Word stress in Arabic
Watson, JCE
Title
Word stress in Arabic
Authors
Watson, JCE
Type
Book Section
URL
This version is available at: http://usir.salford.ac.uk/id/eprint/17633/
Published Date
2011
USIR is a digital collection of the research output of the University of Salford. Where copyright
permits, full text material held in the repository is made freely available online and can be read,
downloaded and copied for non-commercial private study or research purposes. Please check the
manuscript for any further copyright restrictions.
For more information, including our policy and submission procedure, please
contact the Repository Team at: usir@salford.ac.uk.
TBC_135.qxd
7/13/10
19:27
Page 1
135 Word Stress in Arabic
Janet C. E. Watson
1
Introduction
Within phonological theory, Arabic word stress has received arguably more
attention than the word stress of any language other than English; and within
Arabic linguistics, more work has been devoted to stress-related phenomena than
any other single topic. This chapter addresses some of the reasons behind this
phenomenon.
The chapter is structured as follows: §2 considers the characteristics of Arabic
word stress, discussing general features common to different varieties and basic
ways in which modern Arabic dialects differ both from Classical Arabic and from
each other. §3 provides a historical overview of the treatment of Arabic word stress
within generative paradigms, focusing on major contributions in the analysis of
Arabic stress, and cases where data from Arabic has contributed to the development of stress theory. §4 considers stress in Classical Arabic, and then examines
in more detail word stress in three modern dialects – Cairene, San’ani (Yemen),
and Levantine – focusing particularly on phenomena that pose a challenge for
metrical phonology.
2
Characteristics of Arabic word stress
With over 250 million speakers, Arabic is the official language of 18 sovereign
states from Mauritania in the west to Iraq in the east. It is also spoken in parts
of southern Turkey, by Maronite Christians in northern Cyprus, and in parts of
sub-Saharan Africa. Arabic language enclaves are found in the Balkh region of
Afghanistan, parts of Iran, and Uzbekistan. All Arabic dialects exhibit word
stress; however, the socially and geographically diverse area over which Arabic
is spoken leads to differences in the mechanics of word stress assignment. In all
cases stress location is a function of both syllable weight and syllable position,
but dialects differ in the distribution of syllable types, the leftmost extent of stress
(third or fourth syllable from the right), the rhythmic grouping of syllables, the
interaction of stress, syncope and epenthesis, and the degree to which lexical
information may affect stress.
TBC_135.qxd
7/13/10
2
19:27
Page 2
Janet C. E. Watson
Arabic recognizes three weights of syllable: light, heavy, and superheavy.
Light syllables are always open, heavy syllables are open or closed, and superheavy syllables are closed or doubly closed. Examples of these syllable types from
Classical Arabic are given below:
(1)
light
heavy
superheavy
open
closed
CV
wa ‘and’
CVV sa.fara ‘he
traveled’ CVC
min ‘from’
ka.tab.tu
‘I wrote’
CVVC bab# ‘door’
doubly closed
CVCC
bint# ‘girl’
CVVGG1 madd#
‘stretching’
CVVG mad.dun
‘stretching
(nom)’
CV and CVC are unrestricted, although unstressed short vowels in open syllables
are often deleted in modern dialects. CVCC and CVVGG are restricted to wordor utterance-final position.2 In the distribution of other syllable types, however,
dialects vary. Levantine, Sudanese, some Peninsular, and North African dialects
allow CVGG and/or CVVC in derived environments word internally, as in:
/masik-cn/ > [maskcn] ‘holding (masc pl)’, /œaf-ha/ > [œafha] ‘he saw her’,
/mu#allim-cn/ > [m#allmcn] ‘teachers’. Cairene allows CVVC syllables word finally
only, as in: [kitab kibcr] ‘a big book’, but /kitab-na/ > [kitabna] ‘our book’; Cairene
restricts CVCC to utterance-final position, breaking up word-final non-utterance-final
CVCC syllables through epenthesis, as in: /bint Ùawcla/ > [bint[i] Ùawcla] ‘a tall
girl’; CVV occurs only when stressed in Cairene: initial CVV in [’œafit] ‘she saw’
and [’#alam] ‘world’ contrast with initial CV in [œa’fitu] ‘she saw him’ and
[#a’lamu] ‘his world’.
Stress falls on one of the last three syllables, in some dialects one of the last
four syllables, with assignment dependent on the weight and position of the stressed
syllable. Modern dialects follow the assumed rules of Classical Arabic (§3.1) whereby
stress is assigned to a final superheavy (CVVC, CVCC, or CVVGG) syllable, as
in Cairene: [fi’luus] ‘money’, [ma-xa’bazœ] ‘he didn’t cook’, and Palestinian:
[ja’waab] ‘answer’, [bi-’ÚuÙÙ] ‘he puts’. In the absence of a final superheavy, stress
is assigned to a heavy penult (CVV or CVC), as in Cairene: [ka’tabtu] ‘you (pl)
wrote’, [fih’muuha] ‘they understood her’, and Palestinian: [mus’taœfa] ‘hospital’,
[mu’naafis] ‘competitor’. In the absence of either a final superheavy or a heavy
penult, the dialects differ. In words with a heavy antepenult, Cairene stresses the
light penult, while most other dialects stress the antepenult: Cairene [mad’rasa]
‘school’ contrasts with Beirut/Damascene [’madrase].
Modern Arabic dialects differ in their rhythmic grouping of light syllables.
Western and Bedouin-type dialects tend to group light syllables into weak–strong
1
GG denotes geminate.
In Classical Arabic, superheavy syllables occur pre-pausally only, resulting from pre-pausal deletion
of short final vowels or case endings.
2
TBC_135.qxd
7/13/10
19:27
Page 3
Word Stress in Arabic
3
pairs (iambs): Cyrenaican Bedouin stresses the penult in forms such as: *[kitab-at]
> [ik’tibat] ‘she wrote’, [ingital-aw] > [inig’tilaw] ‘they were killed’, and the final
syllable in [ki’tab] ‘he wrote’, [nu’xal] ‘palm-trees’ (Mitchell 1960); eastern urban
dialects group light syllables into strong–weak pairs (trochees), stressing the
antepenult in forms such as Cairene: [’katabit] ‘she wrote’, [in’kasarit] ‘it (fem)
broke’, and the penult in [’katab] ‘he wrote’, [’walad] ‘boy’ (see chapter 46: the
iambic-trochaic law).
All modern dialects differ from Classical Arabic in at least optionally deleting
short vowels in unstressed open syllables (Birkeland 1954). Some dialects delete
short high vowels only, as in Damascene: /fihim-u/ > [’fihmu] ‘they understood’,
/fihim-na/ > [f’himna] ‘we understood’ but /katab-u/ > [’katabu] ‘they wrote’
(Cowell 1964). Other dialects delete short vowels irrespective of their quality, as
in Lebanese Kfar-fiıab: /Òarab-ak/ > [’Òarbak] ‘he hit you’ and /samak-i/ > [’samki]
‘one fish’ (Fleisch 1974).
Several dialects differ from the assumed predictable quantity-based system
of Classical Arabic in that certain morphemes affect stress placement. In Cairene
and Tunisian the 3rd feminine singular perfect inflectional suffix -it attracts stress
on suffixation: Cairene [’ramit] ‘she threw’ becomes [ra’mitu] ‘she threw it
(masc)’, contrasting with other CVCVCV forms, such as ’katabu ‘they wrote’, where
the antepenult is stressed. In Iraqi, the dual suffix -een retracts stress, although
all other cases of word-final CVVC attract stress: [’Œalbeen] ‘two dogs’ contrasts
with [ta#’baan] ‘tired’ (Erwin 1963: 43). In Muslim Mosul Iraqi, stress always
falls on the final syllable of a verbal or nominal stem when it takes a suffix, as
in: [nHx’lHÙ-u] ‘we mix it (masc)’ (Jastrow 2007). In some western dialects and dialects
of Oman, word stress is phonemic in disyllabic noun–verb pairs: initial stress in
[’fihim] ‘understanding’ contrasts with final stress in [fi’him] ‘he understood’
(Janssens 1972).
Finally, modern dialects differ as to whether or not epenthetic vowels count
for stress purposes. In Cairene, a penultimate post-CVC syllable with an epenthetic
vowel is stressed like any other penultimate post-CVC syllable: compare [bin’tina]
‘our daughter’ with [mad’rasa] ‘school’ and [fih’mitu] ‘she understood him’. In
Iraqi and Levantine, by contrast, stress is assigned as if the epenthetic vowel were
not there: penultimate stress in Muslim Mosul [ka’tabit] ‘I wrote’ contrasts with
initial stress in [’katabHt] ‘she wrote’ (Jastrow 2007).
3
Theoretical accounts of Arabic word stress
This section provides a historical overview of theoretical accounts of Arabic
word stress, focusing on ways in which research on Arabic has contributed both
to the development of metrical theory and to a deeper understanding of Arabic
prosodic structure and cross-dialectal differences.
3.1
Pre-generative approaches
Concepts upon which generative studies of Arabic word stress draw have their
roots in early pre-generative approaches. The older traditional studies of Erpenius
(1656), Brockelmann (1907), and Wright (1971) recognized the role of the syllable
and syllable weight in stress assignment, distinguishing between light (CV) and
TBC_135.qxd
7/13/10
4
19:27
Page 4
Janet C. E. Watson
heavy (CVV and CVC) syllables (see chapter 57: quantity-sensitivity). The
analyses of stress in Cairene by Harrell (1957) and Cairene and Cyrenaican
Bedouin by Mitchell (1956, 1960) are based on the position and relative weight
of syllables.
The Prague School (Jakobson 1971) describes stress assignment not in terms of
syllables, but in terms of moras. Moraic accounts of stress in Arabic include
Cantineau (1960: 240, author’s translations), for whom stress in El-Hamme of Gabes
(Tunisia) is placed: “on the third mora of the word . . . on the fourth if the third
corresponds to a consonant,” accounting for penultimate stress in: [k’tibtu] ‘you
(pl) wrote’, and antepenultimate stress in: [’madrasa] ‘school’. The mora is referred
to in the informal expression of Abdo’s (1969) post-SPE (The sound pattern of English;
Chomsky & Halle 1968) account, where Classical Arabic stress is assigned to the
vowel preceding the last two moras (the third or fourth mora from the right-edge).
Within metrical theory, it later returns as a full-fledged element of the representation, firmly embedded within the prosodic hierarchy.
3.2
Generative approaches
Generative approaches to stress in Arabic have followed contemporary approaches
in generative phonology, with a few landmark changes in orientation. In the SPE
segment-based approach adopted by Abdo (1969), Brame (1970, 1973, 1974),
Broselow (1976), Johnson (1979), and Weldon (1980), stress is encoded as a
phonological distinctive feature, [+stress], assigned to a [+syllabic] segment in a
particular segmental context. Essential variables are included in the vocabulary
of phonological rules. Take Palestinian as an example, for which the basic stress
rules are:
(2)
a.
b.
c.
Stress a final superheavy syllable: [ba-’œef] ‘I see’, [bi-’ÚuÙÙ] ‘he puts’.
Otherwise stress the rightmost non-final heavy syllable: [ba-’œefiœ] ‘I don’t
see’, [ka’tabti] ‘you (fem sg) wrote’, [’miktaœif] ‘discovering’.
Otherwise stress the first syllable (up to the antepenult): [’katab] ‘he wrote’,
[’zalama] ‘man’.
Under this approach, stress is assigned by the following rule (Brame 1974),
where C0 indicates an arbitrary number of consonants, including zero, and C 10 either
zero or one consonant.
(3)
Stress assignment
V → [+stress] / ___ C0((VC)VC 10)]
This rule abbreviates three disjunctively ordered sub-rules:
(4)
V → [+stress] / ___ C0VCVC 10]
V → [+stress] / ___ C0VC 10]
V → [+stress] / ___ C0]
3.2.1
e.g. [’zalama], [’miktaœif]
e.g. [ka’tabti], [’katab]
e.g. [’ÚaÙÙ] ‘he put’, [ba-’œef]
The interaction of morphology and word stress
Brame noted that the basic stress rules and their formalism in (2) fail to account
for data such as [ka’tabit] ‘I/you (masc sg) wrote’ (cf. [’katabit] ‘she wrote’),
TBC_135.qxd
7/13/10
19:27
Page 5
Word Stress in Arabic
5
[’?abilna] ‘before us’ (cf. [?’bilna] ‘we accepted’). These forms are derived from
underlying /katab-t/ and /?abl-na/ through epenthesis (see chapter 41: vowel
epenthesis), and are explained by Brame as the assignment of stress prior to
epenthesis, giving intermediate [ka’tab-t] and [’?abl-na]. Epenthesis does not
undo stress assignment, leading to opaque assignment of stress to a light penult
in [ka’tabit], and opaque non-assignment of stress to the heavy penult in
[’?abilna] (§3.2.3; chapter 58: opacity deconstructed). These cases of opaque
assignment, or lack of assignment, of stress were attributed initially by Brame (1970,
1973, 1974) and later by others (e.g. Kenstowicz & Abdul-Karim 1980; Kiparsky
1982, 2000, 2003) to the cycle and the preservation of metrical structure assigned
in earlier cycles ((54); chapter 60: cyclicity). The SPE-type approach to word assignment has since been superseded, but recognition of the role of the cycle and of
the interaction of syncope and epenthesis with word stress assignment has not.
As we shall see below (§4.2.3), within the stratal version of Optimality Theory
(OT), opaque stress is attributed to inter-level constraint masking: if a is the constraint system of domain Y (e.g. stem), and b is the constraint system of a larger
domain Z, then b’s markedness constraints can render a opaque (Kiparsky 2000).
Thus, opacity in dialects such as Levantine is attributed to word-level assignment
of stress and postlexical epenthesis, which renders stress opaque.
3.2.2
The prosodic hierarchy and representation of the syllable
The post-SPE period formed an asyllabic interlude in the analysis of Arabic word
stress. Most pre-generative and non-generative accounts made reference to the
syllable, and within generative phonology there was increasing recognition that
sounds grouped into syllables of differing prosodic weights, and that the syllable
formed part of the prosodic hierarchy (Fudge 1969; Kiparsky 1979; McCarthy 1979;
Selkirk 1980, 1982; Halle & Vergnaud 1987).
At this time it came to be recognized that the syllable formed a unit within a
prosodic hierarchy, a hierarchy that recognized units of prosodic structure above
the syllable – the foot (see chapter 42: the foot) and the prosodic word (see
chapter 51: the phonological word) – and a unit of prosodic structure below
the syllable: the mora. Weight-based, rather than segment-based, models of the
syllable representing the prosodic tier as a series of moras (e.g. McCarthy 1980;
Angoujard 1990; McCarthy & Prince 1990) provide a model that reflects the role
of prosodic weight in stress assignment by accounting for phonological positions,
and by distinguishing between light (monomoraic) and heavy (bimoraic) syllables
(Hayes 1989): short vowels contribute one mora, long vowels two moras, geminate consonants one mora, and coda consonants are assigned a mora through
Weight-by-Position in languages such as Arabic, where CVC syllables count as
heavy (quantity-sensitivity). In San’ani [tiœtai] ‘you (fem sg) want’, for example,
the vowels contribute moras and /œ/ receives a mora through Weight-byPosition (indicated by a dashed line):
q
(5)
t
q
[
[
i
œ
t
[
[
a
i
TBC_135.qxd
7/13/10
19:27
Page 6
Janet C. E. Watson
6
3.2.3
Metrical theory
The prosodic hierarchy plays a pivotal role within metrical theory, which
developed from Liberman (1975) and Liberman and Prince (1977) as part of nonlinear phonology. It shared with the other main branch of that program, autosegmental phonology, the goal of developing alternatives to the non-local devices of
linear theory, such as rule variables and abbreviatory conventions (Kager 1995:
368), and viewed stress as hierarchically organized rhythmic structure. Metrical
theory provided representation of the hierarchical nature of stress independently
of the segmental tier. This was achieved initially through metrical trees, wherein
stress was represented as a hierarchy of binary branching structures, labeled
strong–weak (s–w) or weak–strong (w–s) to mark relative prominence at each layer
(see chapter 43: representations of word stress). The metrical tree representation of Cairene [mux’talifa] ‘different (fem sg)’ (McCarthy 1979) is:
(6)
s
w
m
3.2.4
s
s
w
u
x
s
t
w
a
l
w
i
f
a
Rhythmic organization
The foot types recognized on the basis of rhythmic organization since Prince
(1976) enabled Arabic dialects to be classified in terms of foot shape: dialects
such as Cairene, Levantine, and San’ani, which exhibit initial prominence, organize strings of syllables into trochees (s–w pairs); dialects such as Cyrenaican Bedouin,
Tunisian, and Moroccan, which exhibit final prominence, organize strings of
syllables into iambs (w–s pairs). Compare: Cairene ’bana ‘he built’ and Cyrenaican
Bedouin bi’na:3
(7)
Cairene
Cyrenaican Bedouin
R
s
b
a
R
w
n
a
w
b
i
s
n
a
In early metrical accounts, feet which involved counting more than two were
permitted: for Halle and Vergnaud (1978), the foot in Damascene, a dialect that
exhibits initial prominence, included all syllables (effectively no more than three)
3
All data for Cyrenaican Bedouin are from Mitchell (1960).
TBC_135.qxd
7/13/10
19:27
Page 7
Word Stress in Arabic
7
from the stressed syllable to the right-edge of the word, placing a word like
[’madrase] ‘school’ within a single foot. McCarthy (1979, 1980) restricted the
foot by measuring it in moras: in Damascene, the stressed mora plus, at most,
two following moras. A word comprising three light syllables, such as [’darasu]
‘they learnt’, exhausts the foot, giving [’(darasu)], but in a word comprising a heavy
plus two light syllables, such as [’madrase] ‘school’, the final light syllable is excluded
from the foot: [’(madra)se].
R
(8)
q
[
[
m a
d
q
q
[
[
a
r
s
a
Since these accounts, bounded foot inventories have often excluded feet that require
counting higher than two (but cf. e.g. Burzio 1994). Hayes (1989, 1995) argues
for absolute binarity: the maximal and canonical iamb consists of a light syllable
followed by a heavy syllable, as in Cyrenaican Bedouin [ki’tab] ‘he wrote’:
R
(9)
k
q
q
[
[
[
a
b
i
t
Trochees comprise two equal elements: syllables in the syllabic trochee, moras in
the moraic trochee. In a moraic trochee dialect, [madrasa] ‘school’ comprises two
moraic trochees; cf. (8):
R
(10)
R
q
m
[
[
a
d
r
q
q
[
[
a
s
a
The uneven trochee, which comprises a heavy and a light syllable in moraic trochee
systems, is ruled out by Hayes. It is, however, invoked by Irshied and Kenstowicz
(1984), Angoujard (1990), and Kager (2009) to account for penultimate stress in
TBC_135.qxd
7/13/10
8
19:27
Page 8
Janet C. E. Watson
HLLL forms in trochaic Arabic dialects, as in Jordanian Bani-Hassan [#alla’mato]
‘she taught him’:
R
R
(11)
s
#
3.2.5
a
l
l
w
s
w
a m a
t
o
Constituency and the metrical grid
Metrical trees reflect constituency through sister nodes, but fail to represent in
any transparent way typical stress characteristics of stress clash or alternating rhythm
between strong and weak syllables. Prince (1983) and Selkirk (1984) argued that
the metrical grid could better capture the rhythmic characteristics of stress, and that
constituency into feet could be eliminated. Compare the pure grid representation
of Cairene [mux’talifa] ‘different (fem sg)’ below with the metrical tree representation
in (6):
(12)
*
*
*
* *
*
*
*
m u x t a l i f a
In 1985, data from an Arabic dialect appeared to challenge the effectiveness of
the pure grid. In an account of Bedouin Hijazi Arabic (BHA), Al-Mozainy et al.
(1985) analyzed alternations such as those in (13) as resulting from low vowel
deletion and stress shift.
(13)
’saÚab
’naxal
’salag
‘he pulled’
‘palm trees’
‘hunting dogs’
s’Úabat
n’xalah
s’ligah
‘she pulled’
‘a palm tree’
‘a hunting dog’
They argued the direction of shift was governed by constituent structure and that
vowel deletion in BHA induces left-to-right stress shift to the sister node within
the metrical tree:
(14)
s
s
s
→ w
≥
Ú
a
w
b
a
w
s
t
→
s
Ø
Ú
a
b
a
t
Through eliminating constituency, the pure grid provided no explanatory
account for the direction of stress shift in data such as these. The introduction
of brackets within the grid (Halle & Vergnaud 1987; Halle & Kenstowicz 1991;
TBC_135.qxd
7/13/10
19:27
Page 9
Word Stress in Arabic
9
Hayes 1995), however, enabled the grid to handle such data, showing that on
vowel deletion stress shifts within the bracketed constituent (in this case to the
right):
(15)
a.
(*
(*
*
s a
)
Ú
.)
*
a
b.
(
*)
*
s Ø Ú a
*
) word
foot
syllable
(
b
b
a
t
Further work on BHA showed the initial findings to have been based on an
incorrect analysis of the foot-type: essential paradigms had not been considered
(Angoujard 1992; Paoli 2008), and BHA feet were re-analyzed as iambs rather
than moraic trochees (McCarthy 2003); motivated by the tendency of iambic feet
to maximize quantitative differences between the head and dependent syllables,
deletion targets the unstressed vowel of an iambic foot. The basic conclusions that
stress is preserved on deletion of stressed vowels and that the direction of stress
shift is predictable from constituency remain valid, however, and are supported
by data from other Arabic dialects: Cyrenaican Bedouin (Hayes 1995, data from
Mitchell 1960) and San’ani (Watson 2002: 116) both exhibit stressed vowel deletion and stress shift4 within the foot: leftwards in Cyrenaican, which exhibits iambic
stress, and rightwards in San’ani, which exhibits trochaic stress.
(16)
a.
b.
3.2.6
Cyrenaican [ki’tib] ‘books’ > [’kitbih] ‘his books’
(
* )
( *
)
word
(.
*)
( *
)
foot
*
*
*
*
syllable
k i t i b > k i t Ø b i h
San’ani [’xaœab] ‘wood’ > [x’œabih] ‘a piece of wood’
( *
)
(
*
)
word
(*
.)
(
* )
foot
*
*
*
*
syllable
x a œ a b > x Ø œ a b i h
Unparsed syllables
The findings of many researchers suggest metrical structure need not exhaust the
string of syllables. Hayes (1995) claims syllables left over after foot construction
are either unfooted or form degenerate feet. Degenerate feet are not optimal, but
languages differ as to whether they impose a strong or a weak ban on them.
Languages that permit words smaller than the canonical foot (a single mora in
a moraic trochee system, or a single syllable in an iambic or syllabic trochee
system) tend to invoke the weak ban, while languages that do not permit subminimal words invoke the absolute ban.
Sub-minimal content words are generally unattested in Arabic and subminimal grammatical words are expanded on suffixation to provide a minimal
bimoraic base, as in Cairene:
4
Optional in the case of San’ani: [’xaœabih] ~ [x’œabih].
TBC_135.qxd
7/13/10
19:27
Page 10
10
Janet C. E. Watson
(17)
/min-u/
/#an-kum/
/fi-h/
>
>
>
[minnu]
[#annukum]
[fc(h)]
‘from him/it (masc)’
‘about you (pl)’
‘in it, there is’
Sub-minimal loanwords are typically expanded through vowel lengthening to
match the minimal prosodic word, as in: [bar] ‘bar’ and [baÛ] ‘bus’. Some dialects,
such as San’ani, however, do have stressable monomoraic content words,
including [ab] ‘father’, [ax] ‘brother’, [yad] ‘hand’, [dam] ‘blood’, and a few subminimal function words that contrast with comparable bimoraic words in
Cairene, and never lengthen, including [kam] ‘how many’ (cf. Cairene [kam]),
[man] ‘who’ (cf. Cairene [mcn]), and [ma#] ‘with’ (cf. Cairene [ma#a]) (Watson
2002: 88 –89). Further evidence that degenerate feet are allowed in strong position in San’ani includes the exceptional stressing of peripheral light syllables,
giving optional initial stress in forms such as [’tamam] ‘good’, and optional final
stress in forms such as [œik’mih] ‘party for parturient’ (§4.2.2).
3.2.7
Weak parsing
Many languages stress the third syllable from the word edge. Hayes (1982, 1989,
1995) argues such systems can be accounted for not by expanding the universal
inventory to include ternary feet, but by resorting to the independently motivated
devices of extrametricality at the edge, destressing in clash, and the nonexhaustivity of foot construction. Non-exhaustivity of foot construction means
syllables can be skipped through a device known as weak local parsing, potentially creating ternary alternation in longer strings (Hayes 1995: 308):
(18)
(x .)
(x .)
(x .)
˘ ˘ ˘ ˘ ˘ ˘ ˘ ˘ ˘
This device enables Hayes to provide an account of Arabic dialects which dispenses with the uneven trochee. Bani-Hassan [#alla’mato], seen above in (11), is
analyzed in moraic trochees with weak local parsing (Hayes 1995: 366):
(19)
3.2.8
(x)
(x
.)
¯
˘
˘
˘
# a l l a m a t o
Final consonants, syllables, and feet
The right-edge of the word prompts exception in many languages: extra-long
syllables are often restricted to the right-edge, and syllables that act as heavy
non-finally often fail to attract stress in final position. In Cairene, the sequence
[.tab.] is stressed penultimately in [ka.’tab.tu] ‘you (pl) wrote’, but not finally in
[’ka.tab] ‘he wrote’. To account for the asymmetric behavior of closed syllables
and the invisibility of peripheral elements to stress rules, Liberman and Prince
(1977) introduced the notion of extrametricality (see chapter 45: extrametricality
and non-finality), the rules for which were developed by Hayes (1979, 1982,
1989). Thus, rather than specify for relevant languages that CVC is light finally,
but heavy non-finally, and that only CVCC and CVVC syllables are heavy
finally, the rightmost consonant is analyzed as invisible to stress rules through
extrametricality, making final CVC equivalent in weight to non-final CV (Hayes
1995: 57):
TBC_135.qxd
7/13/10
19:27
Page 11
Word Stress in Arabic
(20)
Final
Non-final
CV
CV<C>
CVC<C>
CVV
CVV<C>
CV
CV
CVC
CVV
CVV
11
In several Arabic dialects, final pairs of light syllables also appear to be invisible
to stress rules. In Palestinian, for example, stress is assigned to the rightmost bimoraic
sequence in these words:5
(21)
’?ana
’katabu
bara’kitna
ka’tabna
‘I’
‘they wrote’
‘our cow’
‘we wrote’
However, in words comprising a heavy syllable followed by two light syllables,
or four light syllables, stress is assigned to the initial syllable:
(22)
’barakito > ’bakarto
’madrasa
’œajaratun (Classical)
‘his cow’ (with High Vowel Deletion)
‘school’
‘tree (nom)’
Hayes (1995) analyses such patterns as resulting from foot extrametricality, subject to the non-exhaustivity condition. Thus, [’madrasa] is parsed as [(mad)(rasa)],
with two bimoraic feet; by not exhausting the word the peripheral foot is eligible
for extrametricality, giving [(mad)<(rasa)>]. Stress is assigned to the rightmost
visible (non-extrametrical) foot: [’(mad)<(rasa)>]. The characteristic Cairene pattern
of stressing a light penult after a heavy antepenult, by contrast, is analyzed as
resulting from lack of foot extrametricality: the rightmost visible foot is stressed
in all dialects, but only in Cairene is the peripheral foot visible to stress rules:
Cairene [(mad)’(rasa)] contrasts with Palestinian [’(mad)<(rasa)>].
3.2.9
CVXC syllables
Stress patterns in several languages, including Arabic, indicate that an extrametrical
consonant does not deprive the rightmost foot of peripherality: by being contained
within the peripheral foot, the peripheral consonant does not intervene between the
foot and the right-edge. In San’ani [’katabatih] ‘she wrote it (masc)’, for example,
extrametrical /h/ falls within the rightmost foot, itself deemed extrametrical. Stress
assignment to the rightmost visible foot gives [’(kata)<(bati<h>)>]. An analysis
of extrametricality in the case of CVVC and CVCC strings, however, predicts the
wrong results: extrametricality would render final C invisible to stress rules; as
an extrametrical element it would fall in the adjacent foot, failing to deprive the
foot of peripherality; as peripheral feet, CVCC and CVVC syllables would be
invisible to stress rules in dialects such as Palestinian for which foot extrametricality holds, but not in dialects where foot extrametricality fails to apply. Such
an analysis would predict a stress difference in words of the pattern CVVCVCC
5
Data from Brame (1973, 1974), Abu-Salim (1980), and Kenstowicz & Abdul-Karim (1980).
TBC_135.qxd
7/13/10
19:27
Page 12
Janet C. E. Watson
12
and CVCCVCC between dialects that exhibit foot extrametricality (23a), and
dialects such as Cairene, which do not (23b):
(23)
a.
Consonant extrametricality:
Foot extrametricality:
Stress rightmost visible foot:
b.
Consonant extrametricality:
Foot extrametricality:
Stress rightmost visible foot:
darrast → darras<t>
(dar)(ras<t>) → (dar)<(ras<t>)>
*’(dar)<(ras<t>)> (= dar’rast)
darrast → darras<t>
n/a
(dar)’(rast)
A significant step in research on Arabic stress concerns the analysis of these
so-called “superheavy” syllables. Superheavy syllables are exceptional on two
counts: they are the only syllable types that are always stressed in final position
(although cf. §4.2.2), and they are restricted to domain-final position (at least
in morphologically simple forms). Thus, in terms of stress rules, they behave
like penultimate CVV or CVC (as do CVVC/CVCC syllables in English; Burzio
1994; Harris & Gussmann 2002). Several analyses of CVVC/CVCC strings have
been proposed around the basic analysis of heavy syllable + element. This was
expressed initially by McCarthy (1979: 453) through Chomsky-adjoining a wordfinal consonant to a heavy syllable:
q
(24)
q
C
q
V [+seg]C# →
C
V [+seg]C#
Halle and Vergnaud (1979) analyze final C in CVXC as the weak element in a
branching foot:
(25)
s
w
k
a
w
s
w
t
a
b
t
In later accounts (Aoun 1979; Angoujard 1981, 1990; Selkirk 1981;6 Burzio 1994),
the rightmost C forms a degenerate syllable, i.e. a syllable with an empty nucleus
or null vowel:
q
(26)
C
6
q
V [+seg] C
D
Only for CVCC syllables. See §135.4.2.2.
TBC_135.qxd
7/13/10
19:27
Page 13
Word Stress in Arabic
13
For Hayes (1995: 126), final C in CVXC is not syllabified at the initial stage: CVXC
is analyzed as heavy syllable + stray consonant, while CVC is analyzed as light
syllable + extrametrical consonant:
(27)
CVX.C
CV<C>
Within OT, Al-Mohanna (2004) analyses final C of word-final CVXC as attached
directly to the prosodic word node. Essentially, then, all accounts show that the
rightmost C in CVXC sequences intervenes between the right-edge and the heavy
syllable, depriving the foot formed by the heavy syllable of peripherality. Thus,
darrast (23) in both Cairene- and Palestinian-type dialects is analyzed as [(dar)’(ras).t].
This section has examined key approaches to Arabic stress within the generative
paradigm, with particular focus on challenges raised by Arabic data for stress
theory. §4 will provide sketches of the stress systems of Classical Arabic and three
modern Arabic dialects, focusing on core similarities and differences between the
dialects, and examining approaches invoked to account for cross-dialectal differences and apparent exceptions to the stress algorithms.
4
4.1
Stress algorithms
Classical Arabic
The early Arab grammarians provided detailed descriptions of segments and
melodic phonological processes characteristic of readings of the Qur’an and certain
Peninsula dialects; however, word stress is never mentioned. This led some
researchers to believe that Classical Arabic exhibited no word stress (Birkeland
1954; Ferguson 1956; Garbell 1958), and others to assume it to have been similar
to the rather fluctuating word-stress system found today in western dialects of
the Arabian Peninsula.
The Classical Arabic stress patterns have since been reconstructed through
comparison of modern dialect stress patterns (Janssens 1972), versification (Weil
1954; Wright 1971), and observation of the non-dialectal pronunciation of Classical
Arabic in some regions (Abu-Fadl 1961; Mitchell 1993). There is now general
consensus that Classical Arabic exhibited predictable stress. Disagreement exists,
however, as to the leftmost limit of stress. It is agreed that penultimate CVC or
CVV bore stress, or, if the penult was light, antepenultimate CVC or CVV. Where
both the penult and antepenult were light, as in /mas?alat-un/, researchers differ: Erpenius (1656), Abdo (1969), Brame (1970), Bohas and Kouloughli (1981), and
Angoujard (1990) argue that stress did not retract beyond the antepenultimate
syllable, giving, in this case, [mas’?alat-un] ‘problem (nom)’. Brockelmann (1907),
Wright (1971), Janssens (1972), and McCarthy (1979), by contrast, claim stress is
assigned to the initial syllable in such cases, giving ’mas?alatun. If this latter holds,
this would mean Classical Arabic, in contrast to the modern dialects, exhibited
unbounded metrical feet, constructing feet from one heavy syllable up to, but not
including, the next heavy syllable. Under both analyses, lexical exceptions exist: dialect
comparison and the non-dialectal pronunciation of Classical Arabic suggest stress
was not assigned to the initial heavy syllable in derived verb forms VII and VIII,
but to the light antepenult, giving: [in’fa#ala] and [if’ta#ala] rather than *[’infa#ala]
TBC_135.qxd
7/13/10
14
19:27
Page 14
Janet C. E. Watson
and *[’ifta#ala].7 Similarly, most particle prefixes are unstressed irrespective of
their relative position (Mitchell 1960: 371, 1975: 77), leading to antepenultimate
or penultimate stress in forms such as: [al-’walad-u] ‘the boy (nom)’ (cf. Wright’s
[’madrasat-un] ‘school (nom)’), [al-’yad-u] ‘the hand (nom)’, (cf. [’maktab-un] ‘office
(nom)’), [wa-’yad-un] ‘and a hand (nom)’ (cf. [’katabat] ‘she wrote’).
The stress algorithm for Classical Arabic is given in (28) (bracketed elements
included in case stress retraction is limited to the antepenult):
(28)
Classical Arabic stress
a.
b.
c.
4.2
Stress a pre-pausal superheavy (CVVC, CVVGG, or CVCC) syllable:
[ki’tab] ‘book’, [’madd] ‘stretching (masc sg)’, [œa’ribt] ‘I/you (masc sg)
drank’.
Otherwise, stress the rightmost non-final heavy (CVV, CVC, or CVVG)
syllable (up to the antepenult): [da’rasna] ‘we learnt’, [Ûa’benun] ‘soap
(nom)’, [’maktabah] ‘library’, [’maddun] ‘stretching (nom)’, [’maktabatun]
‘library’ (non-pause) (or [mak’tabatun]).
Otherwise, stress the leftmost CV syllable (or antepenult): [’kataba]
‘he wrote’, [’katabatuhu] ‘library’ (or [kata’batuhu]).
Arabic dialects
This section presents the basic stress algorithms for Cairene, San’ani, and Levantine,
three dialects analyzed as exhibiting moraic trochaic stress. Each sub-section
considers some of the most significant data that has impacted on metrical theory
and approaches invoked to handle this data. The section is concluded by a table
summarizing the main stress and stress-related typological characteristics of each
dialect and of Cyrenaican Bedouin, aspects of which we considered in §3.
4.2.1
Cairene
More generative accounts of word stress have been provided for Cairene than
any other Arabic dialect. Cairene attracted attention due to its characteristic
avoidance of a heavy antepenult in favor of a light penult, deletion of unstressed
high vowels but, with few exceptions (Woidich 2006), not unstressed low vowels,
reduction of unstressed long vowels, and its exceptions.8 An initial stress algorithm
for Cairene was provided by Harrell (1957, cf. also Mitchell 1956):
(29)
Cairene stress
a.
b.
c.
7
Stress-final CVV(C) or CVCC: [ka’tabt] ‘I wrote’, [?a’be(h)] ‘his father’,
[saka’kcn] ‘knives’, [Ùala’bat] ‘demands’.
Otherwise, stress the antepenult when the penult and antepenult are
light, unless the pre-antepenult is light: [’?abadan] ‘never’, [mux’talifa]
‘different (fem sg)’. Cf. [kata’bitu] ‘she wrote it (masc)’ with preantepenultimate CV.
Otherwise, stress the penult: [yik’tibu] ‘they write’, [#a’malti] ‘you
(fem sg) did’, [mar’taba] ‘mattress’, [’bbtak] ‘your (masc sg) house’.
An exception for Wright (1971), etc., but not for Angoujard (1990).
Descriptive accounts of Cairene in the theoretical literature include Mitchell (1952, 1956, 1960, 1975),
Harrell (1957, 1960), Tomiche (1964), Behnstedt and Woidich (1985), and Woidich (2006).
8
TBC_135.qxd
7/13/10
19:27
Page 15
Word Stress in Arabic
15
Example (29b) predicts penultimate stress whenever the penult, antepenult,
and pre-antepenult are light. However, since long strings of light syllables are
unattested in Cairene, we cannot evaluate the stress pattern of words that exceed
four light syllables by restricting ourselves to dialect data. Mitchell’s (1960, 1975)
study of Classical Arabic pronunciation as taught in centers in Cairo provided
longer strings of light syllables, demonstrating that (29b) held in forms such as
[œaja’ratuhu] ‘his tree’, but not in [œaja’ratun] ‘a tree (nom)’ and [murtabi’Ùatun]
‘connected (fem nom)’. Mitchell provides a comprehensive list of canonical
patterns, for which Langendoen (1968) formulates an algorithm along the following lines.
(30)
Cairene stress
a.
b.
c.
Stress a superheavy ultima.
Otherwise, stress a heavy penult.
Otherwise, stress the penult or antepenult, whichever is separated by
an even number of syllables from the rightmost non-final heavy syllable,
or, if there is no non-final heavy syllable, from the left boundary of the
word.
Seeking a heavy syllable from the word-edge is common, and stressing a light
antepenult in the absence of a heavy penult is explicable on the basis of moracounting (stress the third mora from the right-edge); however, the choice of penult
or antepenult in (30c) depends not only on the weight of the stressed syllable and
that of syllables to its right, but also on the weight of the syllable to its left. McCarthy
(1979) viewed the distinction between odd and even sequences of light syllables
as a tacit alternating pattern, explaining the apparently complex rules as the
grouping of light syllables into left-headed pairs from left to right. Hayes’s (1995)
bracketed-grid account of the facts and of McCarthy’s analysis is below:
(31)
a.
b.
c.
Consonant extrametricality: C → <C> / _____]word.
Foot construction:
Form moraic trochees from left to right.
Degenerate feet are not permitted.
Word layer construction:
Group feet into a right-headed word
constituent: End Rule Right (ERR).
Taking [mak’tabak] ‘your (masc sg) office’, the rules generate the following
metrical structure:
(32)
(.
*
(*)
(*
µ µ
µ
m a k t a
b
)
.)
µ
a <k>
word layer construction: ERR
foot construction: moraic trochees, L>R
consonant extrametricality
This formalism accounts for most of the data, but fails to account for the following cases:
(33)
a.
Penultimate stress in plural nouns of the pattern CiCiCa/CuCuCa: [li’bisa]
‘underpants’, [ıi’riba] ‘crows’, [si’bita] ‘baskets’.
TBC_135.qxd
7/13/10
16
19:27
Page 16
Janet C. E. Watson
b.
Stress of the 3fem sg perfect subject vowel on suffixation: /œafit-u/ >
[œa’fitu] ‘she saw him’, /ramit-u/ > [ra’mitu] ‘she threw it (masc)’,
/katabit-u/ > [kata’bitu] ‘she wrote it (masc)’.
Deletion of short high vowels in CVCVCV(C) sequences in suffixed forms:
/yaxud-u/ > [’yaxdu] ‘they took’ (cf. /#alam-u/ > [#a’lamu] ‘his world’),
/kanakit-u/ > [kanaktu] ‘his coffeepot’ (cf. [kata’bitu] ‘she wrote it
(masc)’).
c.
(33a) and (33b) have been analyzed in various approaches as the exceptions they
are. Penultimate stress in nouns of the pattern CiCiCa/CuCuCa can be explained
by referring to the alternative, older, pattern iCCiCa: [ıi’riba] ~ [iı’riba] ‘crows’.
In the older form, stress is predictably assigned to a light penult following a heavy
antepenult, as in (30c). Woidich (2006) and Watson (2002: 98) then interpret the
apparent exceptional stress of CiCiCa/CuCuCa forms as lexicalization of the older
stress pattern and re-analysis of iCCiCa as CiCiCa.
The unpredictable behavior of the suffixed 3rd feminine singular perfect
inflectional form has been handled in several different, but related, ways.
McCarthy’s metrical tree notation permits a morphologically governed rule to
create a branching node over the verbal suffix -it and any following material, iff
following material exists:
(34)
+it +X
1 2 >> 1
2
Angoujard (1981) initially attributes an ‘indestructible rhyme’ to the -it morpheme.
Later he attributes the heaviness of the morpheme in dialects such as Cairene and
Tunisian to the association of two segments with a three-slotted grid, of which
the first is marked as peak (Angoujard 1990: 120 –121):
(35)
*
*
*
i
*]
t
Watson (2002: 97) invokes exceptional reversal of the direction of stress assignment
from left to right to right to left in the case of suffixation to the -it morpheme.
Forms such as /yaxud-u/ > [yaxdu] ‘they take’ (33c) are problematic for a purely
weight-based model because /yaxud-u/ shares the CVVCVC+V template of
words such as /#alam-u/ (> [#a’lamu]) ‘his world’, differing solely in the quality of the penultimate vowel. Angoujard (1990) deals with these by allowing uneven
trochaic feet consisting of CVV plus CV with a short high vowel, while lone
syllables with short low vowels may, and domain-final syllables must, form a foot
of their own. Compare Angoujard’s representations of /#alam-u/ ‘his world’ and
/?abil-u/ ‘they met’:
R
(36)
#
aa
R
l
a
R
m
u
R
?
aa
b
R
i
Ø
l
u
TBC_135.qxd
7/13/10
19:27
Page 17
Word Stress in Arabic
17
In /#alam-u/ > [#a’lamu], unstressed CVV is reduced to CV. In /?abil-u/ > [’?ablu],
the high vowel in the weak position of the foot is deleted, resulting in a CVVC
syllable, impermissible in Cairene. The long vowel is subject to Closed Syllable
Shortening (CSS) to prevent word-internal CVVC, giving: [’?ablu] ‘they met’. Watson
(2002), who follows Hayes (1995) in disallowing the uneven trochee, analyses forms
such as /safir-it/ > [’safrit] as assignment of metrical structure prior to syncope
with re-assignment of metrical structure after the application of each phonological rule:
(37)
a.
b.
c.
d.
e.
Construction of moraic trochees from left to right: (sa)(firit).
Assignment of stress to the head of the rightmost foot: (sa’(firit)).
Reduction of unstressed CVV to CV: (sa’(firit)).
Refooting: (’(safi)rit).
Syncope of the high vowel: ’(safrit).
Both these approaches deal with this data, but they miss the generalization that,
with the exception of CiCiCa/CuCuCa plurals and the -it morpheme, all short
high vowels in the position CVCVCV(C) are subject to syncope, even if they would
be stressed by the normal stress algorithm (Kenstowicz 1980; Teeple 2009), namely:
/kanakit-u/ > [ka’naktu] ‘his coffeepot’, /kanabit-u/ > [ka’nabtu] ‘his sofa’. This
data suggests word-internal short high vowels are subject to syncope prior to assignment of metrical structure, as long as the resulting syllable is permissible (cf.
Broselow 1992: 36–37): /kanakit-u/ gives [kanaktu], but /mudarris-a/ fails to give
*[mudarrsa] because word-internal [.darr.] (CVGG) is impermissible in Cairene.
Syncope is not restricted to the phonological word in Cairene; it also occurs
within the phonological phrase: high vowels in word-initial CV syllables are
subject to phrasal syncope after a word-final vowel, as in:
(38)
/?ana fihimt/
/Ùardi kibcr/
>
>
’?ana f’himt
’Ùardi k’bcr
‘I understood’
‘my parcel is big’
Deletion fails to occur in (39), however, even though word-initial CV follows wordfinal CV:
(39)
/huwwa fihim/
/huwwa wiÚiœ/
>
>
’huwwa ’fihim
’huwwa ’wiÚiœ
(*fhim)
(*wÚiœ)
‘he understood’
‘he is bad’
In (38), the high vowel falls in an unstressed syllable in the citation form
([fi’himt] ‘I understood’, [ki’bcr] ‘big’). By contrast, in (39), the high vowel falls
in a stressed syllable in the citation form ( [’fihim] ‘he understood’, [’wiÚiœ]
‘bad’). Taken with data such as [ka’naktu] ‘his coffeepot’, this suggests two types
of syncope occur in this dialect: lexical syncope, which targets word-internal CVflanked high vowels prior to the assignment of metrical structure, and phrasal
syncope, which targets unstressed word-initial CV-flanked high vowels after the
assignment of metrical structure.
4.2.2
San’ani
The main interest in San’ani, the dialect of the old city of San’a, Yemen, lies in
its stressing of peripheral light syllables, and the patterning of CVV syllables
TBC_135.qxd
7/13/10
18
19:27
Page 18
Janet C. E. Watson
with syllables ending in the left-leg of a geminate (CVG), but not with CVC
syllables.9
The basic stress algorithm for Classical Arabic (28) applies to most word-types:
(40)
a.
b.
c.
Stress a final CVCC or CVVC syllable: [gi’rct] ‘I/you (masc sg)
read/learnt’, [gam’bart] ‘I/you (masc sg) sat’, [ki’tab] ‘book’.
Otherwise, stress the rightmost non-final heavy syllable (CVV or CVC)
up to the antepenult: [’madrasih] ‘school’, [miı’salih] ‘launderette’,
[da’rastih] ‘I/you (masc sg) recited it’.
Otherwise, stress the leftmost CV syllable: [’libisat] ‘she put on, wore’,
[mak’tabatih] ‘his library’, [’katabatih] ‘she wrote it (masc)’.
(40) fails to apply in San’ani, however, when the penult or antepenult is CVV or
CVG. Here the rightmost non-final CVV or CVG syllable attracts stress from final
CVCC or CVVC (Watson 2002: 81):
(41)
‘soap’
‘clasp’
‘teachers (masc)’
‘going out (masc pl)’
’Ûaben
’xuÙÙaf
mu’darriscn
’xarijcn
Similarly, while the rightmost non-final CVC syllable is stressed iff in penultimate
or antepenultimate position, CVV and CVG are stressable in pre-antepenultimate
position:
(42)
mak’tabatc
mu’sajjilatc
’hakaÆaha
‘my library’
‘my recorder’
‘like this’
(40) often fails to apply in post-pausal position: San’ani exhibits contextually
fluctuating stress (Rossi 1939; Naïm-Sanbar 1994); post-pausally the initial syllable
is usually stressed, irrespective of its weight or that of following syllables. Where
initial CV is stressed before CVXC, this is analyzed in Watson (2002) as a degenerate syllable (§3.2.6):
(43)
’tamam
’katabt
’?usbe#
’baladcyat
‘okay’
‘I/you (masc sg) wrote’
‘week’
‘municipality’
The following revised algorithm accounts for the data:
(44)
San’ani stress
a.
9
Stress the rightmost non-final CVV or CVG syllable: [ma’katib] ‘offices’,
[ba’satcn] ‘gardens’, [’xarijcn] ‘going out (masc pl)’, [mit’?axxirat] ‘late
(fem pl)’, [’hakaÆaha] ‘like this’, [’safart] ‘I/you (masc sg) traveled’,
[’dawwart] ‘I/you (masc sg) looked for’.
Data cited in the theoretical literature include Rossi (1939), Goitein (1970), Naïm-Sanbar (1994), and
Watson (2002).
TBC_135.qxd
7/13/10
19:27
Page 19
Word Stress in Arabic
19
b.
Otherwise, stress final CVVC or CVCC: [?ab’sart] ‘I/you (masc sg) saw’,
[laf’laft] ‘I/you (masc sg) collected’, [diœ’man] ‘rebel’, [ba’nat] ‘girls’.
c. Otherwise, stress the rightmost non-final CVC syllable up to the
antepenultimate: [’laflaf] ‘he collected’, [’maklaf] ‘woman’, [’maklafih]
‘his woman’.
d. Otherwise, stress the leftmost CV syllable: [’katab] ‘he wrote’, [’darasat]
‘she learnt’, [’ragabatih] ‘his neck’, [mak’tabatc] ‘my library’.
Adopting Hayes’ (1995) bracketed-grid approach, Watson accounts for the facts
as follows:
(45)
a.
b.
Consonant extrametricality
Foot construction
C → <C> /__ ]word
Form moraic trochees from left to right.
Degenerate feet are permitted in strong
position.
Foot → <Foot> /__ ]word
ERR
c. Foot extrametricality
d. Word layer construction
The permitting of degenerate feet in strong position (45b) accounts for post-pausal
stressing of CV in words of the structure CVCVXC ([’tamam], [’katabt]), while
(45c) accounts for initial stress in [’maktabih] ‘library’ (cf. Cairene [mak’taba]).
(45) does not, however, account for the most striking characteristic of San’ani: failure of final CVXC to attract stress when a CVV or CVG syllable occurs in the
word, and stress assignment in words ending in CVVC.
Geminates are analyzed as underlyingly moraic (see chapter 3: geminates).
Although non-geminate consonants in the rhyme are assigned a mora through
Weight-by-Position, geminates are moraically distinct from non-geminate consonants
at some stage in the derivation.10 This underlying moraic distinctness accounts
for the asymmetry in San’ani between CVC syllables, on the one hand, and CVG
and CVV syllables, on the other. The distinctness is captured by adopting a twolayered grid within the syllable, where the height of a column depends on the
sonority of the segment associated with it (Hayes 1995: 300, drawing on Prince
1983).11 Underlyingly moraic segments have a mora on each layer, while segments
assigned a mora by Weight-by-Position have a mora on the lower layer only:
(46)
q
q
q
[
[
[
[
[
[
[
[
[
[
[
[
C V
10
[
C V C
q
C V G
C V V
Unstressed vowel shortening in Rural Palestinian also implies a distinction in this dialect between
CVC syllables and CVV and CVG syllables (Younes 1995).
11
A similar two-leveled grid developed by Hyde (2006) distinguishes between prosodic and metrical
weight. Under this approach, CVC, CVV, and CVG in San’ani are bimoraic, but only in the case of
CVV and CVG are both moras projected on the metrical tier.
TBC_135.qxd
7/13/10
20
19:27
Page 20
Janet C. E. Watson
Processes that treat CVC as heavy refer to the lower moraic layer, while processes
that treat CVC as light refer to the upper moraic layer. In San’ani, the weight of
CVC syllables is relative and entirely dependent on the prosodic environment.
Footing is enforced on the upper moraic layer iff underlyingly bimoraic (i.e. CVV
or CVG) syllables occur in the word; elsewhere, footing is enforced on the lower
moraic layer.
This analysis accounts for the failure of CVCC syllables to attract stress, iff
a non-final CVV or CVG syllable falls within the word. Stress is assigned in [’safart]
‘I/you (masc sg) traveled’ as below:
(47)
(
s
x
(x)
q
)
<q>
q
[
[
[
[
[
[
[
a
r <t>
a
f
word layer construction (ERR)
foot construction:
moraic trochees, L > R
consonant extrasyllabicity
Penultimate and antepenultimate stress in words such as ’jazzar ‘butcher’, ’Ûaben
‘soap’ and ’xarijcn ‘going out (masc pl)’ is attributed to a prosodic difference in
Arabic between CVVC and CVCC syllables. In several Arabic dialects, CVVC occurs
in positions from which CVCC is excluded. In Levantine, Sudanese, and Iraqi,
CVVC cannot occur morpheme internally but, unlike CVCC, is attested word internally on suffixation of inflectional affixes, as in Lebanese: /kitab-na/ > [kitabna]
‘our book’ and Sudanese: /masik-cn/ > [maskcn] ‘holding (pl)’ (Broselow et al.
1995, 1997). In Classical Arabic, word-internal CVVG syllables are the regular
outcome of active participle formation, as in [farra] ‘to flee’ > [farr-un] ‘fleeing
(nom)’ (Wright 1971: 26). And in Cairene, CVVC may occur word finally within
the utterance, while words ending in CVCC prompt epenthesis in all but
utterance-final position: compare [ban≠t Ùawclat] ‘tall girls’ with [binti Ùawcla]
‘a tall girl’ (Selkirk 1981).
While CVCC is analyzed as a canonical syllable plus an extra element –
(24)–(27) – final CVVC is analyzed as a single superheavy syllable (Selkirk 1981:
215), which Watson (1999, 2002) interprets as a bimoraic syllable with an extrametrical consonant:
q
(48)
[ [
C
V
<C>
Under this analysis, final CVVC forms a single bimoraic foot in San’ani. Combined
with the moraic grid, the foot formed by word-final CVVC is eligible for extrametricality subject to the Non-exhaustivity Condition, i.e. iff other bimoraic syllables
on the upper moraic layer occur in the word. In assignment of stress in [’Ûaben]
TBC_135.qxd
7/13/10
19:27
Page 21
Word Stress in Arabic
21
‘soap’, two bimoraic feet are constructed on the upper moraic layer; the peripheral
foot is extrametrical, and stress is assigned to the rightmost visible foot:
(49)
(
Û
4.2.3
x
(x)
q
)
word layer construction (ERR)
foot extrametricality
<(x)>
q
[
[
[ [
[
[
[ [
a
b u
<n>
consonant extrametricality
Levantine
Levantine has attracted a great deal of attention within generative phonology
due principally to its complex interactions between the morphology and stress
assignment.12 The dialects basically exhibit the Latin stress rule (McCarthy 1980:
79), with the exception of (50a), which Latin lacks:
(50)
Levantine stress
a.
Stress final CVCC or CVVC: [bi-t-’ÚuÙÙ] ‘she/you (masc sg) put’,
[ja’wab] ‘answer’.
b. Otherwise, stress a heavy penult: [ka’tabti] ‘you (fem sg) wrote’,
[’barak] ‘he blessed’.
c. In disyllables ending in CV or CVC, stress the initial syllable: [’Òarab]
‘he hit’, [’bana] ‘he built’, [’?ana] ‘I’.
d. Otherwise, stress the antepenult: [’darasu] ‘they learnt’, [mut’taÚida]
‘united’, [’#allamat] ‘she taught’, [’madrasa] ‘school’.
Hayes (1995: 128) accounts for Palestinian stress as follows:
(51)
a.
b.
Consonant extrametricality
Foot construction
c. Foot extrametricality
d. Word layer construction
C → <C> /__ ]word
Form moraic trochees from left to right.
Degenerate feet are forbidden absolutely.
Foot → <Foot> /__ ]word
ERR
These rules generate the following metrical structure:
(52)
(*
(*)
m
12
)
.)>
<(*
µ µ
a k t
µ
a
b
µ
a
word layer construction: ERR
foot construction: moraic trochees, L > R,
foot extrametricality
<k>
consonant extrametricality
The many traditional descriptive-analytical accounts of Levantine include Feghali (1919), Bauer
(1926), Cantineau (1939, 1946, 1960), El-Hajjé (1954), Cowell (1964), Jiha (1964), Grotzfeld (1965), Blanc
(1970), Fleisch (1974), and Naïm-Sanbar (1986).
TBC_135.qxd
7/13/10
22
19:27
Page 22
Janet C. E. Watson
Levantine dialects show a few internal differences. Palestinian stresses the fourth
mora from the right-edge in Classical Arabic words such as [œajaratun] ‘a tree (nom)’,
giving [’œajaratun] vs. Beirut/Damascene [œa’jaratun] (Halle & Kenstowicz 1991).
This is analyzed by Halle and Kenstowicz (1991) as a difference in parsing direction and extrametricality: left-to-right in Palestinian (as in (51b), with foot extrametricality), right-to-left in Beirut/Damascene (with syllable extrametricality). Since
Cantineau (1939), Levantine dialects have been classified into “differential” and
“non-differential” dialects, depending on whether only unstressed high vowels
or all unstressed vowels are subject to syncope; cf. (1). Dialects also vary according to whether or not long vowels are shortened in open unstressed syllables,
as in Palestinian /makatib-na/ > [maka’tibna] (Younes 1995) vs. Damascene
/katabHt-o/ > [kata’bHto] ‘she wrote it’ (McCarthy 1980).
4.2.4 The interaction of syncope, epenthesis, and stress
The main interest in Levantine stress data is in the treatment of morphologically
complex forms, where stress is assigned opaquely in some cases to a light penult
(skipping a heavy antepenult or pre-antepenult), in other cases to a light antepenult (skipping a heavy penult). Consider the Palestinian data below (Brame 1974;
Abu-Salim 1980; Younes 1995):
(53)
Penultimate
Antepenultimate
ka’tabit
f’himit
’xubizha
’binitha
’yikitbu
’jabilha
‘I/you wrote’
‘I/you understood’
‘her bread’
‘her daughter’
‘they write’
‘he gave her’
The traditional explanation of opaque stress assignment to the penult in words
such as [ka’tabit] and [f’himit], and to the antepenult in [’binitha] and [’yikitbu]
is that the epenthetic vowel fails to count for stress. Since Brame (1970, 1973, 1974)
this explanation has been translated in rule-based approaches into the relative
ordering of stress, syncope and epenthesis. Stress is assigned in the lexicon, and
within the postlexical component syncope is ordered before epenthesis. This
accounts for opaque stress in Palestinian [ka’tabit] ‘I/you (masc sg) wrote’ and
[’yikitbu] ‘they (masc) write’, where syncope and/or epenthesis take place, and
for transparent stress in [’katabin] ‘they (fem) wrote’, where neither syncope nor
epenthesis occurs:
(54)
first cycle
stress
second cycle
post-cyclic
syncope
epenthesis
output
[[fihm]na]
[katab-t]
[yi-ktib-u]
[katab-in]
’fihm
’fihm-na
ka’tab-t
’yiktib-u
’katab-in
NA
’fihimna
’fihimna
‘our understanding’
NA
ka’tabit
ka’tabit
‘I/you (masc
sg) wrote’
’yiktbu
’yikitbu
’yikitbu
‘they (masc)
write’
NA
NA
’katabin
‘they (fem)
wrote’
TBC_135.qxd
7/13/10
19:27
Page 23
Word Stress in Arabic
23
Levantine-like opacity has continued to attract considerable interest within
constraint-based models. Optimality-theoretic approaches towards opacity include:
invoking constraints on the stressing of epenthetic vowels, maximizing paradigmatic contrasts,13 and translating the notion of the cycle into lexical and postlexical strata. Within parallel OT, Kager (1999) accounts for Levantine opacity by
invoking the constraint Head-Dep(O/I) (‘every vowel in the output prosodic head
has its correspondent in the input’), which prohibits stress on epenthetic vowels.
By dominating constraints responsible for stress, Head-Dep(O/I) rules out penultimate stressed *[fi’himna] (input [’fihm-na]) and *[yi’kitbu] (input [’yiktib-u]), and
other constraints select [’fihimna] and [’yikitbu]. Kiparsky, however, raises two
objections to this constraint: first, its only remit is to prevent epenthetic vowels from
being stressed, but epenthetic vowels are not simply unstressable, they are invisible to stress: words of the form CVCVCV(C) receive antepenultimate stress
unless the final vowel is epenthetic, in which case the penult is stressed, acting
as if the epenthetic vowel were not there, as in: [ka’tabit] ‘I/you (masc sg)
wrote’. Second, Head-Dep(O/I) fails to relate the opacity of stress to other wordlevel prosodic processes, thus missing the generalization “that all processes of
word phonology ignore epenthetic vowels” (Kiparsky 2000: 353). For example,
word-level closed CVVC syllables are shortened even though they are opened
by postlexical epenthesis:
(55)
/œaf-it/
/œaf-t/
>
>
œafit
’œifit (*œafit)
‘she saw’
‘I saw’
Brame’s insight (§3.2.1) that syncope is ordered before epenthesis in dialects such
as Levantine is captured in stratal OT by allowing different constraint rankings
in the lexical and postlexical strata (Kiparsky 2000, 2003). The relevant constraints here are the faithfulness constraint Max-’V, requiring the stressed vowel
of the input to have a correspondent in the output, and the markedness constraints
Reduce, which minimizes the number of non-final light syllables, and Licenseµ, which requires all moras to be licensed by syllables. Syncope takes place at
the word level because Reduce outranks License-µ. At the postlexical level,
epenthesis is prompted by the promotion of License-µ, and Max-’V rules out
candidates in which stress is assigned to the epenthetic vowel. Kiparsky’s stratal
OT azxznalysis of /yiktib-u/ > [’yikitbu] is given in (56) and (57):
(56)
VC dialects: Word level
Input: [’(yik).(ti.bu)] Reduce License-[ …
☞ a. ’(yik).t[.bu
b. ’(yik).(ti.bu)
13
*
*
**!
Broselow (2008), for example, argues that the invisibility of epenthetic vowels in Iraqi Arabic is
motivated by maximization of contrast between stems of different grammatical types.
TBC_135.qxd
7/13/10
Table 135.1 The main stress and stress-related typological characteristics of four Arabic dialects
Palestinian
Cyrenaican Bedouin
Page 24
San’ani
19:27
Cairene
Syllable
structure
CVVC]PW
CVCC]UTT
CVV
CVV: *word-final
CVVC]PW
CVVC: In morphologically
complex environments
CVCC]PW
CVVC]PW
CVVC: In morphologically
complex environments
CVCC]PW
CVV: *word-final
CVVC]PW
CVVC: In morphologically
complex environments
CVCC]PW
Stress
Foot type: Moraic trochees
Direction: L > R
Extrametricality:
<C>: Yes
<Foot>: No
Degenerate feet: No
Weak parsing: No
Stress shift: No
Exceptions: Yes
Foot type: Moraic trochees
Direction: L > R
Extrametricality:
<C>: Yes
<Foot>: Yes
Degenerate feet: Yes
Weak parsing: No
Stress shift: Yes
Exceptions: No
Foot type: Moraic trochees
Direction: L > R
Extrametricality:
<C>: Yes
<Foot>: Yes
Degenerate feet: No
Weak parsing: No
Stress shift: No
Exceptions: No
Foot type: Iambs
Direction: L > R
Extrametricality:
<C>: No
<Foot>: Yes
Degenerate feet: No
Weak parsing: No
Stress shift: Yes
Exceptions: No
Stress-related
phenomena
Syncope: High vowels
Unstressed long vowel
reduction: Yes
CSS: Yes
Opaque stress: Due to
unstressed long vowel
reduction
Syncope: All vowels
Unstressed long vowel
reduction: No
CSS: No
Opaque stress: No
Syncope: High vowels
Unstressed long vowel
reduction: To left of
stressed syllable only
CSS: No
Opaque stress: Due to
syncope and/or epenthesis
Syncope: All vowels
Unstressed long vowel
reduction: No
CSS: No
Opaque stress: Due to
syncope and/or
epenthesis
TBC_135.qxd
7/13/10
19:27
Page 25
Word Stress in Arabic
(57)
25
VC dialects: Post-lexical level
Input: [’yik.t[.bu] License-[ Max-’V Reduce
a. ’yik.t[.bu
b. yi.’kit.bu
☞ c. ’yi.kit.bu
*
*!
*!
**
**
The main stress and stress-related typological characteristics of Cairene, San’ani,
and Palestinian are summarized in Table 135.1. A column is added for
Cyrenaican Bedouin, a dialect analyzed as exhibiting iambic stress. The syllable
structure row shows the distribution of CVV, CVVC, and CVCC syllables in the
dialects: ]pw = restricted to phonological word-final, ]utt = restricted to utterancefinal position, and ’CVV = only when stressed. The stress row summarizes foot
type, direction of foot construction, extrametrical elements (<C> consonant,
<Foot> foot), the permissibility of degenerate feet or weak parsing, and the
occurrence of stress shift or morphologically induced exceptions. The final row
summarizes for each dialect stress-related phenomenon, showing syncope
restrictions, and the occurrence or non-occurrence of long vowel reduction,
closed syllable shortening (CSS) and opaque stress.
4
Conclusion
This chapter has examined firstly how Arabic has contributed to the development
of metrical theory, with its particular contributions in the areas of rhythmic
organization, stress shift, right-edge effects, and the interaction of syncope,
epenthesis, and stress. Second, it has provided sketches of the stress systems
of four Arabic varieties – Classical Arabic, Cairene, San’ani, and Levantine –
focusing on core similarities and differences between dialects, and key
approaches invoked to deal with these. Due to space restrictions I have focused
on a selection of phenomena and data. I have not examined, for instance, notions
of secondary stress – attested in San’ani, for instance, but not in Cairene – stress
conflation, and the more drastic alterations in morpheme shape exhibited by iambic
dialects such as Cyrenaican Bedouin.
REFERENCES
Abdo, Daud A. 1969. On stress and Arabic phonology: A generative approach. Beirut: Khayats.
Abul-Fadl, Fahmi. 1961. Volkstümliche Texte in arabischen Bauerndialekten der ägyptischen Provinz
0arqiyya. Münster: Westfälische Wilhelms-Universität.
Abu-Salim, Isaam M. 1980. Epenthesis and geminate consonants in Palestinian Arabic. Studies
in the Linguistic Sciences 10. 1–12.
Al-Mohanna, Faisal 2004. Paradoxical non-finality: Stress assignment in three Arabic
dialects. Ms., King Saud University. (ROA 735.)
Al-Mozainy, Hamza Qublan, Robert Bley-Vroman & John J. McCarthy. 1985. Stress shift
and metrical structure. Linguistic Inquiry 16. 135–144.
Angoujard, Jean-Pierre. 1981. Contribution à l’analyse prosodique (parlers de Tunis,
du Caire et de Damas). Analyses, Théorie 1. 66–121.
TBC_135.qxd
7/13/10
26
19:27
Page 26
Janet C. E. Watson
Angoujard, Jean-Pierre. 1990. Metrical structure of Arabic. Dordrecht: Foris.
Angoujard, Jean-Pierre. 1992. Syncope et voyelles accentuées (parler du Hedjaz, Bani
Hassan et hébreu biblique). Languages Orientales Anciennes, Philologie et Linguistique 4.
299 –304.
Aoun, Youssef. 1979. Is the syllable or the supersyllable a constituent? MIT Working Papers
in Linguistics 1. 140 –148.
Bauer, Leonhard. 1926. Das palästinische Arabisch. Leipzig: J. C. Hinrichs’sche Buchhandlung.
Behnstedt, Peter & Manfred Woidich. 1985. Die ägyptisch-arabischen Dialekte. Wiesbaden:
Ludwig Riechert.
Birkeland, Harris. 1954. Stress patterns in Arabic. Oslo: Jacob Dybwad.
Blanc, Haim. 1970. The Arabic dialect of the Negev Bedouins. Proceedings of the Israeli Academy
of Science and Humanities 4. 112–150.
Bohas, Georges. 1975. Métrique arabe classique et moderne. Thèse de 3ème cycle. Paris VIII.
Bohas, Georges & Djamel Kouloughli. 1981. Processus accentuels en arabe (parlers du Caire,
de Damas et arabe classique). Analyses, Théorie 1. 36–46.
Brame, Michael K. 1970. Arabic phonology: Implications for phonological theory and historical Semitic. Cambridge, MA: MIT dissertation.
Brame, Michael K. 1973. On stress in two Arabic dialects. In Stephen R. Anderson & Paul
Kiparsky (eds.) A Festschrift for Morris Halle. New York: Holt, Rinehart & Winston. 14–25.
Brame, Michael K. 1974. The cycle in phonology: Stress in Palestinian, Maltese and
Spanish. Linguistic Inquiry 5. 39–60.
Brockelmann, Carl. 1907. Grundriss der vergleichenden Grammatik der semitischen Sprachen.
Vol. 1. Hildesheim: Georg Olm.
Broselow, Ellen. 1976. The phonology of Egyptian Arabic. Amherst: University of
Massachusetts dissertation.
Broselow, Ellen. 1992. Parametric variation in Arabic dialect phonology. In Mushira
Eid & John J. McCarthy (eds.) Perspectives on Arabic linguistics IV. Amsterdam &
Philadelphia: John Benjamins. 7– 47.
Broselow, Ellen. 2008. Stress–epenthesis interactions. In Bert Vaux & Andrew Nevins
(eds.) Rules, constraints, and phonological phenomena. Oxford: Oxford University Press.
121–148.
Broselow, Ellen, Maria Huffman, Su-I Chen & Ruohmei Hsieh. 1995. The timing structure
of CVVC syllables. In Mushira Eid (ed.) Perspectives on Arabic linguistics VII. Amsterdam
& Philadelphia: John Benjamins. 119 –138.
Broselow, Ellen, Su-I Chen & Marie Huffman. 1997. Syllable weight: Convergence of
phonology and phonetics. Phonology 14. 47–82.
Burzio, Luigi. 1994. Principles of English stress. Cambridge: Cambridge University Press.
Cantineau, Jean. 1939. Remarques sur les parlers de sédentaires syro-libano-palesintinens.
Bulletin de la Société de Linguistique de Paris 40. 80–88.
Cantineau, Jean. 1946. Les parlers arabes du Hôrân. Paris: C. Klincksieck.
Cantineau, Jean. 1960. Etudes de linguistique arabe. Paris: C. Klincksieck.
Chomsky, Noam & Morris Halle. 1968. The sound pattern of English. New York: Harper
& Row.
Cowell, Mark W. 1964. A reference grammar of Syrian Arabic (based on the dialect of
Damascus). Washington, DC: Georgetown University Press.
El-Hajjé, Hassan. 1954. Le parler arabe de Tripoli (Liban). Paris: C. Klincksieck.
Erpenius, Thomas. 1656. Grammatica Arabica. Leiden.
Erwin, Wallace M. 1963. A short reference grammar of Iraqi Arabic. Washington, DC:
Georgetown University Press.
Feghali, Michel. 1919. Le parler de Kfar-0abîda (Liban–Syrie). Paris: C. Klincksieck.
Ferguson, Charles A. 1956. Review of Harris Birkeland, Stress patterns in Arabic. Language
32. 384 –387.
Fleisch, Henri. 1974. Etudes d’arabe dialectal. Beirut: Dar el-Machreq.
TBC_135.qxd
7/13/10
19:27
Page 27
Word Stress in Arabic
27
Fudge, Erik C. 1969. Syllables. Journal of Linguistics 5. 253–286.
Garbell, Irene. 1958. Remarks on the historical phonology of an eastern Mediterranean dialect.
Word 14. 303 –337.
Goitein, Shelamo D. 1970. Jemenica: Sprichwörter und Redensarten aus Zentral-Jemen mit
zahlreichen Sach- und Worterläutungen. Leiden: Brill.
Grotzfeld, Heinz. 1965. Syrisch-arabische Grammatik (Dialekt von Damaskus). Wiesbaden:
Harrassowitz.
Halle, Morris & Michael Kenstowicz. 1991. The Free Element Condition and cyclic versus
noncyclic stress. Linguistic Inquiry 22. 457–501.
Halle, Morris & Jean-Roger Vergnaud. 1978. Metrical structures in phonology. Unpublished
ms., MIT.
Halle, Morris & Jean-Roger Vergnaud. 1979. Metrical phonology. Unpublished ms., MIT.
Halle, Morris & Jean-Roger Vergnaud. 1987. An essay on stress. Cambridge, MA: MIT Press.
Harrell, Richard. 1957. The phonology of colloquial Egyptian Arabic. New York: American Council
of Learned Societies.
Harrell, Richard. 1960. A linguistic analysis of Egyptian radio Arabic. In Charles A. Ferguson
(ed.) Contributions to Arabic Linguistics. Cambridge MA: Harvard University Press.
Harris, John & Edmund Gussmann. 2002. Word-final onsets. UCL Working Papers in
Linguistics 14. 1– 42.
Hayes, Bruce. 1979. Extrametricality. MIT Working Papers in Linguistics 1. 77–86.
Hayes, Bruce. 1982. Extrametricality and English stress. Linguistic Inquiry 13. 227–276.
Hayes, Bruce. 1989. Compensatory lengthening in moraic phonology. Linguistic Inquiry 20.
253 –306.
Hayes, Bruce. 1995. Metrical stress theory: Principles and case studies. Chicago: University of
Chicago Press.
Hyde, Brett. 2006. Towards and uniform account of prominence-driven stress. In Eric Bakovic,
Junko Itô & John McCarthy (eds.) Wondering at the natural fecundity of things: essays in
honor of Alan Prince, 139–183. Linguistics Research Center, University of California.
[http://repositories.cdlib.org/lrc/prince/4]
Irshied, Omar & Michael Kenstowicz. 1984. Some phonological rules of Bani-Hassan
Arabic: A Bedouin dialect. Studies in the Linguistic Sciences 14. 109–147.
Jakobson, Roman. 1971. Selected writings. Vol. 1: Phonological studies. The Hague: Mouton.
Janssens, Gerard. 1972. Stress in Arabic and word structure in the modern Arabic dialects. Leuven:
Peeters.
Jastrow, Otto. 2007. Iraq. In Kees Versteegh, Mushira Eid, Alaa Elgibali, Manfred Woidich
& Andrzej Zaborski (eds.) The encyclopedia of Arabic language and linguistics. Vol. 3,
414– 424. Leiden: Brill.
Jiha, Michel. 1964. Der arabische Dialekt von Bi:mizzcn. Beirut: Franz Steiner.
Johnson, C. Douglas. 1979. Opaque stress in Palestinian. Lingua 49. 153–168.
Kager, René. 1995. The metrical theory of word stress. In John A. Goldsmith (ed.) The handbook of phonological theory, 365 –402. Cambridge, MA & Oxford: Blackwell.
Kager, René. 1999. Optimality Theory. Cambridge: Cambridge University Press.
Kager, René. 2009. Stress. In Kees Versteegh et al. (eds.) The encyclopedia of Arabic language
and linguistics. Vol. 4, 344 –353. Leiden: Brill.
Kenstowicz, Michael. 1980. Notes on Cairene Arabic syncope. Studies in the Linguistic Sciences
10. 39 –54.
Kenstowicz, Michael & Kamal Abdul-Karim. 1980. Cyclic stress in Levantine Arabic.
Studies in the Linguistic Sciences 10. 55–76.
Kiparsky, Paul. 1979. Metrical structure assignment is cyclic. Linguistic Inquiry 10.
421– 441.
Kiparsky, Paul. 1982. Lexical morphology and phonology. In Linguistic Society of Korea
(ed.) Linguistics in the Morning Calm, 3–91. Seoul: Hanshin.
Kiparsky, Paul. 2000. Opacity and cyclity. The Linguistic Review 17. 351–365.
TBC_135.qxd
7/13/10
28
19:27
Page 28
Janet C. E. Watson
Kiparsky, Paul. 2003. Syllables and moras in Arabic. In Caroline Féry & Ruben van de Vijver
(eds.) The syllable in Optimality Theory, 147–182. Cambridge: Cambridge University Press.
Langendoen, D. Terence. 1968. The London School of Linguistics: A study of the theories of
B. Malinowski and J. R. Firth. Cambridge, MA: MIT Press.
Liberman, Mark. 1975. The intonational system of English. Ph.D. dissertation, MIT.
Liberman, Mark & Alan Prince. 1977. On stress and linguistic rhythm. Linguistic Inquiry 8.
249 –336.
McCarthy, John J. 1979. On stress and syllabification. Linguistic Inquiry 10. 443–465.
McCarthy, John J. 1980. A note on the accentuation of Damascene Arabic. Studies in the
Linguistic Sciences 10. 77– 98.
McCarthy, John J. 2003. Sympathy, cumulativity, and the Duke-of-York gambit. In
Caroline Féry & Ruben van de Vijver (eds.) The syllable in Optimality Theory, 23–76.
Cambridge: Cambridge University Press.
McCarthy, John J. & Alan Prince. 1990. Foot and word in prosodic morphology: The Arabic
broken plural. Natural Language and Linguistic Theory 8. 209–283.
Mitchell, Terence F. 1952. The active participle in an Arabic dialect of Cyrenaica. Bulletin
of the School of Oriental and African Studies 14. 11–33.
Mitchell, Terence F. 1956. An introduction to colloquial Egyptian Arabic. London: Oxford
University Press.
Mitchell, Terence F. 1960. Prominence and syllabification in Arabic. Bulletin of the School of
Oriental and African Studies 23. 369–389.
Mitchell, Terence F. 1975. Principles of Firthian linguistics. London: Longman.
Mitchell, Terence F. 1993. Pronouncing Arabic 2. Oxford: Oxford University Press.
Naïm-Sanbar, Samia. 1986. Le parler arabe de Ras-Beyrouth: ‘Ayn al-Muraysa. Paris: Geuthner.
Naïm-Sanbar, Samia. 1994. Contribution à l’étude de l’accent yéménite: Le parler des femmes
de l’ancienne génération. Zeitschrift für Arabische Linguistik 27. 67–89.
Paoli, Bruno. 2008. Phonology: metrical. In Kees Versteegh et al. (eds.) Encyclopedia of Arabic
language and linguistics. Vol. 3, 615 –623. Leiden: Brill.
Prince, Alan. 1976. “Applying” stress. Unpublished ms., University of Massachusetts.
Prince, Alan. 1983. Relating to the grid. Linguistic Inquiry 14. 19–100.
Rossi, Ettore. 1939. L’arabo parlato a fian#a. Romea: Istituto per l’Oriente.
Selkirk, Elisabeth. 1980. The role of prosodic categories in English word stress. Linguistic
Inquiry 11. 563 – 605.
Selkirk, Elisabeth. 1981. Epenthesis and degenerate syllables in Cairene Arabic. MIT
Working Papers in Linguistics 3. 111–140.
Selkirk, Elisabeth. 1982. The syllable. In Harry van der Hulst & Norval Smith (eds.) The
structure of phonological representations. Vol. 2, 337–383. Dordrecht: Foris Publications.
Selkirk, Elisabeth. 1984. Phonology and syntax: The relation between sound and structure.
Cambridge MA: MIT Press.
Teeple, David. 2009. Syncope. In Kees Versteegh et al. (eds.) Encyclopedia of Arabic language
and linguistics. Vol. 4, 389 –391. Leiden: Brill.
Tomiche, Nadia. 1964. Le parler arabe du Caire. The Hague: Mouton.
Watson, Janet C. E. 1999. CVVC syllables in Arabic. Languages and Linguistics 4. 1–17.
Watson, Janet C. E. 2002. The phonology and morphology of Arabic. Oxford: Oxford University
Press.
Weil, G. 1954. Das metrische System des al-Xalcl und der Iktus in den altarabischen
Versen. Oriens 7. 304 –321.
Weldon, Ann. 1980. Stress in Cairo Arabic. Studies in the Linguistic Sciences 10. 99–120.
Woidich, Manfred. 2006. Das Kairenisch-Arabisch: Grammatik. Wiesbaden: Harrassowitz.
Wright, William. 1971. A grammar of the Arabic language. Cambridge: Cambridge University
Press.
Younes, Munther. 1995. On vowel shortening in Palestinian Arabic. In Mushira Eid (ed.)
Perspectives on Arabic linguistics VII. 157–172. Amsterdam & Philadelphia: John
Benjamins.