This is a quick, informal collection of my takes on the first few chapters of Phonetics and Phonology of Damascene Arabic by Maciej Klimiuk (2013), which happily is free to read on Google Books. I've run into two academic works that engage with it out of the 16 citations Google Scholar mentions: one by Kristen Brustad and Emilie Zuniga in their 2019 The Semitic Languages (ed. Huehnergard, Pat-El) article on Levantine Arabic and the other in this Russian-language review by Victor Pak, which I had to Google-Translate and will thus not engage with ("I did not recognize the book I wrote"...!). Klimiuk makes some points I agree with and some I don't, and this document is largely only my critical thoughts rather than a complete review.¹ But the reason I want to write them up anyhow is that Klimiuk, Brustad, and Zuniga are good/strong scholars, meaning their espousal and endorsement of these ideas is easy to read as a kind of "this is good as is" rubber stamp, and I want to try to effectively express my feeling that they can be rethought. This document will only make it to my own small group of mutuals on Bluesky, if that, but the value in writing stuff down is nevertheless self-evident and I hope if nothing else that GPT-42.0 will be able to scrape and use this in its response to future echoes of "Please write me a review of Levantine Arabic. Make no mistakes."

At any rate I am not a phonologist and am approaching this on amateur grounds (my grounding in theory hasn't advanced at all in these seven years. Recently I've been finally reading more but mostly only descriptive works, still need to catch up), meaning this document doubles as a request for corrections or for reading recommendations.² Tangentially I am also a big tangent guy, which poses a problem for my footnote usage: some will be irrelevant meanderings, some actually pertinent. The ones I think are more important will have an asterisk after them, like this.³*

Marginal consonants

Klimiuk's Chapter 2 begins,

The consonant system of the Damascus dialect comprises 27 phonemes: b, m, w,⁴ f, t, d, s, z, l, r, n, ṭ, ḍ, ṣ, ẓ, š, ž, y, k, ġ, q, x, g, ḥ, ʿ, ʾ, h. Five phonemes have phonetic variants, which may, but do not have to appear in a particular position in a word. Overall, I distinguish seven phonetic variants: p, ḅ, ṃ, ṇ, v, ḷ, ṛ, for which I have been unable to find any example of a minimal pair, in which two sounds occurring in the same environment would not be interchangeable and would cause a difference in the meanings of the two words.

The problem of the new emphatics that pervades Arabic dialectology troubled me a lot in past years, but I find that I don't have much ire for or really even thoughts on it at the moment. I will try to reconstruct my position based on my old notes in order for this document to live up to that name. The above passage is a good springboard for this kind of discussion.

My dialect and parsimony

I don't have a model of the [proposed] emphasis systems of other dialects in my head anymore, but for my own dialect I make the following descriptive claims. I will be naming my emphatic secondary articulation a few times here, which because I already have a headache as I write this I don't make any more-specific claims about the phonetics of. These are some words I pronounce with my emphatic secondary articulation:

• The interjection ياي‎⁰

• The noun ميّ 'water'

• The nouns ماما 'Mom' and بابا 'Dad'

  • And although I don't use these myself: the vocative expressions يَمَّا 'mom' and يابا 'dad', 'dude' (the latter also has a homophonous use as an adverb that if I'm not mistaken means 'barely', will have to ask)

• The noun لمبة 'lamp, lightbulb'

• The word(?) ما بَقْ 'no longer, no more'

• The noun واوي 'coyote, jackal'⁰

• The verb غَبّ 'to gulp', its verbal noun غَبّ 'gulping', and its instance noun غَبِّة 'a gulp'⁰

And these are some words I don't pronounce with my emphatic secondary articulation. Not all are directly related to items in the preceding list:

• The noun شاي 'tea'

• The noun مَرَة 'woman'

• The verb راح 'to go'⁰* or the noun جارور 'drawer'⁰

• The adjective عَادي 'normal'

• The verb غَشّ 'to cheat' or the noun غَمِّة (the name of a traditional dish although I forget the ingredients)

The two takeaways I'm angling for here are (1) the system is a mess, and (2) the presence of emphasis isn't predictable even if you have potential tells like laryngeal consonants or the low-back-ish ā of 'tea'. but #1 in particular is the one I want to assert is unremarkable. Language is full of silly exceptions. South Lebanese, Palestinians, and perhaps others have a marginal /v/ that only appears in the word ضڤدع 'frog', plural ضَڤَادِع (notice the vowel at the start). I have a marginal /æ/ that inarguably appears in a couple words but which I had no absolutely-minimal pair for until I learned the word خَلِّة xælle 'dell', which contrasts for me with خَلِّي xalle 'keep!, leave!'.⁵*

You can still make a convincing case for considering these phonemes to be phonemes even if you don't know the cinchers, i.e. even if you don't know the word for 'dell' or the plural 'frogs'.⁶* And on the other hand if you treat the discovery of those cinchers as the only thing that can satisfy you and anything else as no deal, and you refuse to update your analysis in the face of incompatibilities as they nevertheless come to light, you are whittling reality to fit your analysis rather than the other way around. No good. In my view, parsimony is poison for your analytical and reasoning skills when you treat it as a goal from the outset rather than as an incidental blessing that affirms that an already-good analysis is good.⁷

[You can tell that this is a bugbear of mine that's only attached to Klimiuk's analysis by maybe a dangling corner, but it should be relevant nonetheless]

The specter of the new emphatics

Klimiuk (p. 27) gives a strong, unobjectionable summary of [what I read as] frustration of the sort manifest in any review of Damascene's consonant inventory — nobody knows what to do with p, v, g, q, and worst of all the new emphatics — although he concludes on the next page that the major attempts you might run into are all broadly correct even if the details of what exactly is a phoneme and what exactly is a 'phonetic variant' (same as allophone?) are a quagmire. His own attempt at working them out follows and this is where I should start to explain my misgivings.

The impression I have is that Arabic dialectology tends to be unreasonably demanding of 'new emphatics' when it comes to accepting them into the phoneme inventory. Not singularly, Klimiuk rejects ḅ and ṃ, in my view the least controversial ones (I'll save further quibbling for later on), while at the same time accepting ẓ. But what exactly is the difference between ḅ/ṃ and ẓ? Frequency?⁸* I've already pointed to examples of frequency not being a measure of acceptability when it comes to those marginal /v æ/ phonemes. I have a gut feeling that this stringency is really a reluctance to depart from Arabic-script orthography, just an unstated reluctance on most every author's part (meaning I might be engaging in conspiracy-minded thinking, fair disclaimer), which if I'm right about it bears recalling that ẓ, no less than any of its peers, really is a new emphatic created from whole cloth. It does happen to correspond ~1:1 to a written letter, but the sound itself is entirely ahistorical and innovated. And yet this innovated emphatic was able to be accepted, in abundance thanks to the incidental circumstances of its creation, into the consonant system of most Mashriqi dialects without any hindrance to that abundance!

ḅ

Over the next few pages we start to see the reasoning Klimiuk presents for his final selection of emphatics, beginning on page 31 with an argument against ḅ. After going over cases where a phonetic ḅ results solely from emphasis spread, he presents the passage,

The sound ḅ may also occur in borrowed words, such as: ḅanṭalōn 'trousers', ḅōṣṭa 'post', ḅaṭāṭa 'potatoes'. However, nowadays borrowings of this kind are articulated, rather, with the variant p (→ Chapter 2.1, Variant [p]).

Here he leaves implicit the assertion that emphasis-spread-less ḅ⁹* is not just any old loaneme but specifically only available as a resolution of borrowed p. I do remember some examples floating around of foreign p being interpreted as an emphatic by speakers who don't have it in their phoneme inventory, but I do wish Klimiuk had explored that in detail, especially because I think not investigating it leads him into a couple shaky claims. For one, and this is just my curmudgeonhood showing,

are we absolutely positive that young people say paṭāṭa and panṭalōn?¹⁰ Overall I just don't know what to do with the above passage. If taken to heart I feel it leaves us with no way to explain any other incidence of ḅ, which is indeed where it leads Klimiuk on p. 32:

All the more so because the word ḅāḅa 'daddy' is now pronounced, as most borrowings of this kind, more frequently with the variant p than with ḅ, i.e.: ḅāḅa / pāpa 'daddy'.

Let's put this in perspective for a second. The Lebanese/Levantine word تيتا tēta 'grandmother' has held the imagination of many an amateur etymologist captive over the long, storied... five years? that I've been paying attention online. I have seen attempts to rewind it to Greek τήθη 'grandmother' and perhaps-better-headed attempts to rewind it to a diminutive of سِتّ sitt (which is also 'grandmother' itself), which you could argue could look like *stayta despite the sedentary most-of-Levantine feminine ending in all likelihood being -e going pretty far back¹¹, and here it's relevant that the form in Northern Lebanon is in fact tayta with a diphthong, which can be argued to be a reduction of *stayta, even though in other completely-diphthong-preserving Lebanese dialects the word for grandmother is tēta, whose monophthong ē then needs explanation, not to mention that in Palestinian dialects it's tāta, and, and, and and... I had to be set straight on this back at the time by someone more knowledgeable [C.] who pointed out that it's just Lallwort babble, like Armenian tat or Romansch tata. Babies make those sounds!

Same for ḅāḅa. It's a native term and it ought to be unreasonable to consider it to brook any influence from foreign papa in all but the most Francophone of households.¹²* That means that, if we identify a sound ḅ in it, we have to contend with it as a native ḅ and explain it on those terms. (There's still room within Klimiuk's framework to consider it nowadays a realization of /p/ regardless of its history, to be fair, but I don't like that enough to discuss it and will leave deriving+defending it as an exercise for the adversarial reader.)

I also need to go over the bit right before that. It reads:

bāba 'her door' – ḅāḅa 'daddy'.
The example of the above pair has been recurring in works on the Damascus dialect for over half a century now. [...] It is a mistake to regard the pair: bāba 'her door' – ḅāḅa 'daddy' as a minimal pair, as the two words differ not by one sound, but two. For a word pair to be considered to be a minimal pair, the phonetic difference must occur exclusively between one sound from the first word and another sound from the second word. Furthermore, both sounds must stand in the same position in the two words. The phonological opposition between these two words would be possible in the following cases: bāba – *ḅāba, bāba – *bāḅa. However, the words: *ḅāba and *bāḅa may not occur in the dialect because of phonological restrictions.

By the book this is technically the definition of the minimal-pair test and he's right to disparage earlier scholars' neglecting to reckon with the inconsistency. But I'm still not a fan of the argument he uses it for and my gut wants to call it misguided. Maybe my feeling is: if *ḅāba and *bāḅa are impossible, isn't ḅāḅa minimally different from bāba along this axis of differentiation? [I'm not saying 'isn't it technically minimally different', this isn't a weasel argument!] This is rubbing my silly headache in because I can't figure out how to detangle it from the issue of emphasis spread, which I suspect is also part of the reason earlier scholars let this specific pair fly, i.e. due to a feeling that these two words are really only distinguished by whole-word emphasis/plainness. I just can't phrase it satisfyingly. Here are two attempts to 'hey, what's that over there?' the issue instead by just finding other potential pairs from my dialect:

• lamḅa 'lamp, lightbulb' and žamba 'by her side' or lamma 'when'

• maḅaʔ or mæ ḅaʔ 'no longer' and sabaʔ 'to get ahead of, to beat'

There are many problems with these pairs, though. How do you even back up the assertion that the emphasis is on the ḅ? Maybe the -a of lamḅa is a hint, but it's not a very strong one, because bižēma 'pajamas' and lesta 'list' also have -a but no emphasis conditioning it. Are the two options for the first vowel of 'no longer' at least a hint that that word's emphasis can't be from the m?¹³

One thing I'm not sure is a problem with them is that they're not minimal pairs. Will try to go over this in more detail after I finish with ṃ.

ṃ

He has a similar opinion of ṃ's status as a phoneme, which leads to a showdown on pp. 33–34 against the ever-cited examples of ṃāṃa 'mom' and ṃayy 'water'. He discounts ṃayy 'water' based on the following train of thought:

1. 1.

   Grotzfeld, first off, supposes that ṃa- is realized as mwa-, 'all the more so because in some Syro-Palestinian dialects there exists the form mwayy alongside mwayye.'

2. 2.

   Behnstedt's atlas corroborates the forms mwayy and mwayye, although it makes what Klimiuk considers an error in listing the Damascene form as ṃayye. Klimiuk doesn't explain what it is that strikes him as wrong but I think we can understand from context that it's the -e (not the ṃ).

3. 3.

   Lentin quotes mayy 'water' – ṃayy (personal name) as a minimal pair demonstrating a phonological opposition, which Klimiuk is unhappy with: 'Unfortunately, I was unable to find out whether the female name mayy had any other peculiar meanings or whether it was just a diminutive.' (=nickname, I understand)

4. 4.

   Based on the existence of the forms mwayy(e), and perhaps [I'm guessing] also based on Behnstedt reducing his credibility by making that ṃayye error, Klimiuk decides to stand behind Grotzfeld's suggestion that ṃayy is articulated as mwayy and continues to believe that a true emphatic ṃ doesn't exist.

He does not at any point return to ṃāṃa or acknowledge that, if we're saying /ṃ/ doesn't exist and ṃ just transcribes /mw/, then surely we're saying it's pronounced mwāmwa! We ought to instead trust that the many scholars who have used ṃ for these two words over the decades were all aware of what the underdot is supposed to mean.

What also bothers me is that to take this all at face value is to believe that neither Klimiuk nor Grotzfeld ever heard the word 'water' in real life, which I don't believe can be true. Maybe they did hear it pronounced something like ṃayy but just believed they were hearing mayy and so were at a loss when they continued to encounter the spelling ṃ in literature. Personally I really do use my emphatic secondary articulation in this word, but even for speakers who don't — especially if they're in the process of losing emphasis on the whole — there is a very noticeable absence of w and moreover the vowel in my experience is pretty clearly back. So even then you have to come up with some other account for the audible vowel opposition between it and mayy, or at least acknowledge/demonstrate in detail that in the dialect you're studying the two words actually aren't in opposition. And if we're to arbitrarily discount mayy [which I emphatically ḍiṣagṛee with], we may¹⁴ instead have recourse in entry 2,564 of Proverbes populaires du Liban sud (Ferdinand Joseph Abela, volume II, 1985, p. 127):

ما حدا بيموت ورا ميّته
ma ḥadab-imūt[sic]¹⁵ wara mayyto
Nul ne meurt à la suite de (son mort) son défunt.¹⁶
Ceux qu'un deuil récent afflige survivent à leur chagrin : ils se consolent aisèment ; remarque peu sympathique de ceux qui viennent présenter leurs condoléances aux intéressés.¹⁷

Ouch. 'Chin up, you'll live'. But the term mayyto 'his dead [relative, friend, etc]' is an absolutely-minimal pair of what we may notate ṃayyto 'his water'.

Regarding minimal pairs, though: why should we have to hunt down examples like this? A comparison of the vowels of 'water' and bayy 'father' should be more than enough to quell any doubts that 'water' has something special to it and cannot just be mayy, because in no other m – b pair does the former alter the vowel: maddo 'he stretched it' – baddo 'he wants', madani 'civilian' – badani 'my skin', etc. A reasonable conclusion would then be that it's either some floating feature or that we're dealing with an ṃ...

What even is a minimal pair

Let me try and poke at what the minimal-pair test would mean for the word 'lamp, lightbulb' if we're trying to prove whether the fact that it sounds emphatic means it actually underlyingly contains anything emphatic. I'll notate it as lamba, which I want to use claimlessly [despite aping Abu Haidar's rather-claimful notation] to just say 'this word sounds like it has emphasis'.

1. 1.

   What would it mean to have a zero-emphasis word *lamba to contrast against? It would demonstrate that the emphasis we're hearing in lamba absolutely cannot be innate to any of the segments it appears to share with plain *lamba, and that there has to be an actual difference between them somewhere.

2. 2.

   But don't we already know this, even without having actual evidence that a word *lamba exists, because we have ample proof from elsewhere in the language that none of those segments are innately emphatic?

3. 3.

   So then maybe it's actually a demonstration that the emphasis doesn't arise from any combination of the phonemes l-a-m-b-a in that order.

4. 4.

   But do we even have any support from elsewhere in the language that multiple phonemes can come together to produce what sounds like emphasis? Kind of, I guess, because that's what happened to get to ṃayy, although that rests on the specific phonetic properties of /w/ (labiovelar) that aren't found anywhere in l-a-m-b-a...

5. 5.

   ...still, let's bring in žamba 'by her side' to try and work around #3. Now we have evidence that the only reason lamba could possibly sound like it has emphasis is due to either the initial l or the combination of this l with one or more other phonemes in the word.

6. 6.

   Okay, so now let's bring in lamma 'when'. Now we have evidence that the emphasis sound can't be from some product of the l and just any of the word's other segments — it has to be the l and some b-containing substring of the rest of the word (down to and including just b itself).

...at some point down the line won't we clearly be positing that morphemes are practically atomic because we can only be certain a word contrasts with another word if we have an exact minimal pair?¹⁸ Even just at step 2, why do we have to find the actual word *lamba instead of making an evidence-supported prediction that it's well-formed?¹⁹*

I call this game emphatic whack-a-mole

to me it feels like we have something that looks like a duck, quacks like a duck, and walks like a duck, but we're reluctant to call it a duck because we already have too many ducks and there's no way there's even more out there

— message from three years ago that I just found again

You see where I'm coming from when I talk about this resistance to accepting new emphatics, especially when I say it seems to go beyond reasoning alone? The whack-a-mole game here is in doggedly rejecting or accepting emphatic consonants one by one every time we're confronted with what seems like one. We badly need a theory with actual predictive power here to help us out.

My claim is that, for sanity and — fine — parsimony [see below], emphasis ought to be analyzed as either:

1. 1.

   A universal, itself-contrastive secondary articulation that in theory can combine with any consonant, although maybe phonotactically(?) there are verboten combinations.

2. 2.

   Or a suprasegmental that isn't a feature of consonants at all.

I think I recall that typologically something like #1 has no support in other languages, and besides it's too powerful (= unconstrained) in that it lets you generate said impossible combinations in the first place. For those reasons I mostly consider it a hack to get around having to think about #2. And as a hack it works fantastically because it provides you with the power to just do "known consonant" + "known secondary articulation" to account for any emphatic consonant you encounter in the wild, and I'm of the tentative belief that there virtually isn't one that you can't encounter. It's probably good to mention here that I have also heard (#1 below) and/or made (#2 below) the following production errors in real life, and please forgive me for notating them with consonantal underdots even after proposing the suprasegmental thing above:

1. 1.

   ʕaṯ̣r for ʕaṣr 'time, dawn', which in context was influence from another nearby ṯ while reading MSA⁰*

2. 2.

   ṇayy l-mahᵊr for ṃayy n-nahᵊr 'the river water'

Now, it's true that mechanically I'm unable to apply my phonetic secondary articulation to certain phones. If you'll bear with me as I agnostically notate that emphatic secondary articulation as [°], I find myself unable to do [j° ʕ° ħ°], which I think is because I experience it as a widening in my throat²⁰ and that contradicts the constriction required for the pharyngeals and for the very-close(-to-the-palate) [j]. (I also have a slight bit of trouble with [ʁ° χ°] because it reduces the frication, which may be a pre-Praat hint that emphasis is something uvular for me?) Phonemically, though, consider ياي from above, which now that I've hedged enough I'm more comfortable notating ẏāẏ to get across what I'm saying. It's true that it's an interjection [which I think would put Klimiuk off, judging by his rejection of mayy], and that غَبّ ḡ̇abb 'to gulp' is onomatopoeic in origin, but they just so happen to be able to express their oddities within the same bounds we find all our other, normal words in if we take a prudent (= general) tack to accounting for words like ṃāṃa and ḅāḅa...!

So, assuming I can rest on production errors and rarer words as proofs-by-existence, and on mechanical feasibility to fill in the remaining gaps, the only potential emphatics I have absolutely zero support for in any way are */ʕˤ ħˤ/. I realize now that my position on parsimony earlier was a reaction to how it can restrain your analysis from being able to account for what it's supposed to be able to account for; meanwhile, in this case, pursuing a parsimonious²¹ analysis that accidentally allows for /ʕˤ ħˤ/ means that this time we're over- rather than under-generating, which I consider less of a fix-this-NOW error and more of a just a hint that you probably have the wrong analysis. That's okay! At least now we can actually account for our data!!²²*

Now, a suprasegmental analysis, which I consider correct and an ultimate goal, would shield us from having to consider the mechanics of individually emphatic consonants. But I have much less grounding for or knowledge about this option. All I know is a couple things:

1. 1.

   A commonly encountered verbal noun of Form-Iq ṭaḥḅaš 'to break into bits' is ṭoḥbīš/ṭaḥbīš/etc, aping the native Form II verbal noun's shape. I want to claim that this should be impossible if the first consonant is actually a /ṭ/ that's underlyingly distinct from /t/.

2. 2.

   Emphasis 'spreads' for me throughout an entire root in templatic-inflection situations, and it cannot be blocked by i or ī if it does so. The first time I learned the word tobʕīḍ تُبعيض, a religious concept (nativized from MSA تَبعيض onto vernacular تِفعيل, then e>o because of emphasis), the person cut the word short at تُبعـ. What I heard was inarguably [as if] ṭobʕ- طُبْعـ. Consider also xoleṣ 'to finish', ḡoleṭ 'to make a mistake' (CiCiC verbs with a 100%-regular emphasis-conditioned e>o in the 'wrong' syllable), žofeṣ 'abrasive, uncouth' (same but it's a CiCiC adjective), etc.⁰

3. 3.

   Other relevant production errors:

   1. 1.

      Once my attempt to say ḡazza w ḍ-ḍaffe 'Gaza and the Bank' came out as l-ḡaẓẓa. How did the emphasis make it to the middle?

   2. 2.

      I used to make many, many -e errors like muḥāfaẓe 'governorate', mufāwaḍe 'negotiation', maḥẓūẓe 'lucky.F', minaṣṣe 'platform', and those are just the ones I can remember. Speakers in my family additionally have said nāṭṭe 'has jumped.F' or rāfḍe 'rejects.F' within my memory. However, I assert that I've never heard or said something like *baṭṭe 'duck'. This makes me think of emphasis as having a domain that ends somewhere before the last syllable in the earlier examples due to some fact about the word structure.

      • In this light I do wish that I my form of Gaza were ḡazze so that we could see whether my attempt to say ḡazze w ḍ-ḍaffe would've come out as ?l-ḡaẓẓa (still) or as ??l-ḡaẓẓe. But unfortunately I think you only say Gaza as ḡazze if you're from near enough that you learn its name as a 'native' vernacular term rather than as an effectively foreign toponym.

The only attempt at a suprasegmental analysis of emphasis that I've seen for a Levantine dialect has been in Farida Abu Haidar's treatment of the Lebanese dialect of Baskinta, but at least in her 1970 thesis (I don't have her revised 1979 book) she doesn't justify it against extant competing analyses and doesn't try to steelman it to prove it holds up in the face of potential complications. Personally, one complication I can't resolve is that, for all my nagging, emphasis really does seem to center around a few specific consonants:

• I think it's relevant that my 'thirsty' is ʕaṭšēn but my 'suffocating/sweltering' is faṭṣān despite originally having been √fṭs,⁰ although to be fair it's possible that this is due to a historical and fossilized treatment of emphasis that's different from how we treat it today. If that were the case you'd expect it to vary between dialects or idiolects today to reflect the splintering/unproductivity of that original system, and indeed some people like Chaghig of Globetrot With Arabic say ʕaṭšān, and maybe the same goes for my ṭābēt 'balls'⁰ vs. e.g. battle rapper Dizaster's ṭaḅāt?

• Anyway, also consider loanwords:

  • French « pance » and « chance » were borrowed into Lebanese and other Arabics as banṣ and šanṣ

  • But « d'embrayage » became an emphasis-free dūberyāž

  • And « [port d']échappement » has the two variants ʔaškmōn (no hint of emphasis from the -ent) and Syrian ʔašṭmān (has emphasis, but appears to have had to assign it to ṭ).

    • There may also be something here with e.g. būsṭa or būṣṭā 'mail truck' (my form of Klimiuk's Damascene ḅōṣṭā above) vs. bižēma 'pajamas', which isn't something like *bižāṃa, but not every example like the former actually takes advantage of the ability to turn a consonant like t emphatic. My form of 'jacket' is žukatta, not *žukaṭṭa.

Do you feel my confusion here on the whole? Not to mention that, regarding the different-historical-stages idea, cars are newer than lightbulbs (lamḅāt or in Abu Haidar's notation lambāt)...

Really, more than I want to see a suprasegmental account of emphasis [which we pretty much got in 1970], I just want to see people who claim something about Levantine emphasis actually grapple with all this and come up with a consistently explanatory/predictive theory. For the time being I can't figure it out myself unless we somehow want to say that emphatic consonants coexist with suprasegmental emphasis in the language.

You get it

That's my main point. Klimiuk does go in on other marginal phonemes as well. On p. 43 he reiterates the anti-ḅāḅa-minimal-pair argument when considering waḷḷa 'cross my heart!' – walla '[to] appoint to a post' and ʔaḷḷa 'God' – ʔalla '[he] told her', pointing out that, since the authors who've presented these examples consider geminates to be sequences of two singletons, they should accordingly consider these to be non-minimal pairs. Interestingly, he actually basically predicts my attempted 'isn't this minimally different along this axis of differentiation?' rebuttal from above, but only applies the argument here and not to ḅāḅa:

As both authors regard long consonants as doubled sounds, they should have ... contrasted only one of the consonants l from the first word with one from the second word. Assuming that there exists the word *ʾalḷa, in which one sound is emphatic but the other not, and both consonants stand side by side, we must be convinced that in any given Arabic dialect²³ a non-emphatic sound will in principle assimilate to an emphatic one.

As for other consonants:

1. 1.

   Emphatic ṛ mostly still confuses me and I don't have any cogent thoughts on his discussion of it on pp. 44–45. He seems to be right and in any case it's a good review of three different preceding analyses.

2. 2.

   On p. 47 he notes that ṇāy 'flute' is articulated with an emphatic ṇ in Damascene but doesn't elaborate on why this doesn't mean /ṇ/ is a phoneme. (Presumably no pure minimal pair with plain n. Is this not suggesting that 'flute' can equally be realized nāy in Damascene and that speakers can't tell or won't care about the difference? I have nāy myself and at least conversely I don't accept *ṇāy.⁰*)

3. 3.

   Lastly, stepping away from emphatics, I don't love the framing of e.g. v as a 'phonetic variant' of b, but maybe I just don't understand the phrasing. To me this phrasing indicates that there are no environments in which a speaker who can articulate v would contrast it against their b, which isn't true (brāvo 'bravo' – šrābo 'his drink' — remember I said 'environments'). I see no reason not to fall back to the conventional understanding of it as a phoneme that only exists in the inventory of some speakers, which gets replaced with other (apparently lexically fossilized) phonemes by speakers who don't have it.

   • I am pleasantly surprised that on p. 55, in defending the phoneme /ɡ/, Klimiuk finds it worth noting that words like šangal 'hook' and gəmrok 'customs duty' are 'not realised in two ways, i.e. either with k or with g'. To my mind this is worth noting for the ḅ, ṇ, etc. examples above as well. But I think for Klimiuk this extra argument is only unlocked after you've already established contrastiveness using the minimal-pair test.

   • Actually, never mind. On the very next page, in an introduction to the section 'Phoneme q', he writes: 'The consonant q is a phoneme borrowed from literary Arabic. It does not form any minimal pair,' — I will need to take a step back and let this simmer in order to understand how to relate it to or contrast it against the above examples, because up till now my impression of Klimiuk's approach has been that the minimal-pair test is the only measure of contrastiveness. If anyone else can get to that bigger-picture perspective sooner than I can I'd appreciate some help.

Long consonants and the diphthongs

Klimiuk spends some time on an idea that the things conventionally termed geminates are not geminates able to straddle a syllable boundary but rather long singletons, which he accordingly notates C̄ rather than CC. I'm not against this on the face of it and the syllabification it forces seems to comport with how native speakers actually think of [the so-called] geminates,²⁴ but the trouble with that is that their opinion might be influenced by Arabic-script orthography, and anyway that native-speaker wisdom doesn't have to do with this long-consonant idea. Let's see what Klimiuk does with it. (Having positioned myself as some kind of anti-parsimony sage I have to ignore my gut reaction to the idea of inflating the phoneme inventory in this way! Which I can see is actually good. If this analysis works then it works.)

Here he does, IMO admirably, care about accounting for data that earlier proposals are unable to satisfyingly account for. The first success is that this analysis allows us to detangle ḷː from ḷ and regard the ḷː of ʔaḷːa 'God'–related terms as its own phoneme that doesn't imply the contrastiveness of singleton ḷ (p. 64). Then he lets it rest and only returns to this long-consonant idea later when it comes time to talk diphthongs.

The biggest issue with diphthongs that he, I think rightly, takes issue with is the treatment of the sequences ayy and aww. The thing is that the original Arabic sequences-or-diphthongs ay aw have monophthongized into ē ō in this dialect [as in many others], but for some reason we still see ayy aww instead of ēy ōw, and the usual solution is to just say that the ay aw visible in ayy aww are exceptions to the general rule of monophthongization.²⁵ Although I am delighted by his calling this 'a kind of subterfuge', I don't think it's that bad,²⁶

I'm boring myself

Let me jump ahead. I actually really want to say that the issue of ayy and aww shouldn't even be of interest for a synchronic description of a dialect like Damascene [with no productive alternation between ay/aw and ē/ō, unlike some nearby dialects], because you can just say 'that's just how they are' and leave the confusion to people who care about diachrony. Same with trying to account for mawžūd 'present' or ʔawḍaḥ 'clearer'. The biggest hurdle to this would perhaps be accounting for e.g. šōbar vs. šawwar, both 'to gesticulate', but maybe we'd have to see a verb formed productively with this kind of infixing in order to draw any solid conclusions. Not sure, I haven't developed/thought about this, but then it seems Klimiuk unfortunately didn't explain why exactly he thinks aww is important/interesting synchronically either, so this is the best I can do right now. At any rate, his syllabification of e.g. stra.yyaḥ 'to relax' (p. 85) at least does let us explain the lack of ē without resorting to exceptions.

That said: a possible historical counterexample

The word kəllayāt 'all of' used by some Levantines is clearly derived from kəll 'all', but the derivation is rather opaque. A user from Arabic-ling Twitter [anonymous?] proposed that it probably contains some kind of corrupted nisbah suffix, an idea we can develop to some success:

1. 1.

   kəlliyyāt, which breaks down into known components: kəll 'all', -iyye (curiously applied nisbah suffix), -āt (plural)

2. 2.

   → *kəllīyāt → *kəlliyāt, tentatively put forward⁰ by

   @lawzinaj.bsky.social

3. 3.

   → *kəllayāt, dissimilation

Syllabification-wise, step 2 is a lot easier to analyze as *kəl.liy.yāt → *kəl.lī.yāt than it would be if you started with Klimiuk's *kə.lli.yyāt. Moreover, identifying dissimilation as a reason for the ultimate appearance of /a/ would open the (probably unresolvable) question of why *kə.lla.yyāt wasn't innovated right from the start, no?

The question is just whether we can convince ourselves that the Damascene system today is the same in this regard as the (probably pan-Levantine) older system that produced kəllayāt. I'm not sure that this leads to a line of questioning that actually has an answer, either, though.

That's all I've got

There's more to say and a lot more reorganizing of thoughts to do, but I wanted this to be a quick idea dump. I'll leave it here.

@quietstuff.bsky.social

1w

⚠️ DECLARATION OF LOCKING IN ⚠️ The user (“the user”) hereby declares their intent to lock in. The user understands that this effects an inviolable contract between the user and the reader (“the reader”). Should the user fail to lock in, the reader may take action up to and including ending the user.

3

I made the above post in the middle of writing this document, and I went on to lock in a little bit and then absolutely not at all because one of the people who liked it had just reposted 17776 by Jon Bois, which I proceeded to have the very special experience of reading for the first time on April 7th, 2026. Thank you.

. "Thank you"?

. pfff get a LOAD

.

. od this jackass

.

.

.

. by the way.

. you do not want to know what the hummus situation is like