24kh-raw [⌂ EvenYrbk]

Mohamad Ali Khalil

On the biconsonantal roots of Lebanese Arabic

This paper examines biconsonantal roots in Lebanese Arabic as a core part of the morphosyntactic paradigm in the language. A better understanding of the templates and patterns by which this type of root manifests allows for a precise description of their behavior and status within the system. Biconsonantal roots show a geminated C2 as well as a glide preceding the inflectional suffixes for the 1st and 2nd persons in form I, typical of final-weak roots, e.g. maddayte ‘you fem. extended’ (biconsonantal) and sallayte ‘you fem. entertained’ (final-weak) . Final-weak roots are triconsonantal roots wherein the third and last consonant is either the palatal glide ‘y’ or the labiovelar ‘w’. The data shows that the glide in biconsonantal roots (maddayte above) is a byproduct of Arabic morphophonology as dictated by phonotactics, unlike final-weak glides (sallayte above), which are a part of the root. The biconsonantal glide appears after a diphthongization process of a long vowel, which in turn is the result of lengthening an epenthetic vowel. The conclusion reached in this paper thus enables biconsonantal roots to be analyzed systemically and classified properly, morphologically and semantically as well, because the templatic morphology of Lebanese Arabic is closely related to its semantics (being a non-concatenative language, a process as simple as vowel ablaut can yield different meanings), paving the way for a more comprehensive and complete picture of the Lebanese Arabic verbal system. More specifically, the analysis allows for a rigorous distinction between what is final-weak and what is biconsonantal with an additional glide.

1. Introduction Lebanese Arabic (LA) is a Levantine variety of modern Arabic spoken in Lebanon. While its templatic morphosyntax is reminiscent of that of Classical Arabic’s, it is by no means identical, nor is LA’s system a continuation of the latter. There are many reasons to believe that LA is not a direct descendant of Classical Arabic (Owens 2006, Kaye 2007, Al-Jallad 2012, Huehnergard 2017 and others). LA, much like other modern varieties of Arabic, does not make use of all of the ‘possible’ forms, binyanim or measures (Kaye 1994, Watson 2002). These forms or templates are based on a root system, containing root radicals or core consonants. These roots are either biconsonantal or triconsonantal.1 The roots are then projected unto a pattern (or template) to 1

Quadriconsonantal roots are possibly an innovation, also cf. El Zarka 2005.

Khalil, On the Biconsonantal Roots of LA 56

yield a certain meaning. The ‘default’ type of root is that which yields a ‘sound’ or strong verb, such a root consists of three obstruents and poses no theoretical issues morphophonologically. Weak verbs, on the other hand, are the ones which contain one or more glides in their triconsonantal root, or are based on a biconsonantal root. While there are proposals that modern varieties of Arabic are based on stems (Farwaneh 2020), this paper proposes a root system for LA, which may not be shared with other varieties. Nonetheless, the strong/weak distinction holds regardless of whether the lexicon is based on roots or stems (Chekayri 2004, Farwaneh 2007). As such, LA must be studied and investigated in its own right, as the data suggests that the language behaves rather differently to what is found in Levantine varieties (see section 3), much less Classical and Modern Standard Arabic. There is a multitude of studies examining weak verbs in both Classical and Modern Standard varieties of Arabic (Schramm 1962, Mitchell 1981, Wightwick & Gaafar 2008 among others) and even some on those of Levantine Arabic (Farwaneh 2020), but LA’s unique behavior as per these verbs means it should be studied separately. LA differs from Syrian, Jordanian and Palestinian Arabic when it comes to weak verbs. The most basic of these templates is the C1aC2aC3/C1iC2iC3 template for strong triconsonantal verbs (Khalil 2023), which is the unmarked binyan 2 of LA termed form I. Form I yields a verb in its bare semantics, that is, without much of the complexity that arises with other forms. Consider the triconsonantal root k-s-r. (1) Root k-s-r ‘related to breaking’

Form I kasar ‘to break something’ This root, for instance, is used to form words ‘related to breaking’. Projecting the root unto form I results in kasar ‘to break something’. In this sense, form I is considered semantically ‘basic’. The regular behavior of this verb in the perfective, attributed to its root having three obstruents, is shown below:

2

Form I, the first binyan, measure, or weight.

Khalil, On the Biconsonantal Roots of LA 57

(2) Perfective form I of a sound triconsonantal root (three obstruents as

root radicals) Root 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl.

k-s-r ‘related to breaking’ kasar-(e)t kasar-(e)t kasar-te kasar kasar-it kasar-na kasar-to kasar-o

The data shows that the conjugation patterns for such a root in form I are regular, leaving the root, with its vocalic melody, intact. Such a strong verb poses no problems with regard to its templatic patterning (see (2) above), as the templates themselves contain three consonantal slots, i.e. C1aC2aC3(-Iinfl), with Is being the inflectional suffix. Weak verbs, however, require special attention to their pattern, as it may not fit into the regular templates.

2. Forms I and II Regular and ‘irregular’ behaviors of forms I and II in LA must be examined in order to be able to identify the internal consonantal and vocal manifestations of each verb. According to Khalil (2023), LA’s forms I and II exhibit a unique morphosyntactic situation. Khalil shows that, in LA, form II is strictly causative, meaning that all other functions which manifest using form II are only homophones thereof and not bona fide form II. Khalil shows that what is traditionally described as denominative, estimative, intensive or other functions of form II are actually allomorphs of form I. Here are the possible cases: (3) If Form I: only CaCaC → No causative form II, intensive allomorph of form I: CaCCaC If Form I: only CiCiC → Form II: CaCCaC If Form I: both CaCaC and CiCiC, and both have the same meaning → Form II: CaCCaC If Form I: both CaCaC and CiCiC, and both have different meanings → No causative form II, intensive allomorph of form I: CaCCaC If Form I does not exist as CaCaC or CiCiC, then CaCCaC (homophonous with form II) is form I, the latter essentially being the traditionally denominative form II verb.

Khalil, On the Biconsonantal Roots of LA 58

The key takeaway from these findings is that not every geminate C2 pattern, or more abstractly a CC medial pattern (i.e. CaCCaC), is to be considered form II. This will prove especially helpful when looking at the data for weak verbs. That said, Khalil only examines triconsonantal roots wherein all three root radicals are obstruents. The weak/biconsonantal patterns are yet to be described.

3. Final-weak and Biconsonantal Verbs Final-weak verbs are verbs which include a glide in their third and final root radical C3. Example (4) illustrates such a verb. (4) Root ɣ-ṭ-y ‘related to covering’ Form I ɣaṭṭā ‘to cover’ 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl.

ɣaṭṭayt ɣaṭṭayt ɣaṭṭayte ɣaṭṭā ɣaṭṭit ɣaṭṭayna ɣaṭṭayto ɣaṭṭo

The fact that this verb is not necessarily form II in LA is consolidated by the above discussion in Khalil (2023), since not every geminate C 2 pattern must be form II. Furthermore, it is clear that ɣaṭṭā is ‘to cover’ and not ‘to make someone cover’, which indicates that this verb is not causative and therefore cannot be considered of form II. The verb in (5) is an example of a biconsonantal root which shares the first two root radicals of (4), leading to a homophonous situation:

Khalil, On the Biconsonantal Roots of LA 59

(5) Root ɣ-ṭ ‘related to immersion, related to falling (by extension, to sleep)’, Form I ɣaṭṭ ‘to dip’ 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl.

ɣaṭṭayt ɣaṭṭayt ɣaṭṭayte ɣaṭṭ ɣaṭṭit ɣaṭṭayna ɣaṭṭayto ɣaṭṭo

If not for the 3rd singular masculine, there would have been no morphophonological distinction between the verbs in (4) and (5), and, taken at face value, one might mistake them for being of the same root. Example (6) presents another verb in order to avoid the homophony that arises in (4) and (5). (6) Root m-d ‘related to extension’ (Biconsonantal/geminate-final root) Form I madd ‘to extend’ 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl.

maddayt maddayt maddayte madd maddit maddayna maddayto maddo

The immediate issue to tackle here is the presence of a palatal glide in a biconsonantal/geminate-final root. If the glide is considered a part of the verb/root, then the pattern obtained here would be C1aC2C2aC3-Cinfl. If the glide, however, is not part of the verb/root, then the pattern is C 1aC2C2aGCinfl. If we argue for the former, i.e. the glide being a part of the verb/root, this would mean that biconsonantal verbs and final-weak verbs are one and the same. Chekayri and Scheer (1996, 2003, 2005) argue for such an analysis for Classical Arabic. Through apophony, they describe the nature of the glides in weak verbs, and then maintaining that these glides are not part of the root, but the result of anarchy. One of their arguments is that some hollow verbs may be found with both /w/ and /y/. Hollow verbs are those which have a glide in the medial (C2) root radical. Chekayri and Scheer talk about free variation in these hollow verbs, that it makes no difference whether one says /w/ or /y/. This is

Khalil, On the Biconsonantal Roots of LA 60

not the case in LA, free variation as such in the verbal system does not exist. Moreover, minimal pairs can be found as in (7): (7) nawwam ‘to hypnotize someone’ nayyam ‘to put someone to sleep’ Whether these two verbs in (7) share the same root is doubtful, but this, at least, tells us that not incorporating glides into roots/stems does not work for LA, as Chekayri and Scheer’s proposal argues that glides do not factor in the surface form of Arabic verbs. This is clearly not the case in LA; whether a verb has a palatal or a labial glide is essential to its semantics: LA glides are phonemic, (7) shows a minimal pair. Farawneh (2020) offers an alternative analysis which shows glides in root radical positions, but shows that these glides do not surface in any form in Levantine Arabic. This is where LA deviates from other Levantine varieties. The glide appears in both final-weak and biconsonantal verbs, which is what this paper attempts to explain. Furthermore, Farawneh’s analysis is predicated on ranking constraints in an Optimality Theoretical framework, most of which do not exist for LA (qua constraints). To cite one example of these constraints, Farwaneh says that Levantine imperative verbs whose stem begins with a consonant cluster such as /ktub/ undergo “mandatory consonantal epenthesis” to become [ʔuktub] (foot binarity). This is not true in the case of LA, no such constraint exists, and as a matter of fact, the imperative of the root k-t-b ‘related to writing’ is indeed ktob. 3.1 Digression on Final-weak Verbs Example (4) above with a final-weak verb, while being almost identical to (5), which shows a biconsonantal verb, does not present the full picture regarding final-weak verbs. It is this geminated medial consonant in a final-weak verb that must be explained. We know that geminating the medial consonant of a verb yields a verb of form II. Khalil (2023) argues that in LA, since form II is a ‘higher’ derived form, it must have a corresponding form I. Any verb which lacks a form I (CaCaC for triconsonantal roots), cannot have a causative form II, and the verb obtained via geminating the middle consonant (i.e. CaCCaC) is an allomorph of form I. Repeated here are the possible form I patterns for triconsonantal roots in LA: (8) a. CaCaC as in ʔatal ‘to kill’ b. CiCiC as in ʔitil ‘to die’ c. CaCCaC (denominative/intensive form I) as in ʔattal ‘to massacre’

Khalil, On the Biconsonantal Roots of LA 61

A triconsonantal root may have all patterns, i.e. (8ab) and (8c), or two of them (8a) and (8c) or (8b) and (8c), or just (8c) in the case of denominative form I (homophonous with form II, the latter of which is said to be exclusively causative). Based on the example in (4), one possible pattern is CaCCā, and CaCCaGIs when conjugated, which means that, since the glide is part of the root, this pattern for final-weak verbs actually corresponds to CaCCaC for triconsonantal roots. Example (9) shows a different final-weak verb with a pattern that does not have a geminate medial consonant: (9) Root ʔ-l-y ‘related to frying’ Form I ʔalā ‘to fry’ 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl.

ʔalayt ʔalayt ʔalayte ʔalā ʔalit ʔalayna ʔalayto ʔalo

As with previous examples, the 3rd person conjugations notwithstanding, this verb is of the pattern CaCaGIs, corresponding to CaCaC. The last type of final-weak verb pattern is given in (10): (10) Root ħ-l-G3 ‘related to sweetness’ Form I ħilē ‘to become sweet’ 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl. 3

ħlīt ħlīt ħlīte ħilē ħilit ħlīna ħlīto ħilo

The identity of the glide, shown here as G, is unclear, it is possible that it could be either, or that we have two different roots with each of them.

Khalil, On the Biconsonantal Roots of LA 62

The verbal pattern in (10) is reminiscent of the CiCiC pattern of triconsonantal verbs in form I. Based on this, we may lay down the possible form I patterns for final-weak verbs: (11)

a. CaCā as in ʔalā ‘to fry’ b. CiCē as in ħilē ‘to become sweet’ c. CaCCā as in ɣaṭṭā ‘to cover’

4. The Problem The possible patterns for final-weak and biconsonantal verbs will be repeated here: final-weak CaCa(C), CiCe, CaCCaC (form I, cf. (4) above), CaCCaC (form II). Some words on the last pattern must be repeated here: According to Khalil (2023), LA verbs can have a form II provided they have a form I (see sections 2 and 3), this means that it is possible to obtain a CaCCaC pattern for finalweak verbs which is form II (distinct from what is found in (4)). Example (12) illustrates the difference between a final-weak form II verb and a form I CaCCaC verb such as the one in (4): (12) Root ħ-l-G ‘related to sweetness’ Form I ħilē ‘to become sweet’ Form II ħallā ‘to make s.t. sweet’ 1st sg. 2nd sg. masc. 2nd sg. fem. 3rd sg. masc. 3rd sg. fem. 1st pl. 2nd pl. 3rd pl.

ħallayt ħallayt ħallayte ħallā ħallit ħallayna ħallayto ħallo

Note the pattern obtained for form I in (12) for this root, which is distinct from the causative form II (CaCCaC). This means that biconsonantal verbs and final-weak ones are homophonous for two morphophonological patterns:

Khalil, On the Biconsonantal Roots of LA 63

(13)

A. Biconsonantal form I (CaCCaC) = Final-weak form I (CaCCaC, (4)). B. Biconsonantal form I (CaCCaC) = Final-weak form II (CaCCaC from a non-geminate form I, see (11)).

(14) illustrates the current situation. (14) Venn diagram showing forms where biconsonantal and final-weak roots are homophonous. Biconsonantal Verbs

Final-weak Verbs

Form I: maddayt → CaCCa(?)Cinfl Form I: maddayt → CaCCa(?)Cinfl

Form I: ʔalayt → CaCaCCinfl Form I: ħlīt → CiCiCinfl

Form I: ɣaṭṭayt → CaCCaCCinfl

Form II: ħallayt → CaCCaCCinfl

A minor issue surfaces here: Biconsonantal verbs of form I were compared to and found to be similar to final-weak verbs of form I and II. What about biconsonantal verbs of form II? While it is out of the scope of this paper to go into the morpho-semantics of biconsonantal verbs, especially ‘higher’ forms, the overarching pattern to be considered here is CaCCaGCinfl (see above), and it will be assumed that, if form II for biconsonantal verbs exists as such, then it will either be of the pattern CaCCaGCinfl, or otherwise be irrelevant to the study at hand. That said, the focus here is why and how the glide appears in biconsonantal verbs.

Khalil, On the Biconsonantal Roots of LA 64

5. Other Varieties of Arabic El Zarka (2005: 373) provides an account of the consonantal composition of the CvCCvC pattern, with different possible C permutations. In her account, she mentions five possible combinations, listed as follows: “katkat (with a reduplicated root or syllable) katkab (with a doubled C1) kattab (with a doubled (i.e., geminated) C2) katbab (with a doubled C3) C1C2C3C4” Whereas this list captures most of the possible patterns, it does not recognize C1vC2C2vG as a manifestation of the same CvCCvC template. El Zarka only discusses biconsonantal roots in light of reduplication, which yields a CvCCvC pattern, but not the aforementioned C1vC2C2vG. This may be due to two reasons. First, glides can be seen as an extension of vocalic realization. For example, the root for names s-m-G appears as sammā in the 3rd singular masculine (taken to be the default pattern). In the latter, the final glide can be seen transforming into a long ‘ā’ and thereby losing its consonantal status. The second reason C1vC2C2vG may have been overlooked is closely tied to the first reason. C1vC2C2vG does not occur as a pattern in isolation, that is to say that it does not occur as a conjugation of the 3 rd singular/plural masculine/ feminine verb, taken to be the default,4 but only manifests when it is conjugated for the 1st and 2nd person singular and plural (cf. (6) above). This means that C1vC2C2vG only appears as C1vC2C2vG-t/ C1vC2C2vGtī/ C1vC2C2vG-na/ C1vC2C2vG-tō, i.e. this glide can only appear pre-consonantally and never word-finally. If the latter was the reason for overlooking C1vC2C2vG patterns, then analysis is still needed to account for the glide that appears in biconsonantal root verbs in the 1st and 2nd singular and plural conjugations. 6. The Epenthetic Vowel To start looking into the issue, let us consider the inflectional suffixes for the 1st, 2nd and 3rd person singular and plural perfective patterns. 4

CvCvC/CvCCvC patterns take the 3rd singular masculine as their default conjugation, in that the inflectional suffix is the zero morph ∅.

Khalil, On the Biconsonantal Roots of LA 65

(15) Lebanese Arabic perfective inflectional suffixes 1. 1st singular

-t

2. 1st plural

-na

nd

3. 2 singular masculine

-t

4. 2nd singular feminine

-te

5. 2nd plural

-to

6. 3rd singular masculine

∅

7. 3rd singular feminine

-it

8. 3rd plural

-o

The input for biconsonantal roots consists of two root radicals, which for the default pattern, i.e. the 3rd singular masculine, yields a C1aC2C2 pattern, since the inflectional suffix is zero. Consider the rest of the suffixes, however. Taken at face value, each of the suffixes in (15), with the exception of the 3 rd person suffixes, would yield a C1aC2C2Cs/C1aC2C2Csv5 pattern. And while a consonant cluster which includes three consonants followed by a vowel is attested in LA nominals (albeit dispreferred), it does not seem to be a licensed phenomenon in LA verbs (also cf. El Zarka 2005: 382). Thus, it appears that a vowel is inserted somewhere in the consonant cluster in order to split it. This is strikingly similar to what is found in Classical Arabic verbs of a biconsonantal root. In Classical Arabic, C1vC2C2Cs is unlicensed and thus the pattern shows a vowel between the geminate/copy C2. This means that Classical Arabic C1vC2vC2Cs(v) pattern looks as if the second vowel is epenthetic and only serves as a cluster-breaker, adhering to the CvCvC pattern found in regular, basic verbs. In LA, however, the vowel appears between the root consonant and the inflectional suffix, yielding a C1vC2C2veCs pattern. This analysis details the mechanism which inserts an epenthetic vowel. What remains to be discussed, therefore, is the exact environments that trigger this vowel. Haddad (1984) distinguishes between two types of epenthesis. The first, which he terms ‘epenthesis’, is an optional epenthetic vowel between a series of two consecutive consonants. The second process, i.e. the one under study in this paper, he calls ‘vowel insertion’. To Haddad, it might have been important to make such a distinction, but the nature of the phonological process, i.e. what qualifies as epenthesis and what does not, is out of the scope 5

‘Cs’ is an inflectional suffix.

Khalil, On the Biconsonantal Roots of LA 66

of this current paper, and thus the vowel will still be considered an instance of epenthesis. Haddad argues that the vowel is inserted in what he calls “stranded syllables”, which is a consonant cluster where the coda of a certain syllable is found as CCC#. Such a cluster is illicit, and the epenthetic vowel is inserted after the penultimate consonant to break it up, leading to an acceptable structure. This, in turn, confirms that the geminate C2 in biconsonantal roots is underlying. If šam ‘to smell’ was underlyingly š-m-Cs, then the verb would have been *šamte/ and only a long /ē/, which could explain the similarity in how biconsonantal verbs function in LA as well as Egyptian Arabic (see (17a)). It is then perhaps unsurprising that when the epenthetic vowel sits in a closed syllable in Egyptian Arabic, it raises to /i/ as in šammitha ‘I smelled it (fem.)’. Furthermore, as to why this epenthetic vowel undergoes lengthening, rather than raising to /i/ is precisely because such a remedy to avoid apocope is preempted by the 3rd sg. fem. ending -it. The epenthetic vowel, then, cannot be /i/ as this would directly clash with another inflectional suffix already present and established for the 3rd sg. feminine. Phonetically, Kelly (2020) finds that unstressed vowels are longer if found in phrase-final positions, as well as if the stressed vowel is phonologically short (cf. (18)), and when in an open syllable (cf. (17)). These findings, while

Khalil, On the Biconsonantal Roots of LA 69

phonetic in nature, show that the epenthetic vowel in question has viable grounds for lengthening. Example (19) repeats (18) and traces the path of the vowel from an epenthetic short unstressed vowel to a long stressed one: (19)

šắmmt (illicit cluster) → *šắmmet (short, unstressed epenthetic vowel)

→ šammḗt (long, stressed epenthetic vowel).

The ensuing glide in Lebanese in (17a), then, is explicable through a process of diphthongization, wherein the long vowel undergoes ‘fission’ and splits into two smaller elements, the short vowel and the glide. There is variety, it appears, even in LA. Speakers of different dialects of Lebanese have different preferences when it comes to the pronunciation (or lack thereof) of the glide. A speaker of Beiruti Lebanese would say šammēt ‘I smelled’ but šammayto/šammayta ‘I smelled it’, whereas a speaker from the North East would pronounce the glide in both words, i.e. šammayt ‘I smelled’ and šammayto/šammayta ‘I smelled it’. Therefore, for Beiruti speakers, the epenthetic vowel can remain intact as a long vowel if it is part of a closed syllable, i.e. šammēt. However, when found as part of an open syllable, the long vowel diphthongizes. The process could then be illustrated as follows: (20) The path of the epenthetic glide in Egyptian, Syrian and Lebanese Arabic ROOT: C1-C2 Conjugation → illicit cluster *C1v́C2C2Cinfl Epenthesis

C1vC2C2v̆́Cinfl(h)v (here’s where Egyptian stops; cf. (17b))

Lengthening

C1vC2C2v̄́Cinfl (here’s where Syrian stops; cf. (17a))

Diphthongization C1vC2C2v́GCinfl (here’s where Lebanese is at when the inflectional suffix gives rise to an open syllable; cf. (17a))

Khalil, On the Biconsonantal Roots of LA 70

This analysis, in turn, has implications for final-weak roots. Since biconsonantal and final-weak roots may be misconstrued as having the same pattern, we must turn to the latter again using the above analysis. To properly see finalweak roots in context, consider the data in (21) for the root s-m-G ‘related to names’. (21) Lebanese 1. sammáyto ‘I named him/it (masc.)’

Syrian sammḗto

Egyptian sammḗto

2. sammáyta ‘I named her/it (fem.)’

sammḗta

sammítha

Here, the exact same process happens but in reverse order. We start off with the original glide, it then undergoes ‘fusion’, in that it forms a diphthong with the preceding vowel, monophthongizes and then shortens. Consistent with the process of epenthesis described above, the process at hand can be illustrated using the same diagram, only inverted. The inversion here means that the glide in the root and the subsequent verbs is canonical (unlike biconsonantal verbs), it forms a diphthong with the preceding vowel (in Lebanese), this diphthong monophthongizes to a long vowel (Syrian) and shortens in closed, non-final syllables (Egyptian). (22) The path of the canonical glide of Lebanese, Syrian and Egyptian Arabic ROOT: C1-C2-G

Canonical glide → diphthong

C1vC2C2v́GCs (here is where Lebanese stops)

Monophthongization

C1vC2C2v̄́Cs (here is where Syrian stops)

Shortening

C1vC2C2v̆́Cs(h)v (here is where Egyptian is at)

The verbal pattern, i.e. C1aC2C2aC3 has no bearing on the process itself, which is found even in C1vC2vC3 patterns. Consider the root ɣ-l-G ‘related to boiling’.

Khalil, On the Biconsonantal Roots of LA 71

(23) The verb ‘to boil’ in Lebanese, Syrian and Egyptian Arabic Lebanese 1. ɣaláyto ‘I boiled him/it (masc.)’

Syrian ɣalḗto

Egyptian ɣalḗto

2. ɣaláyta ‘I boiled her/it (fem.)’

ɣalḗta

ɣalítha

The data in (23) shows that the status of C 2, whether geminated or a singleton, does not affect the outcome ultimately. The fusion and shortening processes still happen according to the scheme illustrated above and repeated for C1vC2vC3 patterns: (24) Final-weak verb with no geminate/double medial consonant ROOT: C1-C2-G

C1vC2vGCs (Lebanese)

C1vC2v̄Cs (Syrian)

C1vC2v̆Cs(h)v (Egyptian)

The retention of the glide and its loss are systemically determined, then, according to the dialect of Arabic spoken. This leads us to broader generalizations: I. II.

III.

Lebanese Arabic retains its glides in open syllables, and has epenthetic vowel diphthongization in open syllables. Syrian Arabic monophthongizes diphthongs formed with glides in closed syllables, and has epenthetic vowel lengthening in open syllables. Egyptian Arabic monophthongizes diphthongs formed with glides and shortens the long monophthongs in closed syllables, and has epenthetic vowels retaining their original short quality in closed syllables, and lengthening in open syllables.

8. Conclusion This paper analyzed the phenomenon of the appearance of a glide which seems to occupy a consonantal position in the biconsonantal roots of Lebanese

Khalil, On the Biconsonantal Roots of LA 72

Arabic. This glide was confirmed not to be a part of the biconsonantal root itself, but the manifestation of an underlying phonological necessity. The significance of this finding lies in the confirmation that the glide of biconsonantal roots is different from the glide of final-weak roots, which is the underlying third root radical in final-weak verbs. For this reason, data from LA was analyzed alongside data from Syrian and Egyptian Arabic. Verbs of biconsonantal roots in all three varieties of Arabic exhibit either an epenthetic vowel or a glide. This phenomenon was interpreted as a process of epenthesis, initially, then lengthening and diphthongization, leading to the formation of the glide ultimately. Epenthesis is a process which breaks up illicit consonant clusters of three consonants, inserting a vowel between the final and penultimate consonant. This epenthetic vowel subsequently lengthens and splits into a short vowel and a glide in LA. This not only reveals key phonological information about LA, but also shows how biconsonantal roots are treated in the morphosyntactic paradigms of the language. Biconsonantal roots can, therefore, be categorized akin to triconsonantal and final-weak roots, which have systemic morphological patterns, i.e. binyanim.

References Abdo, Daud A. 1969. On stress and Arabic phonology: A generative approach. Khayats. Birkeland, Harris. 1952. Growth and structure of the Egyptian Arabic dialect. Oslo: Jacob Dybwad. Brame, Michael K. 1971. Stress in Arabic and generative phonology [Review of On Stress and Arabic Phonology, a Generative Approach, by D. A. Abdo]. Foundations of Language, 7(4): 556–591. http://www.jstor.org/stable/25000563 Chekayri, Abdellah. 2004. Weak verbs in Arabic. In Investigating Arabic, Brill. 65–83. https://doi.org/10.1163/9789047405085_006 Chekayri, Abdellah and Tobias Scheer. 2003. The appearance of glides in Classical Arabic defective verbs. Folia Orientalia, 39: 7–33. Chekayri, Abdellah and Tobias Scheer. 2003. The appearance of glides in Classical Arabic defective verbs. Retrieved from http://tscheer.free.fr/papers/Chekayri%20&%20Scheer %202003%20-%20The%20appearance%20of%20glides%20in%20Classical%20Arabic %20defective%20verbs.pdf Chekayri, Abdellah and Tobias Scheer. 2005. A biliteral approach to weak verbs in Arabic. Handout retrieved from http://tscheer.free.fr/papers/Hdt%20Chekayri-Scheer%20405%20Illinois.pdf Cowan, William. 1970. The vowels of Egyptian Arabic. Word, 26(1): 94–100. https://doi.org/ 10.1080/00437956.1970.11435584 El Zarka, Dina. 2005. On the borderline of reduplication: Gemination and other consonant doubling in Arabic morphology. In B. Hurch (Ed.), Studies on Reduplication, 369–394. Berlin, Boston: De Gruyter Mouton. https://doi.org/10.1515/9783110911466.369

Khalil, On the Biconsonantal Roots of LA 73

Farwaneh, Samira. 2007. Strong/weak verb asymmetry in Arabic: A consequence of OP faithfulness. WECOL 2007, 29. Farwaneh, Samira. 2020. Non-sound verb inflection in Arabic: Allomorphic variation and paradigmatic uniformity. Morphology, 30(1): 61–89. https://doi.org/10.1007/s11525-02009350-w Haddad, Ghassan. 1984. Problems and issues in the phonology of Lebanese Arabic. University of Illinois at Urbana-Champaign. Kaye, Alan S. 1994. Formal vs. informal in Arabic: Diglossia, triglossia, tetraglossia, etc., polyglossia—Multiglossia viewed as a continuum. Zeitschrift für Arabische Linguistik, 27: 47–66. http://www.jstor.org/stable/43525622 Kaye, Alan S. 2007. Classical and colloquial Arabic archaisms. In M. Alhawary & E. Benmamoun (eds.), Approaches to Arabic linguistics: Presented to Kees Versteegh on the occasion of his sixtieth birthday, Brill. 595–605. https://doi.org/10.1163/ej.9789004160156.i-762.166 Khalil, Mohamad A. 2023. On the morphosyntax of Lebanese Arabic: Form II, its manifestations and functions. MA Thesis, Eötvös Loránd University. Retrieved from http://seas.elte.hu/overseas/overseas23khalil.pdf Kelly, Niamh Eileen. 2020. Phrase-final lengthening across segments in Lebanese Arabic. Proc. Mtgs. Acoust. 42(1): 060002. https://doi.org/10.1121/2.0001385 McCarthy, John. 1979. Formal problems in Semitic phonology and morphology. Unpublished PhD dissertation, Cambridge, MA: MIT. Mitchell, Timothy. 1981. The phonology of weak verbs: A simple diagram of rules. al-’Arabiyya, 14: 11–18. Moscati, Sabatino, Anton Spitaler, Edward Ullendorff, and Wolfram Von Soden. 1980. An introduction to the comparative grammar of the Semitic languages (2nd ed.). Wiesbaden: Harrassowitz. Owens, Jonathan. 2006. A linguistic history of Arabic. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199290826.001.0001 Scheer, Tobias and Abdellah Chekayri. 1996. The apophonic origin of glides in the verbal system of Classical Arabic. In Jacqueline Lecarme, Jean Lowenstamm and Ur Shlonsky (eds.), Studies in Afroasiatic grammar, La Hague: Holland Academic Graphics. 62–76. Schramm, Gene M. 1962. An outline of Classical Arabic verb structure. Language, 38(4): 360– 375. https://doi.org/10.2307/410672 Watson, Janet C. E. 2002. The phonology and morphology of Arabic. Oxford University Press. Wightwick, Jane and Mahmoud Gaafar. 2008. Arabic verbs. McGraw-Hill Education. Mohamad Ali Khalil Eötvös Loránd Tudományegyetem ORCID 0000-0002-2038-1651 klaitmohammad@gmail.com