Thursday, February 11, 2021

Review: "Inventing the Berbers"

I finally got a chance to read Ramzi Rouighi's Inventing the Berbers recently; much food for thought.

The book is primarily the history of a name: how did certain people in North Africa come to be called "Berbers", and how did the reference and connotations of this label change over time? Viewed as such, it has a good deal of useful material. He argues that, rather than being derived directly from Latin or Greek "barbari", the label was transferred from East Africa to Northwest Africa as the Arabs moved west; its original associations would be with slavery rather than with barbarism as such. (Traces of the original usage persist: in Nubia, as I first learned on a trip to Aswan, "Berber" is still understood to mean "Nubian"!) In the early medieval period, it was used primarily for rebels and enemies on the fringes; groups with a closer involvement tended to be referred to by more specific terms. Ibn Khaldun's usage is more complex, reflecting Andalusi practice as it emerged in the context of elite competition between Berber and Arab noble families, but shows clear traces of the older tendency to reserve it for "outsiders" to the ruling elite. The modern European usage of the term comes essentially from Ibn Khaldun as filtered through De Slane's essentialism (which turned Berbers into a "race") and subsequent academic and ideological debates, largely in the context of the French colonization of Algeria.

In the penultimate chapter, however, he lays his cards on the table, presenting the term Amazigh as a mere relabelling of the neo-Khaldunian concept of "Berber", constructed with insidious intent and making an already misleading discourse even more ahistorical:

In the early 1950s, a few specialists proposed to replace “Berber” with “Amazigh,” the name some people in northern Morocco had.... “Amazigh” could not fully conceal its colonial birthmark, however. Its rejection of Arab imperialism of centuries past, its search for an authentic indigenous category, and its reliance on the fruits of colonial historiography, epigraphy, and linguistics to do so are all telltale signs. Calling for name change could have led to the realization of the historicity of all names and from there to the historicity of Berberization. It did not... “Amazigh” (indigeneity) was the parting gift of a dying colonialism to the frail nationalisms it had never accepted. Pulling the rug from under “Algeria” and “Morocco,” which as the colons repeated were new and artificial, “Amazigh” dealt a blow to anticolonial nationalism.

The 2-page discussion of “Amazigh” is unacceptably simplistic, especially after multiple chapters of careful examination of the changing semantics of "Berber". The author would have been better off omitting the term entirely than giving it such a caricatural treatment, massively understating the geographic distribution of the term (not just northern Morocco but as far off as northwestern Libya...); his medieval focus cannot entirely excuse the omission, as this term is (less frequently) attested in the medieval period. A proper examination - and, yes, historicization - would have been all the more valuable given that the term was used as an endonym in many regions long before the emergence of the modern trans-national ideology, whereas "Berber" has not been adopted in ordinary Berber speech anywhere, remaining an exonym, and usually an exclusively learned one at that.

Reading as a linguist, I can appreciate the attention given to semantic shifts and to the arbitrariness not only of the sign but of the signified. But as a historical linguist, it feels rather at cross-purposes to the questions of interest to me. Fundamentally, I don't much care which ethnic label people identify or are identified with: for me, "Berber", like "Arabic", is primarily useful as a linguistic category. And its referent has a history starting far earlier than the earliest attestation of "Berber", "Tamazight", or any other label one might choose to apply to it. It is necessary and appropriate to historicize such labels - to be aware that Masinissa or Dihya or Fatma n'Soumer were not acting in the name of some kind of Amazigh nationalism, and may not even have been familiar with "Amazigh" as a name, let alone as an identity. But how this relatively close-knit language family spread, and retreated, remains a historical question, of interest to archeologists and population geneticists as well as linguists, which an exclusive focus on ethnic labels erases.

It should, however, help to provoke reflection on the appropriate choice of label for this language family. "Berber", neutral though it undoubtedly is in English or French, does have a problematic history; the derivation from "barbarian" may be inaccurate, but this book really underscores the extent to which its usage in Arabic has been overwhelmingly negative and "othering" for most of the region's history. "Amazigh" does not have this problem, but is strongly associated with a projection of shared ethnicity into the past which risks distorting our picture of language spread. In an ideal world, one might prefer a purely geographical label ("Northwest African"?), or, better yet, a purely linguistic one (iles-languages, after the usual word for "tongue"?) In practice, however - here as elsewhere - it seems preferable to live with the occasional misunderstandings caused by the use of a well-known "ethnic" term than to confuse the public with a completely novel one.

Tuesday, September 29, 2020

A fable written in Korandje

Yesterday, H. Yahiaoui posted what might be the first continuous story written down in Korandje by a 1st-language speaker (translated from a cynical little fable in Arabic): The Donkey, the Lion, and the Tiger. In this text, we clearly see the "consecutive aorist" used after imperatives but not after perfectives: contrast n-as abəqqạ nə-m-t-as "giveimperative him a slap and tellirrealis him" with a-hh-ana a-tt-asi lit. "he askedperfective and said to himperfective". More crucial among this text's points of interest, however, is the placement of spaces. Word boundaries are surprisingly tricky to determine in Korandje. Plenty of elements could be analysed as bound forms or just as free forms with a somewhat restricted syntactic distribution, and it's hard to decide which is which. A text like this provides suggestive (though certainly not conclusive) data on where speakers perceive them. A few generalizations quickly emerge. In the verb word:
  • Subject markers are written as prefixes to the verb or MAN marker (2Sg n, 3Sg a, etc.)
  • The aspectual auxiliary ba, which turns perfective into perfect and imperfective into progressive, is written as a separate word - but only in contexts where the b is preserved; contrast ənnmər ba bə-kkạ-γ "the tiger is hitting me" with a-(a)-b-kkạ-γəy "he is hitting me".
  • Otherwise, mood, aspect, and negation (MAN) markers are written as prefixes to the verb (Neg s, Prosp (b)aʕam, etc.)
  • Directionals (ti "hither" are written as suffixes to the verb.
  • Object pronouns (2Sg ni, 3Sg ana, etc.) are written as suffixes to the verb.
  • Oblique pronouns (2SgDat nisi, 3SgDat asi, etc.) are usually written as suffixes to the verb word, but in one case (kəs γəys "leave to me") as an independent word, plausibly reflecting its less closely bound status.

In the noun phrase:

  • Genitive n is written as a prefix to the head noun.
  • Possessive pronouns are written as prefixes to the head noun (1Sg ʕan, etc.)
  • The indefinite article (or numeral) fu "a, one" is written as a suffix to the noun it quantifies.
  • "Other" (fyạṭən), despite historically containing "one", is written as a separate word.
  • Demonstratives are written as suffixes to the noun phrase (γu "this", etc.)
  • Dative si and locative ka are written as suffixes to their objects.
  • The focus marker a is written as a suffix to its noun phrase.
  • The identificational copula (aγu "this is", etc.) is written as an independent word, despite historically incorporating the focus marker.

Pending more data, the following cases seem sui generis:

  • səndza-n-a (Neg.Cop-2Sg-Foc) "it's not you who..."
  • mu-kunna-ni (what.Rhet-find-2Sg) "what's wrong with you?"
  • ku-xəd (each-when) "whenever"

For those who can't read the original, here's a transcription of the fable:

  1. Fəṛka a-ddər izmmi-s a-yzʕəf a-hh-ana, a-tt-asi: "Maγạ səndza-n-a lγabət n-uγ bya?"
  2. Izmmi a-tt-asi: "Iyyah… mu-kunna-ni, tuγ ba yzra?"
  3. Fəṛka a-tt-asi: "Nnmər ba bə-kkạ-γ ʕam-mu-ka ku-xəd a-ggwa-γəy, a-m-ti 'Maγạ nə-ss-aʕam-ḍəb taššəyt?', maγạ a-(a)-b-kkạ-γəy kʷəl ana?? Aha tuγa taššəyt-γ ʕamḍəb kʷəl aγəy?"
  4. Izəmmi attasi: "Kəs γəys ləxbạ-γu, nə-s-bə-zzu lhəmm haya."
  5. Aywa ləxʷəddzi(d) izəmmi a-kbʷəy ənnmər a-hh-ana "Tuγ-a taššəyt-γ n-ləxbạ?"
  6. Nnmər a-tt-asi: "ʕa-b-talla γar əssəbbət ndzuγ ʕa-b-kkạṛ-ana wəxḷaṣ.."
  7. Izəmmi a-t ənnmər-si: "Təlla ssəbbət fyạṭən-ka, a-a-ybən… T-as a-m-zu-t-nis əttəffaħ-fu ndzuγ, ndza a-zzu-t-a-nis yạṛạ, n-as abəqqạ nə-m-t-as 'Maγạ nə-ss-aʕam-zu-t-ana tirəy?' Ndza a-zzu-t-a-nis tirəy, n-as abəqqạ nə-m-t-as 'Maγạ nə-ss-aʕam-zu-t-ana yạṛạ?'"
  8. Nnmər a-žžawb-ana a-tt-asi "Lfikrət-f hannu aγu."
  9. Am-bibya ənnmər a-kbʷəy fəṛka a-tt-asi: "Zu-t-γis əttəffaħ-fu."
  10. Fəṛka a-nnəg-aka mliħ a-hh-ana a-tt-asi "Waš ʕa-m-zu-t-ana tirəy wəlla yạṛạ???"
  11. Nnmər a-ttəmtəm an-nin n-tiri a-tti "Tirəy yạṛạ…"
  12. A-ħħərrəm an-kambi ạ-kkạ fəṛka ndza abqa-fu, a-tt-as "Maγạ nə-ss-aʕam-ḍəb taššəyt???"
  13. *** Uγ ba b-iḍləm a-ss-a-bə-ttəlla əssəbbət ndzuγ a-b-yəḍləm.
In English:
  1. The donkey went to the lion angry and asked him: "Hey, aren't you the chief of the forest?"
  2. The lion told him "Yes... what's wrong with you, what has happened?"
  3. The donkey told him "The tiger is hitting me on my face every time he sees me, saying 'Why won't you wear a cap?' Why is he hitting me?? And what cap would I wear anyhow?"
  4. The lion told him "Leave this affair to me, don't worry about it at all."
  5. So when the lion met the tiger, he asked him "What's the issue of this cap?"
  6. The tiger told him "I'm just looking for an excuse to hit him, that's all."
  7. The lion told the tiger: "Look for another excuse, it's (too) obvious... Tell him to bring you an apple so that, if he brings it to you yellow, give him a slap and tell him 'Why won't you bring it red?' If he brings it to you red, give him a slap and tell him 'Why won't you bring it yellow?'"
  8. The tiger replied "This is a good idea".
  9. The next day the tiger met the donkey and told him "Bring me an apple."
  10. The donkey looked hard at him and asked him "Should I bring it red or yellow?"
  11. The tiger mumbled under his breath "Red, yellow..."
  12. He lifted up his hand and slapped the donkey and said "Why won't you wear a cap?"
  13. *** An oppressor doesn't need an excuse to oppress.

Tuesday, March 31, 2020

Nəskibən: "You don't appear any more"

"Nə-s-k-ibən" (2SG-NEG-anymore-appear) "You don't appear any more!"

I heard this sentence several times during my fieldwork in the Saharan oasis of Tabelbala. Parsing it was easy enough, but making sense of it took more flexibility: at first I thought I must have misheard. I've never heard anyone in England or America or France say anything like "You don't appear any more!"; yet there it turned out to be a stock phrase.

It makes sense once you unpack the presuppositions. A man should appear in public regularly - in town, at the market, at the mosque, en route to other places. But, in that slow-paced small town, doing so is an act of socializing, not just a stage in an errand: you don't just pass someone you know by without at least stopping a minute to say hi and share news. Not appearing in public for some time is an event noteworthy in itself, and people can and will criticize you for it if you don't have a valid excuse like illness.

That's not really how it works in Paris or London. You might be obliged to "appear" at your office, but not for strictly social reasons. They might notice your absence at your regular pub or your clubhouse or something, but certainly not on the street. Even in such cities, though, we normally spend much of our day before the gaze of others - if not exchanging greetings and gossip, at least seeing and being seen.

But now things have changed. On Facebook, a friend in Tabelbala recently made a post to urge social distancing, translating the message "Stay at home!" into Korandje: gwạ nən gạ ka! The first response was chaffing from a more frivolous friend, telling him that he's been social distancing so much that "nə-s-k-ibən"!

I imagine the lockdown in Tabelbala is less rigidly enforced than it could be - surrounded by 100 km or more of empty desert in every direction, it is impressively isolated without it. But otherwise, we're all in the same boat now: we don't appear any more. Except online.

What kind of expectations and presuppositions will that create, over weeks that may stretch into months? When we all emerge from our hideouts, will we find it worthy of comment if people don't appear in their usual social media sites or chat forums?

Saturday, March 21, 2020

W-deletion in Arabic

In Arabic, triliteral verbs starting with w- often drop the w- in the imperfect ("present"), and in a few related forms like the verbal noun: وجد wajada "he found" vs. يجد yajidu "he finds", وزن wazana "he weighed" vs. يزن yazinu "he weighs"... But not always: contrast وسن wasina "he fell asleep" vs. يوسن yawsanu "he falls asleep", وجز wajuza "it was brief" vs. يوجز yawjuzu "it becomes brief". Going through a dictionary, it becomes obvious that the primary determining factor is the vowel: verbs with an imperfect in -i- drop the w, while others keep it. (Proviso: verbs which originally had -i- turn it into -a- if the third consonant is "guttural", ie pharyngeal or glottal: thus وقع waqa3a "it happened" vs. يقع yaqa3u "it happens" from *yaqi3u, contrasting with وجع waja3a "it hurt" vs. يوجع yawja3u "it hurts" with original -a-.)

Empirically, this seems to work fine. But it doesn't make sense to me historically. Why should an i in the second syllable correlate with the absence of a w in the first syllable? Any ideas how such a sound change could plausibly have taken place?

Thursday, January 30, 2020

Unifying Mubi -oo- plurals

NB: Sorry, no tone marking today – might throw it in later.

We’ve seen two productive plural allomorphs characterized by round vowels: BVCV > BuCooC vs. BVVCV [-front] > BooCuC. Let’s see where -oo- shows up in the plurals of longer nouns.

Nouns of the form BVCVD(V) [-front] tend to take a plural in BuCooDu (the reduplicative plural horoɗyo > horoɗyuc, discussed last time, seems to be isolated):

  • jorol “fox” > juroolu
  • ɗoloso “lynx” > ɗuloosu
  • kabada “red fig” > kuboodu
  • jubugo “arrow” > juboogu
  • wasaga “thread” > wusoogu

In two cases, a suffix -k is added, with what seems to be dissimilation of *-guk > -yuk:

  • fidak “mat” > fudooyuk
  • cagada “hut” > cugooduk

Formally, despite the shape and the front vowel (which may lead us to rethink the conditioning), the following cases fit this pattern as well:

  • kurri “chicken” (assimilated from *kurɗi) > kurooɗuk
  • urde “granary” > urooduk

In another two cases, both ending in -k, the expected final vowel is omitted:

  • tamak “sheep” > tumook
  • koɗogo “toe-ring” > kuɗook

And, as we saw last time, in one case the final consonant is irregularly reduplicated:

  • bodol “road” > budoolul

If a long vowel is present and this plural type is used, vowel length is normally preserved; thus BVCVVDV yields BuCooDu – sometimes, as above, the final vowel is omitted to end in -k:

  • ɗyubaago “blind” > ɗyuboogu
  • sinyaaro “cat” > sinyooru (the i in the first syllable is probably caused by the following palatal)
  • duwaago “dorcas gazelle” > duwok (with unexpected shortening of the last vowel)

but BVVCVDV yields BooCuD(u):

  • gaayimo “wildcat” > gooyumu
  • kaarumo “fingernail” > koorum

On the other hand, we also find the variant plural gaayimo > guyoomu, suggesting that BuCooDu is starting to be generalized.

What about longer nouns? In those, frontness is irrelevant...

A nasal followed by a voiced stop (except in the Arabic borrowing (a)ngumbul "calabash" > (a)ngunoobul) behaves like vowel length, so BVNDVF(V) > Bo(o)CDuF:

  • tengil “calf” > tongul
  • minjilo “Mubi person” > monjul
  • humbuk “hedgehog” > hoombuk

In the few relevant examples available, BVCVDFV(GV) > BuCoDFu(G), whether D is nasal or not:

  • gomorko “basket” > gumorku
  • suwangot “Arab” > suwongu
  • aranjala “kidney” > uronjul

Otherwise, four-consonant nouns BVCDVF(V), BVCV(V)DVF(V) overwhelmingly (15 out of 24 examples) map to BuCooDuF:

  • ɗurgul “donkey” > ɗuroogul
  • kalman “in-law” > kuloomun
  • sunsuna “tale” > sunoosun
  • kasagar “sword” > kusoogur
  • kodoguno “sorcerer” > kudoogun
  • giraakumo “molar” > gurookum

Now we can finally start to put things together: all of these seem to be mapping to subsets of a notional template CuCo(o/C)CuCu, in a predictable fashion.

If the first syllable is long or includes the first half of a prenasalized stop, you drop the initial Cu:

BVVC > Cu[BooCuC]u
BVVCV > Cu[BooCuC]u
BVVCVD(V) > Cu[BooCuDu]

If the first syllable is open and the second one is closed, you get oC instead of oo:


Otherwise, you just proceed from left to right, always respecting the requirement that the output have at least 2 syllables:

BVCV > [BuCooC]uCu
BVCDV > [BuCooDuk]u
BVCVC(V) > [BuCooDu]Cu
BVCVVD(V) > [BuCooDu]Cu
BVCDVF(V) > [BuCooDuF]u
BVCV(V)DVF(V) > [BuCooDuF]u

So we’re starting to get somewhere. But this opens up a new can of worms: do some geminate-internal plurals belong here too? And where do those BaaCaC plurals fit into the system now? Those questions will have to wait for another time.

Tuesday, January 28, 2020

Reduplicative plurals in Mubi

Yesterday we saw that the dominant plural type for CVC / CVVCV stems, CVVDvD, can be given a unified analysis. How does this generalize to other stem types?

Plurals with final reduplication, dominant for CVC / CVVCV stems, are much less common for other stem shapes. For CV(C)CV stems, however, we do find a reasonable number of reduplicative plurals, this time in -o(o)C:

  • sùwá "calabash" > sùwòw
  • tògò "skin" > tùgók
  • rìwwí "song" > rèwòw
  • màbò "old" > mùbóop
  • cóɓɓì "lance" > cúɓóop
  • lánjá "friend" > lúnjóoc

(There is also one reduplicative plural in -eC - kúrɗyí "buttock" > kòrɗyèc - which seems difficult to reduce to the same pattern. It suggests that height harmony may apply to short vowels too, in which case we might blame sùwòw above on the influence of the semivowel.)

In Mubi, final VVC seems to be rare in simplex words. In monosyllables, it occurs only in loans, while elsewhere, it mainly shows up in imperfectives (cf. Jungraithmayr 2013:31). We may very tentatively suppose that in some common plurals it has been reduced to VC. Reduction of vowel length in final VVC also allows us to revise our analysis from yesterday to take into account bàŋ "mouth" > bòŋúŋ; a surface CVC stem may be underlying *CVC or *CVVC, and the vowel length in the plural reflects that rather than changing the stem.

Apart from the issue of vowel length, the rules outlined so far would yield the following (or, if we assumed that height harmony applies to short vowels too, the forms in brackets):

  • sùwòw (*sòwòw)
  • *tògók
  • *rìwwòw (*rèwwòw)
  • *màbóop (*mòbóop)
  • *cóɓɓóop
  • *lánjóoc (*lónjóoc)

A rule spreading roundness along the lines of a,o > u / _Coo seems feasible, judging by the lexicon, and fixes some of the problems above (not sure what's going on with "song"):

  • sùwòw
  • *tògók
  • *rìwwòw
  • mùbóop
  • *cùɓɓóop
  • lúnjóoc

But that can't be the whole story here. If bòdòl (see below) is a possible word, then so is *tògók; and there's no general ban on geminates in these positions, cp. wíccáak "to jump about while dancing (impf.)". Instead, there seems to be a templatic element here, whereby the vowel before the plural o/oo has been generalized to be u irrespective of the input, and the consonant before it must not be geminate. Insofar as these involve fixed vowels inside the stem, that strengthens the case for a comparable analysis of the previous data, which is feasible for CooDuD plurals. Integrating the -oC plurals seen above, we can thus reformulate the reduplicative plural as follows:

sg. BV(C/D)DV => pl. Bu(C)DooD (with sporadic reduction to -oD)
sg. BVVD(v) [-front] => pl. BooDuD
sg. BVVD(v) [+front] => pl. BVVDaD

Note that the internal vowel lengths and positions of the singular are preserved in the plural in all three of these; only stem-final vowels are affected.

For other stem shapes, reduplicative plurals seem to be sporadic, mostly involving words whose last consonant is a liquid. I can only find 3 clear instances of -uC in Jungraithmayr's lexicon for them, and 1 of -aC:

  • bòdòl "road" > bùdòolúl
  • kòròojó "small calabash" > kòròojúc
  • ɗíngírí "branch" > ɗìngéerúr
  • gúrlí "testicle" > gòrlàl
Once again, the internal vowel positions of the singular are preserved in the plural, if not necessarily the lengths. However, the predictions of our previous hypothesis are thus slightly off in two cases; we would have expected:
  • *bòdòolúl
  • kòròojúc
  • *ɗìngéerár (or, if we force a -uD plural despite the front vowels, *ɗìngúurúr)
  • gòrlàl

For ɗìngéerúr, no explanation for the -uC comes to mind; it is tempting to hope that this one is a typo. For bùdòolúl, however, a templatic explanation for the irregularity is at hand: CuCooCuC is, as we will hopefully see in a later post, the dominant plural template in Mubi for four-consonant nouns, which behave rather differently from what we've seen so far.

Now we've covered final reduplication plurals (glossing over a couple of irregulars, notably gìn "face" > gèenín / gànàn). This leads on naturally into another set of plurals, likewise featuring long vowels but without reduplication; if everything goes well, we can look at those next time.

Monday, January 27, 2020

Pluralizing Mubi biradical nouns

(Attention conservation notice: Kind of technical...)
NB: Updated with important corrections shortly after posting, following discussion with Marijn van Putten.

To better understand the Mubi plural system (following up on last time), let's start with two-consonant nouns of the form CVVCV and CVC (not CVCV, which has different default plural templates). Most of these take a CVVDvD plural with reduplication of the final consonant, with the vowels accounted for by the following correspondences:

  • CeeDi > CaaDaD, e.g. gèébí "horn" > gàabàp (final consonants automatically devoice)
  • CiiDi > CeeDaD, e.g. lìísí "tongue" > lèesàs
  • CooDi, CuuDi, CuD > CooDaD, e.g. fùúdí "thigh" > fòodàt
  • CooDo > either CooDaD or CooDuD, e.g. góoró "throat" > gòoràr; zòoró "tributary river" > zòorúr
  • CaD, CaaD, CeD, CoD, CaaDo > CooDuD, e.g. fáaɓó "breast" > fòoɓúp
(No such nouns end in -a, -e, or -u.)

There seem to be no instances in the lexicon of *iiCa or *uuCa (the only case of iiCaa is in an Arabic loanword); nor are there cases of *aaCi or *aaCu, except in manifest loans like áarìt pl. àwáarìt "devil" (Arabic ʕifriit), with the odd exceptions of gíráakúmò "molar" and káarúmo "fingernail". We may thus assume that there is backwards spread of height from the short vowel to the long vowel preceding it. The latter may be mid but must not be the opposite height to the short one, with conflicts resolved by changing to mid and harmonising for frontness and roundness:

  • ii > ee / _a
  • uu > oo / _a
  • aa > ee / _i
  • aa > oo / _u

Generalizing the latter point, it also looks like roundness spreads backwards to a long vowel in a preceding syllable, i.e.:

  • ee > oo / _o, u
  • ii > uu / _o, u

On a quick glance through the lexicon, the only exceptions I can see to this are vocalised semivowels as in hàaɗáw, pf. héeɗû "knead". This allows us to reinterpret CeeDi above as /CaaDi/, and CooDo as either /CeeDo/ or /CooDo/. We thus get a nice distribution of allomorphs: -aD if the singular contains an underlying front vowel, -uD if not:

  • *CaaDi > CaaDaD
  • CiiDi > *CiiDaD
  • CooDi > CooDaD
  • CuuDi, CuD > *CuuDaD
  • *CeeDo > CooDaD
  • CooDo, CoD > CooDuD
  • CaD, CaaD, CaaDo > *CaaDuD
  • CeD > *CeeDuD

Now we can almost rewrite the rule rather simply to unite all these cases:

sg. CVD, CVVDv [+front] => pl. CVVDaD
sg. CVD, CVVDv [-front] => pl. CVVDuD

The problem with this reformulation is that CooDo < *CeeDo nouns take a plural in CooDaD, not *CeeDaD. Perhaps this is best understood as an effect of the deleted final vowel: the delinked /o/ relinks to the previous vowel. But if so, this only applies for rounding, not for fronting...

This information, missing from the published grammar, gives us some of the necessary background to understand where those pesky CuCooDuC plurals might come from... But that's a story that still needs work, to be continued later perhaps.