?

Quick notes on Cypriot Arabic and masdars

Splitting this out of a larger piece it didn't fully fit into.

April 19, 2026

Alexander Borg's 1985 grammar sketch of Cypriot Arabic has this table on p. 106:

Screenshot of Alexander Borg's 1985 Cypriot Arabic reference grammar sketch: As in many other Arabic dialects, the inventory of morphological patterns available for derived verbal nouns in KA has been greatly reduced. Thus most KA verbal nouns for derived triliteral verbs and for quadriliterals have the shape tCúCCoC or tCúCoC from OA tafaʕʕul and tafaːʕul respectively; the latter occurs mostly with medially weak roots: (101) Verbal noun: Verb θkunnos 'sweeping': kannes tnussof 'sifting (of chaff)': nassef tʕuppo 'filling': ʕappa tsuffo 'cleansing': saffa txuvoš 'collecting': xaveš tsuyož 'fencing in': saež

Notice the amount of innovation on display here. The melody ú⁠…⁠o has taken over many more verbal nouns than it originally had purview over, both within and across verb forms, i.e.

  1. 1.

    'within' verb forms: tʕuppo and tsuffo would instead resemble *⁠tʕVppi and *⁠tsVffi had they directly reflected the Arabic verbal noun shape (defective taCaCCī)

  2. 2.

    'across' verb forms: all of these are Form II verbs that would never have taken this verbal noun in the first place in older or other Arabics (Standard taknīs, taʕbiʔah, taṣfiyah, etc., mainland Levantine taknīs or tiknīs, tiʕbāy(e), tiṣfāy(e), etc.).

    • There's also a rare/unproductive(?) group of direct reflexes of taCCiCah: I believe Latakia-area təʕbe 'filling' from a @lawzinaj.bsky.social Tweet, analogous terba 'raising' in South Lebanon with an IMO-regularishly-hypercorrect(?) a, and a non-region-locked dish called tisʔye [recipe 1] [recipe 2]

    • For what it's worth, Michel Feghali 1919 Le parler de Kfarʿabîda gives a couple more examples:

      • p. 248 tö́ʔ̣dme 'offering, act of offering'

      • p. 248 tūṣye, to me in English 'entrustment' (in the form tūṣāye), for him « recommandation, action de recommander, annonce faite dans les églises »

Reinforcing #2 is the suggestion in the prose above the table that this melody has infected Form III as well (and presumably VI), which we can verify in Borg's 2004 glossary along with yet other forms. I'll do that and also compare them to mainland Levantine, for which Feghali 1919 is especially useful because it might be the only treatment of a Lebanese variety that came out early enough and is also comprehensive enough to describe the genuine native reflexes of a good number of the original masdars.1 Note that there are also mainland Levantine dialects that do use the same type of masdar pattern as Cypriot for certain wazns, but for this doc I'm only interested in the ones where Cypriot and mainland Levantine disagree.

Form III

Borg (2004) p. 379 ṯkupol 'care, solicitude' for p. 379 kapel 'to care for'2. This is a development of Arabic qābala 'to face'.

I have seen ideas that Cypriot tCuCoC continues an older *taCūCuC, but in light of the leveling the entire CyA verbal-noun system has gone through it seems prudent to consider tCuCoC's voweling an independent innovation. This is unless it turns out that there actually is a mainland Levantine form *tCūCuC, which (reiterated below) I'm unaware of.

Curiously, by the way, Cypriot pp. 270–271 tsuʕot~tsuyʕot is specifically glossed as 'mutual help' and listed under p. 270 tsaʕet 'to help one another' rather than under p. 270 saʕet 'to help'. This contrasts with p. 379 ṯkupol's relationship to p. 379 kapel. (Speaking of which, why no *ṯkuypol? Maybe the 'mutual help' gloss is a hint that tsuʕot~tsuyʕot is an older formation with a fossilized older meaning, and in turn that the uy is some vestige of when CyA still had vowel length?)

Mainland Levantine comparison

The would-be cognate, *tʔūbul, is nowhere to be found. The only available Form-III masdars in Levantine varieties are reflexes of Arabic muCāCaCah and CiCāC, albeit each with their own issues to my knowledge (the former having no identifiably native reflexes due to getting muddled with the more recent MSA loan, the latter being rare/vestigial).

Namely, on the latter's part, the only vestige of mine that comes to mind for CiCāC is the saying l-psayne betḥebb l-xnā̈ʔ ('cats like fighting(?)3' but I forget what it means, something about only liking what's bad for you? Forgive my p in 'cat').

However, Michel Jiha 1964 Der arabische Dialekte von Bišmizzīn does put p. 133 ždā̈l, mžā̈dli 'arguing, argument' on equal footing, i.e. without suggesting that for him the former is rare or vestigial.

Form IX

p. 171 ṯpuyoḏ 'whitening' for p. 171 pkyaḏḏa 'to turn white'.

Mainland comparison

I don't use any verbal noun for this wazn, and if I did it would be a technical-domain MSA loan (like ibyiḍāḍ for this verb rather than something like *tbiyyuḍ). I wouldn't be surprised if that were the case for the whole Levant.4

If a native reflex did exist it would probably look like *C(i)CCāC, e.g. *ḥ(i)mrār 'reddening', but the reason that's a 'probably' is that not even Feghali 1919 gives any masdars of this form anywhere that I can find. Maybe they're in his other works, but for now I'll take it as a signal that the masdar doesn't exist. My above stab at what it would've looked like comes from his p. 236 (more below), which indeed suggests e.g. *ḥmṛâṛ 'act of turning red', although even then I have no idea about semivowelful verbs like p. 185 byáḍᵈ̣ 'to turn white' or p. 185 swádᵈ 'to turn black'.

Hassan El-Hajjé 1954 Le parler arabe de Tripoli does list the masdar ḥmiṛā̊ṛ 'reddening' of ḥmaṛṛ 'to turn red' on p. 107, unlike e.g. the formally/formerly Form-VII5 nfaʕal mediopassives on pp. 98–101 where he goes out of his way to say no masdar is used, but ḥmiṛā̊ṛ still has a visibly MSA-loan-y shape.

Meanwhile, Jiha 1964 p. 134 agrees with me in saying that no masdar of this form is used for him.

Form X

p. 444 stuntor 'waiting' for stanter 'to wait'.

There's also a fun p. 368 stuffʕo 'vomiting' for p. 368 staffʕe 'to vomit' (okay, not that fun), which suggests these developments with respect to mainland stafraḡ 'to vomit':

  1. 1.

    Theoretical original Cypriot forms:
    *    stafreḡ vn. *stufroḡ

  2. 2.

    Regular merger ḡ > ʕ:
    *    stafreʕ vn. *stufroʕ

  3. 3.

    Irregular assimilation fr > ff:
    *staffeʕ vn. *stuffoʕ

  4. 4.

    Regular word-final metathesis Vʕ > ʕV:
    staffʕe vn. stuffʕo

(Not necessarily in that order, though. In fact, maybe step 4 actually happened before step 3 in order for the awkwardness of -frʕV to motivate the assimilation into -ffʕV.)

Mainland comparison

Like Form III, no would-be cognate like *stifruḡ of the verbal noun exists. The only verbal nouns we use for Form X reflect Arabic istiCCāC, namely the inherited stiCCāC and the more-recent loan ʔistiCCāC.

That split is a bit artifical, though, because like Form III the two have degenerated into this blob stiCCāC~ʔistiCCāC where it's impossible to tease out the actually native from the loaned. Illustratively, El-Hajjé 1954 gives p. 108 ʔəstəḥsā̈n and p. 110 stəḥsā̈n 'act of preferring, preference' for the same verb without remarking at all on the difference between them. (They do appear in differing contexts but it isn't enough to pin the variation on anything more than facultative preference.)6 I accordingly have no instances of stiCCāC in my dialect, and 'vomiting' would be ʔestefrāḡ, although I can't imagine using it very casually:

❓ ṣaʕb l-ʔestefrāḡ 'vomiting is tough'
✅ ṣaʕᵒb ʔenno l-wāḥad yestafreḡ 'it's tough for one to vomit'

El-Hajjé 1954 also gives p. 113 stəkṛa 'act of renting', which is more-solidly native, or at least shaped like a native inheritance.7 The Cypriot equivalent would be ?stukro, again with no mainland comparison.

Form VIII

Surprisingly, this one is perhaps actually nonexistent. The only originally-form-VIII verb I could find that has any verbal noun at all was p. 290 štra 'to buy' vn. štre 'buying', seemingly reanalyzed as Form I8 (compare rma 'to throw away' vn. rme 'throwing away', grammar sketch (1985) p. 106).

Other than this I looked at glossary (2004) p. 229 xtanak 'to drown (intrans)', p. 198 xtarak 'to burn (intrans)', p. 294 šteʕel 'to work', the unbelievably derived p. 215 ftaxel 'to fall ill', the iffier p. 371 ftaen 'to think, to suppose',9 and the hollow xtar 'to choose', and I could not find a verbal noun listed for any of them, *⁠CtuCoC or otherwise. (I also tried making use of my PDF of the late Antonios Frangiskos's Greek-language textbook on Cypriot Arabic, but not having any Greek except for a tenuous grasp on the alphabet I wasn't able to locate any form tables by skimming or Ctrl+F-ing.10)

For 'the unbelievably derived p. 215 ftaxel' (note for intuition's sake that pxal 'sick' is completely univerbated, f. pxále and pl. pxalín):

Scan of Borg's 2004 glossary of Cypriot Arabic: ḥ-w-l ftaxel, pkyiftxel (v/i) 'fall ill' (Perf. ftxilt, etc.); Form VIII verb from pxal 'sick': ftaxlu l-latuẟkya 'the children fell ill'. [This fusional outcome of the prepositional phrase *bi-ḥāl (> CyA pxal 'sick') finds formal parallels in Ṣaʿd ʾitbaxyar, yitbaxyar 'genesen' (Behnstedt 1987:194, fn 15) < *bi-xayr, and rur Pal Ar ʿuqbāl 'hoffentlich' < ʿuqbā li (Bīr Zēt; Schmidt/Kahle 1918: II, 289); for the same term in Cairene, cf. Vollers (1891:92).]

A mainland cognate would look like *btaḥal.

Mainland comparison

The page of Feghali 1919 I mentioned above blesses us with the mainland vernacular version of this masdar:

  • p. 236 ḥtʔ̣âṛ 'disdain, disdaining' for p. 236 ḥtáʔ̣aṛ 'to disdain'

  • p. 236 ḥtmä̂l 'act of bearing(?)' (« action de supporter ») for p. 236 ḥtámel 'to bear(?)' (« supporter »)0

I daresay these are long extinct by now. Today I'd say the only verbal-noun-y forms used for Form VIII are on the MSA-loaned pattern ʔiCtiCāC, e.g. ḥtmä̂l's cognate ʔeḥtimēl 'possibility'. I think the only verb whose original masdar somewhat survived into the modern-modern day is xtār 'to choose', having developed into the common noun xityār⁠~⁠ʔixtyār 'elderly man':11

someone's avatar
5mo

here i will yap abt خِتْيَار xityār, خِتْيَارة xityāra, عَجُوز ʕajūz, and عَجُوزِة ʕajūze 1) the most common levantine word for 'old man' is خِتْيَار xityār it's a resolution of the weird onset of some form like *خْتْيَار *xtyār, from *خْتِيَار *xtiyār, the form that underlies the arabic word اِخْتِيَار ixtiyār 'choice'

But the fact that a masdar 'CCtCāC' (Feghali notates it qttä̂l and has to leave a preemptive footnote basically going 'don't judge, that's just what the template looks like') existed productively as recently as 1919 helps with my confusion in that thread about how recently xityār could have crystallized.

I have also received the rare verbal-noun-y forms l⁠-⁠(⁠ʔe⁠?⁠)⁠htéra 'the rot' and ʔésteḥa 'shyness' (from htara 'to rot' and staḥa 'to be shy') in 'my' dialect, which seem to be native or nativized to some degree, but the dialect has no discernible broader pattern of defective verbal nouns on this kind of form. Even those two examples I've only ever heard one time each. No clue about CyA.

[As an aside: CyA also shows vestiges(?) of other verbal nouns even on forms that usually(?) take the melody ú⁠…⁠o, like p. 334 il⁠-⁠ʕátirfe 'Confession' from p. 334 ʕatraf12 '(of a priest) to hear confessions' and the perplexing13 p. 257 tirvíke 'breakfast' from p. 256 travek 'to have breakfast' or p. 256 ravek 'to serve breakfast to'.]

Overall mainland comparison

One last thing worth noting is that leveling all the verbal nouns to one single pattern is not at all a rarity across Arabics.

Standard Arabic

The same anonymous Arabist-ling-Twitter user from my needs-to-be-writingly-fixed last post [I'll go with A. as an initial] has pointed out that Standard/Old Arabic itself started to do this between Form⁠-⁠II tiCCāC and taCCāC, Form⁠-⁠III CiCāC < *⁠CīCāC, Form⁠-⁠IV ʔiCCāC, etc., the i⁠…⁠ā melody being common to nearly all higher forms. (For Form⁠-⁠III *⁠CīCāC, see 'Semitic Noun Patterns' by Joshua Fox p. 224 and p. 249. The 'syllabic structure … not allowed in Arabic' bit is a touch harshly phrased (mīlād 'birth, birthday', ʔīmān 'faith') but I figure there are fair ways to qualify it diachronically/morphologically such that it's correct.)

Screenshot of p. 224 of Semitic Noun Patterns by Joshua Fox: Arabic Qitāl > qitāl. [text I've grayed out] Arabic qitāl includes some nouns from PS *qitāl, but also nouns from *qatāl that have undergone dissimilation.[6] Qitāl and qatāl maṣdars are both common[7] and qitāl often occur as biforms[sic?] in Arabic, possibly from two different dialects, only one of which undergoes a dissimilatory shift. [text not grayed out] The L maṣdar qitāl is in this pattern. The vocalic melody of qitāl corresponds to the i-a melody common to most maṣdars of the increased stirpes. **Qītāl would be expected, with the vowel-length marker of the L on the first vowel,[8] but **qītāl, because of its syllabic structure, is not allowed in Arabic.

Yemen

A. has also shared this screenshot of Clive Holes 2018 Arabic Historical Dialectology mentioning Yemeni varieties that even have a Form⁠-⁠II CiCCāC, which is perhaps the same type of leveling:

Screenshot of unknown work: §3.7.3 The verbal noun of pattern II, beginning with the area of SME, is not taCCīC but CaCCīC, more rarely CiCCīC or CiCCēC (Khalafallah 1969: 67 jiddēm). In MeA and in the Shukriyya dialect of Sudan it is CuCCēC (Schreiber 1971: 61ff.; Reichmuth 1983: 253–4), in the Sudanese Nile Valley CiCCēC, and in Yemen CiCCāC.[36]

Following footnote 36 there leads to Holes 2016 Dialect, Culture, and Society in Eastern Arabia vol. III p. 206, which among various other non-MSA patterns also gives Form⁠-⁠II tiCCāC, a prolific Form⁠-⁠II/III CvCCāC, and on the next page a Form⁠-⁠V tiwiCCāC.

Maltese

Maltese has taken the Maghrebi quadriliteral tCaCCīC innovation even further [is that an accurate picture diachronically?] by leveling nearly every verbal noun to a reflex of the melody a⁠…⁠ī. These examples are from Wiktionary and the insight is due to User:Fenakhay. I'll use a circumflex to denote the long vowel even though standard Maltese orthography just uses the bare letter i:

  • stagħġîb 'astonishment' for stagħġeb 'to be astonished'

  • staħrîġ~stħarrîġ 'investigation' for staħreġ~stħarreġ 'to investigate'

  • tregħîd 'thundering' for riegħed 'to thunder' (also 'quivering' for triegħed 'to quiver')

  • trejjîq 'having breakfast' from trejjaq 'to have breakfast'

  • ftakîr 'remembering/remembrance' from ftakar 'to remember'

So: Standard/literary Arabic i⁠…⁠ā, Maltese a⁠…⁠î, Cypriot ú⁠…⁠o.

1.
It's also the worst (Ctrl+)F-ing experience I've ever had in my life because for whatever reason the PDF I have, even after trying to aggressively compress it, grinds my PDF viewer to a halt every time I switch pages. This took so long to write.
2.
Also 'to look at', 'to ponder', 'to consider'
3.
Or is it in an older(?) meaning related to being choked out? I know other dialects use forms like خناقه/خناقها in this saying.
4.
I do want to separate this from the claim that the verbal Form IX itself is infrequent in speech, an assertion I've seen (if itself infrequently) all over the Internet by now. To me this wazn is the only natural way to express the changes of state it's used for, even if in terms of the lexicon it's true that it only gets a handful of words. (My complete set is swädd ⚫, byaḍḍ ⚪, ṣfarr 🟡, zräʔʔ 🔵, ḥmarr 🔴, xḍarr 🟢, smarr 'to (turn) tan', šʔarr 'to turn blond', and randomly ḥlaww 'to become sweeter/prettier'. Feghali 1919 p. 185 adds ʕtáṛʳ̣ 'to turn piebald' and ʕwážᶻ̌ 'to become crooked'.)
5.
[In a just world nfaʕal would've been Form 8 instead of 7 so that this expression could've read 'formally/formerly Form-VIII' and been pronounced 'formally, formerly, Form-V-three'.]
6.
These last two sentences, i.e. "Illustratively [...] preference.)", were formerly a six-paragraph footnote. Sometimes I feel like an LLM looking for the seahorse emoji.
7.
Assuming I'm right that it's native and it's not just a secondarily nativized loan, my probably-uncontroversial take is it was harder for it to submit to the above facultative variation given that it stands more steps away from MSA-loaned ʔistikrāʔ than a non-defective verb does. Now, my impression is that these days -āʔ is taking over by force for most verbs anyway, but some actual digging would be in order before asserting for sure that stəkṛa no longer exists. A semantically related and similarly -āʔ-free word, šarra 'buyer' (< Arabic šarrāʔ), is alive and well in at least my dialect.
8.
Funnily, 'reanalyzed as Form I' is kind of its derivational fate in at least my mainland dialect as well: štara 'to buy' but passive nšara 'to be bought', verbal noun šera or šrēye 'buying' (both < širāʔ). Continuing in the same vein, some dialects even hybridized that verbal noun into a CyA-ish štrā̈ye (not to claim that it was the same process) as seen in El-Hajjé 1954, p. 106, or on Twitter in the spellings ⟨shtreye, chtreye⟩ from users from at least Tripoli and the Beqaa.
9.
Borg's *btayyan guess on that page strikes me as obviously further in the right direction than his ultimate decision to list it under √fṭn.
10.
According to my Discord history, I've just reproduced an older attempt to 'hunt for conjugation tables' on said attempt's four-year anniversary to the day. In the meantime I have apparently made 0 effort to learn Greek.
11.
The *xtyār idea is A.'s from before we knew about Feghali. (Read a bit below for 'A.')
12.
Borg p. 334 supposes that the noun ʕátirfe developed out of iʕtirāf first and that the verb ʕatraf was then backformed, which doesn't click for me. I see it making more sense for the verb ʕtaraf to have first developed into ʕatraf, especially knowing the excrescent vowel CyA inserts after ʕa- in other contexts*, and for the typical [pre-tCuCCoC?] quadriliteral verbal noun to have formed from there.

* See 'to vomit' above. The coolest consequence of this tendency that I've found is its reification in maʕaddút 'counted, added' < *maʕᵃdút.
13.
At least in my dialect it just inexplicably feels like trawwaʔ 'to have breakfast' is the base form and rawwaʔ 'to serve breakfast to' is a derived causative, maybe because the former is much more frequent, so it feels natural to assume that our tirwīʔa 'breakfast' was formed from trawwaʔ after we'd lost the ability to productively create a form like *tarawwuq. I have no idea what to make of Cypriot tirvíke in this light, i.e. of the fact that it doesn't look like *truka or *truvoka or something derived from a verbal noun like *truvok. Previously I just considered it ~evidence of continued migrations to Cyprus and therefore new infusions of lexical items throughout the centuries, but it's just struck me that maybe rawwaq 'to serve breakfast to' was the original verb, leaving 'to have breakfast' to be expressed by something like faṭar. Then trawwaq, originally its (medio?)passive, would only later have eclipsed it in frequency.

This probably agrees with the fact that we share the verb with Iraqi and Gulf Arabic trayyag (I don't trust Maltese), which suggests that the innovation of this verb is old enough that we definitely would've been able to use a form like *truwwqa for 'breakfast' if that had been our intent, and that there's something other than recency behind the formation of tirwīʔa 'breakfast'.

One last thing to note is that there IS one single Lebanese search result on Twitter/X for the form ⟨trew2a⟩, and the simplest explanation, that it's a typo'd transposition of ⟨trwe2a⟩, is tough to swallow because that doesn't match the author's spelling conventions. I quoted it in an old Twitter thread about this verb form in 2022 or 2023 that I think only Haidar saw, and my quote included a quick note like '(if you see this please tell us if it's a typo or not!)' that unfortunately/naturally never went anywhere. [Note that the fact alone that this form appears only once on all Twitter sadly isn't a good signal as to whether or not it's a typo.]
Subscribe to ?
to get updates in Reader, RSS, or via Bluesky Feed
Thoughts on Klimiuk (2013)