρ
1 Αρριστοβου[λα] 325 1st CE Jerusalem
2 Αρριστοβολα 325 1st CE Jerusalem
3 [θεο]δορ{ρ}ω 1051 ? Jerusalem
172 ρ>ø
173 λ>ø
1 παρ]αν[γ]ει<λ>ει 5/6Hev19 ? Judaean Desert
174 ψ > µψ
2 ληµψ[ει] 5/6Hev22 ? Judaean Desert
3 αντιληµψως 848 ? Jerusalem
4 αντιληµψεως 856 ? Jerusalem
175 µψ > µσ
1 αντιληµσεως 2428 ? South Coast
176 µψ > µσ
1 επεµσα 5/6Hev52 ? Judaean Desert
2 πεµσε 5/6Hev52 ? Judaean Desert
3 [επε]µσα Mas741 ? Judaean Desert
- 385 -
9. BIBLIOGRAPHY
ABEGG, MARTIN, JAMES BOWLEY, and EDWARD COOK. 2015. The Dead Sea Scrolls
Concordance. Leiden/Boston: Brill.
ADAMS, J. N. 2003. Bilingualism and the Latin Language. Cambridge: Cambridge University
Press.
AL-ANI, SALMAN H. 2006. “Phonetics.” In Encyclopedia of Arabic Language and Linguistics,
edited by KEES VERSTEEGH, Leiden: Brill.
AL-JALLAD, AHMAD. 2014. “Aṣ-s ̣ādu llatī ka-s-sīn - Evidence for an Affricated Ṣād in
Sibawayh?” Folia Orientalia 51: 51–57.
———. 2015. “Graeco-Arabica I: The Southern Levant.” In Le contexte de naissance de
l’écriture arabe. Écrit et écritures araméennes et arabes au 1er millénaire après J.C.,
edited by F. BRIQUEL-CHATONNET, M. DEBIÉ, and L. NEHMÉ, Louvain: Peeters (Orientalia
Lovaniensa Analecta).
———. forthcoming. The Damascus Psalm Fragment: Middle Arabic and the Legacy of Old
Ḥigāzī.
AL-JALLAD, AHMAD, ROBERT DANIEL, and OMAR AL-GHUL. 2013. “The Arabic Toponyms and
Oikonyms in 17.” In The Petra Papyri II, edited by LUDWIG KOENEN, MAARIT KAIMO,
JORMA KAIMIO, and ROBERT DANIEL, 23–48. Amman: American Center of Oriental
Research.
AL-JALLAD, AHMAD, and ALI AL-MANASER. 2015. “New Epigraphica from Jordan I: A Pre-
Islamic Arabic Inscription in Greek Letters and a Greek Inscription from North-Eastern
Jordan.” Arabian Epigraphic Notes 1: 51–70.
ALEXANDER, PHILIP. 1983. “The Targumim and the Rabbinic Rules for the Delivery of the
Targum.” In Congress Volume: Salamanca 1983, edited by J. A. EMERTON, 14–28.
Leiden: E. J. Brill.
———. 1999. “How Did the Rabbis Learn Hebrew?” In Hebrew Study From Ezra to Ben-
Yehuda, edited by WILLIAM HORBURY, 71–89. Edinburgh: T and T Clark.
ALLEN, W. SIDNEY. 1974. Vox Graeca: A Guide to the Pronunciation of Classical Greek.
Cambridge: Cambridge University Press.
———. 1978. Vox Latina: A Guide to the Pronunciation of Classical Latin. Cambridge:
Cambridge University Press.
ALMBARK, RANA ALHUSSEIN, and SAM HELLMUTH. 2015. “Acoustic Analysis of the Syrian
Arabic Vowel System.” Paper presented at the 18th International Congress of Phonetic
Sciences. Glasgow, United Kingdom. August 10–14.
ALVESTAD, SILJE, and LUTZ EDZARD. 2009. La-h ̣šōḇ , but la-h ̣ăzōr?: Sonority, Optimality, and
the Hebrew ‫ פ״ח‬Forms. Wiesbaden: Harrassowitz.
AMELING, WALTER, HANNAH M. COTTON, LEAH DI SEGNI, WERNER ECK, BENJAMIN ISAAC, ALLA
- 386 -
KUSHNIR-STEIN, HAGGAI MISGAV, JONATHAN J. PRICE, ISRAEL ROLL, ADA YARDENI, MARFA
HEIMBACH, and NAOMI SCHNEIDER, eds. 2010. Corpus Inscriptionum Iudaeae/
Palaestinae: A Multi-Lingual Corpus of the Inscriptions from Alexander to Muhammad.
Berlin/Boston: De Gruyter.
AOYAMA, K., and L. REID. 2006. “Cross-linguistic Tendencies and Durational Contrasts in
Geminate Consonants: An Examination of Guinaang Bontok Geminates.” Journal of
the International Phonetic Association 36: 145–57.
ARSLAN, EDOARDO, ed. 1957. Studi in onore di Aristide Calderini e Roberto Paribeni. Milan:
Ceschina.
BACHER, W. 1903. “Das altjüdische Schulwesen.” Jahrbuch für jüdische Geschichte und
Literatur 6: 48–81.
BAGG, ARIEL M. 2007. Die Orts- und Gewässernamen der neuassyrischen Zeit, 1. Die
Levante. Wiesbaden: Reichert.
BAKER, WENDY, and PAVEL TROFIMOVICH. 2005. “Interaction of Native- and Second-Language
Vowel System(s) in Early and Late Bilinguals.” Language and Speech 48 (1): 1–27.
BAKER, WENDY, PAVEL TROFIMOVICH, MOLLY MACK, and JAMES EMIL FLEGE. 2002. “The
Effects of Perceived Phonetic Similarity on Non-Native Sound Learning by Children
and Adults.” In Proceedings of the 26th annual Boston University Conference on
Language Development, edited by DO. A, L. DOMÍNGUEZ, and A. JOHANSEN, Somerville,
MA: Cascadilla Press.
BALTES, GUIDO. 2014a. “The Origins of the ‘Exclusive Aramaic Model’ in the Nineteenth
Century: Methodological Fallacies and Subtle Motives.” In The Language Environment
of First Century Judaea, edited by RANDALL BUTH, and R. STEVEN NOTLEY, 9–34.
Leiden/Boston: Brill.
———. 2014b. “The Use of Hebrew and Aramaic in Epigraphic Sources of the New
Testament Era.” In The Language Environment of First Century Judaea, edited by
RANDALL BUTH, and R. STEVEN NOTLEY, 35–65. Leiden/Boston: Brill.
BAR-ASHER, MOSHE. 2009. Studies in Mishnaic Hebrew (in Hebrew). Jerusalem: The Bialik
Institute.
———. 2010. “The Verse ‫ש ָׂראֵל‬ ְ ִ ‫שמַע י‬
ׁ ְ (‘Hear, O Israel’) in Greek Transcription on an Ancient
Amulet.” Journal of Ancient Judaism 1 (2): 227–32.
———. 2015. A Morphology of Mishnaic Hebrew (in Hebrew). Jerusalem: The Academy of
the Hebrew Language/The Bialik Institute.
BARDY, G. 1925. “Les traditions juives dans l’œuvre d’Origène.” Revue Biblique 34: 217–52.
BARR, JAMES. 1967. “St. Jerome and the Sounds of Hebrew.” Journal of Semitic Studies 12
(1): 1–36.
BAUMGARTNER, WALTER, and OTTO EISSFELDT, eds. 1950. Festschrift Alfred Bertholet zum 80.
Geburtstag gewidmet von Kollegen und Freunden. Tübingen: Mohr.
BELLEM, ALEX. 2007. “Towards a Comparative Typology of Emphatics Across Semitic and
into Arabic Dialect Phonology (single-spaced version).” diss., University of London.
BEN-ḤAYYIM, ZE’EV. 1958. “Traditions in the Hebrew Language, with Special Reference to
the Dead Sea Scrolls” (in Hebrew). In Scripta Hierosolymitana, IV, edited by CHAIM
RABIN, and YIGAEL YADIN, 200–14. Jerusalem: Magnes.
———. 2000. A Grammar of Samaritan Hebrew: Based on the Recitation of the Law in
Comparison with Tiberian and Other Jewish Traditions. Winona Lake, IN.:
Eisenbrauns.
BERTRAM, G. 1938. “Zur Septuagint-Forschung, III.” Theol. Rundschau, NF 10: 69–80.
BEST, CATHERINE. 1995. “A Direct Realist View of Cross-Language Speech Perception: New
Directions in Research and Theory.” In Speech Perception and Linguistic Experience:
Issues in Cross-Language Research, edited by WINIFRED STRANGE, Baltimore: York.
BEST, CATHERINE, and M. D. TYLER. 2007. “Nonnative and Second-Language Speech
Perception.” In Language Experience in Second Language Speech Learning. In Honor
- 387 -
of James E. Flege, edited by OCKE-SCHWEN BOHN, and MURRAY J. MUNRO, 13–34.
Amsterdam: John Benjamins.
BLAKE, FRANK R. 1911. “Vocalic R, L, M, N in Semitic.” Journal of the American Oriental
Society 31 (2): 217–22.
BLAPP, SAMUEL. 2016. “The Non-Standard Tiberian Hebrew Language Tradition in a
Nutshell.” Paper presented at the annual meeting of the Society of Biblical Literature.
San Antonio, TX. November 21.
BLAU, JOSHUA. 1982. On Polyphony in Biblical Hebrew. Jerusalem: Israel Academy of
Sciences and Humanities.
———. 1984. “A New Book on the Second Column of Origen’s Hexapla” (in Hebrew).
Review of Studies in Hebrew Historical Linguistics Based on Origen's Secunda, by
GERARD JANSSENS. Leshonenu 48–49: 76–80.
———. 1998a. “Non-Phonetic Conditioning of Sound Change and Biblical Hebrew.” In
Topics in Hebrew and Semitic Linguistics, edited by JOSHUA BLAU, 26–35. Jerusalem:
Magnes/The Hebrew University.
———. 1998b. “Hebrew Stress Shifts, Pretonic Lengthening, and Segolization: Possible
Cases of Aramaic Interference in Hebrew Syllable Structure.” In Topics in Hebrew and
Semitic Linguistics, edited by JOSHUA BLAU, 91–106. Jerusalem: Magnes/The Hebrew
University.
———. 2010. Phonology and Morphology of Biblical Hebrew: An Introduction. Winona
Lake, IN: Eisenbrauns.
BLAU, LAJOS. 1894. Zur Einleitung in die heilige Schrift. Budapest: Adolf Alkalay.
BLEVINS, J. 2004. “Klamath Sibilant Degemination: Implications of a Recent Sound Change.”
International Journal of American Linguistics 70 (3): 279–89.
BOLOZKY, SHMUEL. 2013. “Resh: Modern Hebrew.” In Encyclopedia of Hebrew Language and
Linguistics: Volume 3, edited by GEOFFREY KHAN, 389–90. Leiden/Boston: Brill.
BREUER, YOCHANAN. 2013. “Phonology: Rabbinic Hebrew.” In Encyclopedia of Hebrew
Language and Linguistics: Volume 3, edited by GEOFFREY KHAN, 110–13. Leiden/
Boston: Brill.
BRIQUEL-CHATONNET, F., M. DEBIÉ, and L. NEHMÉ, eds. 2015. Le contexte de naissance de
l’écriture arabe. Écrit et é critures araméennes et arabes au 1er millénaire après J.C.
Louvain: Peeters (Orientalia Lovaniensa Analecta).
BRIXHE. 2010. “Linguistic Diversity in Asia Minor During the Empire: Koine and Non-Greek
Languages.” In A Companion to the Ancient Greek Language, edited by EGBERT J.
BAKKER, 228–52. Chichester/Malden, MA: Wiley-Blackwell.
BRØNNO, EINAR. 1943. Studien über hebräische Morphologie und Vokalismus auf Grundlage
der mercatischen Fragmente der zweiten Kolumne der Hexapla des Origenes. Leipzig:
Dt. Morgenländ. Ges.
———. 1956. “The Isaiah Scroll DSIa and the Greek Transliterations of Hebrew.” Zeitschrift
der Deutschen Morgenländischen Gesellschaft 106: 252–58.
———. 1970. Die Aussprache der hebräischen Laryngale nach Zeugnissen des Hieronymus.
Aarhus: Univ. Forl.
BROSELOW, ELLEN. 1988. “Prosodic Phonology and the Acquisition of a Second Language.”
In Linguistic Theory in Second Language Acquisition, edited by SUZANNE FLYNN, and
WAYNE O’NEIL, 295–308. Dordrecht/Boston/London: Kluwer Academic Publishers.
———. 2006. “Phonology.” In Encyclopedia of Arabic Language and Linguistics, edited by
KEES VERSTEEGH, 607–15. Leiden: Brill.
BUBENIK, V. 2007. “Eastern Koines.” In A History of Ancient Greek: From the Beginnings to
Late Antiquity, edited by A.-F. CHRISTIDIS, Cambridge: Cambridge University Press.
BUTCHER, ANDREW, and KUSAY AHMAD. 1987. “Some Acoustic and Aerodynamic
Characteristics of Pharyngeal Consonants in Iraqi Arabic.” Phonetica 44: 156–72.
BUTH, RANDALL. 2014. “The Riddle of Jesus’ Cry from the Cross: The Meaning of ηλι ηλι
- 388 -
λαµα σαβαχθανι (Matthew 27:46) and the Literary Function of ελωι ελωι λειµα
σαβαχθανι (Mark 15:34).” In The Linguistic Environment of First Century Judaea,
edited by RANDALL BUTH, and R. STEVEN NOTLEY, 395–421. Leiden/Boston: Brill.
BUTH, RANDALL, and R. STEVEN NOTLEY, eds. 2014. The Language Environment of First
Century Judaea. Leiden/Boston: Brill.
CAVENAILE, ROBERT. 1958. Corpus Papyrorum Latinarum. Wiesbaden: Harrassowitz.
CEBRIAN, JULI. 2006. “Experience and the Use of Non-Native Duration in L2 Vowel
Categorization.” Journal of Phonetics 34: 372–87.
CELATA, CHIARA, and JESSICA CANCILA. 2010. “Phonological Attrition and the Perception of
Geminate Consonants in the Lucchese Community of San Francisco (CA).” 14 (2): 1–
25.
CERNY, J., PAUL KAHLE, and R. PARKER. 1957. “The Old Coptic Horoscope.” The Journal of
Egyptian Achaeology 43: 86–100.
CHITORAN, IOANA. 2002. “A Perception-Production Study of Romanian Diphthongs and
Glide-Vowel Sequences.” Journal of the International Phonetic Association 32 (2):
203–22.
CHRISTIDIS, A.-F., ed. 2007. A History of Ancient Greek: From the Beginnings to Late
Antiquity. Cambridge: Cambridge University Press.
CHRONZ, M. 2000. “Der Beitrag des Nikolaos von Otranto (Nektarios von Casole) zur
Vermittlung zwischen den Kulturwelten des 13. Jahrhunderts.” In Geistesleben im 13.
Jahrhundert, edited by J. A. AERTSEN, and A. SPEER, 555–73. Berlin: De Gruyter.
CLEMENTS, RUTH. 1997. “Peri Pascha: Passover and the Displacement of Jewish
Interpretation Within Origen’s Exegesis.” Diss., Harvard Divinity School.
———. 2000. “Origen’s Hexapla and Christian-Jewish Encounter in the Second and Third
Centuries.” In Religious Rivalries and the Struggle for Success in Caesarea Maritima,
edited by TERENCE L. DONALDSON, 303–29. Waterloo, Ontario: Wilfrid Laurier
University Press.
COETZEE, ANDRIES W. 1999. Tiberian Hebrew Phonology: Focussing on Consonant Clusters.
Assen: Van Gorcum.
COFFIN, EDNA AMIR, and SHMUEL BOLOZKY. 2005. A Reference Grammar of Modern Hebrew.
COOK, EDWARD M. 2017. “Language Contact and the Genesis of Mishnaic Hebrew.” In The
Edward Ullendorf Lectures in Semitic Philology: Fourth Lecture, Cambridge: Faculty
of Asian and Middle Eastern Studies, University of Cambridge.
COOPER, CRAIG, ed. 2007. Politics of Orality. Leiden/Boston: Brill.
COTTON, HANNAH M., ROBERT G. HOYLAND, JONATHAN J. PRICE, and DAVID J. WASSERSTEIN, eds.
2009. From Hellenism to Islam: Cultural and Linguistic Change in the Roman Near
East. Cambridge: Cambridge University Press.
COWELL, MARK W. 1964. A Reference Grammar of Syrian Arabic (Based on the Dialect of
Damascus). Washington, D.C.: Georgetown University Press.
CRIBIORE, RAFFAELLA. 2005. Gymnastics of the Mind: Greek Education in Hellenistic and
Roman Egypt. Princeton and Oxford: Princeton University Press.
CROSSWHITE, KATHERINE. 2001. Vowel Reduction in Optimality Theory. New York & London:
Routledge.
CROUZEL, HENRI. 1985. Origène. Paris and Namur: Lethielleux and Culture et vérité.
CRUM, W. E. 1942. “An Egyptian Text in Greek Characters.” The Journal of Egyptian
Archaeology 28: 20–31.
CUVIGNY, HÉLÈNE, ed. 2003. La route de Myos Hormos: l’armée romaine dans le désert
oriental d’É gypte. Cairo: Institut français d’archéologie orientale.
DE LANGE, N. R. M. 1976. Origen and the Jews: Studies in Jewish-Christian Relations in
Third-Century Palestine. Cambridge: Cambridge University Press.
DECAEN, VINCENT. 2003. “Hebrew Sonority and Tiberian Contact Anaptyxis: The Case of
- 389 -
Verbs Primae Gutturalis.” Journal of Semitic Studies 48 (1): 35–46.
DECAEN, VINCENT, and WILLIAM J. IDSARDI. 2011. “The Eight Phonemic Vowels of Tiberian
Hebrew: Evidence from Segholate Nouns.” Unpublished ms.: 1–19.
DICKEY, ELEANOR. 2010. “The Creation of Latin Teaching Materials in Antiquity: A Re-
Interpretation of P. Sorb. inv. 2069.” Zeitschrift für Papyrologie und Epigraphik Bd.
175: 188–208.
———. 2012. The Colloquia of the Hermeneumata Pseudodositheana. Cambridge:
———. 2015a. “How Coptic Speakers Learned Latin? A Reconsideration of P.Berol. Inv.
10582.” Zeitschrift für Papyrologie und Epigraphik 193: 65–77.
———. 2015b. “Columnar Translation: An Ancient Interpretive Tool that the Romans Gave
the Greeks.” The Classical Quarterly 65 (2): 807–21.
DICKEY, ELEANOR, and ROLANDO FERRI. 2012. “A New Edition of the Colloquium Harleianum
Fragment in P.Prag. 2.118.” Zeitschrift für Papyrologie und Epigraphik 180: 127–32.
DIRVEN, RENÉ, and MARJOLIJN VERSPOOR, eds. 2004. Cognitive Exploration of Language and
Linguistics. Amsterdam/Philadelphia: John Benjamins Publishing Company.
DMITRIEVA, OLGA. 2012. “Geminate Typology and the Perception of Consonant Duration.”
Diss., Stanford University.
DONALDSON, TERENCE L., ed. 2000. Religious Rivalries and the Struggle for Success in
Caesarea Maritima. Waterloo, Ontario: Wilfrid Laurier University Press.
DONNER, HERBERT, and WOLFGANG RÖLLIG, eds. 1971. Kanaanäische und aramäische
Inschriften. Wiesbaden: Harrassowitz.
DOTAN, ARON. 1985. “Patḥé Ḥaṭfin — a Study in the Evolution of the Tiberian Vocalization”
(in Hebrew). In Avraham Even-Shoshan volume, edited by BEN-ZION LURIA, 157–65.
Jerusalem: Kiryat Sefer.
ECK, WERNER. 2009. “The Presence, Role and Significance of Latin in Epigraphy and Culture
of the Roman Near East.” In From Hellenism to Islam: Cultural and Linguistic Change
in the Roman Near East, edited by HANNAH M. COTTON, ROBERT G. HOYLAND, JONATHAN
J. PRICE, and DAVID J. WASSERSTEIN, Cambridge: Cambridge University Press.
EDZARD, LUTZ. 2001. “Problems with Post-Vocalic Spirantization in Syriac: Cyclic Rule
Ordering vs. ‘Early Phonemization with Paradigmatic Levelling’.” Journal of Semitic
Studies 46 (1): 77–95.
ELDAR, ILAN. 1994. The Art of Correct Reading of the Bible (in Hebrew). Jerusalem: The
Academy of the Hebrew Language.
ELIHAY, J. 2012. The Olive Tree Dictionary: A Transliterated Dictionary of Conversational
Eastern Arabic (Palestinian). Jerusalem: Minerva Publishing House.
ELITZUR, YOEL. 2004. Ancient Place Names in the Holy Land: Preservation and History.
Winona Lake, IN: Eisenbrauns.
———. 2013. “Epigraphic Hebrew: Roman and Byzantine Period.” In Encyclopedia of
Hebrew Language and Linguistics: Volume 1, edited by GEOFFREY KHAN, 843–51.
ELLIOTT, C. J. 1877–1887. “Hebrew Learning among the Fathers.” In A Dictionary of
Christian Biography, edited by WILLIAM SMITH, and HENRY WACE, II, 851–72. London:
J. Murray.
EMERTON, J. A. 1956. “The Purpose of the Second Column of the Hexapla.” Journal of
Theological Studies 7: 79–87.
———. 1971. “A Further Consideration of the Purpose of the Second Column of the
Hexapla.” Journal of Theological Studies 22: 15–28.
EPSTEIN, I., ed. 1935. ˁAbodah Zarah: Translated into English with Notes, Glossary, and
Indices. London: Soncino.
ERWIN, WALLACE M. 2004. A Short Reference Grammar of Iraqi Arabic. Washington, D.C.:
Georgetown University Press.
- 390 -
ESCUDERO, PAOLA, and KATER̆ INA CHLÁDKOVÁ. 2010. “Spanish Listeners’ Perception of
American and Southern British English Vowels.” The Journal of the Acoustical Society
of America 128: 254–60.
ESPY-WILSON, CAROL YVONNE. 1987. “An Acoustic-Phonetic Approach to Speech
Recognition: Application to the Semivowels.” Diss., Massachusetts Institute of
Technology.
FABER, ALICE. 1980. “Genetic Subgrouping of the Semitic Languages.” Diss., The University
of Texas at Austin.
———. 1989. “On the Nature of Proto-Semitic *l.” Journal of the American Oriental Society
109 (1): 33–36.
———. 1992. “Second Harvest: š ibboleθ Revisited (Yet Again).” Journal of Semitic Studies
37: 1–10.
FABRA, LUCRECIA RALLO, and JOAQUÍN ROMERO. 2012. “Native Catalan Learners’ Perception
and Production of English Vowels.” Journal of Phonetics 40: 491–508.
FALLON, PAUL D. 2002. The Synchronic and Diachronic Phonology of Ejectives. New York/
London: Routledge.
———. 2009. “The Velar Ejective in Proto-Agaw.” In Selected Proceedings of the 39th
Annual Conference on African Linguistics, edited by AKINLOYE OJO, and LIOBA MOSHI,
10–22. Somerville, MA: Cascadilla Proceedings Project.
FASSBERG, STEVEN E. 2010. The Jewish Neo-Aramaic Dialect of Challa. Leiden/Boston: Brill.
———. 2012. “Which Semitic Language Did Jesus and Other Contemporary Jews Speak?”
Catholic Biblical Quarterly 74: 263–80.
FERGUSON, CHARLES. 1959. “Diglossia.” Word 15: 325–40.
FIELD, FREDERICK. 1875. Origenis Hexaplorum. Oxford: Clarendon.
FLEGE, JAMES EMIL. 2007. “Language Contact in Bilingualism: Phonetic System Interactions.”
In Laboratory Phonology 9, edited by JENNIFER COLE, and JOSÉ IGNACIO HUALDE, 353–
81. Berlin/New York: Mouton de Gruyter.
FLEGE, JAMES EMIL, OCKE-SCHWEN BOHN, and SUNYOUNG JANG. 1997. “Effects of Experience
on Non-Native Speakers’ Production and Perception of English Vowels.” Journal of
Phonetics 25: 437–70.
FLEGE, JAMES EMIL, IAN R.A. MACKAY, and DIANE MEADOR. 1999. “Native Italian Speakers’
Perception and Production of English Vowels.” The Journal of the Acoustical Society of
America 106: 2973–87.
FLEMMING, EDWARD. 2007. “The Phonetics of Schwa Vowels.” Unpublished ms.: 1–23.
FLESHER, PAUL V. M., and BRUCE CHILTON. 2011. The Targums: A Critical Introduction.
FLORENTIN, MOSHE. 2013. “Samaritan Hebrew: Biblical.” In Encyclopedia of Hebrew
Boston: Brill.
———. 2016. “Samaritan Tradition.” In A Handbook of Biblical Hebrew: Volume 2: Selected
Texts, edited by RANDALL W. GARR, and STEVEN E. FASSBERG, Winona Lake, IN:
Eisenbrauns.
FLUSSER, DAVID. 1988. “At the Right Hand of Power.” In Judaism and the Origins of
Christianity, edited by DAVID FLUSSER, 42–46. Jerusalem: Magnes.
FORDYCE, JAMES F. 1980. “On the Nature of Glottalic and Laryngealized Consonant and
Vowel Systems.” UCLA Working Papers in Phonetics 50: 120–54.
FOURNET, J.-L. 2003. “Langues, écritures et culture dans les praesidia.” In La route de Myos
Hormos: l’armée romaine dans le désert oriental d’Égypte, edited by HÉLÈNE CUVIGNY,
427–500. Cairo: Institut français d’archéologie orientale.
FOX, JOSHUA. 1996. “A Sequence of Vowel Shifts in Phoenician and Other Languages.”
Journal of Near Eastern Studies 55 (1): 37–47.
———. 2003. Semitic Noun Patterns. Winona Lake, IN: Eisenbrauns.
- 391 -
FREEDMAN, H., and MAURICE SIMON, eds. 1961. Midrash Rabba. Hertford, England: Stephen
Austin and Sons, Ltd.
FRIEDRICH, JOHANNES, WOLFGANG RÖLLIG, and MARIA GIULIA AMADASI GUZZO. 1999.
Phönizisch-Punische Grammatik. Roma: Pontificio Istituto Biblico.
GAEBEL, ROBERT E. 1970. “The Greek Word-Lists to Vergil and Cicero.” Bulletin of the John
Rylands Library 52: 284–325.
GAHL, SUSANNE. 2015. “Lexical Competition in Vowel Articulation Revisited: Vowel
Dispersion in the Easy/Hard Database.” Journal of Phonetics 49: 96–116.
GARR, W. RANDALL. 1989. “The Seghol and Segholation in Hebrew.” Journal of Near Eastern
Studies 48: 109–16.
———. 1991. “*ay > a in Targum Onqelos.” Journal of the American Oriental Society 111
(4): 712–19.
GARR, W. RANDALL, and STEVEN E. FASSBERG, eds. 2016. A Handbook of Biblical Hebrew:
Volume 1: Periods, Corpora, and Reading Traditions. Winona Lake, IN: Eisenbrauns.
———. 2016. A Handbook of Biblical Hebrew: Volume 2: Selected Texts. Winona Lake, IN:
Eisenbrauns.
GELLER, M. J. 1983. “More Graeco-Babyloniaca.” Zeitschrift für Assyriologie und
Vorderasiatische Archäologie 73 (1): 114–23.
———. 1997. “The Last Wedge.” Zeitschrift für Assyriologie und Vorderasiatische
Archäologie 87 (1): 43–95.
GENDROT, CÉDRIC, and MARTINE ADDA-DECKER. 2007. “Impact of Duration and Vowel
Inventory Size on Formant Values of Oral Vowels: An Automated Formant Analysis
from Eight Languages.” Proceedings of the 16th International Congress of Phonetic
Sciences 1417–20.
GIANNAKIS, GEORGIOS K., ed. 2014. Encyclopedia of Ancient Greek Language and Linguistics.
GIGNAC, FRANCIS THOMAS. 1976. A Grammar of the Greek Papyri of the Roman and
Byzantine Periods. Milan: Istituto Editoriale Cisalpino - La Goliardica.
GIRGIS, V. 1965. “A New Strategos of the Hermopolite Nome.” Mitteilungen des Deutschen
Archäologischen Instituts 20: 121.
GORDON, CYRUS. 1968. “Northwest Semitic Texts in Latin and Greek Letters.” Journal of the
American Oriental Society 88 (2): 285–89.
GRAFTON, ANTHONY, and MEGAN WILLIAMS. 2006. Christianity and the Transformation of the
Book: Origen, Eusebius, and the Library of Caesarea. Cambridge, MA/London:
Belknap Press of Harvard University Press.
GRAVES, MICHAEL. 2007. Jerome’s Hebrew Philology. Leiden/Boston: Brill.
GREENSTEIN, EDWARD. 1992. “An Introduction to the Generative Phonology of Biblical
Hebrew.” In Linguistics and Biblical Hebrew, edited by WALTER R. BODINE, 29–40.
Winona Lake, IN: Eisenbrauns.
GRENFELL, BERNARD P., and ARTHUR S. HUNT, eds. 1898. The Oxyrhynchus Papyri. London:
The Egypt Exploration Society; Arts and Humanities Research Council by the Egypt
Exploration Society; Egypt Exploration Society with the support of the Arts and
Humanities Research Council and the British Academy.
GRIFFITH, F. LL., and HERBERT THOMPSON, eds. 1904. The Demotic Magical Papyrus of
London and Leiden. London: H. Grevel and Co.
GZELLA, HOLGER. 2015. A Cultural History of Aramaic: Beginnings to the Advent of Islam.
HACKETT, JO ANN. 2008. “Phoenician and Punic.” In The Ancient Languages of Syria-
Palestine and Arabia, edited by ROGER D. WOODARD, 82–102. Cambridge: Cambridge
University Press.
HALÉVY, J. 1901. “L’origine de la transcription du texte hébreu en caractères grecs dans les
Hexaples d’Origène.” Journal Asiatique 9 (17): 335–41.
- 392 -
HAN, M.S. 1992. “The Timing Control of Geminate and Single Stop Consonants in Japanese:
A Challenge for Nonnative Speakers.” Phonetica 49: 102–27.
HANIQUE, IRIS, BARBARA SCHUPPLER, and MIRJAM ERNESTUS. 2010. “Morphological and
Predictability Effects on Schwa Reduction: The Case of Dutch Word-Initial Syllables.”
Proceedings of the 11th Annual Conference of the International Speech Communication
Association (Interspeech 2010) 933–36.
HANSON, R. P. C. 1954. Origen’s Doctrine of Tradition. London: SPCK.
———. 1956. “Interpretations of Hebrew Names in Origen.” Vigiliae Christianae 10: 103–
23.
———. 1959. Allegory and Event: A Study of the Sources and Significance of Origen’s
Interpretation of Scripture. Louisville/London: Westminster John Knox.
HARDISON, DEBRA M., and MIKI MOTOHASHI SAIGO. 2010. “Development of Perception of
Second Language Japanese Geminates: Role of Duration, Sonority, and Segmentation
Strategy.” Applied Psycholinguistics 31: 81–99.
HARRAUER, HERMANN, and PIETER J. SIJPESTEIJN. 1985. Neue Texte aus dem antiken
Unterricht. Wien: Brüder Hollinek.
HARVIAINEN, TAPANI. 1977. On the Vocalism of the Closed Unstressed Syllables in Hebrew.
Helsinki: Finnish Oriental Society.
HATCH, EDWIN, and HENRY A. REDPATH. 1897. A Concordance to the Septuagint and the Other
Greek Versions of the Old Testament (Including the Apocryphal Books). Oxford:
Clarendon.
HAYES-HARB, RACHEL, and KRISTIE DURHAM. 2016. “Native English Speakers’ Perception of
Arabic Emphatic Consonants and the Influence of Vowel Context.” Foreign Language
Annals 49 (3): 557–72.
HAYES, BRUCE. 1989. “Compensatory Lengthening in Moraic Phonology.” Linguistic Inquiry
20 (2): 253–306.
HAYES, F. F. C. 1859. Synopsis of the Grammar of Modern Arabic with a Vocabulary and
Dialogues. Calcutta: Thacker, Spink and Co.
ḤAYYÛḠ. 1897. “Kitāb al-ʔafʕāl ḏawāt ḥurūf al-līn” (in Arabic). In The Weak and Geminate
Verbs in Hebrew by Abû Zakkariyyâ Yaḥyâ ibn Dâwud of Fez known as Ḥayyûḡ, edited
by MORRIS JASTROW, 1–219. Leiden: E.J. Brill.
HEIJMANS, SHAI. 2013. “Greek and Latin Loanwords in Mishnaic Hebrew: Lexicon and
Phonology” (in Hebrew). Diss., Tel-Aviv University.
HENSHKE, YEHUDIT. 2010. “Emphatic Consonants in MS Cambridge of the Mishnah” (in
Hebrew). Leshonenu 72: 421–50.
———. 2013a. “Sephardi Pronunciation Traditions of Hebrew.” In Encyclopedia of Hebrew
Boston: Brill.
———. 2013b. “Tunisia, Pronunciation Traditions.” In Encyclopedia of Hebrew Language
and Linguistics: Volume 3, edited by GEOFFREY KHAN, 860–67. Leiden/Boston: Brill.
HERCUS, L. A. 1986. Victorian Languages: A Late Survey. Canberra: Australian National
University.
HEZSER, CATHERINE. 2001. Jewish Literacy in Roman Palestine. Tübingen: Mohr Siebeck.
HILBERG, ISIDORUS, ed. 1910. Sancti Eusebii Hieronymi Epistulae: Pars I: Epistulae I–LXX.
Vindobonae; Lipsiae: F. Tempsky; G. Freytag.
HILLENBRAND, JAMES M., and MICHAEL J. CLARK. 2001. “Effects of Consonant Environment
on Vowel Formant Patterns.” The Journal of the Acoustical Society of America 109 (2):
748–63.
HOCK, HANS HEINRICH. 1991. Principles of Historical Linguistics. Berlin/New York: Mouton
de Gruyter.
HOFFMAN, LARS MARTIN. 2007. “Der antijüdische Dialog Kata Iudaion des Nikolaos-
Nektarios von Otranto.” Diss., Johannes Gutenberg-Universität.
- 393 -
HONEYBONE, PATRICK, and JOSEPH SALMONS. 2015. The Oxford Handbook of Historical
Phonology. Oxford: Oxford University Press.
HORBURY, WILLIAM, ed. 1999. Hebrew Study from Ezra to Ben-Yehuda. Edinburgh: T and T
Clark.
HORROCKS, GEOFFREY. 2014. Greek: A History of the Language and Its Speakers. Oxford:
Wiley-Blackwell.
HUEHNERGARD, JOHN. 2004. “Afro-Asiatic.” In The Cambridge Encyclopedia of the World’s
Ancient Languages, edited by ROGER D. WOODARD, 138–59. Cambridge: Cambridge
University Press.
———. 2006. “On the Etymology of the Hebrew Relative š ɛ-.” In Biblical Hebrew in Its
Northwest Semitic Setting: Typological and Historical Perspectives, edited by STEVEN
E. FASSBERG, and AVI HURVITZ, 103–125. Jerusalem/Winona Lake, IN: The Hebrew
University Magnes Press/Eisenbrauns.
———. 2011. A Grammar of Akkadian. Winona Lake, Indiana: Eisenbrauns.
———. 2013. “Segholates: Pre-Modern Hebrew.” In Encyclopedia of Hebrew Language and
———. 2015. “Biblical Hebrew Nominal Patterns.” In Epigraphy, Philology, and the
Hebrew Bible: Methodological Perspectives on Philological and Comparative Study of
the Hebrew Bible in Honor of Jo Ann Hackett, edited by JEREMY M. HUTTON, and AARON
D. RUBIN, 25–64. Atlanta: SBL.
———. forthcoming. “Of People and Tongues: Semitic *nis- and *lisān-.”
HUEHNERGARD, JOHN, and CHRISTOPHER WOODS. 2004. “Akkadian and Eblaite.” In The
Cambridge Encyclopedia of the World’s Ancient Languages, edited by R. G. WOODARD,
218–80. Cambridge: Cambridge University Press.
HUSSELMAN, E. M. 1957. “A Palimpsest Fragment from Egypt.” In Studi in onore di Aristide
Calderini e Roberto Paribeni, edited by EDOARDO ARSLAN, 2.453–459. Milan:
Ceschina.
HYMAN, LARRY M. 1985. A Theory of Phonological Weight. Dordrecht, Holland/Cinnaminson,
NJ: Foris Publications.
HYMAN, LARRY M., and IMELDA UDOH. 2007. “Length Harmony in Leggbó: A Counter-
Universal?” In Endangered Languages, edited by PETER K. AUSTIN, and ANDREW
SIMPSON, 73–92. Hamburg: Helmut Buske.
ISAAC, BENJAMIN. 2009. “Latin in Cities of the Roman Near East.” In From Hellenism to
Islam: Cultural and Linguistic Change in the Roman Near East, edited by HANNAH M.
COTTON, ROBERT G. HOYLAND, JONATHAN J. PRICE, and DAVID J. WASSERSTEIN,
JANSSENS, GERARD. 1982. Studies in Hebrew Historical Linguistics Based on Origen’s
Secunda. Leuven: Uitgeverij Peeters.
JASTROW, MARCUS. 1926. A Dictionary of the Targumim, the Talmud Babli and Yerushalmi,
and the Midrashic Literature; with an Index of Scriptural Quotations. New York:
Choreb.
JELLICOE, SIDNEY. 1968. The Septuagint and Modern Study. Oxford: Clarendon.
JESKE, ANDRE R. 2012. “The Perception of English Vowels by Native Spanish Speakers.”
Diss., University of Pittsburgh.
JIN, SU-HYUN, and CHANG LIU. 2014. “Intelligibility of American English Vowels and
Consonants Spoken by International Students in the United States.” Journal of Speech,
Language, and Hearing Research 57: 583–96.
JOSEPH, BRIAN. 2014. “Phonetics.” In Encyclopedia of Ancient Greek Language and
Linguistics, edited by GEORGIOS K. GIANNAKIS, Leiden/Boston: Brill.
JOÜON, P., and T. MURAOKA. 2009. A Grammar of Biblical Hebrew. Roma: Gregorian &
Biblical.
KAHLE, PAUL. 1927. Die Masoreten des Westens. Stuttgart: W. Kohlhammer.
- 394 -
———. 1947. The Cairo Geniza. London: for the British Academy by Oxford University
Press.
———. 1950. “Zur Aussprache des Hebräischen bei den Samaritanern.” In Festschrift Alfred
Bertholet, edited by WALTER BAUMGARTNER, AND OTTO EISSFELDT, 281–86. Tübingen:
Mohr.
———. 1956. “The Masoretic Text of the Bible and the Pronunciation of Hebrew.” Journal
of Semitic Studies 7: 133–53.
KAWAHARA, S. 2007. “Sonorancy and Geminacy.” University of Massachusetts Occasional
Papers in Linguistics 32: 145–86.
KENYON, F. G., and W. E. CRUM. 1937. The Chester Beatty Biblical Papyri. London: Emery
Walker Ltd.
KERR, ROBERT M. 2010. Latino-Punic Epigraphy: A Descriptive Study of the Inscriptions.
Tübingen, Germany: Mohr Siebeck.
KHAN, GEOFFREY. 1987. “Vowel Length and Syllable Structure in the Tiberian Tradition of
Biblical Hebrew.” Journal of Semitic Studies 32: 23–82.
———. 1992. “The Function of the shewa Sign in Vocalized Judaeo-Arabic Texts from the
Genizah.” In Genizah Research after Ninety Years: The Case of Judaeo-Arabic, edited
by JOSHUA BLAU, and STEFAN REIF, 105–11. Cambridge: Cambridge University Press.
———. 1995. “The Pronunciation of the reš in the Tiberian Tradition of Biblical Hebrew.”
Hebrew Union College Annual 66: 67–80.
———. 1997. “Jewish Palestinian Aramaic Phonology.” In Phonologies of Asia and Africa:
Including the Caucasus, edited by ALAN S. KAYE, 103–13. Winona Lake, IN:
Eisenbrauns.
———. 1999. A Grammar of Neo-Aramaic: The Dialect of the Jews of Arbel. Leiden/Boston/
Köln: Brill.
———. 2002. The Neo-Aramaic Dialect of Qaraqosh. Leiden/Boston: Brill.
———. 2004. The Jewish Neo-Aramaic Dialect of Sulemaniyya and Ḥalabja. Leiden/
Boston: Brill.
———. 2005. “Some Parallels in Linguistic Development between Biblical Hebrew and
Neo-Aramaic.” In Semitic Studies in Honor of Edward Ullendorff, edited by GEOFFREY
KHAN, 84–105. Leiden: Brill.
———. 2008a. The Jewish Neo-Aramaic Dialect of Urmi. Piscataway, NJ: Gorgias.
———. 2008b. The Neo-Aramaic Dialect of Barwar. Leiden/Boston: Brill.
———. 2013a. A Short Introduction to the Tiberian Masoretic Bible and Its Reading
Tradition. Piscataway, NJ: Gorgias.
———. 2013b. “Syllable Structure: Biblical Hebrew.” In Encyclopedia of Hebrew Language
and Linguistics: Volume 3, edited by GEOFFREY KHAN, 666–76. Leiden/Boston: Brill.
———. 2013c. “Resh: Pre-Modern Hebrew.” In Encyclopedia of Hebrew Language and
———. 2013d. “Compensatory Lengthening.” In Encyclopedia of Hebrew Language and
———. 2013e. “Vocalization, Babylonian.” In Encyclopedia of Hebrew Language and
———. 2013f. “Vowel Length: Biblical Hebrew.” In Encyclopedia of Hebrew Language and
———. 2013g. “Gaˁya.” In Encyclopedia of Hebrew Language and Linguistics: Volume 2,
edited by GEOFFREY KHAN, 6–10. Leiden/Boston: Brill.
———. 2013h. “Shewa: Pre-Modern Hebrew.” In Encyclopedia of Hebrew Language and
———. 2013i. “Reduction of Vowels: Biblical Hebrew Reading Traditions.” In Encyclopedia
of Hebrew Language and Linguistics: Volume 3, edited by GEOFFREY KHAN, 327–30.
- 395 -
———. 2013j. “Pretonic Lengthening: Biblical Hebrew.” In Encyclopedia of Hebrew
Boston: Brill.
———. 2013k. “Pronominal Suffixes.” In Encyclopedia of Hebrew Language and
———. 2013l. “Epenthesis: Biblical Hebrew.” In Encyclopedia of Hebrew Language and
———. 2016. The Neo-Aramaic Dialect of the Assyrian Christians of Urmi: Volume 1:
Grammar: Phonology and Morphology. Leiden/Boston: Brill.
KIM, D. J. 1972. “A Contrastive Study of English and Korean Phonology (on the Basis of
Aural Perception).” Language Teaching 5: 1–36.
KIPARSKY, PAUL. 2003. “Syllables and Moras in Arabic.” In The Syllable in Optimality
Theory, edited by CAROLINE FÉRY, and RUBEN FLORENTIUS HENDRICUS EDUARDUS VAN DE
VIJVER, 147–82. Cambridge: Cambridge University Press.
KIRK, G. E. 1938. “Three Greek Inscriptions from the Southern Desert.” Palestine Exploration
Quarterly 236–39.
KNOBLOCH, FREDERICK W. 1995. “Hebrew Sounds in Greek Script: Transcriptions and Related
Phenomena in the Septuagint, with Special Focus on Genesis.” Diss., University of
Pennsylvania.
KOGAN, LEONID. 2011. “Reconstructing Proto-Semitic and Models of Classification.” In The
Semitic Languages: An International Handbook, edited by STEFAN WENINGER, 54–151.
Berlin/Boston: Walter de Gruyter Mouton.
KRAHMALKOV, CHARLES R. 2001. A Phoenician-Punic Grammar. Leiden/Boston/Köln: Brill.
KRAMER, JOHANNES. 1977. “Sprachlicher Kommentar zum lateinisch-griechischen Glossar P.
Lond. II 481.” Zeitschrift für Papyrologie und Epigraphik 26: 231–38.
———. 1983. Glossaria Bilinguia in Papyris et Membranis Reperta. Bonn: R. Habelt.
———. 2001. Glossaria Bilinguia Altera (C. Gloss. Biling. II). München/Leipzig: Saur.
———. 2007. Vulgärlateinische Alltagsdokumente auf Papyri, Ostraka, Täfelchen und
Inschriften. Berlin/New York: Walter de Gruyter.
KUTSCHER, EDWARD YECHEZKEL. 1965. “Contemporary Studies in North-Western Semitic” (in
Hebrew). Journal of Semitic Studies 10 (1): 21–51.
———. 1969. “Articulation of the Vowels U, I in Transcriptions of Biblical Hebrew, in
Galilean Aramaic and in Mishnaic Hebrew” (in Hebrew). In Benjamin De Vries
Memorial Volume, edited by EZRA Z. MELAMED, 218–51. Tel-Aviv: Tel-Aviv University.
———. 1974. The Language and Linguistic Background of the Complete Isaiah Scroll.
Leiden: Brill.
———. 1976. Studies in Galilean Aramaic. Ramat-Gan: Bar-Ilan University.
LACAU, P. 1934. “Un graffito égyptien d’Abydos écrit en lettres grecques.” Études de
Papyrologie 2: 229–46.
LADEFOGED, PETER. 2001. A Course in Phonetics. Fort Worth: Harcourt College Publishers.
LADEFOGED, PETER, and IAN MADDIESON. 1996. The Sounds of the World’s Languages.
Oxford/Cambridge: Wiley-Blackwell.
LAMBDIN, THOMAS O. 1985. “Philippi’s Law Reconsidered.” In Biblical Studies Presented to
Samuel Iwry, edited by ANN KORT, and SCOTT MORSCHAUSER, 135–45. Winona Lake,
IN: Eisenbrauns.
———. 2006. Introduction to Classical Ethiopic (Geˁez). Winona Lake, IN: Eisenbrauns.
LAMBDIN, THOMAS O., and JOHN HUEHNERGARD. 2000. The Historical Grammar of Classical
Hebrew. Unpublished ms.
LEVINE, LEE I. 1975. Caesarea Under Roman Rule. Leiden: E. J. Brill.
LIETZMANN, HANS. 1950. The Founding of the Church Universal. New York: Charles
Scriber’s Sons.
MALONE, JOSEPH L. 1971. “Wave Theory, Rule Ordering, and Hebrew-Aramaic Segolation.”
- 396 -
Journal of the American Oriental Society 91: 44–66.
———. 1993. Tiberian Hebrew Phonology. Winona Lake, IN: Eisenbrauns.
MARCOS, NATALIO FERNÁNDEZ. 2001. The Septuagint in Context: Introduction to the Greek
Versions of the Bible. Boston/Leiden: Brill Academic Publishers.
MARGOLIS, MAX L. 1909. “The Pronunciation of the ‫שׁוָא‬ ְ according to New Hexaplaric
Material.” American Journal of Semitic Languages 26: 62–70.
MARTIN, ANDREW, and SHARON PEPERKAMP. 2011. “Speech Perception and Phonology.” In The
Blackwell Companion to Phonology: Volume 4: Phonological Interfaces, edited by
MARC VAN OOSTENDROP, COLIN J. EWEN, ELIZABETH HUME, and KEREN RICE, 2334–56.
Malden, MA: Wiley-Blackwell.
MARTIN, MATTHEW J. 2004. “Origen’s Theory of Language and the First Two Columns of the
Hexapla.” The Harvard Theological Review 97 (1): 99–106.
———. 2007. “Writing Divine Speech: Greek Transliterations of Near Eastern Languages in
the Hellenistic East.” In Politics of Orality, edited by CRAIG COOPER, 251–73. Leiden:
Brill.
MARTÍNEZ, FERNÁNDEZ. 1989. “Vowel Length in Vulgar Latin: Transphonologization of
Quantitative into Qualitative Distinctions.” Itaca 5: 101–13.
MCCARTHY, JOHN J. 1981. “The Representation of Consonant Length in Hebrew.” Linguistic
Inquiry 12: 322–27.
———. 1986. “OCP Effects: Gemination and Antigemination.” Linguistic Inquiry 17 (2):
207–63.
———. 1994. “The Phonetics and Phonology of Semitic Pharyngeals.” In Phonological
Structure and Phonetic Form Papers in Laboratory Phonology III, edited by PATRICIA
A. KEATING, 191–233. Cambridge: Cambridge University Press.
———. 2005. “The Length of Stem-Final Vowels in Colloquial Arabic.” In Perspectives on
Arabic Linguistics XVII-XVIII, edited by MOHAMMAD T. ALHAWARY, and ELABBAS
BENMAMOUN, 1–26. Amsterdam: John Benjamins.
MERCATI, GIOVANNI. 1895. “D’un palimpsesto Ambrosiano contenente i Salmi esapli.” Atti
Acad. Scienze Torino 31:655–676
———. 1947. “Il problema della colonna II dell’Esaplo.” Biblica XXVIII: 1–30, 173.
———. 1958. Psalterii Hexapli Reliquiae. Vatican City: Bybliotheca Vaticana.
———. 1965. Psalterii Hexapli Reliquiae, osservazioni: Commento critico al testo dei
frammenti esaplari. Vatican City: Bybliotheca Vaticana.
MEYER, RUDOLF. 1958. “Bemerkungen zu der hebräischen Aussprachetradition von Chirbet
Qumrān.” Zeitschrift fur die alttestamentliche Wissenschaft 70: 39–48.
MILIK, J.-T. 1967. “Inscription araméenne en caractères grecs de Doura-Europos et une
dédicace grecque de Cordoue.” Syria 44: 289–306.
MILLARD, ALAN. 2013. “Transcriptions into Cuneiform.” In Encyclopedia of Hebrew
Boston: Brill.
MISHOR, MORDECHAY. 2007. “Hebrew in the Babylonian Incantation Bowls” (in Hebrew). In
Sha’arei Lashon: Studies in Hebrew, Aramaic and Jewish Languages Presented to
Moshe Bar-Asher, edited by AHARON MAMAN, STEVEN E. FASSBERG, and YOCHANAN
BREUER, 204–27. Jerusalem: Bialik Institute.
MITCHELL, T. F. 1990. Pronouncing Arabic, I. Oxford: Clarendon.
MOMMSEN, THEODOR, ed. 1908. Eusebius Werke zweiter Band: Die Kirchengeschichte.
Leipzig: J.C. Hinrichs’sche Buchhandlung.
MOR, URI. 2013. “Guttural Consonants: Pre-Masoretic.” In Encyclopedia of Hebrew
Boston: Brill.
———. 2015. Judean Hebrew: The Language of the Hebrew Documents from Judea between
the First and the Second Revolts (in Hebrew). Jerusalem: The Academy of the Hebrew
- 397 -
Language.
MORAG, SHELOMO. 1958. “A Special Type of Evolution.” Proceedings of the VIII
International Congress of Linguistics 425–28.
———. 1963. The Hebrew Language Tradition of the Yemenite Jews (in Hebrew). Jerusalem:
The Academy of the Hebrew Language.
MORENO, FRANCISCO FUENTES. 2008. “‘Brevio/corripio’: tecnicismos prosódicos en los
gramáticos latinos.” Año 19: 127–41.
MOUTSOULAS, E. 1973. “Τὸ ‘περὶ µέτρων καὶ σταθµῶν’ ἔργον Ἐπιφανίου τοῦ Σαλαµῖνος.”
Θεολογία 44: 157–98.
MUNRO, MURRAY J. 1993. “Productions of English Vowels by Native Speakers of Arabic:
Acoustic Measurements and Accentedness Ratings.” Language and Speech 36 (1): 39–
66.
MUNRO, MURRAY J., JAMES EMIL FLEGE, and IAN R.A. MACKAY. 1996. “The Effects of Age of
Second Language Learning on the Production of English Vowels.” Applied
Psycholinguistics 17: 313–34.
MURRAY, MICHELE. 2000. “Jews and Judaism in Caesarea Maritima.” In Religious Rivalries
and the Struggle for Success in North Africa, edited by TERENCE L. DONALDSON, 127–
52. Waterloo, Ontario: Wilfrid Laurier University Press.
MURTONEN, A. 1988. Hebrew in Its West Semitic Setting: A Comparative Survey of Non-
Masoretic Hebrew Dialects and Traditions: Part One: A Comparative Lexicon. Leiden/
New York/København/Köln: E.J. Brill.
MUSSIES, G. 1976. “Greek in Palestine and the Diaspora.” In The Jewish People in the First
Century: Historical Geography, Political History, Social, Cultural and Religious Life
and Institutions, edited by SHMUEL SAFRAI, and M. STERN, Philadelphia: Fortress.
NAEH, SHLOMO. 1992a. “En em lammasoret, or Did the Tannaim Interpret the Biblical Text
Not according to Its Received Reading?” (Hebrew). Tarbiz 62 (1): 401–48.
———. 1992b. “Between Grammar and Lexicography” (in Hebrew). In Language Studies
V–VI: Israel Yeivin Festschrift, edited by MOSHE BAR-ASHER, 297–306. Jerusalem:
Magnes.
———. 1993. “En em lammasoret — Second Time” (Hebrew). Tarbiz 62 (3): 455–62.
———. 2013. “Two Common Issues in Mishnaic Hebrew” (in Hebrew). In Talmudic Studies
II, edited by MOSHE BAR-ASHER, 369–92. Jerusalem: The Hebrew University Magnes
Press.
NAUTIN, PIERRE. 1977. Origène: Sa vie et son œuvre. Paris: Beauchesne.
NAVEH, JOSEPH. 1992a. “Hebrew versus Aramaic in the Epigraphic Finds of the Second
Temple — Bar-Kokhba Period / 135” (in Hebrew). Leshonenu 56: 301–16.
———. 1992b. “Hebrew versus Aramaic in the Epigraphic Finds after the Bar-Kokhba
Period” (in Hebrew). Leshonenu 57: 17–38.
NEWMAN, DANIEL L., and JO VERHOEVEN. 2002. “Frequency Analysis of Arabic Vowels in
Connected Speech.” Antwerp Papers in Linguistics 100: 77–86.
NIMS, CHARLES F., and RICHARD C. STEINER. 1983. “A Paganized Version of Psalm 20:2–6
from the Aramaic Text in Demotic Script.” Journal of the American Oriental Society
103: 261–74.
NÖLDEKE, THEODOR. 1904. Compendious Syriac Grammar. Translated by PETER T. DANIELS.
Winona Lake, IN: Williams & Norgate/Eisenbrauns.
NOY, DAVID. 1993. Jewish Inscriptions of Western Europe. Cambridge: Cambridge University
Press.
ORLINSKY, H. M. 1936. “The Columnar Order of the Hexapla.” Jewish Quarterly Review 27:
137–49.
PALFREYMAN, DAVID, and MUHAMED AL-KHALIL. 2003. “‘A Funky Language for Teenzz to
Use:’ Representing Gulf Arabic in Instant Messaging.” Journal of Computer-Mediated
Communication 9 (1): 23–44.
- 398 -
PARADIA, MARIA, and NAPOLEAN MITSIS. 2013. “The Distinction between laiko and logio as a
Particular Characteristic of the Modern Greek Language: Historical Interpretation,
Contemporary Function and Didactic Usage.” In Greek Research in Australia:
Proceedings of the Biennial International Conference of Greek Studies, Flinders
University June 2011, edited by M. TSIANIKAS, and N. MAADAD, 377–87. Adelaide:
Flinders University Department of Language Studies - Modern Greek.
PEARCE, MARY. 2008. “Vowel Harmony Domains and Vowel Undershoot.” UCL Working
Papers in Linguistics 20: 115–40.
———. 2012. “Effects of Harmony on Reduction in Kera.” Linguistic Variation 12 (2): 292–
320.
PETROUNIAS, E. B. 2007a. “The Pronunciation of Ancient Greek: Evidence and Hypotheses.”
In A History of Ancient Greek: From the Beginnings to Late Antiquity, edited by A.-F.
CHRISTIDIS, 545–55. Cambridge: Cambridge University Press.
———. 2007b. “The Pronunciation of Classical Greek.” In A History of Ancient Greek:
From the Beginnings to Late Antiquity, edited by A.-F. CHRISTIDIS, 556–70. Cambridge:
———. 2007c. “Development in Pronunciation during the Hellenistic Period.” In A History
of Ancient Greek: From the Beginnings to Late Antiquity, edited by A.-F. CHRISTIDIS,
599–609. Cambridge: Cambridge University Press.
PISKE, THORSTEN, JAMES EMIL FLEGE, and IAN R.A. MACKAY. 2002. “The Production of
English Vowels by Fluent Early and Late Italian-English Bilinguals.” Phonetica 59: 49–
71.
PRETZL, OTTO. 1932. “Die Aussprache des Hebräischen nach der zweiten Kolumne der
Hexapla des Origenes.” Biblische Zeitschrift 20: 4–22.
PRICE, JONATHAN J., and SHLOMO NAEH. 2009. “On the Margins of Culture: The Practice of
Transcription in the Ancient World.” In From Hellenism to Islam: Cultural and
Linguistic Change in the Roman Near East, edited by HANNAH M. COTTON, ROBERT G.
HOYLAND, JONATHAN J. PRICE, and DAVID J. WASSERSTEIN, 257–88. Cambridge:
QIMRON, ELISHA. 1986. The Hebrew of the Dead Sea Scrolls. Atlanta, Georgia: Scholars.
QUECKE, HANS. 1997. “Eine griechisch-ägyptische Wörterliste vermutlich des 3. Jh. v. Chr. (P.
Heid. Inv.-Nr. G 414).” Zeitschrift für Papyrologie und Epigraphik Bd. 116: 67–80.
RAHLFS, ALFRED, ed. 2008. Psalmi cum Odis. Göttingen: Vandenhoeck & Ruprecht.
RAHMANI, L. Y. 1994. A Catalogue of Jewish Ossuaries in the Collections of the State of
Israel. Jerusalem: The Israel Antiquities Authority, The Israel Academy of Sciences and
Humanities.
RAY, J. 2007. “Greek, Egyptian, and Coptic.” In A History of Ancient Greek: From the
Beginnings to Late Antiquity, edited by A.-F. CHRISTIDIS, 811–18. Cambridge:
RECASENS, DANIEL, and AINA ESPINOSA. 2006. “Dispersion and Variability of Catalan Vowels.”
Speech Communication 48: 645–66.
RENDSBURG, GARY A. 2013a. “Phonology: Biblical Hebrew.” In Encyclopedia of Hebrew
Boston: Brill.
———. 2013b. “Shibboleth.” In Encyclopedia of Hebrew Language and Linguistics: Volume
3, edited by GEOFFREY KHAN, 556–57. Leiden/Boston: Brill.
REVELL, E. J. 1970. Hebrew Texts with Palestinian Vocalization. Toronto: University of
Toronto Press.
REYMOND, ERIC D. 2014. Qumran Hebrew: An Overview of Orthography, Phonology, and
Morphology. Atlanta: Society of Biblical Literature.
RICHTER, TONIO SEBASTIAN. 2002. “Miscellanea Magica.” The Journal of Egyptian
Achaeology 88: 247–52.
- 399 -
———. 2009. “Greek, Coptic and the ‘Language of the Hijra’: The Rise and Decline of the
Coptic Language in Late Antique and Medieval Egypt.” In From Hellenism to Islam:
Cultural and Linguistic Change in the Roman Near East, edited by HANNAH M. COTTON,
ROBERT G. HOYLAND, JONATHAN J. PRICE, and DAVID J. WASSERSTEIN, 401–46.
ROBERTS, ALEXANDER, JAMES DONALDSON, and A. CLEVELAND COXE, eds. 1885. Ante-Nicene
Fathers. Buffalo, N.Y.: Christian Literature Publishing Co.
RÖLLIG, WOLFGANG. 1960. “Griechische Eigennamen in Texten der babylonischen Spätzeit.”
Orientalia, NOVA SERIES 29 (4): 376–91.
ROSÉN, HAIIM B. 1963. “Palestinian “koine” [Greek] in Rabbinic Illustration.” Journal of
Semitic Studies 8: 56–72.
ROSENHOUSE, JUDITH, NOAM AMIR, and OFER AMIR. 2014. “Vowel Systems of Quantity
Languages Compared: Arabic Dialects and Other Languages.” Proceedings of Meetings
on Acoustics 21: 1–10.
RUBIN, AARON D. 2014. The Jibbali (Shaḥri) Language of Oman: Grammar and Texts.
RYZHIK, MICHAEL. 2008. The Traditions of Mishnaic Hebrew in Italy: According to the
Medieval Jewish Rituals (in Hebrew). Jerusalem: Bialik Institute.
———. 2013. “Italy, Pronunciation Traditions.” In Encyclopedia of Hebrew Language and
SAFRAI, SHMUEL. 1976. “Education and the Study of the Torah.” In The Jewish People in the
First Century: Historical Geography, Political History, Social, Cultural and Religious
Life and Institutions, edited by SHMUEL SAFRAI, and M. STERN, Philadelphia: Fortress.
SAFRAI, SHMUEL, and M. STERN, eds. 1976. The Jewish People in the First Century: Historical
Geography, Political History, Social, Cultural and Religious Life and Institutions.
Philadelphia: Fortress.
SAFRAI, ZE’EV. 1990. “The Origins of Reading the Aramaic Targum in Synagogue.” Immanuel
24/25: 187–93.
SALVESEN, ALISON. 1991. Symmachus in the Pentateuch. Manchester: Journal of Semitic
Studies.
SANKER, CHELSEA. 2015. “Patterns of Misperception of Arabic Consonants.” Proceedings of
the Annual Meeting of the Berkeley Linguistics Society 41: 447–71.
SATZINGER, HELMUT. 1975. “The Old Coptic Schmidt Papyrus.” Journal of the American
Research Center in Egypt 12: 37–50.
SAVILE, HENRY. 1611. Του εν αγιοις πατρος ηµων Ιωαννου αρχιεπισκοπου
Κωνσταντινουπολεως του Χρυσοστοµου των ευρισκοµενων. Etonae, in Collegio Regali:
Ioannes Norton, in Graecis &c., Regius Typographus.
SCAPPATICCIO, MARIA CHIARA. 2015. Artes grammaticae in frammenti. Berlin/Boston: De
Gruyter.
SCHAFF, PHILIP, ed. 1889. A Select Library of the Nicene and Post-Nicene Fathers of the
Christian Church: Saint Chrysostom: On the Priesthood; Ascetic Treatises; Select
Homilies and Letters; Homilies on the Statues. New York: The Christian Literature
Company.
SCHWABE, MOSHE, and BARUCH LIFSHITZ. 1974. Beth She'arim: Volume II: The Greek
Inscriptions. Jerusalem: Masada.
SCHWARZWALD, ORA (RODRIGUE). 2013. “Consonant Clusters: Modern Hebrew.” In
Encyclopedia of Hebrew Language and Linguistics: Volume 1, edited by GEOFFREY
KHAN, 572–79. Leiden/Boston: Brill.
SEDERHOLM, V. H. 2006. Papyrus British Museum 10808 and Its Cultural and Religious
Setting. Leiden: Brill.
SHARVIT, SHIMON. 2016. A Phonology of Mishnaic Hebrew: Analyzed Materials (in Hebrew).
Jerusalem: The Academy of the Hebrew Language.
- 400 -
SIEGFRIED, CARL. 1884. “Die Aussprache des Hebräischen bei Hieronymus.” Zeitschrift für
die alttestamentliche Wissenschaft 4: 34–83.
SILVERMAN, DANIEL. 2011. “Schwa.” In The Blackwell Companion to Phonology: Volume 1:
General Issues and Segmental Phonology, edited by MARC VAN OOSTENDROP, COLIN J.
EWEN, ELIZABETH HUME, and KEREN RICE, 628–42. Malden, MA: Wiley-Blackwell.
SISINNI, BIANCA, PAOLA ESCUDERO, and MIRKO GRIMALDI. 2014. “The Perception of American
English Vowels by Salento Italian Adult Listeners: Longitudinal Development in the
Classroom Context.” Proceedings of the International Symposium on the Acquisitions
of Second Language Speech Concordia Working Papers in Applied Linguistics 5: 709–
21.
SJÖRS, AMBJÖRN. 2015. “The History of Standard Negation in Semitic.” Diss., Uppsala
University.
SOKOLOFF, MICHAEL. 2011. “Jewish Palestinian Aramaic.” In The Semitic Languages: An
International Handbook, edited by STEFAN WENINGER, 610–19. Berlin/Boston: Walter
de Gruyter Mouton.
SOLI, SIGFRID D. 1981. “Second Formants in Fricatives: Acoustic Consequences of Fricative-
Vowel Coarticulation.” The Journal of the Acoustical Society of America 70: 976–84.
SOLLBERGER, EDMOND. 1962. “Graeco-Babyloniaca.” IRAQ 24 (1): 63–72.
SPARKS, H. F. D. 1959. “The Symbolical Interpretation of ‘Lebanon’ in the Fathers.” Journal
of Theological Studies 10 (2): 264–79.
SPEISER, EPHRAIM A. 1925–1926. “The Pronunciation of Hebrew according to the
Transliterations in the Hexapla.” Jewish Quarterly Review 16: 343–82.
———. 1932–1933. “The Pronunciation of Hebrew Based Chiefly on the Transliterations in
the Hexapla (Continued).” Jewish Quarterly Review 23: 233–65.
———. 1934–1935. “The Pronunciation of Hebrew Based Chiefly on the Transliterations in
the Hexapla (Continued).” Jewish Quarterly Review 24: 9–46.
SPERBER, ALEXANDER. 1937. “Hebrew Based upon Greek and Latin Transliterations.” Hebrew
Union College Annual 12–13: 103–274.
———. 1966. A Historical Grammar of Biblical Hebrew: A Presentation of Problems with
Suggestions to Their Solution. Leiden: E.J. Brill.
SPERBER, DANIEL. 2012. Greek in Talmudic Palestine. Ramat-Gan: Bar-Ilan University.
STAPLES, W. E. 1939. “The Second Column of Origen’s Hexapla.” Journal of the American
Oriental Society 59: 71–80.
STEINER, RICHARD C. 1982. Affricated Ṣade in the Semitic Languages. New York: The
American Academy for Jewish Research.
———. 2005. “On the Dating of Hebrew Sound Changes (*Ḫ > Ḥ and *Ġ > ˁ) and Greek
Translations (2 Esdras and Judith).” Journal of Biblical Literature 124 (2): 229–67.
———. 2007. “Variation, Simplifying Assumptions, and the History of Spirantization in
Aramaic and Hebrew.” In Sha'arei Lashon; Studies in Hebrew, Aramaic and Jewish
Languages Presented to Moshe Bar-Asher, edited by AHARON MAMAN, STEVEN E.
FASSBERG, and Y. BREUER, 52–65. Jerusalem: Bialik Institute.
———. 2011. “Ḫ > Ḥ in Assyria and Babylonia.” In A Common Cultural Heritage: Studies
on Mesopotamia and the Biblical World in Honor of Barry L. Eichler, edited by GRANT
FRAMGE et al., 195–206. Bethesda, MD: CDL.
STEINER, RICHARD C., and CHARLES F. NIMS. 2017. The Aramaic Text in Demotic Script: Text,
Translation, and Notes. Unpublished ms.
SUCHARD, BENJAMIN D. 2016. “The Development of the Biblical Hebrew Vowels.” Diss.,
Universiteit Leiden.
TAYLOR, C. 1900. Hebrew-Greek Cairo Genizah Palimpsests From the Taylor-Schechter
Collection. Cambridge: Cambridge University Press.
TEODORSSON, SVEN TAGE. 1977. The Phonology of Ptolemaic Koine. Milan: Istituto Editoriale
Cisalpino - La Goliardica.
- 401 -
THOMASSEN, EINAR, and JOHANNES VAN OORT, eds. 2009. The Panarion of Epiphanius of
Salamis: Book I. Translated by Frank Williams. Leiden/Boston: Brill.
THOMPSON-PANOS, KARYN, and THOMAS-RUŽIĆ. 1983. “The Least You Should Know about
Arabic: Implications for the ESL Writing Instructor.” TESOL Quarterly 17 (4): 609–23.
THOMPSON, EDWARD MAUNDE. 1966. A Handbook of Greek and Latin Paleography. Chicago:
Ares Publishers.
TROFIMOVICH, PAVEL, WENDY BAKER, and MOLLY MACK. 2001. “Context- and Experience-
Based Effects on the Learning of Vowels in a Second Language.” Studies in Linguistic
Sciences 31 (2): 167–86.
TSUKADA, KIMIKO, DAVID BIRDSONG, ELLEN BIALYSTOK, MOLLY MACK, HYEKYUNG SUNG, and
JAMES EMIL FLEGE. 2005. “A Developmental Study of English Vowel Production and
Perception by Native Korean Adults and Children.” Journal of Phonetics 33: 263–90.
TURNAGE, MARC. 2014. “The Linguistic Ethos of the Galilee in the First Century C.E.” In The
Linguistic Environment of First Century Judaea, edited by RANDALL BUTH, and R.
STEVEN NOTLEY, 110–81. Leiden/Boston: Brill.
VAN BERGEM, DICK R. 1991. “Acoustic and Lexical Vowel Reduction.” In ESCA Workshop on
Phonetics and Phonology of Speaking Styles, 1–5. Barcelona, Spain: ISCA Archive.
———. 1995. “Experimental Evidence for a Comprehensive Theory of Vowel Reduction.” In
4th European Conference on Speech Communication and Technology EUROSPEECH
‘95, 1319–22. Madrid, Spain: ISCA Archive.
VAN OORT, JOHANNES, and EINAR THOMASSEN, eds. 2013. The Panarion of Epiphanius of
Salamis: Books II and III. De Fide. Translated by Frank Williams. Leiden/Boston: Brill.
VAN OOSTENDORP, MARC. 2005. “Mora Theory.” Unpublished ms.: 1–8.
———. 2013. “Shewa: The Term schwa in Modern Linguistics.” In Encyclopedia of Hebrew
Boston: Brill.
VERMES, G. 1966. Review of Les devanciers d’Aquila, by DOMINIQUE BARTHÉ LEMY. Journal
of Semitic Studies 11: 261–64.
VERSTEEGH, KEES, ed. 2006. Encyclopedia of Arabic Language and Linguistics. Leiden: Brill.
VON SODEN, WOLFRAM. 1995. Grundriss der akkadischen Grammatik. 3rd ed. Roma:
Pontificio Istituto Biblico.
WATSON, JANET C. E. 2002. The Phonology and Morphology of Arabic. Oxford: Oxford
University Press.
———. 2007. “Syllabification Patterns in Arabic Dialects: Long Segments and Mora
Sharing.” Phonology 24: 335–56.
———. 2012. The Structure of Mehri. Wiesbaden: Harrassowitz.
WATSON, JANET C. E., and ALEX BELLEM. 2010. “A Detective Story: Emphatics in Mehri.”
Proceedings of the Seminar for Arabian Studies: Papers from the Forty-Third Meeting
of the Seminar for Arabian Studies Held at the British Museum 40: 345–55.
WENINGER, STEFAN, ed. 2011a. The Semitic Languages: An International Handbook. Berlin/
Boston: Walter de Gruyter Mouton.
———. 2011b. “Reconstructive Morphology.” In The Semitic Languages: An International
Handbook, edited by STEFAN WENINGER, 151–78. Berlin/Boston: Walter de Gruyter
Mouton.
———. 2011c. “Old Ethiopic.” In The Semitic Languages: An International Handbook,
edited by STEFAN WENINGER, 1124–41. Berlin/Boston: Walter de Gruyter Mouton.
WESTENHOLZ, AAGE. 1991. “The Phoneme /o/ in Akkadian.” Zeitschrift für Assyriologie und
Vorderasiatische Archäologie 81: 10–19.
———. 2007. “The Graeco-Babyloniaca Once Again.” Zeitschrift für Assyriologie und
Vorderasiatische Archäologie 97 (2): 262–313.
WISE, MICHAEL OWEN. 2015. Language and Literacy in Roman Judaea: A Study of the Bar
Kokhba Documents. New Haven and London: Yale University Press.
- 402 -
WUTZ, FRANZ. 1914. Onomastica Sacra. Leipzig: J. C. Hinrichs.
———. 1925a. Die Psalmen textkritisch untersucht. München: Kösel and Pustet.
———. 1925b. Die Transkriptionen von der Septuaginta bis zu Hieronymus. Stuttgart: W.
Kohlhammer.
———. 1933. Die Transkriptionen von der Septuaginta bis zu Hieronymus. Stuttgart: W.
Kohlhammer.
YA’AKOV, DORON. 2013. “Yemen, Pronunciation Traditions.” In Encyclopedia of Hebrew
Boston: Brill.
YAHALOM, JOSEPH. 2016. “Palestinian Tradition.” In A Handbook of Biblical Hebrew: Volume
2: Selected Texts, edited by RANDALL W. GARR, and STEVEN E. FASSBERG, Winona Lake,
IN: Eisenbrauns.
YANG, BYUNGGON. 1996. “A Comparative Study of American English and Korean Vowels
Produced by Male and Female Speakers.” Journal of Phonetics 24: 245–61.
YEIVIN, ISRAEL. 1985. The Hebrew Language Tradition as Reflected in the Babylonian
Vocalization (in Hebrew). Jerusalem: The Academy of the Hebrew Language.
YENI-KOMSHIAN, GRACE H., and SIGFRID D. SOLI. 1981. “Recognition of Vowels from
Information in Fricatives: Perceptual Evidence of Fricative-Vowel Coarticulation.” The
Journal of the Acoustical Society of America 70: 966–75.
YUDITSKY, ALEXEY. 2005. “Reduced Vowels in the Transcriptions from Hebrew in the
Hexapla” (in Hebrew). Leshonenu 67: 121–41.
———. 2006. “New Readings of MS O39 from the Ambrosiana Library” (in Hebrew).
Leshonenu 68: 63–71.
———. 2007a. “The Grammar of the Hebrew of Origen’s Transliterations” (in Hebrew).
Diss., Ben-Gurion University.
———. 2007b. “On Origen’s Transliterations as Preserved in the Works of the Church
Fathers” (in Hebrew). Leshonenu 69: 301–10.
———. 2008a. “The Weak Consonants in the Language of the Dead Sea Scrolls and in the
Hexapla Transliterations.” In Conservatism and Innovation in the Hebrew Language of
the Hellenistic Period (Studies on the Texts of the Desert of Judah 73), edited by JAN
JOOSTEN, and JEAN-SÉBASTIEN REY, 233–39. Leiden: Brill.
———. 2008b. “Gutturals in the Tradition of Origen’s Transcriptions” (in Hebrew). In
Iggud: Selected Essays in Jewish Studies, edited by TAMAR ALEXANDER-FRIZER, ZIVA
AMISHAI-MAISELS, DAN LAOR, ORA SCHWARTWALD, and YOSEF TOBI, 3–13. Jerusalem:
World Union of Jewish Studies.
———. 2013. “Transcription into Greek and Latin Script: Pre-Masoretic Period.” In
Encyclopedia of Hebrew Language and Linguistics: Volume 3, edited by GEOFFREY
KHAN, 803–22. Leiden/Boston: Brill.
———. 2016. “Hebrew in Greek and Latin Transcriptions.” In A Handbook of Biblical
Hebrew: Volume 2: Selected Texts, edited by RANDALL W. GARR, and STEVEN E.
FASSBERG, 99–116. Winona Lake, IN: Eisenbrauns.
———. 2017. A Grammar of the Hebrew of Origen’s Transcriptions (in Hebrew). Jerusalem:
The Academy of the Hebrew Language.
ZEMÁNEK, PETR. 1996. The Origins of Pharyngealization in Semitic. Praha: Enimga
Corporation.
ZUNTZ, G. 1943. “On the Opening Sentence of Melito’s Paschal Homily.” The Harvard
Theological Review 36 (4): 299–315.
- 403 -

The Second Column Secunda of Origens Hex

Uploaded by

Copyright:

Available Formats

The Second Column Secunda of Origens Hex

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Second Column Secunda of Origens Hex

Uploaded by

Copyright:

Available Formats

COPYRIGHT

BENJAMIN PAUL KANTOR

THE SECOND COLUMN (SECUNDA) OF ORIGEN'S HEXAPLA

NA'AMA PAT-EL, SUPERVISOR

JOHN HUEHNERGARD, CO-SUPERVISOR

BENJAMIN PAUL KANTOR

THE UNIVERSITY OF TEXAS AT AUSTIN

FOR THE DEGREE OF

THE UNIVERSITY OF TEXAS AT AUSTIN

I truly regard it as a privilege to have been one of his ﬁnal students.

the Biblioteca to give me such an opportunity has contributed greatly to my dissertation

diverse parts of the world.

she never threatened to burn my manuscripts (see note 332).

BEǊAMIN PAUL KANTOR, PH.D.

THE UNIVERSITY OF TEXAS AT AUSTIN, 2017

SUPERVISOR: NA'AMA PAT-EL

CO-SUPERVISOR: JOHN HUEHNERGARD

(Secunda) of Origen's (185–254 CE) Hexapla, which constitutes a Greek transcription of

roughly contemporary Greek transcription conventions for other languages.

ancient Palestine, provides a reconstruction of contemporary Greek pronunciation. Chapter 5,

based on a linguistic analysis of comparative transcription material, surveys typical Greek

and orthography, linguistic studies on cross-language perception, and moraic phonology.

CITATIONS OF THE SECUNDA

been retained in the representation of the Secunda transcriptions. A transcribed word is

followed by one asterisk (*) when it indicates a reasonably hypothesized emendation, in

CITATIONS OF ANCIENT GREEK AND LATIN AUTHORS

Thesaurus Linguae Graecae® Digital Library (http://www.tlg.uci.edu) (TLG). Usually, TLG

such as or Corpus Scriptorum Ecclesiasticorum Latinorum (CSEL) for Jerome's letters

CITATIONS OF ANCIENT PAPYRI

/www.papyri.info), Trismegistos (http://www.trismegistos.org), or POxy: Oxyrhynchus

Online (http://www.papyrology.ox.ac.uk/POxy). Note that names of papyri in italics signify

that they belong to a multi-volume collection.

OTHER ONLINE/ELECTRONIC TOOLS UTILIZED

Yerushalmi, and the Midrashic Literature (1926) (http://www.tyndalearchive.com/tabs/

jastrow), and the Comprehensive Aramaic Lexicon (CAL) (http://cal1.cn.huc.edu). Finally, a

EMPHASIS (BOLD OR ITALICS) IN QUOTATIONS

1943; JANSSENS 1982; and YUDITSKY 2017).

in-depth study of contemporary Greek phonology and orthography. Third, a phonetic

transcription of the pronunciation of the Secunda has yet to be produced.

1.2. PREVIOUS RESEARCH

long after, a number of scholars worked on the material.

grammatical, morphological, and phonological categories, while discussing and explaining

Tiberian Hebrew and thus creates confusion even in historical-linguistic matters—a

shortcoming pointed out by both JANSSENS and YUDITSKY.

The next monograph to be published on the Secunda was JANSSENS's Studies in

Hebrew Historical Linguistics Based on Origen's Secunda (1982). JANSSENS specifically

material with a more historically-grounded linguistic approach, the historical linguistics in

JANSSENS 1982, 25–36; YUDITSKY 2013, 803–804; 2017, 5–14).

quotations form external sources. Methodologically, YUDITSKY emphasizes the importance of

composition. However, orthography is only conservative when there is an established

spelling tradition. Transcription, by nature, is far more indicative of current pronunciation.

approach for analyzing the transcriptions. Moreover, a comparison with transcription

greater detail in the body of the dissertation.

models of cross-language perception, and moraic phonological theory.

First, an analysis of the phonology of the Secunda is undergirded by principles of historical

innovative development is apparent (for a more detailed discussion, see 6.2).

the second column is regarded as a transcription and not a transliteration. It is necessary,

transcriptions emerged out of a centuries-old tradition of transcribing continuous Hebrew

described in more detail in chapters 4 and 5 (4.2; 5.2).