User:M. I. Wright/Lebanese Arabic verbs

In progress.

This is a rundown on verbs and their function in Lebanese Arabic. A large portion of the page also applies to North Levantine and even Levantine Arabic in general, but it only explicitly concerns Lebanese for now.

Introductory notes
This page doesn't discuss phonology unless it has a direct effect on verb morphology or derivation. So, in the pursuit of 'not having to care about phonology specifics', terms are Romanized using a sort of Hans Wehr system that's been modified with the concerns outlined in WT:CON AR:
 * are used rather than Hans Wehr.
 * Stress is marked explicitly, albeit with bolding rather than with an accent.
 * The letter is used to represent the tā marbūṭc, a feminine suffix with a couple of realizations and whose particulars aren't a focus of this document. (It can be either  or  depending on its environment, as well as  if syntactically the possessee in a genitive construction.)
 * A ligature,, is used to indicate "any semivowel": either w or y. (It's actually a ligature of v and y, but the resemblance to wy is too convenient to pass up.)
 * The dotless is used for, an "indeterminate" high/mid vowel that can either be rounded or not. (In other words, alteration between damma and kasra.) While the choice of vowel is consistent per speaker, it can vary wildly between different speakers and regions — and, despite being a phonemic distinction, it carries zero semantic load. All of that makes it best to cover both variants with one stone without worrying about specifics.
 * In keeping with that, and with the Hans Wehr convention of representing emphatic consonants with an underdot, is used to represent a high vowel that some speakers round because of its proximity to an emphatic. The difference between the previous case and this one is that the former alteration is completely arbitrary and not phonologically conditioned at all, whereas this rounding only occurs as a sort of emphasis-assimilation.
 * is used for all other historic high vowels, and the schwa is used for epenthetics if needed. (Cowell's A Reference Grammar of Syrian Arabic differs here: for example, he transcribes the basic triliteral past verb in -i- as fiʕel. At least in Lebanese, however, it's impossible for i to be stressed without becoming e, and since both come from the same vowel historically anyway, it seems like a safer bet to make it fiʕil and avoid possible inaccuracies.)
 * Note that the is not intended to represent a sound anything like . The actual vowel it represents is something like  depending on its phonemic context, but (as with ) the rules for this variation can themselves vary enough to make it not worth representing its actual value in transcription. More importantly, these epenthetic vowels share the characteristic that they're only inserted to break up a coda consonant cluster and can be deleted under certain constraints, so it feels apt to represent them with a character often connoting "no well-defined vowel in particular".
 * The historic diphthongs and  are transcribed as  and . This document avoids the "either aw or ō" dance.

Verbs are demonstrated using the dummy root. The esoteric verb-form numbers are supplemented as much as possible with actual template examples, although they're used as is when referring to fuṣḥa. Speaking of which, English doesn't have a 100%-accurate way to refer to the pre-Modern stage of Arabic that gave way to the contemporary varieties: "Classical Arabic" is apparently a sister evolution, "Qur'ānic Arabic" is a specific dialect of antiquity, "Old Arabic" is too old, and "Modern Standard Arabic" is today’s artificial-amalgam reconstruction that has no business being an ancestor of anything. To sidestep all of that, this page uses the generic Arabic term fuṣḥa.

Overview
Lebanese Arabic is still very much in touch with the "root system" of derivation that characterizes Arabic and Semitic languages: verbs are created by slotting three or four discontinuous consonants into a template that realizes them as an actual verb, and every such template is distinct in terms of conjugation and meaning. An individual template is called a "verb form", and its traits are gone over below.

As in Modern Standard Arabic, the properties of the root itself also have some bearing on its realization as a verb. Some roots consist of only two consonants rather than three, with the second being doubled to mimic a three-consonant root; this causes some interesting things to happen, because those two identical consonants like to stick together when the root is realized. Roots can also have a "weak consonant", w or y, which for historical reasons gets elided in certain contexts (with unexpected results), or they can contain a glottal stop to which the same happens. Each distinct 'kind' of root is given separate consideration below.

Generalities of conjugation
Verbs are conjugated for five things: Conjugation can be summarized by looking at how the final three items are conjugated for in each tense, and additionally comparing the nonpast subjunctive to the nonpast indicative conjugation. (The "tenses" may actually be perfective/imperfective aspects, but it's confusing, and nothing really changes when they're referred to as tenses instead.)
 * 1) Tense (past, nonpast)
 * 2) Mood (subjunctive, indicative) — only on nonpast verbs
 * 3) Person (first, second, third)
 * 4) Number (singular, plural)
 * 5) Gender (masculine, feminine) — not in the first person or in the plural

Conjugations are formed using suffixes in past-tense verbs.

Some speakers lower the u and i vowels to o and e, which sometimes renders stress the only thing telling apart a 3sg.f conjugation from a 1sg/2sg.m conjugation. However, the elision of the latter two's epenthetic (in contrast to the reluctant deletion of the first's i) can still serve as a distinguishing mark.

All first- and second-person suffixes morphologically begin with a consonant, but the rest are vowel-initial (besides the 3sg.m suffix, as there isn't one). This has noticeable effects on stress and on the form of certain verbs, both of which will be discussed in detail later.

In the Arabic script, the plural suffixes -u are counterintuitively spelled ـوا rather than ـو for historical (but not etymological) reasons. Many speakers, or even most speakers, do in fact use the latter spelling. On the other hand, the 2sg.f form is spelled ـتِي rather than ـتِ in defiance of historical convention — this is because the contemporary Arabic dialects don't preserve the old word-final short vowels (which were represented using diacritics), in their place shortening the original long vowels. Any ostensible word-final "short vowel" in a contemporary dialect is therefore really a "dialectal long vowel" underlyingly, so it should be written as such.

The subjunctive conjugations, which are the least-marked nonpast conjugations, are formed using both prefixes and suffixes:

Notice that, besides the anomalous disappearance of the 1sg glottal stop, all prefixes are vowelless when on a verb that starts with a single consonant (like ) or a vowel (like ). But if the verb starts with a consonant cluster, all prefixes surface with a vowel ᴉ-. About this vowel:
 * If the verb's first consonant is, the ᴉ- is allowed to become a-, presumably due to some quality of the pharyngeal that likes to induce openness in the vowel. (An interesting phenomenon related to this is looked at below, when discussing voweling patterns on verbs.)
 * Examples:
 * Additionally, if the verb's first consonant is w, the result of *iwC is actually an initial vowel ū, and the result of *iwV is an initial sequence uwV.
 * Examples: ,
 * Diachronically, however, this might instead be compensated for by assimilating the semivowel, which here results in . This process is much more common in Syria than in Lebanon.
 * The rounded variant of the ᴉ- occurs when there's any emphatic consonant in the verb, not just one immediately following.
 * Examples:, , . (See the conjugation tables below for where this rounding occurs in other verb forms.)
 * Some verbs with a historic initial (or a hamza analyzed as one) may also induce the rounded variant.
 * Finally, some speakers exhibit rounding without any surrounding emphasis to condition it at all, resulting in conjugations like . This is rare, however.

Lastly, the indicative mood is indicated with a prefixed accompanying the subjunctive conjugation:

The same notes as above about the ᴉ- vowel apply. Notice the nasal assimilation of the prefix in the plural forms. Also, note that marking the indicative mood is only one of this prefix's purposes; see its section below for a more-thorough description.

Verb forms
Epenthetics are not represented below (that is, clusters are written as clusters), but they can be inferred regularly.

Biliteral & triliteral
As mentioned earlier, biliteral roots end up acting triliteral anyway, because the second radical gets doubled to mimic a third one. However, this results in noticeably different conjugation paradigms between biliteral and triliteral verbs: the doubled consonant likes to stay geminate when realized in an actual verb. This is even-more prononounced in dialects of contemporary Arabic than in fuṣḥa, as this geminate consonant is no longer allowed to break up into a vowel-separated pair when the base verb is required not to end in a final vowel: compare to, not. This innovation does result in ambiguity, however: is the 1sg/2sg.m conjugation of both  and, whereas the former would be  in a more-fuṣḥa-ish conjugation paradigm.

Triliteral:
(Section anchor: I_Triliteral)


 * Nonpast:
 * Passive:
 * Active participle:
 * , rarer (etymologically distinct)
 * Passive participle:
 * Verbal noun:
 * , rarer (etymologically distinct)
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:
 * Verbal noun:
 * Verbal noun:

This is the only verb form with semantically significant variation in voweling, where in each tense there are two possible patterns that a verb can take. These are...
 * Past:
 * , henceforth a
 * , henceforth ᴉ
 * Nonpast:
 * , henceforth ı
 * There's also specific kind of verb that can only look like (notice the singly dotted i). This is henceforth i.
 * , henceforth a

Usually, the correspondence between past–nonpast voweling is a–ı and ᴉ–a, but:
 * a–i in particular (notice the i rather than an ı, indicating no regional variation whatsoever between ) can occur when the verb is from a diachronic Form IV verb, half-preserving its original voweling in this way. See Form IV below.
 * a–a mostly exists in three cases...
 * Rarer verbs or MSA loans, like, , and.
 * However, the last case (originally hamza-final verbs) is being or has been regularized to ᴉ–a: a common variant of that particular verb is.
 * Verbs with a final pharyngeal. This is the same reason why the ᴉ- vowel of conjugational prefixes is allowed to become a- when a verb's first consonant is ; some quality of the pharyngeal seems to induce openness in a preceding vowel, which causes both tenses to tend toward the variant in -a- here. There still are i-a cases here like and, but the only case where a pharyngeal-final verb can follow the nonpast i pattern is when it's from a historic Form IV verb; again, see below.
 * ᴉ-ı isn't as easy to predict. It exists sporadically, appearing in random verbs like, and it varies by speaker at that. It may be becoming more common with final-weak verbs [TODO: explain]

The shining star of all this variation is the verb, which can take both voweling patterns in each tense, as well as optionally conjugating with nonpast prefixes a- rather than ᴉ- due to the initial pharyngeal!

Sound conjugations (no weak radicals)
(Section anchor: CCC)

Notice that the unstressed nonpast -ı- always collapses to -i- when there's a following syllable, after which it's subject to the "forbidden sequence" rule and may be deleted. On the other hand, the nonpast -a- remains no matter its environment.

Hamzated conjugation (initial radical ʔ)
(Section anchor: 2CC)


 * Nonpast:, ᴉ–a unattested
 * Passive:
 * Active participle:
 * Active participle:
 * Active participle:

This only applies to two verbs in the language: and, as these are the only two natively preserved hamza-initial verbs. Compare, which retains the initial glottal stop in all conjugations due to being a later MSA reborrowing.

Hollow conjugations (middle radical ꝡ)
(Section anchor: CWC)


 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * , rarer
 * Verbal noun:
 * Passive participle:
 * , rarer
 * Verbal noun:
 * Verbal noun:
 * Verbal noun:

These verbs are characterized by an in the past tense, and any of the three long vowels    in nonpast conjugations. The nonpast verb in is much rarer than the other two, but it can arise from either root semivowel. Otherwise, a medial y in the root will yield a verb with nonpast, and a root-medial w will yield a verb with nonpast.

As shown in the tables, however, the past-tense of a hollow verb is untenable in would-be-superheavy syllables that are induced by any consonant-initial conjugational suffix. It's therefore replaced by a short vowel ᴉ.
 * Regionally and historically, this is actually split into two cases based on the verb's nonpast form: a verb with nonpast would have a short vowel u in the past tenses, and the short vowel ᴉ would only surface for a verb in nonpast  or . One of the shifts that distinguishes North Levantine from its neighbor is a near-ubiquitous merger of both of these two short vowels to ᴉ, regardless of the corresponding root semivowel.
 * Examples: but ;  but
 * The shortening of the ā, by the way, has no basis in synchronics — it's a relic of fuṣḥa ' s phonotactics, which explains why the 3sg.m form is allowed to retain a long ā despite still being a just-as-superheavy syllable. That is, it descends from what was diachronically only a heavy syllable: in fuṣḥa, verbs along the lines of had a mandatory final a, giving  with the structure CVV.CV. So, because the initial syllable wasn't a superheavy *CVVC but instead just a heavy CVV, it was fine keeping its long vowel, and this transfered into contemporary Arabic as well. But contemporary Arabic dialects often lose some marks of fuṣḥa ' s aversion to superheaviness, so the CVVC syllable in  that results from the loss of the final short a is allowed to remain superheavy  — and only originally-shortened superheavy syllables continue to surface short.

For some reason, it appears that all verbs with a root-medial semivowel are traditionally considered "hollow" and thus "weak" in Arabic. This is silly, because there are verbs that simply treat this semivowel as a proper root consonant — like — and therefore conjugate as sound verbs. As such, they have no reason not to be considered sound verbs. Such verbs follow the conjugation tables above rather than these ones.

Final-weak conjugations (final radical ꝡ)
(Section anchor: CCW)


 * Passive participle:
 * Verbal noun:
 * Verbal noun:

Unlike in fuṣḥa, where roots with a final radical w yield -final verbs like, contemporary Arabic dialects level all weak final radicals to y. The reflex of that particular verb in Lebanese, for example, is.

Biliteral:
(Section anchor: CC)


 * Verbal noun:
 * , rarer
 * Nonpast:
 * Passive:
 * Nonpast:
 * Passive:

Recall the above note about the final geminate remaining geminate here no matter what. That means that, in all past-tense conjugations that use suffixes, this verb form looks identical to the final-weak triliteral Form I verb above. The conjugation in present-tense a is quite rare, by the way: for speakers who do use it, it can only be found in the verbs, , and , in ascending order of rareness.

(Section anchors: لقى and لقي)


 * Nonpast:
 * Passive:
 * Active participle:
 * , ambiguous alone but more-common when suffixed
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:

This verb, which descends from of Form III, has partway jumped paradigms: it remains Form III in nonpast conjugations, but in the past tense it's shortened the initial alif to become Form I. (As such, its original dialectal form is  in past-tense a, but many regions sport a raised variant in i.)

Not to be confused with the entirely-Form-1 verb.

(Section anchor: إجا)


 * Nonpast: ,
 * Active participle:
 * Verbal noun:
 * , rarer
 * Verbal noun:
 * , rarer
 * Verbal noun:
 * , rarer
 * , rarer

This verb descends from, but the regular process of hamza-loss rendered it doubly weak. Plenty of contemporary varieties are fine with the resulting, but it shifted in North Levantine Arabic to be bisyllabic in all third-person past conjugations. Some speakers also slightly lengthen the first vowel in nonpast conjugations, giving something like.

As in most or all contemporary Arabic varieties, this verb uses a suppletive imperative. In Lebanese, the chosen suppletion is derived from the imperative of, which see.

(Section anchor: ضل)


 * Nonpast: ,
 * Active participle:
 * Verbal noun: ?

This verb is from. It's remarkable enough on its own for being one of the only biliteral Form I verbs in Lebanese Arabic to conjugate with a nonpast a for some speakers, but what's even weirder is that it's the only verb in the language able to conjugate using object pronouns for subject agreement; the object pronouns are just stuck onto the end of the normal conjugations. Additionally, the second-person forms of the pseudoverb supplete the imperative here for some speakers.

The normal conjugation can be figured out without much trouble by looking at the generic biliteral tables above. However, the object-pronoun conjugations are as follows:

(Section anchor: سطع)


 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:

This verb is uncertain in origin and it's rarer nowadays anyway. For some speakers, it conjugates as a past-tense a verb in third-person conjugations, but as a past-tense ᴉ verb in other persons. (More specifically, this has to do with the form of the suffix used for conjugation: the third-person past conjugations use either a vowel-initial suffix or no suffix at all, which conditions the pronunciation, while conjugations in other persons use a consonant-initial suffix that conditions the pronunciation .)

(Section anchor: كان)


 * Nonpast:
 * Passive:
 * Active participle:
 * Verbal noun:
 * Verbal noun:
 * Verbal noun:

This verb is a fairly straightforward reflex of. For some speakers, the final n assimilates into the t-initial suffixes used by some past-tense conjugations.

(Section anchor: عطى)


 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:
 * Verbal noun:

This verb descends from. For some speakers, its glottal-stop-initial conjugations (that is, the imperatives and the first-person subjunctive) conjugate with initial rather than the expected.

Form II
There are no hollow verbs in this form. A weak medial radical is simply treated as a regular consonant. Similarly, biliteral Form II verbs are indistinct from triliteral verbs, as the two duplicate radicals are treated separately.

Sound conjugation (no weak radicals)

 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:

(Section anchor: CaCCaC)

Final-weak conjugation (final radical ꝡ)

 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:

(Section anchor: CaCCa)

Form III
The same provisions regarding hollow and biliteral verbs apply here as for Form II.

Sound conjugation (no weak radicals)

 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:

(Section anchor: CaCCaC)

Final-weak conjugation (final radical ꝡ)

 * Nonpast:
 * Passive:
 * Active participle:
 * Passive participle:
 * Verbal noun:
 * Passive participle:
 * Verbal noun:
 * Verbal noun:

(Section anchor: CaaCa)

Form X:
While many stafʕal verbs are in common Lebanese use, the form's once-productive meanings ("to seek X" and "to consider X") are fossilized, and new coinages are unlikely to be understood or accepted unless they're borrowed from MSA. (However, the "to consider" meaning would be slightly more familiar to speakers on this front.) The additional productive meaning that stafʕal has gained or retained in some other dialects ("to act like X") does not seem to exist in Lebanese: consider, which maps to Lebanese , or other examples like which have no snappy Lebanese equivalent.

Obsolete or exceedingly rare
These forms were once commonplace and productive, but they've been left behind in Lebanese Arabic today.

[TODO: add stfa33al, stfaa3al]

Form IV:
The initial hamza, like most initial hamzas, was deleted in a lot of varieties, rendering this form vestigial and otherwise obsolete. Its meaning was generally a causative of the corresponding faʕal verb. There are five possible outcomes in for it in Lebanese Arabic, listed in descending order of conservativeness:


 * 1) Preserved outright.
 * : rare, marked (religion)
 * : more common and not particularly marked (but still an MSA reloan, not native vocabulary)
 * 1) Regularized as being of different verb form.
 * , reanalyzed as being of form fāʕal.
 * 1) * Most other examples, like, are reanalyzed as being quadriliteral verbs: the original root gets an additional initial hamza to make four consonants.
 * 2) Merged into or made to subsume the corresponding faʕal verb.
 * 3) * Certain historic pairs of / have reflexes /. These aren't universal, however, and some speakers use the form with both meanings.
 * 4) ** /, vs. /
 * 5) ** /, vs. /
 * , presumably from a merger of and
 * 1) * from, and similar examples of subsuming
 * 2) * The vowel of a hollow verb is a very easy tell for this process, because it causes verbs whose hollow consonant is w to ostensibly diverge from their root. Beyond examples like (historic root ṣ-w-b, compare ) and  (historic root š-w-l), the following pairs exist where the latter is historically of Form IV:
 * 3) ** and
 * 4) ** and
 * 5) ** and
 * 6) ** and
 * 7) Replaced by the corresponding  causative, causing it to lose any conflicting original meaning.
 * , not
 * , not
 * 1) Or gone altogether, replaced by an unrelated synonym.

This form also survives in active participles and verbal nouns, even if the corresponding verbs are either no longer extant or no longer identifiably Form IV.


 * , active participle of, vestige of . However, many speakers use the levelled participle.
 * , vestige active participle of . However, the forms and  exist as well.
 * ,, , all in similar situations but less likely to have a levelled equivalent. (Re-loaned, not native.)

Pseudoverbs
[TODO: badd/bidd, fi; 3ind, ma3, 2il-; 7all-, Sarr-/Sall-}

Serial verb construction
Verbs can be stacked together in a, with each serialized verb indicating a purpose or consequence of the first. When an ostensible SVC has an idiomatic meaning not immediately parseable as this kind of “indicating a purpose or consequence” relationship, it becomes fair game to call the first verb in the sequence an auxiliary verb. See.

The typical SVC consists of an initial fully-conjugated verb followed by any number of verbs in the subjunctive mood, all of the latter of which share the same subject. It’s in fact directly equivalent to archaic English, and therefore ultimately analogous to an English -infinitive construction. However, unlike with the English analogues, stacking does not imply recursion:



Note that later terms in the SVC can only occur if the preceding verb has been given all its arguments. If a verb anywhere in the sequence (including the initial term) takes a subjunctive verb for an argument, then this will override the SVC syntax.