Cambridge Semitic Languages and Cultures

Studies in Semitic Vocalisation

and Reading Traditions
EDITED BY AARON D. HORNKOHL AND GEOFFREY KHAN
 https://www.openbookpublishers.com

© 2020 Aaron D. Hornkohl and Geoffrey Khan. Copyright of individual chapters is

maintained by the chapters’ authors.

This work is licensed under a Creative Commons Attribution 4.0 International license
(CC BY 4.0). This license allows you to share, copy, distribute and transmit the text; to
adapt the text and to make commercial use of the text providing attribution is made to the
authors (but not in any way that suggests that they endorse you or your use of the work).
Attribution should include the following information:

Aaron D. Hornkohl and Geoffrey Khan (eds.), Studies in Semitic Vocalisation and Reading
Traditions. Cambridge, UK: Open Book Publishers, 2020, https://doi.org/10.11647/OBP.0207

In order to access detailed and updated information on the license, please visit, https://
doi.org/10.11647/OBP.0207#copyright

Further details about CC BY licenses are available at, https://creativecommons.org/

licenses/by/4.0/

All external links were active at the time of publication unless otherwise stated and have
been archived via the Internet Archive Wayback Machine at https://archive.org/web

Updated digital material and resources associated with this volume are available at
https://doi.org/10.11647/OBP.0207#resources

Every effort has been made to identify and contact copyright holders and any omission or
error will be corrected if notification is made to the publisher.

Semitic Languages and Cultures 3.

ISSN (print): 2632-6906 ISBN Paperback: 978-1-78374-935-5

ISSN (digital): 2632-6914 ISBN Hardback: 978-1-78374-936-2
ISBN Digital (PDF): 978-1-78374-937-9
DOI: 10.11647/OBP.0207

Cover image: Detail from a bilingual Latin-Punic inscription at the theatre at Lepcis
Magna, IRT 321 (accessed from https://it.wikipedia.org/wiki/File:Inscription_Theatre_
Leptis_Magna_Libya.JPG). Leaf of a Syriac prayer book with Western vocalisation signs
(source: Wikimedia Commons). Leaf of an Abbasid-era Qurʾān (vv. 64.11–12) with red,
yellow, and green vocalisation dots (source: Wikimedia Commons). Genizah fragment of
the Hebrew Bible (Gen. 11–12, Cambridge University Library T-S A1.56; courtesy of the
Syndics of Cambridge University Library). Genizah fragment of a Karaite transcription
of the Hebrew Bible in Arabic script (Num. 14.22–24, 40–42, Cambridge University
Library T-S Ar. 52.242; courtesy of the Syndics of Cambridge University Library). Greek
transcription of the Hebrew for Ps. 22.2a in Matt. 27.46 as found in Codex Bezae (fol. 99v;
courtesy of the Syndics of Cambridge University Library).
Cover design: Anna Gatti
PHONOLOGICAL ADAPTATION AND THE
BIBLICAL ARAMAIC AND BIBLICAL
HEBREW REFLEXES OF *I AND *U*

Benjamin D. Suchard
————————————————————————————

1.0. INTRODUCTION
For over a century, historical linguists have been guided by the
Ausnahmslosigkeit der Lautgesetze, the principle that sound
changes affecting a language are phonetically regular and excep-
tionless, as put forward by the nineteenth-century German
philologists and linguists known as the Neogrammarians.
Hermann Paul (1880, 69) formulates this principle as follows:

*
I am very grateful to Geoffrey Khan for having invited me to come
present the contents of this paper in Cambridge. I also thank the attend-
ing audience for their comments, especially Shai Heijmans, who pro-
vided me with numerous helpful suggestions. Any remaining errors are
my own.
The occasional transliterations of Tiberian Hebrew words and vowel
signs follow the conventions outlined in Johnson and Goerwitz (1995).
Phonetic transcriptions, given in the International Phonetic Alphabet,
are enclosed in [square brackets]; phonemic representations are pre-
ceded and followed by a /forward slash/.

© Benjamin D. Suchard, CC BY 4.0 https://doi.org/10.11647/OBP.0207.05

172 Benjamin D. Suchard

Hence, if we speak of the consistent operation of sound

laws, this can only mean that a sound change will treat
every individual case in which the same phonetic condi-
tions present themselves within the same dialect in the
same fashion. Thus, where one and the same sound for-
merly occurred, this must either stay the same sound in the
later stages of development as well, or, where a split into
several different sounds has taken place, a specific cause
should be indicated which explains why this sound arose
in one case and that sound in the other, and this cause
should be purely phonetic in nature, such as the influence
of surrounding sounds, stress, syllable structure, etc.1

Adhering to this principle has pushed linguists beyond

merely identifying tendencies operating in a certain language and
allowed them to discover phonetically conditioned sound
changes that would otherwise have gone unnoticed. As the regu-
larity of sound change is a universal principle, it can also be
shown to apply to Biblical Hebrew (Suchard 2019). In this lan-
guage, however, we are faced with a small number of phenomena

1
Wenn wir daher von konsequenter Wirkung der Lautgesetze reden, so kann
das nur heissen, dass bei dem Lautwandel innerhalb desselben Dialektes alle
einzelnen Fälle, in denen die gleichen lautlichen Bedingungen vorliegen,
gleichmässig behandelt werden. Entweder muss also, wo früher einmal der
gleiche Laut bestand, auch auf den späteren Entwicklungsstufen immer der
gleiche Laut bleiben, oder, wo eine Spaltung in verschiedene Laute eingetreten
ist, da muss eine bestimmte Ursache und zwar eine Ursache rein lautlicher
Natur wie Einwirkung umgebender Laute, Akzent, Silbenstellung u. dgl. an-
zugeben sein, warum in dem einen Falle dieser, in dem andern jener Laut
entstanden ist.
 The Reflexes of *i and *u 173

that seem impervious to an explanation through regular sound

laws. The topic of this paper is one such problem: the reflexes of
Proto-Northwest-Semitic short *i and short *u in the Tiberian vo-
calisation of the Hebrew Bible, which vacillate between i, u and
ɛ, ɔ in some environments in Biblical Hebrew and Aramaic and
between i, u and e, o in other environments in Biblical Aramaic
only.
I will suggest that the solution for this irregularity lies in a
process of phonological adaptation in the reading tradition. Pho-
nological adaptation is the process where linguistic material from
one language (the source language) is adapted to fit the phonol-
ogy of another language (the target language) (Hock 1991, 390–
97). A common occurrence with loanwords, this usually involves
the replacement of source language phonemes that do not occur
in the target language with their closest approximations in the
target language. Crucially, this substitution is not always regular.
Speakers may even vary in their adaptation of the same foreign
material from one token to the next; Cohen (2009, 93) provides
the example of an Israeli basketball player variously realising the
English loan block shot [sic: blocked shot] /blɒk ʃɒt/ as /blak ʃat/,
/blok ʃot/, and /blak ʃot/. I will argue that this kind of irregular-
ity lies behind the varying reflexes of *i and *u in Biblical Ara-
maic and Biblical Hebrew.
As phonological adaptation depends on the phonologies of
the languages involved, the following section will discuss the
phonemic inventories of Biblical Hebrew at different points in
time. We will then first consider the variation between i, u and e,
174 Benjamin D. Suchard

o, which is limited to Biblical Aramaic, before examining the var-

iation between i, u and ɛ, ɔ in both Biblical Aramaic and Biblical
Hebrew. Phonological adaptation can be held responsible for
both of these irregularities: adaptation of Aramaic texts to He-
brew phonology in the first case and adaptation of the biblical
reading tradition to the phonology of an unidentified language,
possibly Greek, in the second case.

2.0. PHONEMES AND ALLOPHONES

As the concept of the phoneme is crucial to the process of pho-
nological adaptation, let us consider it first. A phoneme is the
smallest contrastive unit in the sound system of a language, as is
commonly accepted. But what exactly do we mean by contrastive?
If we find variation between two sounds in a language, I
will assume that this variation is contrastive unless there is evi-
dence to the contrary. If sounds are not contrastive, they are re-
ferred to as allophones. Evidence for allophony can be of two
kinds.
First, the allophony may be phonetically conditioned,
which is to say that it is completely predictable from the phonetic
environment in which two sounds occur. The textbook example
for this kind of allophony is the variation between aspirated and
plain voiceless plosives in most varieties of English. Aspirated
voiceless plosives like [pʰ] occur only in syllable-initial position.
Plain voiceless plosives like [p] do not occur in syllable-initial
position, but do occur everywhere else. [pʰ] and [p] are thus in
complementary distribution: we can completely accurately pre-
dict whether a particular word has [pʰ] or [p] based solely on
 The Reflexes of *i and *u 175

phonetic environment. Therefore, the two sounds are not con-

trastive at a deeper level and can both be represented as one and
the same phoneme /p/, with the position in the syllable deter-
mining whether this phoneme is realised with or without aspira-
tion.
The other case in which variation between two sounds is
not contrastive is if it is completely unconditioned by linguistic
factors. The English word pit, for instance, can be realised as both
[pʰɪt̚], with an unreleased alveolar stop at the end of the word,
and [pʰɪʔ], with a glottal stop (again, in many varieties). Both
realisations are equally valid and the variation is not conditioned
by phonetic, morphological, syntactic or lexical factors. Hence,
the two allophones are said to be in free variation and can once
again be ascribed to one and the same underlying phoneme, e.g.
/t/.
Practically, then, we can say that variation between two
sounds is contrastive if and only if it is conditioned at any of the
non-phonetic levels mentioned above: if it is conditioned by mor-
phological, syntactic, or lexical features. This conditioning may
yield one or more minimal pairs, pairs of morphologically or lex-
ically distinct words that differ only in the presence of one or the
other sound under consideration, but these may also coinci-
dentally not occur. Hence, minimal pairs prove a phonemic con-
trast, but their absence does not prove a lack of contrast.
Let us turn to some illustrations from Tiberian Biblical He-
brew. The phonemic realisations are based on the description of
the Tiberian pronunciation given by Geoffrey Khan, e.g., in Khan
176 Benjamin D. Suchard

(1996).2 For our first example, we see variation between [iː] and
[eː], as in ‫[ יָ ִׂשים‬jɔˑˈsiːm] ‘he will put’ and ‫[ יָ ֵׂשם‬jɔˑˈseːm] ‘let him
put’. As the occurrence in a minimal pair shows, this variation is
not phonetically conditioned: both sounds can occur in exactly
the same phonetic environments. Nor are the sounds in free var-
iation: ‘he will put’ would always be read with [iː] while ‘let him
put’ would always be read with [eː] (and the same goes for all
other words where these sounds occur). Thus, [iː] and [eː] are
phonemically contrastive: they belong to two different pho-
nemes.
For a second example, there are the various ways the vowel
sign shewa is realised. In Tiberian, it is realised as a vowel if it
stands between two consonants that would otherwise be syllable-
initial. This vowel is [i] before y; a short vowel with the same
quality as the next vowel before gutturals; and [a] elsewhere. In
other positions, shewa is realised as zero, i.e., no vowel is read.
These realisations are not in free variation, but we clearly see a
purely phonetic conditioning. Hence, they belong to one and the
same phoneme—or in this case, the lack of a phoneme, as the
vocalic realisations can all be interpreted as allophones of zero.
By conducting this kind of analysis for every sound in the
Tiberian pronunciation of Biblical Hebrew, we arrive at a vocalic
phonemic inventory as presented in Table 1 (Suchard 2018). The
analysis underlying this phonemic system is based on Tiberian
Biblical Hebrew, but it also holds for Tiberian Biblical Aramaic.

2
See now also Khan (2020).
 The Reflexes of *i and *u 177

Table 1. The vocalic phonemes of Tiberian Biblical Hebrew

(and Tiberian Biblical Aramaic)
front central back
short un- long short un- long short un- long
marked marked marked
close /i/ /ī/ /u/ /ū/
close- /ē/ /ō/
mid
open- /ɛ̆/ /ɛ/ /ɔ̆/ /ɔ/ /ɔ̄/
mid
open /ă/ /a/

Based on evidence from historical phonology (Suchard

2017, 211–12) and Latin and Greek transcriptions (see, e.g., Kan-
tor 2017), earlier stages of Hebrew appear to have had a simpler
phonemic inventory, presented in Table 2. The labelling as ‘pre-
Tiberian Biblical Hebrew’ is admittedly vague, but given the long
period for which this system seems to have been in place, no
more precise appellation suggests itself.

Table 2. The vocalic phonemes of pre-Tiberian Biblical

Hebrew
front central back
short long short long short long
close /ī/ /ū/
mid /e/ /ē/ /o/ /ō/
open /a/ /ā/

The main difference with the Tiberian phonology is that

later /i/ and /ɛ/ are still one phoneme /e/, just as later /u/ and
/ɔ/ are still one phoneme /o/. Tiberian /ɔ̄/ still has its older
value, /ā/, and the underlyingly short ḥaṭef vowels of the Tibe-
rian pronunciation have not yet become separate phonemes.
178 Benjamin D. Suchard

Bearing these phonemic inventories in mind, let us consider

the irregular reflexes of *i and *u, starting with the Biblical Ara-
maic interchange between i, u and e, o.

3.0. BIBLICAL ARAMAIC I : E AND U : O

Stressed Proto-Aramaic *i and *u (normally deriving from Proto-
Northwest-Semitic *i and *u, respectively) are reflected in two
different ways in Biblical Aramaic. Stressed *i surfaces either as
i, as in *wa-baṭṭílū > ‫ּוב ִִּׂ֫טלּו‬
ַ ‘and they stopped (m)’ (Ezra 4.23), or
as e, as in *hawθíb > ‫הֹותב‬
ֵׂ ‘he settled’ (Ezra 4.10). Similarly,
stressed *u surfaces either as u, as in *wa-yisgúd > ‫‘ וְ יִׂ ְסגֻּ֑ד‬and he
prostrates himself (pause)’ (Dan. 3.6), or as o, as in *gúddū > ‫ִּ֫גּדּו‬
‘cut down (mpl)’ (Dan. 4.11). These different reflexes can even
occur in what would otherwise seem to be the same word: cf.
*yitʕabíd > ‫‘ יִׂ ְת ֲע ֵׂבד‬it (m) will be made’ (Ezra 6.11; 7.23; Dan.
3.29) beside ‫‘ יִׂ ְת ֲע ִׂ ֻּ֑בד‬idem (pause)’ (Ezra 6.12; 7.21).
As discussed in §2.0, these different reflexes are phonemi-
cally contrastive in Tiberian Biblical Aramaic. Seemingly contra-
dicting the principle of Ausnahmslosigkeit der Lautgesetze, how-
ever, no conditioning factor is apparent that can explain “why
this sound arose in one case and that sound in the other.” As the
examples cited above suggest, the reflexes as *i and *u seem to
be associated with pausal position. This was already noted by
Bauer and Leander (1927, 23) and confirmed by Amos Dodi
(1989). The reflex in non-pausal forms, however, remains unpre-
dictable. Vincent DeCaen (2004) tries to explain the variation in
this position on prosodic grounds, too, but his account ultimately
 The Reflexes of *i and *u 179

leaves a number of forms unexplained. The irregularity thus re-

mains.
Taking a closer look at the occurrence of each reflex, we
find that the variation is limited to closed syllables. Leaving the
less frequent forms with *u aside for the moment, we see short *i
in closed syllables reflected as i in words like ‫‘ ַת ִּׂדק‬it (f) will crush’
(Dan. 2.40, 44), ‫‘ ְמ ֵׂשיזִׂ ב‬saves (m)’ (Dan. 6.27), or ‫‘ יָ ִׂכל‬able (ms)’
(Dan. 3.17; 4.37) beside a reflex as e in words like ‫‘ ְש ֵׂלט‬he had
power’ (Dan. 3.27), ‫‘ ְש ִֵּׂ֫א ְלנָ א‬we asked’ (Ezra 5.9, 10), or ‫‘ ָכ ֵׂהל‬able
(ms)’ (Dan. 2.26; 4.18). In open syllables, however, we find only
i reflexes, as in ‫‘ ְס ִּ֫ ִׂלקּו‬they (m) went up’ (Ezra 4.12; Dan. 2.29),
‫‘ ְש ִּ֫ ִׂלטּו‬they (m) had power’ (Dan 6.24), ‫‘ וְ ַה ְל ִִּׂ֫בישּו‬and they (m)
clothed’ (Dan. 5.29), and ‫‘ ִׂה ְת ְר ִִּׂ֫חצּו‬they (m.) trusted’ (Dan. 3.28).
This distribution becomes meaningful if we consider it from
the point of view of pre-Tiberian Hebrew phonology. Due to a
combination of sound changes, the Hebrew non-low stressed
short vowels *e and *o had been preserved only in closed sylla-
bles. Stressed instances of short *e and *o had lost the stress in
open syllables and later underwent reduction (Blau 2010,
§3.5.12.2.6). That the distribution of the reflexes of *i and *u in
Biblical Aramaic matches a pattern attested in the phonology of
Biblical Hebrew suggests that the irregularity we are dealing with
is due to some kind of interaction between these two strata of the
biblical corpus.
In fact, we may explain the Biblical Aramaic situation
through a difference in phonetics between the precursors of
Biblical Aramaic and Biblical Hebrew. As was mentioned above,
different types of evidence suggest that Proto-Northwest-Semitic
180 Benjamin D. Suchard

*i and *u had normally shifted to *e and *o in pre-Tiberian

Biblical Hebrew. There is no indication, however, that this sound
change affected the Aramaic dialect underlying the Biblical
Aramaic reading tradition. Let us assume that this variety of
Aramaic preserved Proto-Aramaic *i and *u unchanged. Once the
Biblical Aramaic texts became an integral part of the Hebrew
Bible, this difference in phonology between the Aramaic and
Hebrew portions would have formed an unstable situation. Since
the vast majority of the biblical texts are in Hebrew, it would be
natural for readers to adapt the tiny Aramaic part of the corpus
to the dominant Hebrew phonology, especially considering the
fluid transitions between both languages in the actual text. In
doing so, Aramaic *i and *u could either be changed to the
corresponding short vowels, *e and *o, or to the corresponding
long vowels, *ī and *ū. As phonological adaptation is not bound
by regularity, this then yielded the irregular outcomes we have
observed. The process is illustrated in Table 3.

Table 3. Phonological adaptation of pre-Biblical Aramaic

forms to Hebrew phonology
Original Adapted form Biblical meaning
Aramaic Aramaic
*gúddū *góddū ‫‘ ִּ֫גּדּו‬cut down (mpl)’
*wa-yisgúd ́
*wa-yesgūd ‫‘ וְ יִׂ ְסגֻּ֑ד‬and he prostrates him-
self (pause)’
*yitʕabíd *yetʕabéd ‫‘ יִׂ ְת ֲע ֵׂבד‬it (m) will be made’
*yitʕabíd *yetʕabīd́ ‫‘ יִׂ ְת ֲע ִׂ ֻּ֑בד‬idem (pause)’

In pausal position, *i and *u were more likely to be associ-

ated with Hebrew *ī and *ū due to the crosslinguistic effect of
 The Reflexes of *i and *u 181

prepausal vowels being phonetically lengthened (Nooteboom

1997, 658). This explains why the Biblical Aramaic words with
major disjunctive accents all occur with i and u, not with e or o
(Dodi 1989). In open syllables, stressed *i and *u were always
incorporated as long *ī and *ū, as stressed *e and *o in this posi-
tion did not occur in the receiving Hebrew phonology. Thus, e.g.,
́ >
*salíqū ‘they (m) went up’ was necessarily adapted to *salīqū
‫ס ִּ֫ ִׂלקּו‬.
ְ

4.0. BIBLICAL HEBREW AND BIBLICAL ARAMAIC I : Ɛ AND

U:Ɔ

The interchange of stressed i : e and u : o is limited to the Aramaic

part of the biblical corpus. Another alternation characterises the
entire corpus. In closed, unstressed syllables, we find two short 3
front vowels, written i and ɛ, and two short back vowels, written
u and ɔ. Their distribution is largely predicted by phonetic envi-
ronment. With the front vowels, ɛ normally occurs next to gut-
turals, e.g., ‫‘ ֶח ְדרֹו‬his room’, while i occurs elsewhere, e.g., ‫ִׂב ְטנֹו‬
‘his belly’, ‫‘ ִּׂד ְמכֶֶם‬your (mpl) blood’. With the back vowels, u nor-
mally occurs before geminates, e.g., ‫‘ כּלֹו‬all of it (m)’, while ɔ oc-
curs elsewhere, e.g., ‫‘ ָק ְדשֹו‬his sanctuary’, ‫‘ ָהגְ ָלה‬he was exiled’.
However, we also find these vowels occurring in the ‘wrong’ en-
vironment. Unconditioned ɛ occurs in words like ‫‘ ֶמ ְמ ָש ָלה‬author-
ity’ and ‫‘ יֶ ְד ֶכם‬your (mpl) hand’. Similarly, unconditioned u occurs

3
Technically, these are unmarked for length according to the analysis
put forward in Suchard (2018). In this environment, they are realised
as short.
182 Benjamin D. Suchard

in words like ‫‘ ש ְל ָחן‬table’ and ‫‘ מגְ ִׂלים‬exiled (mpl)’. As the distribu-

tion is not completely phonetically conditioned and the different
vowels are not in free variation—the same word in the same con-
text always being read with the same vowel—the contrast be-
tween these vowels must be considered phonemic for Tiberian
Biblical Hebrew and Aramaic (as argued in more detail in
Suchard 2018, 204).
These four short vowels derive from only two different pho-
nemes in the pre-Tiberian Hebrew phonology: /e/ and /o/. We
are therefore dealing with an unconditioned phonemic split sim-
ilar to the one in Biblical Aramaic discussed in §3.0. Perhaps,
then, a similar explanation based on phonological adaptation can
be found.
The usual transcription in alphabetic scripts as mid vowels
and the historical relatedness with long /ē/ and /ō/ support a
representation of the phonemes we are dealing with as /e/ and
/o/. Given the absence of other short, non-low vowels in pre-
Tiberian Hebrew phonology, however, it is likely that the pho-
netic realisations of these phonemes covered the entire non-low
part of the vowel space. That is to say that the phoneme we rep-
resent as /e/ could have realisations ranging from [i], [ɪ], or [e]
to [ɛ] and the phoneme that we represent as /o/ could be realised
as anything from [u], [ʊ], or [o] to [ɔ].
To readers who were well accustomed to the phonology of
the biblical reading tradition, this variation would go unnoticed,
as it was non-contrastive. Speakers are not typically conscious of
allophony of this type. Suppose, however, that one of the readers
in the chain of transmission that would eventually lead to the
 The Reflexes of *i and *u 183

Tiberian reading tradition already had a contrast between /i/ :

/ɛ/ and /u/ : /ɔ/ in closed, unstressed syllables. This contrast
could have been imported, for instance, from the reader’s native
language. In this case, our reader would be hypersensitive to the
different allophonic realisations of /e/ and /o/. When hearing a
higher realisation, he would assign it to /i/ or /u/; lower realisa-
tions would be assigned to /ɛ/ and /ɔ/. Thus, what were origi-
nally allophones—with phonetic factors largely determining the
distribution, but ultimately in free variation—could split into dif-
ferent phonemes as they were mapped onto a pre-existing con-
trast taken from another language. This scenario is illustrated in
Table 4, where Teacher represents the older stage of the reading
tradition, where the variation is allophonic, and Student repre-
sents the stage where the phonemic contrast was imposed on the
originally allophonic variants.

Table 4. Phonemicisation of /i/ : /ɛ/ in the reading tradition

Teacher Teacher Student Tiberian meaning
thinks… says… thinks… Biblical
Hebrew
beṭnṓ /e/ biṭnṓ /e/ biṭnṓ /i/ ‫ִׂב ְטנֹו‬ ‘his belly’
‘your (mpl)
yed̠kɛ̠ m
́ /e/ yɛd̠kɛ̠ m
́ /e/ yɛd̠kɛ̠ m
́ /ɛ/ ‫יֶ ְד ֶכם‬
hand’
heḡlā ́ /e/ hiḡlā ́ /e/ hiḡlā ́ /i/ ‫ִׂהגְ ָלה‬ ‘he exiled’
heḡlā ́ /e/ hɛḡlā ́ /e/ hɛḡlā ́ /ɛ/ ‫ֶהגְ ָלה‬ ‘idem’

Once the contrast had become phonemic in the mind of the

reader, he would consistently produce realisations very close to
[i] and [u] in words with /i/ and /u/ and [ɛ] and [ɔ] in words
184 Benjamin D. Suchard

with /ɛ/ and /ɔ/. This distinction was then passed on in the read-
ing tradition until it was fixed in writing by the Tiberian vocalis-
ers.
In the case of the purely Biblical Aramaic problem dis-
cussed in §3.0, the close match with the independently recon-
structed pre-Tiberian Hebrew phonology made the somewhat
speculative solution more plausible. In the case of i : ɛ and u : ɔ,
however, the suggestion of phonological adaptation holds a
purely hypothetical other language responsible, whose only
known characteristics are a contrast between /i/ : /ɛ/ and /u/ :
/ɔ/ in closed, unstressed syllables. Can we identify a language
that could plausibly have caused this phonological split in the
biblical reading tradition?
The first suspect would be Jewish Palestinian Aramaic. As
the vernacular language of the Tiberian Masoretes and their di-
rect precursors (as attested by its use in the masoretic notes), at
least, we may expect it to have influenced the reading tradition
in some way. But the phonology of Jewish Palestinian Aramaic
does not match the profile we are looking for. While Jewish Pal-
estinian Aramaic distinguishes between /i/ : /e/ and /u/ : /o/
and could thus plausibly have split a mid vowel phoneme into
two, it seems that only /e/, /a/, and /o/ occurred in closed, un-
stressed syllables (Fassberg 1991, 34–41). Thus, imposing Jewish
Palestinian Aramaic phonology on the Hebrew reading tradition
would have preserved /e/ and /o/ in this position, not split them.
Looking further east does not solve our problem either.
While influence from Jewish Babylonian Aramaic is historically
possible, its vowel inventory was apparently even poorer than
 The Reflexes of *i and *u 185

that of Jewish Palestinian Aramaic. As it probably did not distin-

guish between /u/ and /o/ (Morag 1961), it cannot be blamed
for the split of /o/ into /u/ and /ɔ/ in the reading tradition.
Beyond Semitic, we find a final candidate in Palestinian
Greek, the phonology of which has been admirably described by
Benjamin Kantor (2017). According to Kantor’s description and
analysis (110–31), this variety of Greek featured the /i/ : /ɛ/ con-
trast that we are looking for. In the back vowels, however, we
find /u/ contrasted with /o/, not with /ɔ/. This is not what the
Hebrew situation would lead us to expect a priori, but on further
reflection it may explain some curious facts of Hebrew historical
phonology. As we have seen, the distribution of /i/ and /ɛ/ dif-
fers from that of /u/ and /ɔ/. With the front vowels, /i/ has the
less restricted distribution, while with the back vowels, /ɔ/ does.
Perhaps this can be attributed to the asymmetry in the Greek
vowel system: Hebrew /o/ was normally mapped to Greek /o/
and to Greek /u/ only in more limited cases; this default value
/o/ in the reading tradition later shifted to /ɔ/ in Tiberian. In the
front vowels, on the other hand, Hebrew /e/ was more commonly
adapted to Greek /i/, with /ɛ/ being the largely conditioned var-
iant. There would thus seem to have been a hierarchy for the
preferred vowel matching the Hebrew close-mid vowels, with a
Greek close-mid vowel being the best choice when available, fol-
lowed by a close vowel and then an open-mid vowel.4

4
Further evidence for the asymmetry between /e/ and /o/ in this regard
comes from the pausal consecutive imperfect forms of some weak verbs.
As described in Blau (1981), the forms with an *i vowel developed like
186 Benjamin D. Suchard

5.0. CONCLUSION
The irregular reflexes of *i and *u in Biblical Hebrew and Biblical
Aramaic challenge the principle of regular sound change. I have
argued that the solution is not to be sought in sound change at
all, obviating the need for regularity. The conditioning of the Bib-
lical Aramaic split discussed in §3.0 reflects features of pre-Tibe-
rian Biblical Hebrew phonology. This suggests that phonological
adaptation is at play, a process that could also explain the similar
split discussed in §4.0. As phonological adaptation is often char-
acterised by irregularity, this provides us with an explanation
from generally accepted principles of historical linguistics.
The phonology causing the adaptation was seen to be pre-
Tiberian Biblical Hebrew in the case of Biblical Aramaic stressed
*i and *u and was suggested to be Palestinian Greek in the case
of Biblical Hebrew and Aramaic unstressed *e and *o in closed
syllables. The influence of these languages on the biblical reading
tradition is compatible with what we might call the least surpris-
ing model of the oral transmission of the biblical texts. First, Bib-
lical Hebrew and Biblical Aramaic texts came to be combined in
a shared, biblical corpus, leading to the adaptation of the Ara-
maic material to Hebrew phonology. Based on grammatical fea-
tures of the Aramaic variety underlying the Biblical Aramaic

*wayyḗlek > *wayyḗlɛk > *wayyēlɛ́k > *wayyēlák > ‫‘ וַ יֵׂ ַלְֻּ֑ך‬and he de-
parted (pause)’. Forms with an *u vowel like ‫‘ וֶַיָ ֻּ֑מת‬and he died (pause)’
do not reflect the parallel lowering of *o. Based on the account sketched
in the main text, we may now understand the development of these
forms as *wayyām ́ ot (with *o preserved in unstressed position as it
matched Greek /o/?) > *wayyāmót > ‫וַ יָ ֻּ֑מת‬.
 The Reflexes of *i and *u 187

reading tradition, I have argued elsewhere (Suchard forthcom-

ing) that this fixing of the combined reading tradition should be
placed in first-century CE Palestine. The later influence of Pales-
tinian Greek, the most likely culprit behind the split discussed in
§4.0, then supports a continuing transmission in Roman Pales-
tine; historical considerations suggest that the tradition was
maintained in the centres of Jewish learning in Galilee (Geller
1998, 562–65). While the involvement of Palestinian Greek, es-
pecially, remains speculative, the account offered here provides
one more example of how the results of historical linguistics and
linguistic reconstruction can help to illuminate the history of the
ancient world as it is known to historians from more direct
sources.

6.0. REFERENCES
Bauer, Hans, and Pontus Leander. 1927. Grammatik des Biblisch-
Aramäischen. Halle: Max Niemeyer.
Blau, Joshua. 1981. ‘On Pausal Lengthening, Pausal Stress Shift,
Philippi’s Law and Rule Ordering in Biblical Hebrew’. He-
brew Annual Review 5: 1–15.
———. 2010. Phonology and Morphology of Biblical Hebrew. An
Introduction. Linguistic Studies in Ancient West Semitic 2.
Winona Lake, IN: Eisenbrauns.
Cohen, Evan-Gary. 2009. ‘The Role of Similarity in Phonology:
Evidence from Loanword Adaptation in Hebrew’. PhD dis-
sertation, Tel-Aviv University.
188 Benjamin D. Suchard

DeCaen, Vincent. 2004. ‘The Pausal Phrase in Tiberian Aramaic

and the Reflexes of *i’. Journal of Semitic Studies 49 (2):
215–24.
Dodi, Amos. 1989. ‘Pausal Forms in Aramaic’. In Studies in the
Hebrew Language and the Talmudic Literature Dedicated to the
Memory of Dr. Menaḥem Moreshet, edited by Menaḥem Zevi
Kaddari and Shim‘on Sharvit, 63–74. Ramat-Gan: Bar-Ilan
University Press (in Hebrew).
Fassberg, Steven E. 1991. A Grammar of the Palestinian Targum
Fragments from the Cairo Genizah. Atlanta: Scholars Press.
Geller, Barbara. 1998. ‘Transitions and Trajectories: Jews and
Christians in the Roman Empire’. In The Oxford History of
the Biblical World, edited by Michael D. Coogan, 560–96.
Oxford: Oxford University Press.
Hock, Hans Heinrich. 1991. Principles of Historical Linguistics. 2nd
edition. Berlin: Mouton de Gruyter.
Johnson, Robert M., Jr., and Richard Goerwitz. 1995. ‘A Simple,
Practical System for Transliterating Tiberian Hebrew Vow-
els’. Hebrew Studies 36: 13–24.
Kantor, Benjamin P. 2017. ‘The Second Column (Secunda) of Or-
igen’s Hexapla in Light of Greek Pronunciation’. PhD dis-
sertation, University of Texas at Austin.
Khan, Geoffrey. 1996. ‘The Tiberian Pronunciation Tradition of
Biblical Hebrew’. Zeitschrift für Althebraistik 9: 1–23.
———. 2020. The Tiberian Pronunciation Tradition of Biblical He-
brew. 2 vols. Cambridge Semitic Languages and Cultures 1.
Cambridge: University of Cambridge and Open Book Pub-
lishers.
 The Reflexes of *i and *u 189

Morag, Shelomo. 1961. ‘Notes on the Vowel System of Babylo-

nian Aramaic as Preserved in the Yemenite Tradition’. Pho-
netica 7: 217–39.
Nooteboom, Sieb. 1997. ‘Prosody of Speech: Melody and
Rhythm’. In The Handbook of Phonetic Sciences, edited by
William J. Hardcastle and John Laver, 640–73. Oxford:
Blackwell.
Paul, Hermann. 1880. Prinzipien der Sprachgeschichte. Halle: Max
Niemeyer.
Suchard, Benjamin D. 2017. ‘A Triconsonantal Derivation of the
Lamed-He Paradigm’. Kleine Untersuchungen zur Sprache des
Alten Testaments und seiner Umwelt 22: 205–21.
———. 2018. ‘The Vocalic Phonemes of Tiberian Hebrew’. He-
brew Studies 59: 193–207.
———. 2019. The Development of the Biblical Hebrew Vowels. Stud-
ies in Semitic Languages and Linguistics 99. Leiden: Brill.
———. Forthcoming. ‘The Origins of the Biblical Aramaic Read-
ing Tradition’. Vetus Testamentum.