The Linguistic Review 2023; 40(1): 43–75

Radwa Fathi*
Ablaut in Cairene Arabic
https://doi.org/10.1515/tlr-2022-2105
Published online December 16, 2022

Abstract: The vowel alternations relating the perfectives and imperfectives of

Form I verbs in Cairene Arabic are examined in this article. It is argued that such
alternations are driven by a coherent system of Ablaut (or apophony), the same
system as was proposed by Guerssel and Lowenstamm for Classical Arabic. Mounting
evidence from unrelated languages strongly suggests that Ablaut systems are
organized in similarly rigid fashion cross-linguistically. The question of the impact of
linguistic change on Ablaut thus arises. One possibility is that change may culminate
in the abandonment of apophony. Another possibility is that apophony itself drives
change. It is argued in this article that several past and ongoing developments in the
evolution of the vocalization of verbs in Cairene took place under the control of the
apophonic system.

Keywords: Ablaut; apophony; Cairene Arabic; Classical Arabic; diachronic change;

Form I verbs

1 Introduction
This article is devoted to the study of a set of vowel alternations in the verbal system
of Cairene Arabic. Following Cantineau (1950) and McCarthy (1979), familiarity with
which I assume, I accept the idea that Semitic verbs arise from the combination of
three ingredients: a consonantal root, a vocalic melody, and an array of prosodic
patterns. The Cairene Arabic verbal system is a typical example of the gamut of verb
types produced by the combination of these ingredients, and its 10 types will be
illustrated in due course. The topic of this article concerns one only of these verb
types, the type sometimes called “basic”, or “Form I”, or “Measure I”. In Form I, the
perfective/imperfective aspectual alternation is accompanied, as we will see in
detail, by a vowel change which varies from one verb to the other, e.g., silim/yi-slam

*Corresponding author: Radwa Fathi, Laboratoire de Linguistique de Nantes, UMR 6310, Nantes,
France, E-mail: radwa06@gmail.com
44 Fathi

‘be safe’ versus ṭalab/yi-ṭlub ‘ask’ versus ʁasal/yi-ʁsil ‘wash’.1 Earlier discussions of
the vocalization of Form I verbs in Arabic include Cantineau (1949), Fleisch (1954),
Brame (1970), Kuryłowicz (1972).
On the basis of a corpus of more than twelve hundred verbs of Cairene Arabic,2 I
will argue, against the general consensus, that the perfective/imperfective vowel
changes are controlled by a coherent apophonic mechanism. This mechanism makes
it possible to derive the vocalization of imperfectives from that of their respective
perfectives. I will first identify the underlying vocalic makeup of Cairene Arabic
perfectives (the input to apophony).3 I will then show that the apophonic mechanism
actioning those Cairene inputs is exactly the same as the one proposed by Guerssel
and Lowenstamm (1996) for Classical Arabic.
As well, I intend to show, for the first time I believe, how that theory can
be harnessed to explain linguistic change, past or ongoing. Consider for instance
well-known pairs of verbs such as tiʕib ‘tire, intransitive’ versus taʕab ‘tire, transi-
tive’, both from the same root √tʕb. Here, the differential vocalization of the verbs is
the vehicle of a difference in argument structure. Such cases will not be discussed in
this paper. But next to pairs of that type, numerous doublets are attested, e.g., nigiħ ∼
nagaħ both from the same root √ngħ ‘succeed’, this time with no difference in
meaning.4 Because such doublets cannot be traced back to older forms of Arabic, I
assume that they are Egyptian developments. I will show that they are not randomly
distributed. Rather, pockets of doublets arose in very specific contexts and, I will
argue, as a direct consequence of apophonic pressure on the vocalization system.
While I do not assume that Cairene Arabic is a direct descendant of Classical
Arabic, much of the discussion of the former will be conducted against the back-
ground of the latter.5 The reasons for this are: a) the two forms of Arabic are

1 I transcribe verbs such as silim ‘he became safe’ and kubur ‘he grew up’, etc. with high vowels i and
u. This is a phonological notation. Those verbs are actually pronounced with mid-vowels e and o,
hence [selem] and [kobor], etc. Similarly, imperfectives such as yi-ṭlub ‘he asks’ or yi-ʁsil ‘he washes’,
are pronounced [ye-ṭlob] and [ye-ʁsel], etc. See Fathi (2013) for discussion.
2 My corpus results from an exhaustive search of Badawi and Hinds (1986) supplemented with data
known to me as a native speaker or supplied by my informants. In the course of the past 10 years, I
have interviewed dozens of speakers of Cairene Arabic. Interviews were conducted in Cairo with
speakers of both sexes (aged between 4 and 74) and different social classes though primarily from the
middle class. For their patience and cooperation, I am especially indebted to Ahmed Abdelkader,
Kamelia Abdelkader, Amal Amir, Marwa El Maimouny, Marwa Mahmoud Saleh and Nadia Ahmed
Saleh.
3 In this article, I use ablaut and apophony interchangeably.
4 All doublets mentioned in this article are attested in Badawi and Hinds (1986). Their existence was
confirmed by my consultants.
5 For the complexity of the formation of the Egyptian Arabic dialect area, see the eponymous article
by Behnstedt and Woidich (2018), and Woidich (1994) on Cairene Arabic specifically.
 Ablaut in Cairene Arabic 45

obviously related genetically, and close enough to warrant parametric comparison,

b) the anteriority of one over the other provides diachronic perspective, c) much
more altogether is known about Classical Arabic than about Cairene Arabic, and d)
the theory of apophony I will argue for was developed on the basis of Classical Arabic
evidence.
The discussion is organized in 9 sections. In Section 2, I introduce and compare
relevant aspects of Classical Arabic and Cairene Arabic evidence. In Section 3, I lay
out the basics of Guerssel and Lowenstamm’s theory of apophony. Section 4 is a
preliminary assessment of the facts of Cairene Arabic in that light. In Section 5, I offer
an analysis of the organization of Cairene Arabic perfectives. In Sections 6 and 7,
I demonstrate the operation of apophony in the Cairene dialect. In Section 8, I
speculate about the future evolution of the Cairene system. Section 9 contains several
remarks on the theory of apophony. In the conclusion, I sum up my results.

2 Classical Arabic and Cairene Arabic

Classical Arabic Measure I verbs are classified into four major classes according to
the vocalization of their perfective and imperfective.6 The first class involves
perfective FaʕaL which alternates with imperfective ya-FʕiL,7 e.g., ʁasal/ya-ʁsil. The
second involves again perfective FaʕaL which alternates this time with imperfective
ya-FʕuL, e.g., ṭalab/ya-ṭlub. The third class involves perfective FaʕiL which alternates
with imperfective ya-FʕaL, e.g., salim/ya-slam. The fourth and last class involves
perfective FaʕuL which alternates with imperfective ya-FʕuL, e.g., kabur/ya-kbur.
These four classes are displayed in (1).8 The alternations just observed are known as
“Ablaut”.

(1) Perf./Imperf. Perf. Imperf. Root ‘gloss’

i. FaʕaL/ya-FʕiL ʁasal ya-ʁsil √ʁsl ‘wash’
ii. FaʕaL/ya-FʕuL ṭalab ya-ṭlub √ṭlb ‘ask’
iii. FaʕiL/ya-FʕaL salim ya-slam √slm ‘be safe’
iv. FaʕuL/ya-FʕuL kabur ya-kbur √kbr ‘grow up’

6 All verbs discussed in this article are verbs built from triliteral ‘strong’ roots, i.e., roots involving no
glides or adjacent identical consonants.
7 Following the practice of the medieval Arab grammarians, I use F-ʕ-L as the prototypical root. Thus,
ʁasal and ṭalab are token verbs. FaʕaL denotes the type of which ʁasal and ṭalab are tokens. All
examples are 3rd person masculine singular. Perfectives are quoted as FvʕvL and imperfectives as
ya-FʕvL where the prefix marks 3rd person masculine singular.
8 A fifth class exists, FaʕaL/ya-FʕaL. Unlike the four classes mentioned above, the alternating
pattern FaʕaL/ya-FʕaL is conditioned by the consonantal environment. Its discussion is taken up in
Section 6.3.
46 Fathi

Ablaut is also present in the Cairene dialect. But Cairene Arabic displays six classes of
alternations and not only four as in Classical Arabic. This is illustrated in (2). Unlike
Classical Arabic which exhibits three types of perfectives: FaʕaL, FaʕiL and FaʕuL,
Cairene Arabic is reputed to exhibit only two: FaʕaL and FiʕiL.9 Each of these two
perfectives, FaʕaL and FiʕiL, may alternate with three imperfective patterns, yi-FʕaL,
yi-FʕiL and yi-FʕuL.10
Thus, Cairene perfectives kasar, ʁasal and ṭalab, all sharing the same pattern
FaʕaL, alternate with yi-FʕaL, yi-FʕiL and yi-FʕuL, hence yi-ksar, yi-ʁsil, and yi-ṭlub,
respectively. Likewise, perfectives silim, misik and ṣibir, all sharing the same pattern
FiʕiL, alternate with different imperfective melodies, yi-FʕaL, yi-FʕiL and yi-FʕuL,
hence yi-slam, yi-msik, and yi-ṣbur, respectively.

(2) Perf. Imperf. Root ‘gloss’ Perf. Imperf. Root ‘gloss’

i. kasar yi-ksar √ksr ‘break’ iv. silim yi-slam √slm ‘be safe’
ii. ʁasal yi-ʁsil √ʁsl ‘wash’ v. misik yi-msik √msk ‘take hold of’
iii. ṭalab yi-ṭlub √ṭlb ‘ask’ vi. ṣibir yi-ṣbur √ṣbr ‘act with patience’

The complexity of the facts just laid out has led most if not all investigators to
conclude that the alternating patterns in (2) are lexical, i.e., not rule-governed.
Mitchell, for instance, notes in his manual (Mitchell 1962: 76) “[…] the imperfective of
any verb should be learnt in conjunction with the perfective as it is met.” Abdel-
Massih’s manual (Abdel-Massih 1975: 94) also indicates that the vocalization of the
imperfective is not predictable from the perfective. While Broselow (1976: 139) sees
the vocalization of verbs as an arbitrary lexical fact, Gadalla (2000: 63–64) notes that
the stem vowel alternations are unsystematic. So does Abboud-Haggar (2003).
It is true that there is something forbidding in the data in (2). Because every
logical possibility seems to be attested, we have no clue even as to whether a
directional system relates the vowel of one aspect to the vowel of the other. Never-
theless, a specific reason leads me to strongly doubt that the general consensus is
justified. I will assume that Ablaut serves as the exponent of aspectual derivation. If
the derivation of aspect is systematic, so must be the realization of its exponent. On
this view, the data in (2) cannot just be an inert morass of forms. Rather, it must be
organized in the form of a directional system, the very same directional system
whereby aspect itself is derived. Guerssel and Lowenstamm (1990, 1996) developed
such a proposal for Classical Arabic. It is presented in the next section.

9 This is how the Cairene Arabic system is usually presented in the literature. Refinements will be
proposed shortly.
10 When the imperfective stem vowel is u, the prefix vowel can optionally be realized as i or u, hence
yi-ṭlub ‘he asks’ or yu-ṭlub, yi-xrug ‘he goes out’ or yu-xrug, etc.
 Ablaut in Cairene Arabic 47

3 The perfective/imperfective vowel alternations

as a vehicle of derivation of aspect in Classical
Arabic (Guerssel and Lowenstamm 1996)
Guerssel and Lowenstamm (1990, 1996) argue that alternations such as in (1) are “[…]
part of a genuine apophonic system serving in synchronically active fashion as the
vehicle of the derivation of aspect.” I briefly reproduce their proposal here.
In (3), we can see anew the four major classes of alternations. The first a vowel
of the perfective stem is an invariable ingredient in all the perfective forms, so is
the a vocalizing the prefix in the imperfective. These two ‘invariable’ components,
italicized in columns (a) and (b) of (3), are not involved in the derivation of aspect.11
When they are factored out, there remains the two series of vowel alternations in
(c) and (d) of (3), isolated in (4). The hypothesis is that these vowels are involved in the
derivation of aspect.

(3) a. Perf. b. Imperf. c. Perf. d. Imperf. e. Root ‘gloss’

i. ʁasal ya-ʁsil ʁ_sal y_- ʁsil √ʁsl ‘wash’
ii. ṭalab ya-ṭlub ṭ_lab y_-ṭlub √ṭlb ‘ask’
iii. salim ya-slam s_lim y_-slam √slm ‘be safe’
iv. kabur ya-kbur k_bur y_-kbur √kbr ‘grow up’

(4) a. Perfective b. Imperfective

i. a i
ii. a u
iii. i a
iv. u u

Guerssel and Lowenstamm (1996) identify two main analytical challenges that
hinder the establishment of the directionality of the derivation (Perfective →
Imperfective or Imperfective → Perfective): opacity and partial polarity.
First, regarding opacity, if the derivation goes from the perfective to the
imperfective, a perfective input a will have two different imperfective outputs, i and
u, (4.i) and (4.ii). This is illustrated in (5a). Conversely, if the derivation goes from the
imperfective to the perfective, an input u will give rise to two different perfective
outputs, a and u, (4.ii) and (4.iv). This is shown in (5b).

11 According to Guerssel and Lowenstamm (1990, 1996), both vowels signal ‘active voice’ in Classical
Arabic.
48 Fathi

(5) a. Derivation b. Derivation

Perfective i a u Imperfective i u a
   
Imperfective a i u Perfective a u i

Regarding partial polarity, as shown in (6), regardless of the direction of the

derivation, an input a has an output i while an input i has an output a; in either case:
a ↔ i.

(6)

Guerssel and Lowenstamm blame this vexing situation on a failure to detect a fourth
input object, they call it X. Suppose X is introduced in such way that it eliminates the
ambiguity inherent in (5a) and (5b). I illustrate this with the example involving /a/
in (5a). The ambiguity of a stems from the fact that an input a has two different
outputs, i and u. That ambiguity can be reduced in two ways: a → i and X → u or X → i
and a → u. The inclusion (in that manner) of X in the set of inputs in (5) gives rise to
the four possibilities in (7) – two for each version of directionality – all of which are
free from any ambiguity.
(7) a. Perf. i a X u c. Imperf. i u X a
         
Imperf. a i u u Perf. a a u i

b. Perf. i X a u d. Imperf. i X u a
         
Imperf. a i u u Perf. a a u i

When all four schemes in (7) are compared, a salient feature of (7b) distinguishes it
from the other three: unlike the other three, the boxed scheme in (7b) does not
involve the a ↔ i polarity noted above. For that reason, Guerssel and Lowenstamm
see it as the most promising of all four hypotheses in (7). Selecting (7b) carries two
consequences: a) the vowel of the imperfective is derived from the perfective;12 b)
the alternating pattern involving X is FaʕaL/ya-FʕiL. Concretely, the difference
between perfectives ṭalab and ʁasal is the presence of underlying /a/ in the former
and underlying /X/ in the latter, both neutralizing on the surface as [a]. Given that
difference, their respective roles in the apophonic scheme is as in (8).

12 This is also the view of the medieval Arabic grammarians (see Bohas and Guillaume 1984).
 Ablaut in Cairene Arabic 49

(8) a. ṭalab /ṭalab/ → ya-ṭlub b. ʁasal /ʁasXl/ → ya-ʁsil

ṭal/a/b ya-ṭlub ʁas/X/l ya-ʁsil

Guerssel and Lowenstamm proposed that /X/ is ø (zero). If so, the [a] vocalizing
the second root consonant in [ʁasal] can be viewed as a result of the propagation of
the a vocalizing the first root consonant. This is shown in (9a). This is because
vacuous vocalization of the second root consonant of a perfective – *ʁasl in the case
at hand – would be rejected by the grammar of Classical Arabic. ṭalab, on the other
hand is assumed to be equipped with lexical vowel /a/ which vocalizes its second root
consonant, (9b). The idea behind the view of X as “zero” is that unlike the perfectives
of the verb classes represented by ṭalab, salim and kabur which are assigned lexical
vowels {a, i, u} by the grammar, the perfectives of the class of ʁasøl is not assigned
any lexical vowel.

(9) a. /ʁasøl/ >> [ʁasal] b. /ṭalab/ >> [ṭalab]

ʁ s l ṭ l b

/a/ /ø/ /a/ /a/

The alternations given earlier in (3) and (4) can consequently be redefined as in (10)
where the input perfective vocalizations and the corresponding output vowels in the
imperfectives have been italicized.

(10) a. Perf. input b. Imperf. output c. Perf. → Imperf.

i. ʁasal /ʁasøl/ → ya-ʁsil ø→i
ii. ṭalab → ya-ṭlub a→u
iii. salim → ya-slam i→a
iv. kabur → ya-kbur u→u

With the minimal re-organization of the verbs in (10) given in (11), the full picture of
the proposal emerges. Guerssel and Lowenstamm (1996) highlight the striking formal
property of the individual vowel alternations in (11): each output vowel is the input to
another, (11c).

(11) a. Perf. input b. Imperf. output c. Perf. → Imperf.

i. ʁasal /ʁasØl/ → ya-ʁsil ø→i
ii. salim → ya-slam i→a
iii. ṭalab → ya-ṭlub a→u
iv. kabur → ya-kbur u→u

That is, the apophonic mechanism can be construed as the linear path in (12).
50 Fathi

(12) The apophonic chain: ø → i → a → u → u

Guerssel and Lowenstamm (1996) conclude that the vowel alternations attested
in Classical Arabic Measure I verbs are not lexically recorded as such. Rather, each
root is associated with one of the four “vocalizers” {ø, i, a, u}. This results in exactly
four types of perfectives. They note: “[…] Classes of vowel correspondences are a
mere by-product of the operation of the ablaut function.”
The descriptive potential of the apophonic chain in (12) is further explored in
Chekayri and Scheer (1996, 2003) for Classical Arabic, Faust (2017) for Palestinian
Arabic, Ségéral (1995; 1996) for Ge’ez, Ségéral (2000) for Semitic nouns in general. It
has also been tested against evidence from Berber, see Bendjaballah (1998, 1999, 2001,
2005) for Kabyle; Lahrouchi (2011), Lahrouchi and Ségéral (2009) for Tashelhyt,
Idrissi (2000) for Berber nouns in general. For extensions to German, see Ségéral
(1995), Ségéral and Scheer (1998). As well, see Bendjaballah (2003) for Beja, Baturay
Meral (2010) for Turkish, and Boyé (2014) for Malay.
The next section is devoted to a preliminary assessment of the Cairene Arabic
evidence in light of the apophonic mechanism in (12).

4 Towards an apophony-based account of the

‘aspectual’ alternations in Cairene Arabic
Let us examine once more the perfective/imperfective vowel alternations of Cairene
Arabic.

(13) Perf. Imperf. Root ‘gloss’ Perf. Imperf. Root ‘gloss’

i. kasar yi-ksar √ksr ‘break’ iv. silim yi-slam √slm ‘be safe’
ii. ʁasal yi-ʁsil √ʁsl ‘wash’ v. misik yi-msik √msk ‘take hold of’
iii. ṭalab yi-ṭlub √ṭlb ‘ask’ vi. ṣibir yi-ṣbur √ṣbr ‘act with patience’

If we recast the data in (13) in terms of the token root FʕL as in (14), it becomes clear
that the alternations therein are also plagued by partial polarity and even more
severely by opacity than their Classical Arabic counterparts.

(14) Perf. Imperf. Perf. Imperf.

FaʕaL yi-FʕaL FiʕiL yi-FʕaL
FaʕaL yi-FʕiL FiʕiL yi-FʕiL
FaʕaL yi-FʕuL FiʕiL yi-FʕuL

The system appears opaque as to whether the derivation goes from the perfective to
the imperfective or the other way round. In (15), I have represented two possibilities
 Ablaut in Cairene Arabic 51

regarding directionality. In (15a), the perfective is the base of the derivation and in
(15b) the imperfective serves as the base. In either case, the system is opaque: no
single output can be traced back to a unique base.

(15) a. b.
Perfective → Imperfective Imperfective → Perfective

Faʕal Fiʕil yi-Fʕal yi-Fʕil yi-FʕuL

yi-Fʕal yi-Fʕil yi-FʕuL Faʕal Fiʕil

Moreover, polarity plagues both schemes: under either version of directionality, it

will be the case that an i in the input corresponds to an a in the output while at the
same time an a in the input corresponds to an i in the output. In other words, at all
times, a ↔ i, see (16).

(16) a. b.
Perfective → Imperfective Imperfective → Perfective

Faʕal Fiʕil yi-Fʕal yi-Fʕil yi-FʕuL

yi-Fʕal yi-Fʕil yi-FʕuL Faʕal Fiʕil

As we see, Cairene Arabic offers the same two challenges as Classical Arabic, namely
opacity and partial polarity. Moreover, the terms of polarity are exactly the same in
the two languages, viz. a and i. Consequently, I pursue the idea that the apophonic
mechanism in (12) can shed light on Cairene Arabic, too.
For better visibility, let me first factor out the imperfective prefix yi-, invariable
in all forms, only retaining the imperfective vowel and one only of the two identical
vowels of the perfective. This is done in (17).

(17) Perfective Imperfective Perfective Imperfective

a a i a
a i i i
a u i u

Under the assumption that the vowel of the perfective is derived from that of the
imperfective, one alternation appears to conform with the apophonic mechanism
established in (12): the boxed cell in (18) manifests the (i → a) step.
52 Fathi

(18) Perfective Imperfective Perfective Imperfective

a a i a
a  i i i
a u i u

On the alternative view that the perfective melody is mapped into the imperfective via
apophony, both boxed cells in (19) comply with the apophonic mechanism, specif-
ically with steps (i → a) and (a → u).

(19) Perfective Imperf. stem Perfective Imperfective

a a i  a
a i i i
a  u i u

Comparing the results of (18) and (19), I decide to pursue the hypothesis that it is the
imperfective melody that is derived from that of the perfective by apophony.
In (20), I illustrate (19) with examples of actual verbs, one for each case. 306 verbs
pattern like ṭalab → yi-ṭlub, thus exemplifying the (a → u) apophonic step. 177 verbs
exemplify the (i → a) step, patterning as silim → yi-slam.

(20) Perf. Imperf. No. of verbs Perf. Imperf. No. of verbs

i. kasar yi-ksar iv. silim yi-slam 177
ii. ʁasal yi-ʁsil v. misik yi-msik
iii. ṭalab yi-ṭlub 306 vi. ṣibir yi-ṣbur

This preliminary result is highly encouraging for two reasons. First, with two
perfective melodies, the a of FaʕaL and the i of FiʕiL, the presence of two matching
imperfective types, yi-FʕuL and yi-FʕaL as their respective outputs is exactly what the
apophonic scheme predicts. Second, the two classes boxed in (20) are well populated:
together, they amount to a big third of all verbs in my corpus.13
But, by the same token this encouraging result begs a question. With just two
perfective melodies – FaʕaL and FiʕiL – as sole inputs to apophony, two classes of
alternations only are expected, viz. FaʕaL/yi-FʕuL and FiʕiL/yi-FʕaL. In other words,
the verb types exemplified in (20) should pattern uniformly (but counterfactually) as

13 A reader points out that one third (two classes out of six) is hardly surprising if verbs are allotted
more or less evenly to the different classes. But, we have no a priori reason to expect an even
distribution. In Classical Arabic, to give just one example, the FaʕuL/ya-FʕuL class is very small in
comparison with other classes. Even at this early stage of my demonstration (that is, when not all
classes have been identified yet), the careful reader can note that the class in (20.iii) is almost twice as
large as the class in (20.iv).
 Ablaut in Cairene Arabic 53

in (21). The question is why does the actual system display a significantly wider range
of alternations?

(21) Perf. Imperf. Perf. Imperf.

i. kasar *yi-ksur iv. silim yi-slam
ii. ʁasal *yi-ʁsul v. misik *yi-msak
iii. ṭalab yi-ṭlub vi. ṣibir *yi-ṣbar

What is needed at this point is a richer set of perfective types, one which will make it
possible to account for the attested range of alternations. Identifying the fine
structure of perfective verbs is therefore necessary. This is one of the topics taken up
in the next section.

5 The ingredients of the Cairene Arabic perfective

and their deployment
A comparison between the perfectives of Cairene and Classical Arabic (22) brings to
light two clear differences. The first element of difference is the supposed absence in
Cairene Arabic of an u-Perfective corresponding to the FaʕuL type of Classical Arabic,
e.g., kabur in (22.iv).

(22) a. Classical. b. Cairene c. Classical d. Cairene Root ‘gloss’

Ar. Ar. Ar. Ar.
i. ʁasal ʁasal FaʕaL FaʕaL √ʁsl ‘wash’
ii. ṭalab ṭalab FaʕaL FaʕaL √ṭlb ‘ask’
iii. salim silim FaʕiL FiʕiL √slm ‘be safe’
iv. kabur kibir FaʕuL FiʕiL √kbr ‘grow up’

The second element of difference has to do with the i-Perfective: both languages do
have an i-Perfective but whereas the i is confined to the rightmost vocalic position of
the stem in Classical Arabic salim, it occupies both positions in Cairene silim.

5.1 The u-perfectives

In Section 2, I hinted at the fact that the presentation of the Cairene Arabic evidence
as given so far would be in need of refinement. Indeed, Cairene Arabic in reality
possesses an u-Perfective. Its existence often remains unacknowledged because
where an u-Perfective is possible, it stands in free variation with an i-Perfective.
Badawi and Hinds (1986) mention 51 FuʕuL verbs, 48 of which are in free variation
54 Fathi

with FiʕiL. Even though the FuʕuL variants are much less frequently used than their
FiʕiL counterparts, they can still be used without causing surprise or perplexity.14
Examples are given in (23), and their existence is recorded in (24). Section 7 below is
entirely devoted to a discussion of such doublets and how they arose.

(23) kubur ∼ kibir ‘grow up; increase in size/age, √kbr’

ʁulub ∼ ʁilib ‘suffer inconvenience or hardship, √ʁlb’
ʁuluṭ ∼ ʁiliṭ ‘err; make a mistake, √ʁlṭ’
xuluṣ ∼ xiliṣ ‘come to an end; be finished, √xlṣ’
suxun ∼ sixin ‘become hot, √sxn’
kutur ∼ kitir ‘become numerous; proliferate, √ktr’

(24) a. b. Root ‘gloss’

i. ʁasal FaʕaL √ʁsl ‘wash’
ii. ṭalab FaʕaL √ṭlb ‘ask’
iii. silim FiʕiL √slm ‘be safe’
iv. kibir ∼ kubur FiʕiL ∼ FuʕuL √kbr ‘grow up’

When Cairene FuʕuL perfectives are acknowledged, it becomes clear that i, a, and
u too are involved in the vocalization of the perfectives of Cairene and Classical
Arabic, (25).

(25) a. Cl. Ar. b. Cai. Ar. c. Cl. Ar. d. Cai. Ar. Root ‘gloss’
i. ʁasal ʁasal FaʕaL FaʕaL √ʁsl ‘wash’
ii. ṭalab ṭalab FaʕaL FaʕaL √ṭlb ‘ask’
iii. salim silim FaʕiL FiʕiL √slm ‘be safe’
iv. kabur kubur ∼ kibir FaʕuL FuʕuL ∼ FiʕiL √kbr ‘grow up’

Extrapolating from that observation, I formulate the strong claim in (26).15

(26) Cairene Arabic and Classical Arabic Measure I perfectives manage the same
vocalic material.

14 The tolerant attitude I am describing with respect to kubur versus kibir and the like can be
compared to the tolerance surrounding the use of alternative (and unquestionably grammatical) past
verbal forms in English, e.g., dreamed ∼ dreamt, learned ∼ learnt, etc. Similarly, French nouns are
normally Masculine or Feminine. Yet, the nouns of a group of modest size can be assigned to either
gender: un après-midi or une après-midi ‘afternoon’, un clop or une clop ‘cigarette’, etc. cf. Fathi and
Lowenstamm (2016) for discussion.
15 The claim here is that the range of possible vocalizations for perfective verbs is the same in
Classical Arabic and in Cairene Arabic. This does not mean that any single Cairene Arabic verb is
necessarily vocalized exactly as the Classical Arabic verb from the same root. Compare for instance
Classical Arabic ṭalaʕ ‘climb’ versus Cairene Arabic ṭiliʕ.
 Ablaut in Cairene Arabic 55

The differential arrangement of the same ingredients in the two languages remains
to be accounted for: why does Cairene Arabic have silim and kubur with two identical
vowels where Classical Arabic has salim and kabur with both a and then the lexical
vowel?

5.2 The u-perfectives and the i-perfectives

In Section 3, Classical Arabic perfectives were shown to be equipped with the

following vowel material: /ṭalab/ for ṭalab, /ʁasøl/ for ʁasal, /salim/ for salim, and
finally /kabur/ for kabur. This material was organized as in (27).

(27) a. b. c. d.
ṭalab ʁasal salim kabur

ṭ l b ʁ s l s l m k b r
a a a ø a i a u

The conjecture in (26) implies the presence in Cairene Arabic of the ingredients
present in (27), i.e., the lexical vowel, italics in (27), and the a to its left. The difference
as I will now show is in the way they deploy.
Being superficially identical in both varieties, the two verbs ṭalab and ʁasal do
not clearly indicate whether their organization in Cairene Arabic is as in (27a) and
(27b) or otherwise. By contrast, Cairene Arabic silim and kubur (compared to salim
and kabur) point to the organization shown in (28) whereby the lexical vowel has
propagated leftward, and vowel a, parenthesized in (28), has remained latent, as I
show in the next subsection.
(28) a. b.
silim (and not *salim) kubur (and not *kabur)

s l m k b r

(a) i (a) u

5.3 The a-perfectives

I submit that the organization described in (28) is one of the parameters charac-
terizing the Cairene dialect. The rule is thus: latency of vowel a and leftward
propagation of the lexical vowel. This applies across the board, i.e., not only in the
case of silim and kubur but also to ṭalab and ʁasal. However, because of the vacuity of
V2 in ʁasal, vowel a is the only local resource available for the vocalization of the
56 Fathi

stem, as seen in (29b).16 The emptiness of V2 creates the conditions for the overt
manifestation of the “latent” vowel a.17 We will soon see that the underlying /ø/
whose presence follows from the hypothesis in (26) is apophonically active.

(29) a. b. c. d.
ṭalab ʁasal silim kubur

ṭ l b ʁ s l s l m k b r

(a) a a ø (a) i (a) u

In the account just offered, ʁasal (29b) is identically represented in Classical and
Cairene Arabic, cf. (9a) above. In the next section, we will see how ʁasal and forms of
that type behave exactly in Cairene as they do in Classical Arabic in the apophonic
derivation of the vowel of their corresponding imperfectives.
My proposal is not as abstract (or difficult to learn) as it might seem. Indeed, the
idea noted in (29) that a is a basic ingredient of every single verb finds justification in
the Cairene data in (30): the ubiquity of vowel a (italicized) strongly suggests that it is
part of every verbal form in Cairene Arabic. If so, Measure I verbs, as represented in
(29), are no exception.18

(30) Measure II FaʕʕiL fahhim ‘he made X understand, √fhm’

Measure III FāʕiL sāfir ‘he travelled, √sfr’
Measure IV ʔaFʕaL ʔaħrag ‘he embarrassed, √ħrg’
Measure V ʔitFaʕʕiL ʔitħakkim ‘he kept under control, √ħkm’
Measure VI ʔitFāʕiL ʔitfāhim ‘he arrived at a mutual
understanding, √fhm’
Measure VII ʔinFaʕaL ʔinfagar ‘he exploded, √fgr’
Measure VIII ʔiFtaʕaL ʔištaʁal ‘he worked, √šʁl’

16 Because in this case “latent a” ceases to be latent, I have removed the parentheses surrounding it
in (29b).
17 Lowenstamm (2011) offers evidence from Moroccan Arabic of a latent vowel a similarly surfacing
under favorable circumstances.
18 We do not know why vowel a became latent. While this issue lies outside the scope of this
discussion, a promising line of research can nevertheless be indicated: I noted in Section 3 that /a – ø/
is the underlying vowel melody of verbs such as in (27b) and I interpreted the ø as reflecting the
absence of a lexical vowel. If that is correct /a – ø/ is the melody shared by all verbs prior to the
assignment of their lexical vowel. According to Guerssel and Lowenstamm (1990), this minimal
melody is the input to the derivation of the passive perfectives. Indeed, apophony directly derives u –
i the passive melody common to all verbs from a – ø by apophonic steps (a → u) and (ø → i), hence
ṭulib ‘it was asked’, ʁusil ‘it was washed’, šurib ‘it was drunk’, etc. Cairene Arabic lost such internal
passives. Quite possibly, the latency of vowel a is a correlate of the loss of its relevance in the
derivation of Voice.
 Ablaut in Cairene Arabic 57

Measure IX ʔiFʕaLL ʔiħmarr ‘he became red, √ħmr’

Measure X ʔistaFʕiL ʔistangid ‘he asked for rescue, √ngd’

To sum up, in this section I reintroduced the FuʕuL perfective into the discussion,
only to emphasize how severely marginalized it has become. Section 7 is devoted to
FuʕuL. There, I will argue that apophony is responsible for its demise. The other point
made in this section had to do with the organization of the vocalic ingredients of
perfective forms. The various classes of ablaut can now be addressed.

6 The ablaut classes of Cairene Arabic

6.1 The most direct cases

With more than one third of all verbs falling out, we saw in Section 4 encouraging
indications of how the apophonic mechanism might account for aspectual alterna-
tions in Cairene Arabic. Those were the cases implementing apophonic steps (a → u)
and (i → a) and exemplified by (ṭalab → yi-ṭlub) and (silim → yi-slam). These al-
ternations are boxed in (31), repeated from (20) for convenience. They are attested by
a total of 483 verbs out of the 1,256 of my list.
(31) Perf. Imperf. No. of verbs Perf. Imperf. No. of verbs
i. kasar yi-ksar iv. silim yi-slam 177
ii. ʁasal yi-ʁsil v. misik yi-msik
iii. ṭalab yi-ṭlub 306 vi. ṣibir yi-ṣbur

According to the view of the perfective established in Section 5, ṭalab/yi-ṭlub and

silim/yi-slam can be broken down into the ingredients appearing in (32). Each of the
two perfectives is made up of a root, a lexical vowel in V2 (a in ṭalab and i in silim), and
a “latent” a vowel in V1 (in brackets). The perfective stem emerges upon leftward
propagation of the lexical vowel (32a), and the vowel of the imperfective by
apophony (32b).

(32) Step (a  u): ṭalab/yi-ṭlub Step (i  a): silim/yi-slam

Root ṭ l b s l m
a.
(a) a (a) i
Perfective ṭalab silim
b.

Imperfective yi-ṭlub yi-slam

In the next subsection, we look at the implementation of apophonic step (ø → i).

58 Fathi

6.2 The (ø → i) alternation

Two alternating patterns in (31) display an i-Imperfective. These are yi-ʁsil in the pair
ʁasal/yi-ʁsil in (31.ii) and yi-msik in the pair misik/yi-msik in (31.v). I will argue that
both patterns instantiate apophonic step (ø → i).
The first of these two alternations, ʁasal/yi-ʁsil, can be accounted for in
straightforward fashion. The derivation involves underlying /ʁasøl/ where /ø/ serves
as the input to the derivation of the imperfective; then, because of the vacuity of V2,
“latent” a eventually spreads rightward (33). As we will soon see, the behavior
of “latent” a is what distinguishes the two patterns discussed in this subsection, ʁasal/
yi-ʁsil and misik/yi-msik.

(33)
ʁ s l yi -ʁ s l

a ø i
/ʁasøl/  [ʁasal] yi-ʁsil

457 verbs behave as ʁasal→ yi-ʁsil. This is recorded in (34.ii).

(34) Perf. Imperf. Perf. Imperf.

i. kasar yi-ksar iv. silim yi-slam 177 (i  a)
ii. ʁasal yi-ʁsil 457 (ø  i) v. misik yi-msik
iii. ṭalab yi-ṭlub 306 (a  u) vi. ṣibir yi-ṣbur

At this point, the vocalization of the imperfectives of 940 verbs out of 1,256 (i.e., ¾ of
the cases) has been credited to one of the apophonic steps.
Our straightforward assessment of the ʁasal/yi-ʁsil pattern in (34.ii) was based
on the idea that an imperfective melody yi-FʕiL can only be the output of apophony
applying on /FaʕøL/. What do we make then of the pattern in (34.v) where yi-FʕiL is
matched by a FiʕiL perfective?
Indeed, what we expect is for “masak” (not misik) to be the lawful perfective of
yi-msik. In the rest of this subsection, I explore the content of the class to which misik
belongs, eventually showing how the pattern in (34.v) can be absorbed into (34.ii).
Examining the 21 verbs which that class contains brings to light a crucial finding:
the FiʕiL perfectives of 16 out of those 21 verbs are in free variation with a FaʕaL
variant.19 Examples are given in (35a). For the five verbs in (35b) only, is FiʕiL the only
possible perfective.

19 Of course, such FaʕaL variants are part of the 457 verbs in (34.ii).
 Ablaut in Cairene Arabic 59

(35) a. FiʕiL ∼ FaʕaL → yi-FʕiL (16 verbs out of 21)

i. wiris ∼ waras/yi-wris ‘inherent, √wrs’
ii. ħilif ∼ ħalaf/yi-ħlif ‘swear, √ħlf’
iii. filit ∼ falat/yi-flit ‘slip away, √flt’
iv. ʕinid ∼ ʕanad/yi-ʕnid ‘behave stubbornly, √ʕnd’
b. FiʕiL→ yi-FʕiL (5 verbs out of 21)
i. misik/yi-msik ‘take hold of, √msk’
ii. libis/yi-lbis ‘dress, √lbs’
iii. nizil/yi-nzil ‘descend, √nzl’
iv. biʕid/yi-bʕid ‘be distant, √bʕd’
v. wilid/yi-wlid ‘give birth, √wld’

The 16 FaʕaL variants – an example of which is waras in (35a) – are derived exactly as
ʁasal/yi-ʁsil in (33). But, what makes the 21 FiʕiL versions, e.g., wiris, possible?
My answer is simply that, in such cases, instead of behaving as in (29b), the
“latent” vowel a (part of the basic equipment of all perfectives) has remained latent,
(36a). This triggers the epenthetic system of which the default vowel is i, (36b).

(36) a. (a) remains latent b. Epenthesis vocalizes the stem

w r s w r s

Epenthesis /i/ Epenthesis /i/

(a) ø
(a) ø

To sum up this discussion of the 478 verbs participating in the (ø → i) alternation, the
breakdown is as follows: a) a large group of 457 verbs, with ʁasal/yi-ʁsil as an
example, involves the overt manifestation of latent vowel a, b) a very small group of 5
verbs, exemplified in (35b) results from the inertia of the latent vowel, finally, c) a
small group of 16 verbs, documented in (35a), displays a pattern of wavering between
overt manifestation and inertia. Because the underlying vocalic endowment in all
three groups is the same, their apophonic behavior is uniform. I illustrate the case of
misik in (37) for comparison with ʁasal in (33).

(37)
a. b.
m s k yi -m s k

Epenthesis /i/ Epenthesis /i/

(a) ø i
a remains latent

The boxed cell in (38.v) should thus be seen, not as a genuine type of alternation, but
rather as a special case of the (ø → i) class. For clarity, I continue representing the 21
verbs patterning as misik/yimsik as a group of its own, viz. (38.v).
60 Fathi

(38) Perf. Imperf. Perf. Imperf.

i. kasar yi-ksar iv. silim yi-slam 177 (i  a)
ii. ʁasal yi-ʁsil 457 (ø  i) v. misik yi-msik 21 (ø  i)
iii. ṭalab yi-ṭlub 306 (a  u) vi. ṣibir yi-ṣbur

At this point, three of the apophonic steps of the chain have been identified for the
Cairene system: (a → u) as in ṭalab/yi-ṭlub (38.iii), (i → a) as in silim/yi-slam (38.iv), and
(ø → i) as in both ʁasal/yi-ʁsil (38.ii) and misik/yi-msik (38.v). Before identifying the last
step (u → u), in the next subsection I analyze the FaʕaL/yi-FʕaL pattern in (38.i).

6.3 The FaʕaL/yi-FʕaL group – what is it?

With 288 items, almost ¼ of the entire corpus, FaʕaL/yi-FʕaL verbs form a sizeable
group. Examples are given in (39).

(39) kasar/yi-ksar ‘break, √ksr’

baʕat/yi-bʕat ‘send, √bʕt’
masaħ/yi-msaħ ‘sweep, √msħ’
nahab/yi-nhab ‘loot, √nhb’
zaraʕ/yi-zaraʕ ‘plant, √zrʕ’
xaṭaf/yi-xṭaf ‘kidnap; snatch, √xṭf’
ʔabaḍ/yi-ʔbaḍ ‘cash, √ʔbḍ’
nasax/yi-nsax ‘copy, √nsx’
ladaʁ/yi-ldaʁ ‘sting, √ldʁ’, etc.

The existence of this group challenges Guerssel and Lowenstamm’s theory of

apophony because, as the reader will recall, it does not include an (a → a) step.
Three options suggest themselves in relation to this apparent challenge, (40).

(40) a. The theory of apophony must be revised so as to include an (a → a) step.

b. The theory of apophony per se is not in need of revision along the lines of
a. above, but its scope must be reappraised: apophony selectively applies
to specific sectors of the verbal system and ignores entire portions of it
altogether; perhaps, the FaʕaL/yi-FʕaL group is one such portion.
c. The theory of apophony does not need amending; rather, apophony
applies unrestrictedly but its output can be obfuscated by the subsequent
application of a further process.

Option (c) opens a promising path. Indeed, the verbs of the FaʕaL/yi-FʕaL group share
a significant property: the second and/or third consonant of their root is a guttural,
 Ablaut in Cairene Arabic 61

an emphatic, a uvular or an r, i.e., {ʕ, ʔ, h, ħ, ẓ, ṣ, ṭ, ḍ, ʁ, x, r}.20 On the view that such

consonants can exert a lowering influence on vowels, we have to ask whether the a of
yi-FʕaL does not mask a canonical apophonic output. Two such possibly altered
outputs must be considered. Indeed, we established that surface FaʕaL expresses two
different underlying objects, /FaʕøL/ and /FaʕaL/. /FaʕøL/ derives yi-FʕiL and /FaʕaL/
derives yi-FʕuL. In both cases, a high vowel – i or u – is the output. It stands to reason
that in a “lowering” consonantal environment both high vowels could lower to a as
per the derivations in (41).
(41) a. FaʕøL → yi-FʕiL >> eventually lowered to yi-FʕaL
b. FaʕaL → yi-FʕuL >> eventually lowered to yi-FʕaL

If so, the FaʕaL/yi-FʕaL set does not even constitute a unitary group. Rather, it is the
sum of the results of the two processes depicted in (41).21
This reasoning was anticipated by the medieval Arabic scholars in their work on
Classical Arabic. Alongside the alternations reviewed in Sections 2 and 3, Classical
Arabic also has a FaʕaL/ya-FʕaL group of verbs. Early on, the medieval Arabic
scholars noted the presence of guttural consonants in the roots of those verbs and
explained the outstanding vocalization of their imperfectives in terms of the
lowering effect of such consonants. Of interest, is the impeccable argument devel-
oped by the Arabic grammarians to the effect that the FaʕaL/ya-FʕaL alternation
reflects the interference of lowering consonants and is therefore derivative and not
genuine. Their position is most clearly summed up by Raḍī d-Dīn al-Astārābāđī, a 13th
century scholar, in his Šarḥ Šāfiyat Ibn al-Ḥāğib (Al-Astārābāđī 1975, vol. I: 117):

The masters of morphological derivation say: FaʕaL/yaFʕaL is derived from FaʕaL/yaFʕuL or

FaʕaL/yaFʕiL. Here is why: when they saw that a only appears when a guttural is present, they
found in the guttural the reason for the presence of a in the imperfectives of FaʕaL. They became
convinced that the guttural is the cause of the presence of a, and as it is impossible to find a
except in the vicinity of a guttural, they concluded that this a has no other motivation. Conse-
quently, they became convinced that a is not free and unconditioned as u or i are, or it would
occur freely where no guttural is present.

Lowering in Cairene – as in Classical Arabic – does not apply with full regularity.
Sometimes, it simply does not apply at all. Sometimes, it applies optionally, allowing
doublets to arise. In such cases, i or u emerges.

20 The fact that Cairene Arabic emphatics, unlike their Classical Arabic counterparts, participate in
the lowering of neighboring vowels is nothing new. That fact was recognized in the earliest attempts
at describing Cairene Arabic in systematic fashion (Spitta-Bey 1880: 205, 206; Vollers and Burkitt 1895:
47; Willmore 1905: 134); and confirmed by subsequent investigators (Mitchell 1956: 34, 73; Jomier and
Khouzam 1973: 111; Broselow 1976: 140; Woidich 2006: 63, 67).
21 A third scenario giving rise to the FaʕaL/yi-FʕaL pattern can be considered: FaʕaL, in this case,
masks an underlying (and lowered) FiʕiL, whereas yi-FʕaL merely reflects the i → a apophonic step.
62 Fathi

A sample of verbs which are not affected by lowering is given in (42).

(42) ṣaraf/yi-ṣrif ‘spend (money), √ṣrf’

šaʁal/yi-šʁil ‘occupy, √šʁl’
ʁaṣab/yi-ʁṣib ‘oblige, √ʁṣb’
zaħam/yi-zħim ‘crowd, √zħm’
laxam/yi-lxim ‘fluster; confuse, √lxm’
ħaḍan/yi-ħḍun ‘hug, √ħḍn’
ʔamar/yi-ʔmur ‘order, √ʔmr’
zaʁad/yi-zʁud ‘poke, √zʁd’
salaʔ/yi-sluʔ ‘boil, √slʔ’
šarax/yi-šrux ‘crack, √šrx’

“Hesitating” application of the rule gives rise to two classes of doublets, each of
seven verbs. In the first class, the imperfective vowel may be a or i. Examples are
given in (43).

(43) faḍaħ/(yi-fḍaħ ∼ yi-fḍiħ) ‘subject to scandal, √fḍħ’

bahar/(yi-bhar ∼ yi-bhir) ‘dazzle, √bhr’
gaħad/(yi-għad ∼ yi-għid) ‘receive with ingratitude, √għd’
rabaʕ/(yi-rbaʕ ∼ yi-rbiʕ) ‘gallop (of a donkey), √rbʕ’

In the other class, it may be a or u.

(44) ṣalaħ/(yi-ṣlaħ ∼ yi-ṣluħ) ‘be suitable, √ṣlħ’

ṭabaʔ/(yi-ṭbaʔ ∼ yi-ṭbuʔ) ‘take a constricting hold, √ṭbʔ’
šaʔaṭ/(yi-šʔaṭ ∼ yi-šʔuṭ) ‘catch (a flying object), √šʔṭ’
gaħar/(yi-għar ∼ yi-għur) ‘bury in a hole, √għr’

In sum, the behavior of verbs from triliteral roots including a lowering consonant in
second and/or third position breaks down into three groups: a) 288 lower a vowel to
yi-FʕaL, b) 14 lower a vowel in wavering fashion, c) the rest of them, 429 verbs, do not
lower a vowel; 177 are part of group (45.ii) and 252 are part of group (45.iii).
If the yi-FʕaL imperfectives of FaʕaL verbs can be traced back to yi-FʕiL or yi-
FʕuL as per the sequences in (41),22 the entire group in (45.i) reduces to special cases of
either (45.ii) or (45.iii).

22 Note the heterogeneous status of the evidence just reviewed. Where doublets are involved, the
underlying identity of surface FaʕaL is easily retrieved: /bahør/ in the case of bahar/(yi-bhar ∼ yi-bhir)
and /ṣalaħ/ in the case of ṣalaħ/(yi-ṣlaħ ∼ yi-ṣluħ). But where no doublets are involved, as with kasar/
yi-ksar, it is impossible to know which of /kasør/ or /kasar/ underlies surface [kasar]. In Section 7, I
describe how the grammar reacts in the face of this uncertainty.
 Ablaut in Cairene Arabic 63

(45) Perf. Imperf. Perf. Imperf.

i. kasar yi-ksar 288 LOWERING iv. silim yi-slam 177 (i  a)
(ø  i) or (a  u)
ii. ʁasal yi-ʁsil 457 (ø  i) v. misik yi-msik 21 (ø  i)
iii. ṭalab yi-ṭlub 306 (a  u) vi. ṣibir yi-ṣbur

Up to this point, I have shown how apophony accounts for all but one of the alter-
nating patterns in Cairene Arabic. To sum up: the alternating patterns in (45.ii),
(45.iv) and (45.iii) follow the three regular steps (ø → i → a → u) respectively. I have
argued that the class in (45.v) is a special case of the pattern in (45.ii), and that the
class in (45.i) a special case of either pattern (45.ii) or (45.iii).
We can now turn our attention to the last group (45.vi) and proceed to its
autopsy.

6.4 The FiʕiL/yi-FʕuL group

According to the apophonic mechanism, yi-ṣbur should be the output of either ṣabar
(a → u) or ṣubur (u → u), but not ṣibir as there is no (i → u) apophonic step.
The examination of the class to which ṣibir/yi-ṣbur belongs, which contains all in all
seven verbs, reveals a striking finding: all such verbs have their FiʕiL perfectives
alternating in free variation with either FaʕaL or FuʕuL or both. This is shown in (46).

(46) sikin ∼ sakan/yi-skun ‘inhabit, √skn’

ʁirib ∼ ʁarab/yi-ʁrub ‘to set (of the sun), √ʁrb’
fitir ∼ fatar/yi-ftur ‘become tepid, √ftr’
milik ∼ malak/yi-mluk ‘posses, √mlk’23
gibin ∼ gubun/yi-gbun ‘act cowardly, √gbn’
sikit ∼ sakat ∼ sukut/yi-skut ‘become silent, √skt’
ṣibir ∼ ṣabar ∼ ṣubur/yi-ṣbur ‘act with patience, √ṣbr’

It is obvious that the imperfective melody characterizing this class of verbs is derived
from either perfective FaʕaL or perfective FuʕuL, not FiʕiL. The question therefore is
why FiʕiL figures as a possible perfective for a yi-FʕuL imperfective. In the next
section, I examine FuʕuL verbs. As part of this discussion, I will explain how FiʕiL
came to gradually replace FuʕuL.

23 The imperfective yi-mluk alternates in free variation with yi-mlik, hence yi-mluk ∼ yi-mlik.
64 Fathi

7 The 4th apophonic step (u → u) in Cairene

Arabic – what happened to it?
In this section, I review the group of 51 verbs whose perfectives are vocalized as
FuʕuL. The expectation is that their imperfectives will be of the yi-FʕuL type as per
the 4th apophonic step, (u → u). Such is not the case, as we will see. Indeed, they are in
their overwhelming majority of the yi-FʕaL type, much as if apophony had played no
role in the vocalization of those verbs. I will argue, on the contrary, that the
numerous departures from the FuʕuL/yi-FʕuL pattern are cases of back formation
induced by strict application of apophony.
I will first review FuʕuL perfectives, then their corresponding imperfectives.

7.1 FuʕuL perfectives

In Subsection 5.1, I indicated that FuʕuL verbs alternate with variants. Thus, the verb
from root √kbr ‘grow’ can be realized as kubur or kibir with no difference in
meaning. The set of logical possibilities for FuʕuL to occur or not to occur with
variants is as in (47).

(47) a. FuʕuL alone (i.e., with no variants)

b. FuʕuL ∼ FaʕaL
c. FuʕuL ∼ FiʕiL ∼ FaʕaL
d. FuʕuL ∼ FiʕiL

All four patterns are instantiated, though with a striking imbalance.

Two verbs exemplify the pattern in (47a), ħuduʔ ‘become too salty, √ħdʔ’
and ħurum ‘be forbidden, √ħrm’). This pair of verbs can be readily dismissed as
insignificant because the first verb is most commonly realized in Form IX, ʔiħdaʔʔ
(i.e., not in Form I, ħuduʔ) whereas the second verb primarily appears in idiomatic
expressions (cf. Badawi and Hinds 1986: 200–201).
One single verb exemplifies the pattern in (47b), ḍumur ∼ ḍamar ‘shrivel;
atrophy, √ḍmr’.
Five verbs only follow the pattern in (47c):

(48) ṣubur ∼ ṣibir ∼ ṣabar ‘act with patience, √ṣbr’

sukut ∼ sikit ∼ sakat ‘become silent, √skt’
ṭuhuʔ ∼ ṭihiʔ ∼ ṭahaʔ ‘become thoroughly fed up, √ṭhʔ’
kufur ∼ kifir ∼ kafar ‘deny God’s existence, √kfr’
ṣuʔuṭ ∼ ṣiʔiṭ ∼ ṣaʔaṭ ‘fall; fail, √ṣʔṭ’
 Ablaut in Cairene Arabic 65

In sharp contrast, 43 verbs exemplify the pattern in (47d). Examples were given
earlier in (23). More examples appear in (49):

(49) ṣuʕub ∼ ṣiʕib ‘become more difficult, √ṣʕb’

zuhuʔ ∼ zihiʔ ‘get bored, √zhʔ’
ʁumuʔ ∼ ʁimiʔ ‘darken, √ʁmʔ’
xurub ∼ xirib ‘be destroyed; be ruined, √xrb’
xumur ∼ ximir ‘ferment; rise (of dough), √xmr’
In other words, except for the isolated cases just reviewed, the generalization in (50)
holds:

(50) a FuʕuL verb can also be realized as FiʕiL.

Note that the generalization in (50) is asymmetrical: while just about every FuʕuL
verb can be realized as FiʕiL, the reverse is not true. For instance, kubur can be
realized as kibir as we saw, but realizing libis ‘dress, √lbs’ as *lubus, or silim ‘be safe,
√slm’ as *sulum, is out of the question.
The conclusion is inescapable: the territory of FuʕuL was gradually invaded by
FiʕiL.
Although no systematic surveys were conducted on that particular point over
time, we can get a sense of the gradual nature of the phenomenon when late 19th
century descriptions are compared with contemporary accounts.
Spitta-Bey (1880), introducing Measure I triliteral perfectives, recognizes three
versions: FaʕaL, FiʕiL, FuʕuL. The very fact of acknowledging the existence of
FuʕuL is in itself highly significant as we will soon see. Simultaneously, Spitta-Bey
also acknowledges the presence of frequent FuʕuL ∼ FiʕiL doublets. Spitta-Bey
(1880: 194) offers kutur ∼ kitir ‘become numerous; proliferate, √ktr’ as an example:
“Frequently [emphasis mine] both forms FiʕiL and FuʕuL are found next to each
other, e.g., kitir and kutur (…)”.24 Note that Spitta-Bey, a writer very careful in his
choice of words, says “frequently”, not “always” nor “regularly” nor “systemati-
cally”. This suggests that in the 1870’s the point has not yet been reached where
every FuʕuL verb can equally well be realized as FiʕiL. In Spitta-Bey’s description,
the FaʕaL and FiʕiL patterns are illustrated by means of very common verbs, katab
‘he wrote’ and misik, respectively. kutur is the verb picked by Spitta-Bey (1880: 204)
to illustrate the FuʕuL pattern. The fact that kutur sits as a bona fide representative
of its own pattern alongside other items of everyday use such as katab and misik
strongly suggests, in turn, that its marginalization by kitir has not been completed,
yet.

24 Spitta-Bey (1880: 194) gives two more examples: fiḍil ∼ fuḍul ‘remain, √fḍl’, sikit ∼ sukut ‘become
silent, √skt’.
66 Fathi

Willmore (1905: 116), also acknowledging the existence of the pattern in (47d)
writes: “Many verbs (…) take the form birik or buruk optionally,25 though the latter is
perhaps more common”. One of his examples of such FuʕuL ∼ FiʕiL doublets is
precisely the verb from root √ktr, quoted as kutur ∼ kitir. Willmore only provides
one example explicitly documenting a definite preference on the part of speakers for
a FuʕuL variant over the FiʕiL alternative: ʕuṭus ‘sneeze, √ʕṭs’ over ʕiṭis. It is difficult
to know how representative Willmore’s single example is. But, we can assume that
his remark rests on the observation that the FuʕuL pattern was still well alive at the
time of his writing.
As we reach the middle of the 20th century, descriptions change, attesting to the
progressive marginalization of the FuʕuL pattern.
On the FuʕuL pattern, Mitchell (1956: 32, 1962: 71) writes: “The sequence u-u
occurs for i-i with some speakers, e.g., xuruṣ ‘he was struck dumb’. i-i (xiriṣ), however,
is much commoner and may always be used”. Mitchell’s statement makes it possible
to put a date on (50): by mid-20th century, the invasion by FiʕiL of the domain of
FuʕuL has been completed.
Subsequent decades show something new: in sharp contrast with Spitta-Bey and
Willmore, authors such as Tomiche (1964) or Jomier and Khouzam (1973) do not even
mention the FuʕuL pattern any more. Gadalla (2000) goes as far as to explicitly deny
its existence.
Woidich (2006), is an interesting exception.26 Noting its residual status, Woidich
acknowledges the FuʕuL pattern. He illustrates each of the three vocalization clas-
ses – FaʕaL, FuʕuL, FiʕiL – by means of single examples. That is, the possible
membership in doublets or triplets of his verbs is not indicated. The verb from root
√ktr is one of the illustrative examples. Significantly, it serves to illustrate pattern
FiʕiL, i.e., kitir. If one version only is to be given for each verb, then surely Woidich is
correct in selecting kitir as the more popular form.
This rapid survey documents the gradual infringement of FiʕiL over the domain
of FuʕuL verbs. Prima facie, this is intriguing. Indeed, a FuʕuL perfective is supposed
to produce a yi-FʕuL imperfective. On that basis, we can well see how FaʕaL might
come to compete with FuʕuL in the form of a FaʕaL/FuʕuL doublet: both FaʕaL
and FuʕuL are legitimate apophonic sources for yi-FʕuL. FiʕiL, by contrast, produces
a yi-FʕaL imperfective and it is not immediately clear what caused it to compete with
FuʕuL. But a review of the imperfectives of FuʕuL verbs answers this question.

25 Willmore uses root √brk as a token root, i.e., as √FʕL is used in this paper.
26 The discrepancy on this point between Mitchell (1956, 1962) and Jomier and Khouzam (1973) on the
one hand, and Woidich (2006) should come as no surprise. Woidich’s book is a scholarly reference
grammar (the most complete and thorough to this day) whereas Mitchell’s and Jomier and Khouzam’s
are merely practical manuals for language instruction.
 Ablaut in Cairene Arabic 67

7.2 The imperfectives of FuʕuL verbs and the apophonic

backlash

Unexpectedly, a tiny minority of only 4 (out of 51) FuʕuL verbs are paired with
the expected yi-FʕuL imperfective.27 The other 47 have a yi-FʕaL imperfective. This
a-vocalization is clearly related to the fact that 43 of those 47 verbs are from roots
including a lowering consonant in C2 or C3 position, or in both.28 We will see how this
majoritarian imperfective vocalization is key to understanding the rise of FiʕiL.
Consider the verb from root √ṣʕb ‘become more difficult’, our first example in
(49). As we know, its perfective comes in two variants, ṣiʕib ∼ ṣuʕub and its imper-
fective is yi-ṣʕab. We cannot precisely identify a period when ṣuʕub was the sole
perfective of that verb but I assume that such was the case at some point. At
that point, ṣuʕub/yi-ṣʕab can be interpreted in light of the apophonic mechanism as
follows. The input being ṣuʕub, the relevant apophonic step is (u → u), and the output
is yi-ṣʕub. An additional event is therefore required for the derivation of yi-ṣʕab.
Given the presence of a lowering consonant in one of the relevant positions of the
root, Lowering can be credited for the completion of that last step. In sum, ṣuʕub and
yi-ṣʕab are related as shown in (51).

(51) ṣuʕub → yi-ṣʕub → yi-ṣʕab

apophony Lowering

I submit that at some further point in the course of the transmission of the language
from one generation to the other, speakers ceased to analyze the a of yi-ṣʕab as
resulting from Lowering. Rather, they took that a at face value. I represent that
reanalysis as in (52), which reproduces the configuration in (51) with the difference
that yi-ṣʕab is no longer construed as the lowered guise of yi-ṣʕub.

(52) ṣuʕub → yi-ṣʕub → yi-ṣʕab

apophony Lowering

Severing the link between the initial and the final point of the derivation wrecks
the entire scheme. To see this, consider (53) where the strikethroughs in (52) have
been omitted.

(53) ṣuʕub → yi-ṣʕab

apophony

27 (sikit ∼ sakat ∼) sukut/yi-skut ‘become silent, √skt’, (ṣibir ∼ ṣabar ∼) ṣubur/yi-ṣbur ‘act with
patience, √ṣbr’, (ḍamar ∼) ḍumur/yi-ḍmur ‘shrivel; atrophy, √ḍmr’, (gibin ∼) gubun/yi-gbun ‘act
cowardly, √gbn’.
28 I have nothing to say about the intriguing overrepresentation of roots with lowering consonants
in FuʕuL verbs.
68 Fathi

The diagnosis is inescapable: the reanalysis in (52) culminates in the apophonic

dead end in (53). Indeed, because there is no (u → a) apophonic step, ṣuʕub cannot
possibly be the apophonic antecedent of yi-ṣʕab. In the face of this dead end, I
submit, apophony steps in and enforces the existence of a new form, one capable of
fulfilling that role. The theory of apophony is unambiguous as to what the rightful
input to yi-FʕaL should be: FiʕiL is the only apophonic source of yi-FʕaL. In the case
at hand, therefore, ṣiʕib must be the new form. I call apophonic backlash this type of
back formation whereby apophony creates new, suitable input forms (further
examples will be adduced and discussed in the next section).
The creation of a new form such as ṣiʕib hardly entails the instantaneous
disappearance from the linguistic environment of the already existing ṣuʕub.
Indeed, it cannot be the case that the entire speech community performs a rean-
alysis such as in (52) as a block and at once. For the generation which initiates the
change, ṣuʕub is no longer apophonically motivated but it is still present in the
linguistic environment because the older generations continue operating along
the lines of (51) with ṣuʕub as an active apophonic player. The diachronic scheme I
have advocated thus predicts the existence of same-meaning doublets. But, it also
predicts the asymetrical distribution whereby perfectives (not imperfectives)
come in doublets: if, as I claim, doublets arise as a result of back formation, their
distribution is exactly that of apophonic inputs, i.e., perfectives.
We are now in a position to characterize the compelling competitive advantage
of FiʕiL over FuʕuL. Consider once more the micro-system formed by yi-ṣʕab and its
two perfectives, ṣuʕub and ṣiʕib.

(54) a. ṣuʕub → yi-ṣʕub

yi-ṣʕab
b. ṣiʕib

The competitive edge of FiʕiL is twofold. On the one hand, the derivation in (54b) is
apophonically optimal in the sense that the output of apophony has not been
rendered opaque by the subsequent application of Lowering. On the other hand,
(54b) brings the tiny and motley group of FuʕuL verbs into the fold of a richly attested
pattern, the FiʕiL/yi-FʕaL pattern (silim/yi-slam, etc.). This answers the question
inherent in the title of this section. We can at last construe the existence of the
mysterious class in (45.vi): the pressure exerted by FiʕiL is so strong that it replaces
FuʕuL where the imperfective is yi-FʕaL, but also in the handful of cases where it has
remained yi-FʕuL.
In this section, I described how the apophonic system can explain the gradual
demise of the FuʕuL pattern over time. In the next section, I speculate about the
future evolution of the vocalization pattern of the language.
 Ablaut in Cairene Arabic 69

8 Changes, past and future

My account of the competitive advantage of FiʕiL with respect to FuʕuL predicts
that, given the opportunity, FiʕiL will proliferate in other sectors of the verbal
system, too.
In the preceding section, we saw how ṣiʕib came to be viewed as a more
attractive input to yi-ṣʕab than ṣuʕub. Outputs with the same yi-FʕaL profile as yi-ṣʕab
abound, offering FiʕiL additional opportunities to increase its territory. In Section
6.3, we established that all the imperfectives of the 288 verbs of the FaʕaL/yi-FʕaL
type could be construed as ‘lowered’ cases of either yi-FʕuL or yi-FʕiL. For conve-
nience, I illustrate this anew, this time with the dual analysis of verb nagaħ/yi-ngaħ
‘succeed, √ngħ’. nagaħ is intrinsically ambiguous as to whether it expresses
underlying /nagaħ/ (55a) or underlying /nagøħ/ (55b). Either way, apophony will
assign a high vowel to the corresponding imperfective, viz. yi-nguħ (a → u) or yi-ngiħ
(ø → i), respectively. Both vowels are equally likely to fall prey to Lowering and
neutralize into yi-ngaħ.

(55) a. nagaħ → yi-nguħ

apophony Lowering

yi-ngaħ
Lowering
b. nagøħ → yi-ngiħ
apophony

As already noted in 6.3., a derivation such as in (55) is not only opaque because the
output of apophony is modified by Lowering, but also ambiguous because it is
impossible to retrieve what the input of the derivation was, /nagaħ/ or /nagøħ/. But,
note that an alternative scheme is possible, viz. the reanalysis culminating in back
formation, (56c): with its a, yi-ngaħ (exactly as yi-ṣʕab) can be analyzed as the output
of another of the steps of the apophonic chain, namely (i → a). If so, nigiħ must be the
perfective and indeed that form is an attested variant of nagaħ.

(56) a. nagaħ → yi-nguħ

apophony Lowering

yi-ngaħ
Lowering
b. nagøħ → yi-ngiħ
apophony
back formation

c. nigiħ

Evidently, when all available analytical options are compared, /nigiħ/ comes out as a
much more attractive input to the derivation than the obscure and undecidable
70 Fathi

“/nagaħ/ or /nagøħ/” disjunction. This verb is not an isolated example. Indeed, 24 of

the 288 FaʕaL/yi-FʕaL verbs have a FiʕiL alternant. Some examples are given in (57).29

(57) waṣal ∼ wiṣil/yi-wṣal ‘arrive, √wṣl’

harab ∼ hirib/yi-hrab ‘escape, √hrb’
waʔaʕ ∼ wiʔiʕ/yu-ʔaʕ (</yi-wʔaʕ/) ‘fall down, √wʔʕ’
raḍaʕ ∼ riḍiʕ/yi-rḍaʕ ‘suckle, √rḍʕ’

At the time of this writing, two forms coexist for the perfectives of the verbs in (57).
My account of what I have called the competitive advantage of the FiʕiL variant leads
to the expectation that it will evict the FaʕaL variant over time. The prediction is that,
while nagaħ and nigiħ for example are still both recognized as the legitimate
perfectives of yi-ngaħ, their prospects are quite different: nigiħ will outlive nagaħ,
eventually causing nigiħ/yi-ngaħ to join the FiʕiL/yi-FʕaL class.
The conjecture that the days of nagaħ are numbered (and more generally the
FaʕaL perfectives of the FaʕaL/yi-FʕaL group) is more than just speculation. That
trend has been under way for some time as I now show.
Aro (1964) compiled lists of vocalization patterns for Measure I verbs in Semitic
languages. While the focus of his book is on classical languages, a portion of his
chapter on Classical Arabic includes a substantial discussion of Cairene Arabic.
There, Aro notes that the bulk of Cairene FiʕiL/yi-FʕaL verbs includes verbs from the
same roots as in Classical Arabic. e.g., Classical Ar. fariħ/ya-fraħ and Cairene Ar. firiħ/
yi-fraħ ‘rejoice’, both evidencing the same i–a alternation. But he observes that in
addition the Cairene FiʕiL/yi-FʕaL group has also absorbed a significant number of
verbs which in Classical Arabic belong to the FaʕaL/ya-FʕuL and FaʕaL/ya-FʕiL
classes.30 An example of that “class shift” appears in (58) with the verb from root √ṭlʕ
‘rise’, a FaʕaL/ya-FʕuL verb in Classical Arabic though a FiʕiL/yi-FʕaL verb in Cairene
Arabic.
(58) Classical Arabic ṭalaʕ/ya-ṭluʕ Cairene Arabic ṭiliʕ/yi-ṭlaʕ

Beyond a brief remark to the effect that such class-shifters seem to be of the
“intransitive-fientive” type, Aro does not discuss the phenomenon any further. A
crucial fact seems to have escaped his notice: such verbs are from roots including a

29 Six of the 24 verbs which pattern as exemplified in (57) are of the FaʕiL type in Classical Arabic, for
instance karah ∼ kirih/yi-krah ‘hate, √krh’ (karih in Classical Arabic) and laħaʔ ∼ liħiʔ/yi-lħaʔ ‘catch
up with, √lħʔ’ (laħiq in Classical Arabic). I do not know why or when those 6 verbs developed a FaʕaL
version in Cairene (perhaps as a result of Lowering). In any case my claim stands: the FaʕaL version
of those 24 verbs is under permanent threat of being ousted by the FiʕiL alternative.
30 Aro assumes a historical filiation between the two versions of Arabic. For the sake of the argu-
ment, I accept that view without discussion or apology.
 Ablaut in Cairene Arabic 71

lowering consonant. When that factor is taken into account, the change exemplified
in (58) can be explained. My argument will be that the verb from root √ṭlʕ did not
simply migrate from one class to another. Rather, the change comprises distinct steps
described in (59), all of which mobilize two independent phenomena already iden-
tified earlier in this paper, viz. Lowering, then the apophonic backlash effect. The
point I wish to establish is this: the sequence of steps necessarily includes a transitory
stage at which a FiʕiL ∼ FaʕaL perfective doublet exists, then disappears only leaving
the FiʕiL version.
Stage 1 is the representation of the verb as it was passed on to Cairene Arabic as
part of the legacy of an earlier form of Arabic. Stage 2 shows what I assume to be an
Egyptian development, viz. the lowering of the vowel of the imperfective (in bold).
This causes the temporary membership of the verb in the FaʕaL/yi-FʕaL group. Stage
3 shows the back-formation of an innovative, alternative perfective (in bold) as a
result of what I have called “backlash”. Both perfective forms coexist, but only for as
long as the more conservative generation is present. Stage 4 represents the situation
after the conservative generation has disappeared. At that stage, the FaʕaL variant
has been totally ousted.
(59) Stage 1 Legacy: the verb from root √ṭlʕ ṭalaʕ/ya-ṭluʕ
Stage 2 Lowering ṭalaʕ/yi-ṭlaʕ
Stage 3 Back-formation of a new perfective ṭiliʕ ∼ ṭalaʕ/yi-ṭlaʕ
Stage 4 Loss of FaʕaL ṭiliʕ ∼ ṭalaʕ/yi-ṭlaʕ

In (60), I provide a sample of such verbs (Aro 1964: 91). Strikingly, no FaʕaL
perfectives are attested for those verbs anymore, even though (by hypothesis) they
were passed on as such from Classical Arabic.
(60) Classical Arabic Cairene Arabic
Attested Non-attested31
ṭalaʕ/ya-ṭluʕ ‘climb’ ṭiliʕ/yi-ṭlaʕ ṭalaʕ
ʕaθar/ya-ʕθur, ‘stumble; find’ ʕitir/yi-ʕtar ʕatar
xalaṣ/ya-xluṣ ‘escape’ xiliṣ/yi-xlaṣ xalaṣ
faṭar/ya-fṭur ‘breakfast’ fiṭir/yi-fṭar faṭar
ʕaraf/ya-ʕrif ‘know’ ʕirif/yi-ʕraf ʕaraf
ʕaṭas/ya-ʕṭis ‘sneeze’ ʕiṭis/yi-ʕṭas ʕaṭas

31 This is not to say that such forms cannot be heard in northern Egypt: while such forms are no
longer attested in the Cairene dialect (see e.g., Badawi and Hinds 1986), they are well alive in the other
cities, e.g., ṭalaʕ, ʕaraf (Alexandria), faṭar, ʕaṭas (eastern delta cities).
72 Fathi

I conclude this section with a return to its initial point: my prediction is that the
currently extant nigiħ ∼ nagaħ doublet (and others of the same kind) awaits the same
development as was undergone earlier by the verbs in (60).
The next (and last) section is devoted to an assessment of what the Cairene
Arabic evidence discussed in this article can tell us about the theory of apophony.

9 A note on the theory of apophony

In this article, I have shown how FuʕuL/yi-FʕuL alternations had essentially
disappeared from the Cairene system. How does this absence square with my claim
that apophony is the central device responsible for the perfective/imperfective
alternations in Forms I verbs?
I address this question through the prism of proposals put forth by Guerssel
2003, 2014).
Guerssel argues that the apophonic chain comprises 3 (not 4!) steps:
1) ø → i, 2) i → a, 3) a → u.
That is, Guerssel claims there is no 4th apophonic step (u → u). I briefly review
his demonstration on the basis of Classical Arabic evidence.32
Guerssel proposes that the vocalization of Form I verbs proceeds as follows.
First, the melody of the perfective is copied onto the bare imperfective template
shown in (61b). When that first step is completed, the perfective and the imperfective
have identical melodies (61a) and (61b). Then, apophony takes the (copied) imper-
fective melody as its input and modifies it as expected, (61c).
In the case of FaʕuL verbs such as kabur, the perfective melody is copied onto the
imperfective as in all other cases. But, because there is no 4th apophonic step, the
imperfective simply surfaces with the vowel which was copied from the perfective.

(61) a. b. c. d.

Perf. y_Fʕ_L apophony output

i. ʁasøl yaʁsøl yaʁsøl yaʁsil yaʁsil

ii. salim yaslim yaslim yaslam yaslam

iii. ṭalab yaṭlab yaṭlab yaṭlub yaṭlub

iv. kabur yakbur DNA yakbur

32 Lahrouchi (2011), on the basis of nominal evidence from several Afroasiatic languages, develops a
novel argument crediting the Obligatory Contour Principle for ruling out a 4th (u → u) apophonic step.
 Ablaut in Cairene Arabic 73

On this view, FaʕuL/ya-FʕuL alternations such as kabur/ya-kbur manifest apophonic

inertia. In other words, they are not part of the apophonic system. Guerssel’s
proposal has two important consequences.
The first consequence is the rationalization of an intriguing and hitherto
unexplained property of the 4th apophonic step: unlike all other apophonic steps, it
does not induce a change. On the view that there is no such thing as a 4th apophonic
step, the puzzle vanishes.
The second consequence is that a language heavily relying on apophony for
the vocalization of its verbal or nominal forms may or may not display (u – u)
alternations. Such alternations are extraneous to apophony and their presence or
their absence cannot found an apophonic diagnosis. Cairene Arabic is an interesting
case in this respect. We saw how the language lost (u – u) verbal alternations. That
loss in no way indicates that apophony has lost any of its vigor. On the contrary, I
have shown that with its backlash effect, apophony can even “correct” the lexical
vocalization of perfectives.

10 Conclusion
In this article, I have examined the vowel alternations relating the perfectives and
imperfectives of Form I verbs in Cairene Arabic. I have argued that the exact same
system of Ablaut which Guerssel and Lowenstamm (1990, 1996) identified for
Classical Arabic is at work in Cairene, too. However, I have gone further than this.
Handling the evidence of a living language involves facing the specific challenge of
understanding the direction of its continuous evolution. In the case of Cairene
Arabic, one such challenge is the puzzling proliferation of doublet forms in the
perfective. I have argued that such doublets arise as the result of apophonic pressure
and that their existence represents a transitory stage of ongoing linguistic change:
for each of the doublets discussed in this article, one member is imposed as a back-
formation effect of apophony; it will co-exist for some time with the “old” member of
the doublet, but it is bound to eventually oust it. That the apophonic mechanism can
be viewed as an agent of linguistic change constitutes a novel kind of argument in its
favor.

References
Abboud-Haggar, Soha. 2003. Introducción a la dialectología de la lengua árabe. Granada: Fondacion El
Legado andalusi.
Abdel-Massih, Ernest. 1975. An Introduction to Cairene Arabic. Ann Arbor: Center for Near Eastern and North
African Studies, University of Michigan.
74 Fathi

Al-Astārābāđī, Raḍī d-Dīn. 1975. In Muḥammad Nūr al-Ḥasan, Muḥammad Al-Zafzāf &
Muḥyī l-Dīn ʿAbd al-Ḥamīd (eds.), Šarḥ Šāfiyat Ibn al-Ḥājib, vol. 4. Beirut: Dār al-Kutub al-ʿilmiyya.
Aro, Jussi. 1964. Die Vokalisierung des Grundstammes im Semitischen Verbum. Helsinki: Societas Orientalis
Fennica.
Badawi, El-Said & Martin Hinds. 1986. A dictionary of Cairene Arabic. Beirut: Librairie du Liban.
Baturay Meral, Semra. 2010. Apophony in Turkish. Journal of Turkish Studies International Periodical for the
Languages, Literature and History of Turkish or Turkic 5/4. 201–220.
Behnstedt, Peter & Manfred Woidich. 2018. The formation of the Egyptian Arabic dialect area. In
Clive Holes (ed.), Arabic historical dialectology: Linguistic and sociolinguistic approaches, 64–95. Oxford:
Oxford University Press.
Bendjaballah, Sabrina. 1998. Aspects Apophoniques de la Vocalisation du Verbe Berbère (Dialecte Kabyle).
In Patrick Sauzet (ed.), Langue et grammaire II & III – Phonologie, 5–24. Paris: Université Paris 8.
Bendjaballah, Sabrina. 1999. Trois figures de la structure interne des gabarits. Paris: Université Paris 7
dissertation.
Bendjaballah, Sabrina. 2001. The negative preterite in Kabyle Berber. Folia Linguistics 34. 185–223.
Bendjaballah, Sabrina. 2003. The internal structure of the determiner in Beja. In Jacqueline Lecarme (ed.),
Research in afroasiatic grammar II, 35–53. Amsterdam: John Benjamins.
Bendjaballah, Sabrina. 2005. Apophony. In Kees Versteegh (ed.), Encyclopedia of Arabic language and
linguistics, 64–68. Leiden: Brill.
Bohas, Georges & Jean-Patrick Guillaume. 1984. Etudes des théories des grammairiens arabes. Damascus:
Institut français de Damas.
Boyé, Gilles. 2014. Apophony and chiming words in Malay. In Sabrina Bendjaballah, Noam Faust,
Mohamed Lahrouchi & Nicola Lampitelli (eds.), The form of structure, the structure of form: Essays in
honor of Jean Lowenstamm, 57–65. Amsterdam: John Benjamins.
Brame, Michael K. 1970. Arabic phonology: Implications for phonological theory and historical semitics.
Massachusetts Institute of Technology dissertation.
Broselow, Ellen. 1976. The phonology of Cairene Arabic. Amherst: University of Massachusetts dissertation.
Cantineau, Jean. 1949. Le vocalisme radical de l’inaccompli dans le parler arabe d’Alep. Bulletin de la Société
de linguistique XLV–1. 61–73.
Cantineau, Jean. 1950. La notion de schème et son altération dans diverses langues sémitiques. Semitica
III. 73–83.
Chekayri, Abdallah & Tobias Scheer. 1996. The apophonic origin of glides in the verbal system of classical
Arabic. In Jacqueline Lecarme, Jean Lowenstamm & Ur Shlonsky (eds.), Studies in afroasiatic grammar,
62–76. The Hague: Holland Academic Graphics.
Chekayri, Abdallah & Tobias Scheer. 2003. The appearance of glides in Classical Arabic defective verbs.
Folia Orientalia 40. 7–33.
Fathi, Radwa. 2013. La longueur vocalique en arabe égyptien : une nouvelle conception. Paris: Université Paris
7 dissertation.
Fathi, Radwa & Jean Lowenstamm. 2016. The gender assignment pattern of French nouns – selection and
allomorphy. Morphology 26. 477–509.
Fathi, Radwa. In press a. Quelques mythes à propos du système vocalique de l’arabe du Caire. Canadian
Journal of Linguisticsin press.
Fathi, Radwa. In press b. Vowel length and prominence in Cairene Arabic. In Florian Breit, Yuko Yoshida
and Connor Youngberg (eds.),. Elements, licensing and GovernmentLondon: UCL Press.
Faust, Noam. 2017. The apophonic chain and the form of weak and strong verbs in Palestinian Arabic. The
Linguistic Review 34(1). 83–121.
Fleisch, Henri. 1954. La première forme du verbe arabe dans un parler libanais (Maâsser Beit ed-dîne).
Mélanges de l’Université Saint Joseph XXXI–5. 287–313.
 Ablaut in Cairene Arabic 75

Gadalla, Hassan. 2000. Comparative morphology of standard and Cairene Arabic. Munich: Lincom Europa.
Guerssel, Mohand. 2003. Why Arabic guttural assimilation is not a phonological process. In Stefan Ploch
(ed.), Living on the edge, 581–599. Berlin: Mouton de Gruyter.
Guerssel, Mohand. 2014. On the licensing of glides. In Sabrina Bendjaballah, Noam Faust,
Mohamed Lahrouchi & Nicola Lampitelli (eds.), The form of structure, the structure of form: Essays in
honor of Jean Lowenstamm, 85–102. Amsterdam: John Benjamins.
Guerssel, Mohand & Jean Lowenstamm. 1990. The derivational morphology of the verbal system of Classical
Arabic. Ms. Université du Québec à Montréal & Université Paris 7.
Guerssel, Mohand & Jean Lowenstamm. 1996. Ablaut in classical Arabic measure I active verbal forms. In
Jacqueline Lecarme, Jean Lowenstamm & Ur Shlonsky (eds.), Studies in afroasiatic grammar, 123–134.
The Hague: Holland Academic Graphics.
Idrissi, Ali. 2000. On Berber plurals. In Jacqueline Lecarme, Jean Lowenstamm & Ur Shlonsky (eds.),
Research in Afroasiatic grammar, 101–124. Amsterdam: John Benjamins.
Jomier, Jacques & Joseph Khouzam. 1973. Manuel d’arabe égyptien, parler du Caire. Paris: Klincksieck.
Kuryłowicz, Jerzy. 1972. Studies in semitic grammar and metrics. Wrocław: Zakład Narodowy im.
Ossolińskich.
Lahrouchi, Mohamed. 2011. Co-occurrence restrictions in the vocalic patterns of afroasiatic plurals. Folia
Orientalia 47. 89–103.
Lahrouchi, Mohamed & Philippe Ségéral. 2009. Morphologie gabaritique et apophonie dans un langage
secret féminin (taqjmit) en berbère tachelhit. Canadian Journal of Linguistics/Revue canadienne de
linguistique 54(2). 291–316.
Lowenstamm, Jean. 2011. The phonological pattern of Phi-features in the perfective paradigm of
Moroccan Arabic. Brill’s Annual of Afroasiatic Languages and Linguistics 3. 140–201.
McCarthy, John. 1979. Formal problems in semitic phonology and morphology. Massachusetts Institute of
Technology dissertation.
Mitchell, Terence F. 1956. An introduction to Cairene Colloquial Arabic. Oxford: Clarendon Press.
Mitchell, Terence F. 1962. Colloquial Arabic: The living language of Egypt. London: The English Universities
Press.
Ségéral, Philippe. 1995. Une théorie généralisée de l’apophonie. Paris: Université Paris 7 dissertation.
Ségéral, Philippe. 1996. L’Apophonie en Ge’ez. In Jacqueline Lecarme, Jean Lowenstamm & Ur Shlonsky
(eds.), Studies in afroasiatic grammar, 360–391. La Hague: Holland Academic Graphics.
Ségéral, Philippe. 2000. Théorie de l’Apophonie et Organisation des Schèmes en Sémitique. In
Jacqueline Lecarme, Jean Lowenstamm & Ur Shlonsky (eds.), Research in afroasiatic grammar,
263–299. Amsterdam: John Benjamins.
Ségéral, Philippe & Tobias Scheer. 1998. A generalized theory of ablaut: The case of modern German
strong verbs. In Fabri Ray, Ortmann Albert & Teresa Parodi (eds.), Models of inflection, 28–59.
Tubingen: Niemeyer.
Spitta-Bey, Wilhelm. 1880. Grammatik des arabischen vulgärdialectes von Aegypten. Leipzig: J.C.
Hinrichs’sche Buchhandlung.
Tomiche, Nada. 1964. Le parler Arabe du Caire. The Hague: Mouton.
Vollers, Karl & Francis C. Burkitt. 1895. The modern Egyptian dialect of Arabic. Cambridge: Cambridge
University Press.
Willmore, John S. 1905. The spoken Arabic of Egypt. London: David Nutt.
Woidich, Manfred. 1994. Cairo Arabic and the Egyptian dialects. In Dominique Caubet & Martine Vanhove
(eds.), Actes des premières journées internationales de dialectologie arabe de Paris, 493–507. Paris.
Woidich, Manfred. 2006. Das kairenisch-arabische. Wiesbaden: Harrassowitz Verlag.