PhDniX's blog

North-American Conference of Afro-Asiatic Linguistics 44: 13-14 February 2016

In the weekend of 13 and 14 February 2016 I attended the North-American Conference of Afro-Asiatic Linguistics 44.  There were a variety of very interesting talks, and I was able to give two talks at the conference (Gene Gragg was unable to come, and a I ended up filling up the gap in the program with a last minute, but I believe well-received presentation). The Program can be found here.

It is difficult to highlight any of the talks, because they were all of high quality. But let me just mention a few.

Simone Mauri's talk was highly interesting. He convincingly shows that in the Ayt Atta Tamazight dialect the 'sequential aorist' is not only used to express a sequentiality, which is how it is often described. Instead the sequential aorist should be considered as a 'default tense', which simply agrees with the established TAM marker earlier in the sentence. Verbs that break the chain of Aorists, by for example a flashback formulated by a perfective, in fact do not even break the aorist chain:

da di nnkkr ṣbaḥ isul lḥal illas nddu s amrdul

'We  used to get up (IMPF) in the morning, it was still dark (PF), we would go (AOR, continuing the chain of the IMPF not PF!) to the desert...'

James Allen's keynote address Ancient Egyptian as a Real Language was a highly informative and funny, and was a sobering account on how much work still needs to be done in treating Ancient Egyptian as a real spoken language spoken by actual people, rather than a highly theoretical construct.

Alexander Magidow's talk on Diachronic Classification fo Arabic Dialects via Personal Pronouns was a fascinating talk which shows just how difficult it is to make any sense of the diachronic development of the Arabic dialects spoken today, due to the many overlapping isoglosses in every way. His work on the Database of Arabic Dialects is becoming more impressive and useful as he continues his work on filling in the data. Hopefully with time we may untangle at least some of the messy tangle that is the modern dialects.

The sobering account of the state of Afro-Asiatic from a Semitic perspective by Na'ama Pat-El and Aren Wilson-Wright lead to an animated discussion on what the second A in NACAL should stand for if there isn't an Afro-Asiatic. Their conclusion is the suggestion that Afro-Asiatic languages may be a Sprachbund rather than a Language Family.

I agree that Afro-Asiatic comparison still has a long way to go. At the same time I don't see much explanatory value in the term Sprachbund. In the way that Sprachbund is usually defined (for example when talking about the ever-elusive Altaic, or the striking similarities between Japanese and Korean) is that you have a large group of languages that are structurally incredibly similar, but lack any (convincing) cognate vocabulary or morphology. This is clearly not what is going on in Afro-Asiatic (and it does not seem to be the definition of Sprachbund that Pat-El and Wilson-Wright seem to mean). (Parts of) Afro-Asiatic have very clearly striking morphological similarities (Prefix conjugation in Semitic, Berber and (some of) Cushitic; Feminine suffix -t; probably apophonic plural formations, etc.), but the amount of vocabulary that we can easily reconstruct with regular sound correspondences is rather small.

The absence of regular sound correspondences, is, however, not a (a priori) failing of the methods used by Afroasiaticists. It is simply that the amount of good cognates that we find is often so small, that we simply cannot find enough cognates to establish a sound correspondence. I would start feeling comfortable about a sound correspondence if you have maybe 4 or 5; usually we have about 2. But I still think we can probably write up some:

Proto-Afro-Asiatic *s (?): PBerb. *s; PSem. *s; PCentral-Chad. *ɬ

PAA *l-s 'tongue' > PBerb. *a-iləs < *a-Hĭlĭs; Semitic *lis-ān
PAA *s-m 'name' > PBerb. *a-isəm < *a-Hĭsĭm; Semitic *sim; PCentral-Chad. *ɬɨmɨy ʸ
PAA *s-n 'tooth' > PBerb. *a-sen; Semitic *sinn-; PCentral-Chad. *ɬɨyɨn ʸ
PAA *s-m 'ear' > PBerber *esəm < *a-yĭsĭm; PCentral-Chad. *ɬɨmɨy ʸ (Perhaps cf. Semitic s-m-ʕ 'to hear'?)

Some exceptions, to this correspondence are PBerb. *ăssən < *ăwsən 'to know' ~ PCentral-Chadic *sɨn. But we have to realize that even though that appears to be an irregular correspondence, Proto-Afro-Asiatic is really far back if it existed. There may easily have bee conditioning factors that split Proto-Afro-Asiatic *s into *ɬ and *s which are now completely unrecoverable (e.g. PAA *ws > s but PAA *s > *ɬ ?). There is a limit to the comparative method. Only by studying it really carefully, and even some more of the doubtful cognates, can we understand what the language may have been like. And even if we never conclusively prove it with the comparative method, this doesn't mean that it once was a (unprovable) real thing.

Proto-Afro-Asiatic *b: PBerb. *β; PSem. *b; PCentral-Chad. *v

PAA *l-b 'heart' > PBerb. *ul(ə)β < *a-Hŭlŭβ; PSem. *libb- (or *lubb-?)
PAA *b-ṭ 'night'> PBerb. *a-aβăṭ < *a-Hăβăḍ; PCentral-Chad. *viɗ
PAA *ʔ-b-n '(mill)stone' > PBerb. *ta-ʔ(v)βun-t; PSem. *ʔabn-; PCentral-Chad. *wɨvɨn
PAA *b-n 'hut' > PBerb. *a-βăn; PCentral-Chad. *vɨn ʸ; cf. Sem. *b-n-y 'to build'?

We may count the isolated word for 'rain' in Awjili here avún and compare it to PCentral-Chad. *vɨn 'rain'. I think we should avoid isolated nouns in language families mostly, but in this case, Awjili does seem to be one of the first languages to branch off from Proto-Berber (no significant shared innovations with any other Berber language, and a couple of (almost) isolated nouns which don't look niew, this being one fo them). The sound correspondences to check out, so this may be taken as a true cognate, just not as evidence for the sound correspondence.

Proto-Afro-Asiatic *m: PBerb. *m; PSem. *m;

PAA *s-m 'name' (see above)
PAA *s-m 'ear' (see above)
PAA *m-ṭ-r 'rain' > PBerb. *a-m(v)ẓar; PSem. *maṭar- (I don't know of the emphatic ṭ in PAA is actually ṭ or some other emphatic. I Believe Takács reconstructs č̣ for this root...)
PAA *m-w-t 'to die' > PBerb. *ămwət 'to die'; PSem. *m-w-t-; PCentral-Chad. *mɨts < PChad. mɨt?
PAA *m-y 'water' > PBerber *a-aman < *a-Hămā̆̆y-ăn?; PSem. *may-; PCentral-Chad. *yɨyɨm
PAA *m-y 'mouth' > PBerber *a-me < *a-măy? PCentral-Chad. *may

Proto-Afro-Asiatic *n: PBerb. *n; PSem. *n; PCentral-Chad. *n

PAA *s-n 'tooth' (see above)
PAA *ʔ-b-n '(mill)stone' (see above)
PAA *b-n 'hut' (see above)

Proto-Afro-Asiatic *l: PBerb *l; PSem. *l; PCentral-Chad. *l

PAA *l-s 'tongue' (see above)
PAA *l-b 'heart' (see above)

(Some correspondences like Proto-Afro-Asiatic *t > PBerb. t~H; PSem. *t almost go without saying, due to the many parallel morphemes in Berber and Semitic and Cushitic that share *t as well as some lexical items such as 'to die' above).

I think if we remain really conservative and strict about our correspondences (something previous Afro-Asiaticists haven't really done), we might find more than we think in terms of good correspondences.

Pat-El and Wilson-Wright make a very strong case that basically everything about Afro-Asiatic needs to be reevaluated with fresh eyes, and that we should seriously consider that at least some of the traditionally considered Proto-Afro-Asiatic languages may not be Proto-Afro-Asiatic. But I personally think (and I haven't  always thought this) that if you keep the scope small enough that you can actually have an informed opinion about the Proto-languages of some of the members, very promising correspondences will show up.

Now onto my own papers.

The first was titled The Secondary Origins of the Berber e Vowel, in which I argued that a large amount of the e vowels found in Berber (i in most dialects) should be considered allophones of Proto-Berber *a. I have been writing on the subject of the origin of this vowel, and other vowels in Berber for quite some time. and the paper that I have been writing on it is becoming bigger and bigger. There is a good chance that the eventual paper might end up as a small monograph, rather than an article.

This monograph would complement Maarten Kossmann's Essai sur la phonologie du proto-berbère well. That book mostly focuses on the Proto-Berber consonantism, while my monograph would focus on the Proto-Berber vocalism.

I have uploaded both the slides and the handout provided with this talk on Academia. My handout provides several little tidbits of attractive similarities both in cognates and formation between Berber and Semitic.

The second talk was titled A feminine formation in Arabic & Berber, where I compare some of the semantic and morphological peculiarities of Berber feminines with the suffix -e and -a with Arabic feminine nouns with the feminine suffix -ay and -āʔ. This is hopefully a good step to more formally studying the quite obvious close connections between Berber and Semitic.

Rather than attempting to reconstruct all of Afro-Asiatic, I think it would be useful to focus our efforts first on the obviously promising connections, and should explore smaller possible subbranches first, like Berbero-Semitic.

 

 

Is Thamud a triptote?

The word ṯamūd refers to an ancient civilisation, which is mentioned several times in the Qurʔān.

Like many place names, in Classical Arabic, ṯamūd is normally treated as a so called 'diptote'. Diptotes are a special class of nouns that lack nunation in the form without a definite article (which is not necessarily indefinite, Names are definite per definition even without the definite article, for example). Moreover, they do not distinguish between the accusative and the genitive which are only marked with -a. In the definite and and construct form diptotes declines along three cases.

nom. ṯamūdu acc./gen. ṯamūda; constr.nom. ṯamūdu constr.gen. ṯamūdi constr.acc. ṯamūda all spelled as ثمود <ṯmwd>.

Notice that the spelling of a definite-articleless diptote accusative is different from a triptote accusative, which is marked with a final ا <ʔ>, read as -an (and -ā in pause).

Interestingly, we find four instances of ṯamūd- in the Qurʔān, where the accusative is marked with a final ا <ʔ>.

Q11:68 كان لم يغنوا فيها الا ان ثمودا كفروا ربهم الا بعدا لثمود

kaʔan lam yaġbaw fī-hā ʔa-lā ʔinna ṯamūda kafarū rabba-hum ʔa-lā buʕdan li-ṯamūda

'As if they had never prospered therin. Unquestionably, Thamud denied their Lord; then, away with Thamud'

The spelling in this case is much better understood taking ṯamūd as a Triptote, rather than as a traditional diptote. The topicalising particle ʔinna (or at least, a particle that causes left-extraposition, which is very Latin-script normative term) takes the accusative case, while the preposition li- takes the genitive case. If we read ṯamūdan and ṯamūdin, we would exactly get this spelling. In this line, therefore, ṯamūd is clearly treated as a triptote rather than a diptote.

Q25:38 وعادا وثمودا واصحب الرس وقرونا بين ذلك كثيرا

wa-ʕādan wa-ṯamūda wa-ʔaṣḥāba (a)rrassi wa-qurūnan bayna ḏālika kaṯīran

'And [We destroyed]  `Aad and Thamud and the companions of the well and many generations between them'

ṯamūd appears here in a string of indefinite accusative, and the reading ṯamūdan is once again suggested by the spelling.

Q29:38 وعادا وثمودا وقد تبين لكم من مسكنهم وزين لهم الشيطن اعملهم فصدهم عن السبيل وكانوا مستبصرين

wa-ʕādan wa-ṯamūda wa-qad tabayyana lakum min masākini-him wazayyana lahumu (a)š-šayṭānu ʔaʕmāla-hum fa-ṣadda-hum ʕani (a)ssabīli wa-kānū mastabṣirīna

'And [We destroyed]  `Aad and Thamud, and it has become clear to you from their [ruined] dwellings. And Satan had made pleasing to them their deeds and averted them from the path, and they were endowed with perception.'

Same context as 25:38, so can be understood in the same way.

Q53:50-51 وانه اهلك عادا الاولى وثمودا فما ابقى

wa-ʔannahū ʔahlaka ʕādan al-ʔūlā, wa-ṯamūda fa-mā ʔabqā

'And that He destroyed the first [people of] `Aad || And Thamud -and He did not spare [them]'

Again, essentially the same context.

So so far this all looks like ṯamūd is indeed declined triptotically in the Qurʔān. What about attestations where ṯamūd is not spelled with a final Alif?

Q7:73 Q11:61 wa-ʔilā ṯamūda <wʔly ṯmwd> 'And to the Thamud'

ʔilā takes the genitive, so if we assume ṯamūdin it has the expected spelling.

Q9:70 Q14:9 Q40:31 qawmi nūḥin wa ʕādin wa-ṯamūda <qwm nwḥ wʕʔd wṯmwd> 'The people of Noah and [the tribes of] `Aad and Thamud.'

In the second part of construct phrase, and thus in the genitive, this spelling is the expected spelling for ṯamūdin.

Q17:59 wa-ʔātaynā ṯamūda (a)n-nāqata <wʔtynʔ ṯmwd ʔlnʔqḧ> 'And we gave Thamud the she-camel'

In this context the noun is expected to be in the accusative. The absence of an alif is unexpected if ṯamūd was a triptote.

Q26:141 Q54:23 kaḏḏaba ṯamūdu (a)lmursalīna/bi-n-nuḏuri/bi-ṭaġwā-hā <kḏb ṯmwd ʔlmrslyn/bʔlnḏr/bṭġwyhʔ> 'Thamud denied the messengers/the warning/their transgression'

Nominative position, the triptotic reading ṯamūdun would have the same spelling.

Q27:45 wa-la-qad ʔarsalnā ʔilā ṯamūda <wlqd ʔrslnʔ ʔly ṯmwd> 'And we had certainly sent to Thamud ...'

ʔilā ṯamūdin would be spelled the same.

Q38:13 wa-ṯamūdu <wṯmwd> 'And Thamud'

wa-ṯamūdun would be spelled the same.

Q41:17 Q69:5 wa/fa-ʔammā ṯamūdu <wfʔmʔ ṯmwd> 'And/so as for Thamud'

wa/faʔammā ṯamūdun would be spelled the same.

Q51:43 wa-fī ṯamūda <wfy ṯmwd> 'and in Thamud'

fī governs the Genitive, ṯamūdin would be spelled like this.

Q69:4 wa-kaḏḏabat ṯamūdu wa-ʕādun bi-l-qāriʕati <kḏbt ṯmwd wʕʔd bʔlqʔrʕḧ> 'Thamud and `Aad denied the Striking Calamity'

Nominative position: ṯamūdun would be spelled the same.

Q89:9 wa-ṯamūda <wṯmwd> 'and in the cities and (in) Thamud'

bi as found in Q89:6, which this noun refers back to takes the genitive, tamūdin would be spelled the same.

 

To sum up, four out of five times that ṯamūd appears in the accusative, it is treated as a triptote. In all other cases an interpretation as a triptote or a diptote is equally likely. Considering this evidence, it seems to suggest that ṯamūd was originally a triptote in the language of the Qurʔān, and it was only reinterpreted as a diptote in the reading tradition later (Turns out I'm not the first to come up with this).

One might expected that in older Qurʔān codices we might find Q17:59 written with the 'expected' extra alif, but this is never the case. Not even in very early documents like DAM 01-27.1 (carbon dated to 606-649).

However, the other way around is attested, e.g. the Paris Qurʔān is lacking the alif in 11:68, but other early manuscripts like Or. 6814 Carbon dated to come from 680-798, simply have the alif. The Samarkand codex also has alif.

The Sanaa Palimpsest, which often deviates from the modern Cairo edition, does have the alif in 25:38 (11:68 is not available/not present).

Q29:38 is attested without its alif in the Chester Beatty Library Is. 1615 I fragment, but it is present in the carbon dated (649-675) Ms. Tübingen Ma VI 165 manuscript.

 While it requires more in-depth research, it seems that diptotic spellings are often present, but that in the older manuscript the triptotic spelling is common.

A conservative solution to the variation of diptosy versus triptosy of this noun would be to assume that the noun could be both diptotic and triptotic. However, the fact that many early documents simply retain the triptotic spelling, suggests to me that this is older. The letter that follows the graphically separate alif in Q17:59 is also a graphically separate alif. Is its absence in this environment perhaps simply the result of an haplography?

So what triggered the innovation in the reading tradition to read ṯamūd as a diptote while the orthography suggests it's a triptote? Perhaps this can be understood as the result of a 'correction'. In Classical Arabic grammar all masculine proper names that are not of the shape faʕl, fiʕl, fuʕl, faʕal or a participle/adjective are diptotic. ṯamūd is a quite marginal adjectival pattern, and perhaps this is the source of the confusion. Either the pattern wasn't conceived of as adjectival enough unlike faʕīl patterns, or simply its obvious non-adjectival meaning triggered it to be treated as a diptote.

Why does Arabic spell 'hundred' so weird?

As has hopefully become clear in these past blog posts, Arabic, and especially Arabic of the Qurʔānic text is spelled in a really weird way. Many of these weird spellings have fascinating explanations, and while a definitive explanation is always elusive, there are plenty of reasonable arguments to make.

However, one spelling that has always baffled me is the Arabic word miʔah 'hundred', which is spelled ماية <mʔyḧ>. Normally <ʔ> markes either the sequence aʔ(a) in internal position, and <y> generally marks y or ī, while it only marks ʔ if it is directly preceded or followed by the short vowel i. That is technically the case here, but why is <ʔ> written in front of it? The spelling of the word rather looks like māyah, or perhaps maʔyah.

I filed this in my mental repository of "weird things that will become relevant when I find something that looks like it", and just the other day, I found just that!

It turns out, in the Qurʔān, we find several sequences of surprising/unusual spellings where ʔ and y or i have changed places from both a historical perspective, and in the way it is conventionally read in the Qurʔān.

First of all, the word for miʔah appears consistently in this weird spelling.

Besides this, a almost identical environment appears in Q51:47 where a sequence iʔa is spelled <ʔy> rather than the perhaps expected <yʔ> or simply <ʔ>. The noun ʔayd 'strength' normally written ايد <ʔyd> is preceded by the preposition bi-, this sequence in classical Arabic orthography would simply be written as بايد <bʔyd>, however that is not what we find in Q51:47, there we find it spelled with the rather foreign-looking bi-ʔaydin باييد <bʔyyd>. This is essentially the exact same 'weird' spelling that we find in miʔah, but even accross a morpheme boundary, the sequence iʔa is being written with <ʔy>, even if it yields the unusual looking sequence <yy>.

But this isn't the only environment where we find this, also the sequence yʔ yields the same unusual spelling, as found in Q18:23 li-šayʔin لشاي <lšʔy>. What is interesting about this spelling of the word šayʔ, is that it is one of the typical 'markers' of "Middle Arabic" (I would argue Colloquial-like) writing in the very first centuries of Islam in the Papyri studied by Hopkins (1984). The normal spelling of šayʔ is attested in many instances, and in the exact same context in Q16:40. The Sana'a Qurʔān in fact has the spelling show up in another place (Q19:9).

But this isn't all, also the verb yaʔisa 'to despair' shows up in this spelling in the imperfect stem -yʔas- with this same metathesis in spelling, which leads to rather enjoyable spelling 'corrections' in the vocalized text, where the <ʔ> is acquiesced, and an extra glottal stop sign is written on top of an empty bit of line to 'fix' the metathesized yʔ. Attested in these spellings we have:

Q12:87 wa-lā tayʔasū <wlʔ tʔyswʔ> 'and despair not'

Q12:87 lā yayʔasu <lʔ yʔys> 'he despairs not'

Q13:31 ʔa-fa-lam yayʔas(i) <ʔflm yʔys> 'and he despairs not' (translated as 'and he doesn't know, I am not sure how that works).

These three are the only instances of the imperfective stem of this verb, so the spelling is 'regular'.

A final environment is an example of an etymological sequence *īʔa which is once again spelled in this unusual way, in the passive of ǧāʔa 'to bring': ǧīʔa

Q39:69 wa-ǧīʔa <wǧʔy> 'and he will be brought'

Q89:23 wa-ǧīʔa <wǧʔy> 'and it will be brought'

These are the only two examples of this verb in this conjugation, and thus the spelling is 'regular' within the Qurʔān.

So these spellings appear quite regular, how can we understand them? In the case of sequences yʔ being spelled as <ʔy>, the answer appears fairly straightforward, a metathesis yʔ > ʔy has regularly taken place in the Arabic dialect of the QCT. This explanation however does not work for sequences of iʔa and īʔa as examined above. Here we must assume that the 'palatal/front' feature from the consonant was somehow transferred to behind the ʔ, or alternatively assume the ʔ was palatalized to ʔʸ (which is extremely exotic). A general rule can be formulated: iʔ, īʔ, ayʔ > i, īʔy (or iʔy?), aʔy.

It was pointed out to me by Lameen Souag that this metathesis (or palatalization or whatever you want to call it) is still found in modern Arabic dialects, e.g. Algerian Arabic layas 'despair' < *al-ʔayās not **al-yaʔās, which implies that the metathesis was generalized, and even used for the stem outside of the imperfect stem.

So now the question we've been asking ourselves throughout this series of articles: Is this rule part of the language of the orthography or part of the language the Qurʔān was composed in?

One thing is for sure: the spelling for 'hundred' is solidly part of the language of the orthography. But in the much more common word šayʔ 'thing', the "metathesized" spelling is actually uncommon. This is the only form that occurs often enough to really make any judgement about what the orthographic standard is. And it is quite clear that the spelling شي <šy>  rather than شاي <šʔy> is the orthographic standard for this word in the Qurʔān. This might suggest that the metathesized spellings are rather slips of the pen, shedding some light on the pronunciation of the Qurʔān. This would then suggest that this metathesis is indeed part of the Qurʔānic lanuage.

However, every instance of this metathesis other than the word šayʔ is spelled as if the shift had applied. It then is regular rather than a unusual deviation from the 'norm'. This might suggest that those spellings were originally the orthographic norm, and that somehow the 'colloquial' spelling شي <šy> for *šaʔy sneaked into the Qurʔānic orthography as the normal form.

However we analyze this, that there is something going on with y/i+ʔ sequences in the Arabic language is very clear. I would love to hear my readers' opinion on how this problem would be solved.

---

[UPDATE] Phillip Stokes informed me of the highly interesting book Qur'ans of the Umayyads by François Déroche (2013), which talks about older Qur'an manuscripts and their spelling variation. On page 22-23, Déroche mentions the spelling of the phrase bi-ʔāyāt(i-nā) 'with (our) signs', being consistently written as <bʔyyt(nʔ)> rather than the spelling <bʔyt(nʔ)> that we find in the the Cairo edition of the Qur'an today.

This is then once again an example of this same spelling convention that we find in miʔah, but this time, interestingly, before a long ā. This shows that this spelling was clearly more widespread in earlier editions of the Qurʔān. Déroche fails to note that this spelling closely resembles the spelling of miʔah, and tries to attribute it to a form of imalah, (vowel raising near an i vowel), which I don't find particularly convincing.

Déroche (pg. 23) also mentions that šayʔ is consistently spelled <šʔy> unless it was in the indefinite accusative, where it was written <šyʔ>, as per the standard orthography. It is difficult to decide what to make of that variation, but looks somewhat similar to a spelling convention mentioned earlier by Déroche where 3ibād ' servants'  is spelled <3bʔd> unless it is an indefinite accusative  3ibādan <3bdʔ>. 

This might suggest a orthogaphic convention that avoids a sequence <ʔCʔ>, which in Kufic writing takes up a lot of space, but alternatively, we may imagine that it represents some kind of linguistic reality and the the sequence āCā undergoes a shortening to aCā (Thank you Phillip for this useful suggestion). If the latter interpretation is correct, that would suggest that the spelling <šʔy> whether from original *šaʔy or not, was already pronounced šāy by these scribes.

---

[UPDATE] A similar looking but opposite example of the phenomenon we have looked at above is found twice in the Qurʔān for the spelling fa-ʔin فاين <fʔyn> 'so if' where the yāʔ appears to be used as a mater lectionis for a short vowel. (Q3:144, Q21:34). This is almost certainly a different phenomenon, similar to how ʔulāʔika is spelled اوليك <ʔwlyk>, even in Classical Orthography. Some examples of this are found for word-final vowels as well, which will be discussed in a future post on the final short vowels.

---

[UPDATE] Sometimes old Qurʔān manuscripts provide some more examples, e.g. the Tübingen manuscript (carbon dated to 649-675) which has <bʔyytnʔ wlqʔ ʔlʔxrh> where the standard text has <bʔytnʔ wlqʔy ʔlʔxrh> bi-ʔāyāti-nā wa-liqāʔi (a)l-ʔāxirati (Q30:16). 

---

[UPDATE] For those not swayed by Maghrebi Arabic layas to show that the metathesis of the *yʔ > *ʔy happened and let to generalization of the stem in ʔayisa,  there is evidence for it in the Early Islamic Arabic Papyri (in the 3rd century A.H.) cited by Hopkins (1984: 59-60): "§56 It is of intereest to note that the metathesized root ʾys (< yʾs) "despair", which is typical of (but not confined to) later Arabic, also occurs in the Papyri: Giess 11, 8 (3rd cent. A.H.), where read ايست "I despaired"."

---

[UPDATE] Yet another example of this spelling 'metathesis' is found in the phrase bi-ʔayyāmi (a)llāhi باييم الله<bʔyym ʔllh> 'in the days of Allah' (Q14:5). We also find it in bi-ʔayyi-kum <bʔyykm> 'which of you' (Q68:6).

After discussing this quite a lot with colleague lately, of which you will find some in the comments. I think my idea that this is purely phonetic is probably incorrect. the ayʔ > aʔy > āy metathesis, is probably real, and explains some of the forms that we find. In the cases of iʔa sequences being spelled as <ʔy>, I think we might be dealing with an orthography convention instead.

If originally (just post-Nabatean Arabic). One can imagine that at some point the original miʔah, spelled <mʔh>, but a shift turns it into being pronounced as miyah, a compromise spelling was created, retaining the etymological <ʔ>, followed by a <y> to denote that it is pronounced miyah not miʔah, or even māh, which would probably be the most natural reading of the spelling <mʔh> in this orthography.

This shift of course also affected the sequences of iʔa/ā > iyya/ā in other environments, and the most common environment for this to happen is in cases of the prepositions li- and bi- (Although, somewhat surprisingly I have found no examples of this after the preposition li-). The spelling convention of miyyah '100' was transferred as a spelling convention to write this sequence with the prepositions, yieldings <bʔyyt> for bi-yāyātin, <bʔyyd> for bi-yaydin, <bʔyym> for bi-yayyām,  <bʔyy> for bi-ʔayy

This argument may also be reversed. Perhaps The original spelling of <ʔy> for iyya derived from environments where a preposition bi- clashed with a noun that started with a vowel a or ā. It is orthographic convetion in Arabic to retain the writing of this initial vowel, with <ʔ> in Arabic orthography (although in fact broken by li- + definite article al- written <ll> not the expected <lʔl>. Perhaps the spelling <ʔy> was simply a way to retain the orthographic convention of keeping initial <ʔ> of the word after a preposition, while also accounting for the pronunciation which was iya. This spelling convention which then derived from 'orthographic sandhi' could have been generalized as a way of writing i/īya/ā and was then spread to words like miyah <mʔyh> 'one hundred' and ǧīya <ǧʔy> 'he brought'. One wonders if there's not some amount of etymological/historical spelling at work though. riʔāʔ, presumably pronounced riyā(ʔ) is always spelled <ryʔ>, not <rʔy> as you might expect. Likewise fiʔah 'company, group' is always spelled <fyh>, never <fʔyh>, despite having the almost exact same structure as miʔah. A look at older Qurʔān documents, reveals that also in the earlier documents the word was already spelled <fyh>.

One also wonders if there is some significance to the fact that all examples where bi-ʔa sequences wit the spelling <bʔy> that are found in Qurʔān documents are orthographically followed by another <y>, so orthographically these words always have the sequence <bʔyy>. I don't know what this means, but it seems significant.

Revision of Proto-Berber reconstructions of my MA Thesis

I was leafing through my MA thesis the other day. My first real work on Proto-Berber reconstruction. While I was pleasantly surprised that I'm not complete embarrassed about my idead on Proto-Berber reconstruction at the time, my views have certainly developed. I thought it would be a nice exercise to make a list of corrections the reconstructions, as I would currently reconstruct them as opposed to how I reconstructed them in 2011.

(1) *ta-ʔ(v̆̆)ḍuβ-t 'wool'.

Basically unchanged. I now admit the possibility that there may have been a vowel in between the first two radicals. This solves the problem that there are quite a lot of *ʔCVC stems, but vanishingly few *CCVC stems, instead we find many *CvCVC.

(2) *a-ʔ(v̆̆)fud pl. *i-ʔ(v̆̆)fadd-ăn 'knee'

See (1). On the i vowel in Awjili see Van Putten (2013). The short ă for *a in Ghadamsi is one of the few returning Eastern Berber Isoglosses, it shared it with  Nefusi ufəd pl. ifəddən and Siwi fud pl. ifəddən, but surprisingly not with Sokna and El-Foqaha which are usually close to Siwi, and not with Awjili.

(3) *a-ʔ(v̆̆)βyur 'moon'

See (1). Certainly reconstructible for Proto-Berber, but it's a but unclear where the glottal stop in Zenaga and Ghadamsi come from has come from. Zenaga points to *a-ʔyur pl. *i-βyar-ăn, Ouargli points to *a-βyur pl. *i-βyar-ăn. 

The Tuareg plural orăn is bizarre. I suspect it is somehow a *i-CŭCC-ăn plural. A fairly rare plural, most prominently attested in Tamazight argaz pl. irəgzən. e.g.: *i-ʔŭyr-ăn > (i-)urăn > orăn. 

The alternation between *ʔ and *β remains unusual and unexplained.

 (4) *a-ʔ(v̆̆)ḱal 'earth'

See (1). Awjili looks like a loan from a Zenatic language.

(5) *ta-ʔ(v̆̆)mar-t pl. *ti/ə-ʔ(ĭ)mir-ăʔ 'beard'

See (1).For the *ti/ə- prefix see Van Putten (forthcoming.a).

The plural suffix certainly contains a glottal stop, but it is ambiguous whether it is -aʔ or *-ăʔ. The reason why I prefer reconstructing *-ăʔ is due to the alternation between -a and -Ø in this plural formation in Tuareg dialects, e.g.:

Verbal noun of əgər 'to throw'

Mali teǧere pl. tiǧir (T-ka), ti-gur (A-grm)

Ahaggar teǵere pl. tiǵor

Iwellemmeden tăggor pl. šigur

Ayer tegăre pl. tigura

Verbal noun of əwət 'to hit'

Mali tewete pl. tiwit

Iwellemmeden tewăte pl. šiwit

Ayer tewăte pl. tiwita

Ayer has -a where the other Tuareg dialects have no suffix. Word-final accented *-ắʔ yields -a in Tuareg while unaccented *-ăʔ yields -Ø. So assuming there was some form of accent alternation in the plural (the details are not fully understood) we can reconstruct:

*ta-gar-e pl. *ti/ə-wit-ắʔ ~ *ti/ə-wít-ăʔ

*ta-wăʔt-e pl. *ti/ə-wəʔt-ắʔ ~ *ti/ə-wə́ʔt-ăʔ

This still needs some more evidence, and perhaps a study on the Proto-Berber accent in the noun. A more careful reconstruction would simply say that the suffix is either *-aʔ or *ăʔ.

(6) *ta-ʔ(v̆̆)riḱ-t pl. *ti/ə-ʔ(ĭ)riḱ-ăʔ 'saddle'

See (1) and (5).

(7) *ta-ʔ(v̆̆)fuk-t 'sun'

See (1). There is some alternation between uk ~ u among the dialects which is not fully understood but is certainly related to the verb with w in the root, e.g. Tuareg ifaw 'to be light'. But it is not fully understood.

(8) *aʔ(v̆̆)luḍ 'mud'

See (1).

(9) *(t)a-ʔĭrĭg/ǵ(-t) ; stone'

Ghd. érəǵ and Zng. tiʔrgiT are a little difficult to reconcile. We know that the sequence *ăʔĭ can yield e in Ghadames. It is not clear if it would yield iʔ in Zenaga. More research into the vocalism of Zenaga is necessary.

(10) *a-ʔ(v̆̆)βun 'flour'

See (1).

(11) ?*(a-)iʔsəm pl. *(i-)iʔsəm-aw-ăn 'name'

I am still hesitant to reconstruct this for Proto-Berber. It would be impossible to tell if it was an early loan from Arabic. But if I had to reconstruct it, it would look like the form given above.

(12) ?*a-ḍăʔăr pl. *i-ḍăʔar(r)-ăn 'foot/leg'

I now believe that the short vowel in Tuareg aḍăṛ needs to be considered. But I am not sure how to analyse it yet. Medial unaccented *ăʔ(ă) yields ă in Tuareg. But CăC stems regularly shift the prefix from *a- > e- (Van Putten forthcoming.b). This shift is absent, which suggest some interfernig factor. A clear solution is not yet obvious. There are some other nouns with a similar distribution, e.g. Tuareg ašăl 'daylight' ~ Tashl, azal 'day'. The exact reconstruction needs further examination. before a proper reconstruction can be made.

Kossmann (2001) suggested reconstructing *a-ḍʔar, which I followed in my thesis, but this does not solve the Tuareg issue.

(13) *a-ʔ(v̆̆)dif 'marrow'

This noun, along with alim~lum 'hay' and asif~suf 'river'have i in Kabyle, Tamazight and Tashelhiyt, but an u in Zenatic and Tuareg. Already in my thesis I noted this distribution I elaborated on this in Van Putten (2014), and suggested a reconstruction *a-diʔf.

Notice however that in all three cases (see 79 and 80, below) these nouns have a stable initial vowel in the singular (and not in the plural, except in Tamazight for this word, but that's surely secondary). This is normally a sign of a stem-initial glottal stop. This is somethnig I failed to notice both in my thesis and in my article, but certainly suggests perhaps some kind of metathesis:

Tamazight/Tashelhiyt/Kabyle *a-ʔdif

Zenatic/Tuareg *a-diʔf

What is causing this metathesis in Tamazight/Tashelhiyt/Kabyle?

Here's a new thought: In Ghadames initial *a-ʔ yields o (followed by shortening of the following vowel), final *ắʔ likewise yields o-. This latter result is an isogloss it shares with 'Inner Zenatic' (which excluse varieties like Riffian, Zwara and Tunesian Zenatic languages, but certainly include Ouargla and Mzab). Perhaps Inner Zenatic had a development as follows:

*a-ʔdif, *a-ʔsif, *a-ʔlim > *o-dif, *o-sif, *o-lim >  (regular metathesis of rounding feature) *a-duf, *a-suf, *a-lum > (Typical Zenatic irregular loss of the prefix vowel before CV) a-duf, suf, lum.

This same metathesis of the rounding feature would then also have happened in Tuareg.

If we accept this solution however, we have to accept that the Zenaga äḏiʔf underwent metathesis of the glottal stop. This is not uncommon, and certainly happens in other situations, but obviously makes the solution less attractive.

(14) *(a-)uβăʔr 'lion'

This word poses some issues for the reconstruction, see also Kossmann (2001). Like (12) the Tuareg cognate has a short *ă, while other languages have a full vowel: ahăr. I somehow missed this cognate in my thesis.

(15) *a-fuʔəs pl. *i-fass-ăn 'hand'

*CuCəC stems exist, *CuCC stems do not. The loss of *ʔ in the plural of the Zenaga cognate ävuʔš pl. uvässän, uṿässän is unexpected. The solution will come down to how we understand these formations with a infix and long final consonant.

Perhaps an assimilation of the glottal stop to the next s (a solution already suggested by Prasse, but not without its problems, such assimilation are not found in the verbal system, ắʔker 'to steal' yields akər, not əkkər. But one wonders if roots that have ʔs as the first two root consonants exist.

(16) *(a-)iləs pl. *(i-)iləs-aw-ăn 'tongue'

I think the final glottal stop that appears to be present in Zenaga is an epenthetic vowel, because the final cluster tʸš was not permitted.

The plural Zng. ätʸšūn ~ Tamazight/Tashelhiyt alsiwn seem similar, and sufficiently weird to perhaps warrant reconstruction for Proto-Berber. In my forthcoming article on the vowel e I argue that this i~a variation is originally e~a variation. If we reconstruct the alternation for this word, that would be the only counterexample.

I will present this in Austin Texas at the North American Conference of Afro-Asiatic Linguistics. I will post my slides after the presentation on my Academia.

This word is of course a very nice cognate to the Semitic ls 'tongue' root.

(17) *a-ɣăss pl. *i-ɣs-an ~ *(a-)aɣăss pl. *(i-)aɣăss-ăn 'bone'

The reconstruction of this word is difficult. Some languages have emphasis in the final long consonant. Awjili somehow has a cluster st.

It's also not clear if this word had a stem initial vowel or not. It depends on the language whether it is the case.

For the stem initial vowel, see my NACAL presentation (as 16)

(18) *a-ḍad ~ *a-ḍawd ~ *a-ḍakkəd pl. *i-ḍəwd-an

The Awjili taqt pl. taqqíd may not be a cognate at all. Zenaga aḏ̣aġḏi pl. iḏuġḏän and Tuareg ăḍaḍ pl. iḍəḍwan, iḍăḍwan both seem to point to some element *w but on different sides of the d. Ghadames has kk, which probably related to this element *w in some way. Reconstruction remains difficult. The final i in Zenaga may be an epenthetic vowel to break up the word-final cluster rather than a reflex of *əʔ.

(19) *ta-βăyn-e 'dates'

The prefix was ta- that underwent the *a > e shift as discussed in Van Putten forthcoming.b.

(20) *(a-)aβăya pl. *(i-)aβăyaw-ăn 'nephew, grandchild'

Proto-Berber probably already underwent *aw# > *a. Forms that have final -aw presumably introduced it from the plural. Tuareg retains the old situation ahăya pl. ihayawăn.

(21) *(a-)aβwal pl. *(i-)aβwal-ăn 'speech'

Reconstruction unchanged.

(22) ta-βur-t  pl. ti/ə-βur-ăʔ 'door'

Forgot to reconstruct the plural. This seems the best reconstruction. The Awjili  təv(v)úrt pl. tvurr has a puzzling plural formation. Not that Mali Tuareg tăhort pl. tihor is once again missing the final -a, as expected in this plural type.

(23) *a-βăḍ pl. *i-βăḍ-an ~ *(a-)aβăḍ pl. *(i-)aβăḍ-an /  *(i-)uβḍ-an 'night'

The plural suffix -an usually is only added to a CŭCC stem, cf. *a-ǵăllid pl. *i-ǵŭld-an 'king' (cf. also fāris pl. fursān plurals in Arabic). This does not seem to be the case in this word in most Berber languages, which is somewhat unexpected. Mzab and Ouargla iyḍan is probably from such a formation < *i-βŭḍ-an. The forms Kabyle/Tashelhiyt aḍan ~ Zenaga āḍan point to a stem initial vowel. Perhaps the stem initial vowel reconstruction is better for these languages. But both formations seem present in Berber. It is difficult to decide which is original. The reflexes with e vowel in the stem-initial form is discussed in an article currently in preparation, see my NACAL presentation (see 16).

(24) *ta-ʔβun-t 'big stone, anvil'

Should have reconstructed this with *ʔβ cluster rather than *ʔ/β.

(25) *aβv̆̆rn 'flour'

Reconstruction unchanged. Awjili vrún remains strange, presumably a different formation.

(26) The word for 'chameleon' remains difficult to reconstruct. The different words cannot really be reconciled. t(a)-aβa(w) ~ t(a)-ata(w) pl. t(a)-at-iw-en ~  t-awəḍ-en??

(27) *(a-)anβăr pl. (i-)anβar-ăn 'eyebrow, eyelash'

Not many cognates. See my NACAL presentation (see 16)

(28) *(a-)uləβ pl. *(i-)ulβ-aw-ăn 'heart'

This noun seems to have a uCC shape, otherwise rare. *(a-)uləβ is sooner expected, but we would expect this to show up in some languages as i or u, but it only does in Siwi uli < *uləβ?

This is almost certainly cognate to the Semitic root *lb.

(29) *a-ɣirβ pl. *i-ɣirβ-an? ~ *a-ɣăβăr pl. *i-ɣăβăr-iw-an 'shoulder'

This stem seems to have a vowel i in the stem for Tuareg. But the Zenaga, Tashelhiyt and Tamazight forms may point to internal a~e variation. This is normally due to a conditioned shift as in my a > e paper that I will present at NACAL (see 16). The variation is unexpected in this noun from the rules that I have set up in this article.

A suggestion might be that this noun has a different formation in Zenaga, Tashelhiyt and Tamazight, something like *a-ɣăβăr with metathesis would yield the expected results Proto-Atlas Berber *e-ɣer pl. *i-ɣar-iw-ăn.

CiCC is not a normal stem formation, but a *ə before *β would give i or u reflexes.

(30) *i-damm-ăn 'blood'

Unchanged. Compare Semitic *dm.

(31) *ta-fer-t pl. *ti-fer-en ~ *ta-făr-t pl. *ti/ə-fa/ăr-en ? 'reward'

Zenaga and Ghadames clearly have a different formation with a ă vowel in the stem. The reason for this difference in formation is not well understood.

(32) *ta-meḍ-t pl. *ti-meḍ-en 'umbilical cord'

I do not understand the Zenaga form tṃaḌ pl. tuṃṃuddäyn, tṃuddäyn.

(33) *a-ɣil pl. i-ɣall-ăn 'arm'

The forms iɣil appear to be secondary, see Tamazight iɣil ~ aɣil.

(34) *ta-ɣər-t ~ *ta-ɣar-t  'drought'

I originally ignored the Kabyle taɣəṛt form. Not broadly attested.

(35) *t(a)-aɣəy-t pl. t(i)-uɣay  ~ ta-ɣuyye-tt pl. ti-ɣuya 'shout'

The Kabyle and Zenaga formations are one formation the Tamazight and Tuareg reflect the other.

(36) *(a-)udəm pl. *(i-)udəm-aw-ăn 'face'

Unchanged.

(37) *(a-)eḍəs 'sleep'

One of the few noun formations that have a stem-initial e. This is ultimately secondary on a Pre-Proto-Berber level. This is a verbal noun formation from *C1=C2 CCC verbs, e.g. *ăṭṭəs 'to sleep', because of their causatives *əssuṭəs and because one obvious Afro-Asiatic cognate, it seems that these nouns had an original stem initial w: Pre-Proto-Berber *a-wĭḍĭs.

(38) *(a-)afăr(r)  pl. *(i-)afăr-iw-ăn ~ *(a-)afər pl. *(i-)afər-iw-ăn ? 'wing'

This word is difficult to reconstruct. Multiple formations appear to play a role.

(39) *t(a)-urəʔ-t ~ *ta-ruʔ-t pl. t(i-)uraʔ 'lung'

The Singular of Ouargle, Foqaha and Tuareg are probably originally plurals. The loss of final -a in Tuareg, despite this almost certainly being a final sequence *aʔ, is a counter-example to my theory that the plural suffix discussed in (5) is *ăʔ rather than *aʔ. The forms tarut appear to be a different formation (with metathesis?) Perhaps *ta-ruʔ-t.

Compare the obvious Semitic cognate *riʔ-at- 'lung'

(40) *(a-)urəɣ 'gold'

Unchanged. In Pre-Proto-Berber presumably from something like *a-wŭrŭɣ. The connection with the root wrɣ 'to be yellow' is obvious, also compare the Semitic root *wrq.

(41) *(a-)aɣăf pl. *(i-)aɣăf-iw-ăn ~ *(a-)iɣəf pl. *(i-)ɣəf-aw-an 'head'

Two different formations appear to play  a role in this noun.

(42) *(a-)asăf pl. *(i-)usf-an 'day'

Reconstruction unchanged. Compare the similar plural formation in (23).

(43) *a-lăm pl. *i-lăm-aw-ăn 'skin'

Kabyle tallumt 'screens made of thin strips of skin' is obviously a different derivation from the same root. Ghadames and Zenaga forms may point to a formation with a stem-initial vowel *(a-)alăm pl. *(i-)alămaw-ăn, but this might be an individual reanalysis in these languages.

(44) *(a-)alăf pl. *(i-)alăf-iw-ăn ~ *(i)-ilf-an 'swine'

Despite its very marginal distribution, the new findings on the vowel e in Berber, allow for quite a specific reconstruction. Obviously, its limited distribution may make it not reconstructible for Proto-Berber. But it is not unlikely that a word, consider the status of pork in Islam, was simply lost in all other Berber languages.

(45) *t(a)-adăkt / *t(a)-edək-t / *t(a)-idv̆̆k-t 'mastic tree'

Due to its limited distribution, we cannot decide between these different reconstructions.

(46) *(a-)adal 'alga'

Unchanged.

(47) *(i-)ad-an 'bowels'

Basically unchanged. Clearly originally a Plurale Tantum.

(48) *a-ḍil 'grape'

The lack of a prefix i- in these languages confirms that the stem vowel is i, not e.

(49) *(a-)ugəl pl. *(i-)ugl-an 'tooth'

Presumably from Pre-PB *a-wŭgŭl, see (40).  Awjili has a different formation *a-wăgil/*a-wăgel. Ghadames also has a different formation *ta-wəgle-t?.

(50) *a-ʔ(v̆̆)gum 'trunk'

Glottal stop added. See Van Putten (2015).

(51) *a-gur pl. *i-gur-ăn 'castrated goat'

Unchanged, Very marginal distribution.

(52) *ta-β(v̆̆)gus-t pl. *ti-β(v̆̆)gus-en 'saddle strap'

Fixed a type. Otherwise unchanged.

(53) *(a-)aguẓẓ/s pl. *(i-)aguẓẓ/s-ăn / *a-gaẓẓ pl. *i-gŭẓẓ-an 'cheek'

Unusual alternation between ẓ and s. I'm quite sure now that the Tamazight word belongs to the etymon.

Ghadames and Tuareg point to a different formation: *a-gaẓẓ pl. *i-gŭẓẓ-an (see 23).

(54) Quite sure that this is an Arabic loanword.

Tuareg form has a typo. Should be ta-krurəy-t pl. ti-krurəy-en. This is probably not a loanword, but I am not sure if it has cognates outside of Tuareg. Appears to be onomatopoeic.

(55) *a-ʔ(v̆̆)kuz 'mealworm'

Glottal stop added. See Van Putten (2015).

(56) t(a)-aməm-t 'honey'

For the i vowel in Awjili see Van Putten (2013).

(57) a-m(m)ud 'prayer'

Unchanged.

(58) *a-m(m)ud 'door post'

In some weird way related to (57)?

(59) *ta-mur-t pl. *ti-mur-ăʔ 'earth'

See (5).

(60) *ti-mẓ-en ~ *t(i)-umẓ-en  'barley'

The u vowel has found in Foqaha is more widespread. Related to *aməẓ 'to grab' somehow?

(61) *(a-)inəɣ pl. *(i-)inəɣ-an / *(a-)anŭɣ pl. *(i-)anŭɣ(-iw)-ăn 'palate'

I believe the Kabyle forms anəɣ and inəɣ are two separate formations. The Tashelhiyt form *anɣa is unusual.

(62) *ta-ʔnas-t pl. *ti/ə-ʔnis-ăʔ 'key'

Unchanged.

(63) *(a-)eɣəd 'ashes'

See (37).From *ăqqəd 'to burn' Pre-PB *a-wĭɣĭḍ. Compre Semitic *wqd.

(64) *ta-răy-t 'heap of grain'

Due to insights into the e vowel in the prefix, the vocalism can be established without the need of languages that distinguish e~i and ă~ə.

(65) *(a-)erəd 'grain of weat' pl. *i-yərəd-ăn 'wheat'

For a long time I have been puzzled by the Awjili and Foqaha plurals yə́rdən.

Perhaps this confirms my hypothesis of (37) that Pre-Proto-Berber *a-wĭ > *a-yĭ > *e. Without the presence of the *a- prefix in the plural, the yə sequence resurfaces. Other languages would have either analogically corrected it, or perhaps underwent a development *iyə > i (not unlikely, of course).

 (65) *(a-)uraw 'cupped hands'

The tamazight singular uru is somewhat surprising.

(67) *(a-)aruy pl. *(i-)aruy-ăn  'porcupine'

Limited distribution, reconstuction basically unchanged.

(68) *ta-ʔ(v̆̆)saf-t pl. *ti/ə-ʔ(ə)suf-ăʔ '(acorns of the) holly oak'

It's somewhat unclear whether this noun had an initial glottal stop. But Kabyle points to this. Tashelhiyt might be a generalization of the stable initial vowel from the singular to the plural.

(69) *a-sen pl. *i-sen-ăn 'tooth'

Prefix is normal e- due to the a > e shift.

This has a good cognate in Semitic *sinn-

(70) *t(a)-esəm-t 'salt'

The shape of this noun suggests a Pre-PB *ta-wĭsĭm-t

(71) *(a-)esəm / *ta-ssəm-t 'animal fat'

The shape of the masculine noun may go back to Pre-PB *a-wĭsĭm. The feminine form 'confirms' the semi-vowel, as it the cluster *ws yields *ss, Pre-PB *ta-wsəm-t (cf. *ăwɣed > *ăqqəd, 37 & 63).

(72) *ta-βyăt-t pl. *ti-βyat 'shoulder'

This word is somewhat difficult to reconstruct, as the *βy cluster clearly gives unexpected reflexes here and there. This seems like the most sensible solution.

(73) *a-zyeḍ pl. i-zŭyḍ-an or i-zăyḍ-an 'donkey'

There is some ambiguity about the realization of the plural and the position of the y in the stem. I think reconstructing this nouns with an existing plural pattern that shifts the position of the y in the stem, explains much of the variation.

(74) *a-zăʔăl?

See discussion under (12). I don't know how to reconstruct these forms yet, but I expect that a glottal stop plays a role. The plural in Tuareg i-šil-an should perhaps be understood as a *i-CŭCC-an: *i-zŭʔl-an with an accented stem vowel.

Reconstruction of the singular remains unclear.

(75) *t(a)-aẓ(ẓ)ul-t 'kohl'

Unchanged.

(76) *(a-)i/ezv̆̆m pl. *(i-)i/ezv̆̆maw-ăn 'lion'

Unchanged.

(77) *a-ẓar pl. i-ẓur-an 'root'

Languages that have aẓur in the singular probably analogically transferred the plural stem to the singular.

(78) *a-ză/aβu 'hair'

Perhaps the Awjili form is a spearate formation, but perhaps the sequence ăβu simply yields aw outside of Awjili. The Tuareg form is clearly unrelated.

(79) *a-ʔ(v̆̆)sif pl. *i-ʔ(v̆̆)saff-ăn 'river'

See (13).

(80) *a-ʔ(v̆̆)lim 'straw'

See (13) and (79).

(81) *a-făffəʔ pl. *i-făfʔ-an? 'teat'

I missed the Tuareg cognate e-făff pl. i-făf-an and the Zenaga cognate əvəffi /əfəffəʔ/ pl. uffän /əffan/ in my MA thesis.

This noun may have had a stem-initial vowel, as in Kabyle and Tamazight.

Tashelhiyt uff pl. uffan appears to be a different formation: *(a-)ufəf pl. *(i-)uff-an ?

(82) ta-răm-t pl. ti-ram 'meal'

If the stable initial vowel in Tashelhiyt is original rather than the regular prefix in Kabyle, the reconstruction may be *t(a)-irv̆̆m-t or *t(a)-erv̆̆m-t instead.

(83) *a-des pl. *i-dis-an 'side'

Prefix now shown to be *a-

(84) *a-les pl. *i-les-ăn 'fleece'

Prefix now shown to be *a-. It has a stable initial vowel in some dialects. This is presumably secondary. This is a verbal noun formation of the verb *ăləs 'to shear (wool)'. In Tuareg verbal nouns of CC verbs regularly have the form  e-CeC. It is more rare outside of Tuareg.

(85) *ta-ssir-t ~ ta-sser-t pl. *ti-syar  / *ti-sir-ăʔ 'grand mill'

The long consonant in Kabyle almost certainly has something to do with the y showing up in its plural. But the exact conditioning of this type of alternation is still not well understood. Compare also forms like inəlli pl. inəlwa 'thread'.

(86) *a-ẓed 'milling'

Prefix now shown to be *a-

(87) *ta-dis-t pl. ti-dis-en 'belly'

Probably the same root, but a different formation from (83).

(88) *a-ɣăyd pl. *i-ɣăyd-ăn 'billy goat'

Prefix now shown to be *a-

(89) *(a-)a/e/ifef 'sieving'

Stem initial *a, *e and *i cannot be distinguished in this context with just Northern berber cognates. Derived from the stative verb *ifif 'to sieve'

(90) *(a-)afắ/aʔ(ă)r 'cynodont'

Presumably a stem initial a and a stem-internal glottal stop.

(91) *a-fes pl. *i-fes-ăn 'hyena'

Many of the forms of this noun seem mutuall exclusive. Kabyle has a stem initial vowel, Tamazight and Tashelhiyt clearly do not. The Tashelhiyt plural ifasiwn is, hopefully, an innovation in Tasehlhiyt, as my rules of e~a alternation don't allow for the vowel a to shift to e in this context.

(92) *(a-)agăr pl. *(i-)igr-an 'field'

If this is a loanword from Latin ager 'field', reconstructing it with the a vowel before the a > e shift is very attractive. Especially the presence of many other agricultural words in Berber from Latin makes this likely.

If you do not accept the Latin origin, rhe forms *(a-)egər, *(a-)igər or *(a-)igăr are equally likely.

Note that 'reconstructing' a Latin loanword for Proto-Berber is of course difficult, but Latin influence may have predated the a> e shift.

(93) *ta-ʔ(v̆̆)luf-t pl. ti/ə-ʔ(ə)luf-ăʔ 'sorrow'

The distribution of the prefix vowel in Kabyle suggests the presence of a *ʔ. But this is of course quite indirect evidence.

(94) *a-mur pl. i-mur-ăn 'part, portion'

It seems possible that this is related to *ta-mur-t 'land, earth' as in 'portion (of land)'.

---

That's all of them! I'm not dissatisfied at all with my MA thesis' content, despite some mistakes and oversights. But by now, I certainly have changed my views on the reconstruction of many of these nouns. Any comments and thoughts are welcome of course.

Final diphthongs in the Qurʔān

One of the first quirks of Classical Arabic orthography one learns is the alif maqṣūrah or 'Broken Alif', this is, in modern orthography a word-final yāʔ without two dots below it, which you are to pronounce as an ā, e.g. ḥattā 'until, even' حتى <ḥtỳ>, kubrā 'bigger/biggest (fem.) كبرى <kbrỳ> and ramā 'he threw' رمى <rmỳ>. While this sign is graphically distinct from the y now, it originally was not, and was not in the Qurʔān. As you might expect, this spelling is a historical spelling. These words originally contained a y, which was lost at some point in the history of Arabic.

At some point in the history the word-final sequence -ay (or -aya, depending on when you think the word-final a was lost) shifted to -ā, but the spelling was never adjusted for this. A similar development happened with word-final -aw(a), as in *ʔatawa 'he came' > ʔatā, but for the w, this development before the establishment of the orthography, and therefore the verb is simply written with a final alif and not with a final waw: اتا <ʔtʔ>. The form with final w is attested in Pre-Islamic Arabic.

In Classical Arabic orthography, whenever the ā written with alif maqṣūrah is followed by a clitic, as for example in ramā-hu 'he threw it', the orthography stops writing the ā with an yāʔ and instead writes it as normally as an alif: رماه <rmʔh>. This is different in the QCT, where the yā remains written رميه <rmyh>. This has lead people to suggest that in the original language of the Qurʔān's composition,  the sound that was marked by the alif maqṣūrah in this context was still a diphthong or perhaps a ē (e.g. Thackston 1994: 275).

While it is tempting to allow for this archaic pronunciation to be part of the language of the QCT, I don't think the QCT conclusively points to this if we remain conservative in our approach. It is certainly true that the QCT orthography has a different norm than the latter Classical orthography, but if we accept that word-final <y> can be read as ā in certain archaic spellings, I don't think it's a stretch that this archaic spelling would also be employed word-internally.

There are, however, some reading traditions of the Qurʔān that actually pronounce all cases of ā < *ay(a) as ē, so certainly this pronunciation was of someone's language, but there's no reason to assume that this more archaic reading tradition is the more correct one for the language. Moreover, it should be noted that the readition tradition that Lameen Souag talks about even pronounces some instances as ē where the QCT actually has a final ʔ, namely in šāʔa شا <šʔ>, which originally came from *šayaʔa, so certainly this archaic pronunciation is not fully equal to the QCT orthography. This makes assuming that final y-verbs where pronounced with ē or ay(a) even more problematic.

There is yet another reading tradition, namely the Warsh reading tradition, that has ē in environments that is slightly more akin to the QCT orthography. Verbs that end in final <y> have an ē as well as the original sequence aya in nouns like Tawrāh (in documents that mark the Warsh pronunciation, a dot below the consonant preceding the vowel, marks the pronunciation ē). Interestingly, the prepositions that are written with a word-final <y> like ʔilā الى <ʔly>, ʕalā على <ʕly> and ḥattā حتى <ḥty> are pronounced with an ā. The difference between these prepositions and the verb can be understood from the shape in Proto-Arabic. The verbs of this type originally had a final triphothong /aya/, e.g. *ramaya > ramē whereas the prepositions of this type ended in final /ay/, e.g. *ʕalay > ʕalā . It seems that the dialect of the reading tradition underwent a shift *aya > ē but *ay > ā. [1]

 Finally, if we look at early Qurʔānic codices, such as the Samarkand Codex (1st word on the 8th line, Q7:38) as well as the Ṣanʕāʔ Qurʔān (Q2:193) we find clear examples of the alif maqṣūrah being mistakenly written with an alif, in ḥattā 'until, even'. This suggests that to the scribes of this manuscript, the alif maqṣūrah was clearly pronounced as an [ā]. One wonders if the fact that this word is pronounced with ā in the Warsh reading tradition is related to this.

But other explanations are also possible. For example, ḥattā never appears in an environment where the etymological y resurfaces, unlike with verbs such as ramā which has it reappear in the paradigm: ramaytu رميت <rmyt> 'I threw'.

In summary, the QCT orthography allows for reading final y verbs as ē more easily than the Classical Orthography. However, just because the orthographic practices differ form the later Classical Standard, does not mean it necessarily reflects the actual spoken form. It is therefore possible that, despite forms like <rmyh>, it was still pronounced ramā-hu, and that the <y> was simply spelled etymologically.

We must conclude that there is no convincing evidence in favour of reading these forms either with ē or with ā. The Samarkand Codex shows one etymological *ay written as ā, but in a position where the Warsh reading tradition, which generally has ē would also read it as ā. If we were to find example of y final verbs spelled with a final alif in old Qurʔān manuscripts, that would be good evidence in favour of it being pronounce ā, but without those examples, we just have an absence of evidence for an ē or ā pronunciation.

---

The vast majority of the modern dialects follow the developments of the traditional reading of the QCT, with the exception that final w verbs have generally merged completely with final y verbs. However, in Yemen in the Jabal Rāziḥ dialect, all cases where QCT has an alif maqṣūrah, the dialect pronounces it as ē, so: ramē, ḥattē, ʕalē, ʔūlē etc. And it does not shift the vowel as in šāʔa to **šēʔ. The dialect therefore closely mirrors the QCT orthography (but not necessarily the QCT language) (Behnstedt 1987: 133-134).

 

EDIT:

In the comments Ahmad Al-Jallad pointed out a very good argument that clearly swings the reading of final -ay as ay or ē over final ā, namely: Rhyme in the Qurʔān. There are several Surah's or parts of Surah's that rhyme on -ay, these never interchange with final -ā. Some examples are:

Q79:15-32 ending in (rhyme is broken at Q79:33, but continues again on Q79:34-46) :

mūsā, ṭuwan, ṭaġā, tazakkā, fa-taxšā, al-kubrā, wa-ʕaṣā, yasʕā, fa-nādā, al-ʔaʕlā, wa-(a)lʔūlā, yaxšā, baḏā-hā, fa-sawwā-hā, ḍuḥā-hā, daḥā-hā, wa-marʕā-hā, ʔarsā-hā

 Now look at the QCT spelling:

<mwsy, ṭwy, ṭġy tzky, ftxšy, ʔlkbry, wʕsy, ysʕy, fnʔdy, ʔlʔʕly, wʔlʔwly, yxšy, bḍyhʔ, fswyhʔ, ḍḥyhʔ, wmrʕyhʔ, ʔrsyhʔ>

In Arabic rhyming schemes the -hā feminine direct object clitic and -hu the masculine direct object clitic are often 'invisible' to the rhyme. All the words preceding the -hā end in a -ay, (which in Classical orthography would actually be written with <ʔ>, rather than <y>). This is a practice already attested in Pre-Islamic Old Arabic, as found in KRS 2453 as discussed in Al-Jallad 2015.

All examples are: Q20:2-13, 15-24, 36-38, 40-84, Q53:1-56, Q70:15-18, Q75:31-40, Q79:15-46 (without 33), Q80:1-10, Q87, Q92, Q93:1-8 Q96:6-14.

The only consistent deviation I see in the Qurʔān, is femine elatives with y as their third root consonant.

Feminine elative in Arabic have the pattern CuCCā, e.g. kubrā 'biggest'. This final ā is from an original ay sequence, Proto-Arabic *kubrayu, as such, it is transcribed with a <y>: كبري <kbry>. The only exception to this are feminine elatives that have y just before this -ay suffix, in those cases it is written with <ʔ>, e.g. ʕulyā عليا <ʕlyʔ>, dunyā دنيا <dnyʔ> etc.

This is interpreted most easily explained as a dissimilatory rule where the sequence *-yay becomes *-yaw and subsequently -yā. This rule is certainly active in the Quranic Consonantal Text, where dunyā is written as in the regular Classical Arabic orthography. However, it is only these dissimilated -ay sequences that show up in these rhymes. This suggests that the language that these rhyme Surah's were composed in did not undergo the dissimilation that is present in the QCT orthography and that dunyā was simply pronounced dunyē or dunyay.

Examples of these kinds of words "breaking" the rhyme are:

Q53:29 dunyā (also Q20:72, Q79:38, Q87:16)

Note: In the Warsh reading tradition these words are actually dotted to be read as ē, which seems to confirm a form as dunyē.

Note: In the Warsh reading tradition, that verse is Q53:28! In the standard Cairo edition, Q53:28 ends at a word that breaks the rhyme: šayʔan شيا <šyʔ>. This is the only exception to -ay rhyming schemes not accounted for. I'm tempted to take the Warsh unsplit form as the original (this also means that the numbering of verses in the Warsh reading is different!).

Q53:44 ʔaḥyā
Note: Likewise dotted for ē in Warsh.

So my new conclusion is that the QCT absolutely provides evidence that the old diphthong -ay was pronounced not as -ā but as ē or simply -ay. Even nouns that have an -ā in the Qurʔānic orthography but have -ay etymologically, dunyā and ʔaḥyā, were probably pronounced with ē. This agrees with the Warsh reading tradition, and deviates from the QCT orthography.

This close match between the Warsh reading tradition and the clues from the QCT make me lean strongly to accepting the pronunciation of Warsh in this aspect as original, ḥattā and ʕalā were probably pronounced with ā, which also explains the spelling variants mentioned.

It is worth adding that in the Jabal Rāziḥ Yemeni dialect simply has dunyā, and thus also here once again corresponds closely to the dialect of the QCT orthography (but not with that of the Qurʔānic language!).

---

[1] It is interesting to note that the word matā متى <mty> 'when?' is pronounce matē in the Warsh reading tradition, which suggests a form *mataya. Outside of Arabic, e.g. in Hebrew this form also yields an unexpected reflex if we assume an original *matay, as it it mātāy in Hebrew. These things may be related in some way. Note that other question words also take a final -a, e.g. ʔayna 'where?'  and kayfa 'how'.

 References

Behnstedt, P. 1987. Die Dialekte der Gegend von Ṣaʿdah (Nord-Jemen). Wiesbadne: Harrassowitz.

Thackston, W. M. 1994. An Introduction to Koranic and Classical Arabic. An Elementary Grammar of the Language. Bethesda, MD: IBEX Publishers.