Tigre between ejectives and pharyngealization

There is some debate over the original pronunciation of the "emphatic" consonants (Arabic ط ض ظ ص ق) in Semitic and more generally in Afroasiatic: were they ejective as in Amharic, or pharyngealized/uvular as in Arabic? For a number of reasons, such as that in proto-Semitic they did not show a voicing contrast, the general opinion is that they were glottalized. Yet pharyngealized consonants show up not just in Arabic and neo-Aramaic but even in Berber, which would on the face of it suggest that the feature predates proto-Semitic. Either we have to suppose independent parallel development, or we must assume that Berber ejectives turned into pharyngealized consonants under the influence of Arabic. The latter seems more probable, but only if we can show that it is indeed plausible for a language to make such a change as a result of widespread bilingualism in Arabic.

It turns out that Tigre, the main language of northern Eritrea, offers a concrete example of just that. The inland plateau dialect of the Mansa`, commonly considered as standard, is described by Raz (1983) as having four ejectives k' (usually [ʔ]), t', s', and č̣ , and no pharyngealized or uvular consonants. You can hear an example of standard Tigre here, which seems consistent with his description. The coastal Hirgigo dialect spoken around Massawa, however - as heard in these Learn Tigre YouTube videos, however, show a rather different situation. ḳ is simply [q] (as in "elbow", "neck", "thigh"), ṭ is [tˤ] (as in "goat"), ṣ is [sˤ] (as in "white", "black", "back"); only for č̣ can you occasionally hear a slightly ejective realization [tʃ] ~ [tʃ'] (as in "fingers" or "fingernails"). The result is a good deal easier for an Arabic speaker to pronounce! This should not be too surprising: the port of Massawa has had extensive contact with Arabic speakers for many centuries. In fact, it's said to be the place where some of the first Muslims, seeking refuge from the persecution they were suffering in Mecca, landed on their way to the Abyssinian court. Such a diversity of emphatic consonant realizations within a single language confirms in turn that it is plausible for the habit of pharyngealizing emphatic consonants to be transferred from a language to its neighbors.

In search of the missing radical: a piece of Berber historical morphology

Berber normally has no glottal stops (ء = ʔ) – in fact, Chafik suggested that this was why North Africa favours the Warsh reading of the Qur'an, in which most glottal stops are omitted. However, it turns out* proto-Berber did have glottal stops - and you can still see their footprints on the verbal system.

Berber languages normally have three basic aspect/mood forms:

• the “aorist” (or “simple imperfect”), used mainly for hypothetical events (“eat!”, “I will eat”, “I would eat”...);
  
• the “preterite” (or “simple perfect”), used mainly for past events conceived of as wholes (“I ate”, “I have eaten”);
  
• the “intensive” (or “intensive imperfect”), used for events ongoing at the time being referred to, irrespective of tense (“I eat”, “I am eating”, “I was eating”, “keep eating!”)

Usually, you can predict the preterite and intensive from the aorist. For three-consonant roots – eg lmd “learn”, a widespread Phoenician loanword – this is how it works in Tuareg (Tahaggart):

• Aorist: ǎlməd “learn!”
  
• Preterite: (y)-əlmǎd “(he) learned” (change the vowel pattern)
  
• Intensive: (i-)lammǎd “he is learning” (double the middle consonant)

Tuareg has kept a distinction between two short vowels, ǎ and ə; but most varieties have just merged the two, so there is no difference in three-consonant roots between the aorist and preterite. So in Siwi, for example, you get:

• Aorist: əlməd “learn!”
  
• Preterite: (y)-əlməd “(he) learned”
  
• Intensive: (i)-ləmməd “he is learning”

(Students of Akkadian/Assyrian/Babylonian will be getting a sense of déjà vu now...)

But some verbs have two consonants rather than three. Looking at Siwi I noticed that, if the verb had two consonants and no long vowels, there seemed to be two possibilities for the intensive, not just one; contrast:

• Aorist: fəl “leave!”
  
• Preterite: (y)-əfla “(he) left”
  
• Intensive: (i)-təffal “he is leaving”

vs.

• Aorist: ləs “wear!”
  
• Preterite: (y)-əlsa “(he) wore”
  
• Intensive: (i)-ləss “he is wearing”

So why the split?

Well, looking at the intensive forms, you see that in fəl you double the first consonant, while for ləs you double the second one. If you wanted to try to relate these to three-consonant verbs, you might think of something like:
- fəl < *Xfl
- ləs < *lsX

But if you look at Siwi on its own, there seem to be a lot of problems with this idea: in particular, why would the preterite of fəl end in -a?

Looking wider provides some answers. It turns out that in Tuareg – like Kabyle, and Tashelhiyt, and Ghadamsi, and a few other varieties – these verbs are distinct in the preterite too, and they are distinguished in exactly the way you'd expect from that little piece of internal reconstruction:

• Aorist: əfəl “leave!”; əǵən “kneel!”
  
• Preterite: (y)-fǎl “(he) left”; (y)-ǵǎn “(it) knelt”
  
• Intensive: (y)-ffal “he is leaving”; (y)-ǵǵan “it is kneeling”

vs.

• Aorist: ǎls “wear!”; əsəl "hear!"
  
• Preterite: (y)-lsa “(he) wore”; (y)-sla "he heard"
  
• Intensive: (y)-lass “he is wearing”; (y)-sall "he is hearing"

It's just that in Siwi – and Mzabi, and Chaoui, and Tarifit, and all the other Zenati Berber languages – the preterites of these two verb classes are merged, so they both end in -a. So our internal reconstruction is looking good... but what consonant might have been lost?

Zenaga, the Berber language of Mauritania, gives us part of the answer. In Zenaga, they look like this:

• Aorist: ägun “kneel!”
  
• Preterite: (y)-ugän “(it) knelt”
  
• Intensive: (y)-uggan / (yə)-ttugun “it is kneeling”

vs.

• Aorist: ätyši “wear!”, ätyšaʔ-m “wear! (to a group)”
  
• Preterite: (y)-ityša “(he) wore; ityšäʔ-n “they wore”
  
• Intensive: (yi)-yässä “he is wearing”; yässäʔ-n “they are wearing”

Notice that glottal stop ʔ that shows up when you add a consonant. That isn't automatic in Zenaga: contrast y-ugrah “he heard”, ugrān “they heard”. So it looks as though the original conjugation of “wear” was something like:

• Aorist: *ǎlsəʔ “wear!”
  
• Preterite: *(y)-əlsǎʔ “(he) wore”
  
• Intensive: *(yə)-lassǎʔ “he is wearing”

We can also see that the missing first consonant in verbs like fəl, if they had one, was not ʔ – as far as I know, no Berber language has preserved evidence of what it may have been. (The t showing up in Siwi is probably not original, but rather borrowed from the intensive of vowel-initial roots.)

But there's still a problem here: why is *-ǎʔ reflected differently in the intensive vs. the preterite? A full answer for that would require a look at reflexes of the glottal stop in general, not just in the verbal system. But in several Berber languages, in fact, it's reflected identically. Compare, from opposite ends of the Berber world:

Tashelhiyt (southern Morocco):

• Aorist: ls “wear!”
  
• Preterite: (i)-lsa “(he) wore”
  
• Intensive: (i)-lssa “he is wearing”

Awjila (eastern Libya):

• Aorist: əsəl “hear!”
  
• Preterite: (yə)-sla “(he) heard”
  
• Intensive: (i)-səlla “he is hearing”

Clearly, Tashelhiyt and Awjila are not likely to form a subgroup! So my tentative interpretation would be that the form with -a is regular, and the form without -a found in Siwi, and Tuareg, and Kabyle, and almost every other Berber language between southern Morocco and Awjila is analogical – the intensive is always formed from the aorist, and it must have felt wrong to have one that looks as though it's based on the preterite. I've been looking at the always problematic subgrouping of Berber lately, and this would have interesting implications for that – it would suggest that Kabyle is more closely related to Zenati than to Moroccan Atlas Berber, since they share this innovation. But in Berber a lot of innovations seem to have spread areally, so it's scarcely conclusive.

* (All but the last bit of this post is an introductory summary of work by Prasse, Kossmann, and Taine-Cheikh that I've recently been digesting. It offers an interesting small-scale parallel to the story of Saussure's laryngeals.)

The unreliability of Afroasiatic etymologies

The fact that Semitic, Egyptian, Berber, Cushitic, and Chadic all belong to a single family - Afroasiatic - is fairly secure, based on striking correspondences in basic morphology. However, it is often not appreciated just how difficult it is to find reliable lexical comparisons between these families, and just how primitive the current state of AA reconstruction is. The easiest source of AA etymologies online is Militarev's database on Starling, so I'm going to pick on it for this post (Orel & Stolbova and Ehret reveal similar issues, but the latter doesn't even include Berber, and I'm focusing mainly on Berber entries here for convenience.)

Suspiciously many entries are listed as having a cognate in only one Berber language (eg earth, hide, skin, run away); given the general closeness of different Berber varieties, you would expect valid proto-Berber terms to be reflected in more than one place. However, these could always be right. Other issues are more serious.

In several cases, a single proto-Berber root is split across several AA ones, due to mistaken sound correspondences. For example:

• Proto-Berber *i-qăs "bone, (fruit) pit" is split between PAA *ʔayš/ʔawš- "ripened grain, corn" with Zenaga iʔssi (quoted without the glottal stop) "os; grain, graine, baie; comprimé, pilule, cachet, pastille; perle" (Taine-Cheikh), and *ḳ(ʷ)as "bone", with all other reflexes of *iqăs, even though Berber γ (<*q) commonly corresponds to Zenaga ʔ.
• Proto-Berber *ta-Hăli (> *ti-Həli) "sheep" is split between pAA *ʔayl "ram" and *bawil "ram", although Ghadames-Awjila v corresponds regularly to Tuareg h and other Berber Ø. (A couple of forms, like Figuig tili mistakenly glossed as "ram", have even somehow found their way into a third etymon, "proto-Berber" *laH!) The issue is alluded to in a cryptic comment under the Berber section of PAA *waʔil "wild goat/ram; antelope": "Pr. H No. 220 (and Kössm. 193): Ghdm., Audj. Hgr etc. te-hele < *tiHeli, which, on the contrary, is to be connected with *ʔayl- 'ram' 3061 (together with Brb. forms of the t-ili type), as *ʔ > h in Hgr, while *ʕ > Hgr 0".
• 
  Most reflexes of pan-Berber ikərri / akrar "ram" are assigned to PAA *kar(w)- "ram, goat; lamb; kid". (The Semitic parallels listed for this word are rather interesting.) But Zenaga ǝgrǝrh, pl. gurănh 'bélier' (Nic. 156), on its own, is given a supposed proto-Berber form *gur- "ram", corresponding to an AA form *(ʔa-)gʷar "kind of antelope; ram; goat". In fact, however, there is a common correspondence of Zenaga g followed by a sonorant to proto-Berber k (eg ägärgur "chest" = Siwi ikərkər, əməgyih "dine" = Kabyle iməkli etc), and this word is obviously related to the other Berber forms.

Another case is listed as doubtful, eg:

• Most reflexes of Proto-Berber *a-lăqŭm "camel" are under PAA *ʕalVḳ/g- ˜ *lVḳ/gum- ˜ *ḳalVm- "camel"; but the Zenaga one äyiʔm, with regular *l > y (in his source's transcription ǯ) and common *γ > ʔ as seen previously, ends up as PAA *gam-al- (?).
  

Similarly, unrelated forms may be grouped together due to accidental similarity, eg:

• Under PAA *kʷay(-t)- "hen; partridge; dove; chick" is listed a "proto-Berber" form *i-kaHi; but the Ahaggar form listed corresponds regularly to Niger Tuareg tekažit, Mali Tuareg tekazzit, Awjila təkažit "hen" (see Kossmann 2005:60), and as such is unrelated to the Ayr and Tawllemmet forms takəyya quoted.

Another problem is undetected loans; this applies especially in sub-Saharan Africa, where little work has been done on their impact. PAA *ʔa/iw / *waʔ "bull, cow" is supported by Tawellemmet hawu "cow", isolated in Berber and obviously borrowed from Songhay, cp. Zarma haw, Tadaksahak hawú; removing this from the etymology leaves only pan-Tuareg iwan "cows", with no evidence for the desired *H. PAA *bar "cereal, corn" is supported by Zenaga būru "bread"; but this word is isolated in Berber and widespread in West Africa (eg Wolof mbuuru, Soninke buuru, Bambara nbuuru, Peul mbuuru, Zarma buuru), and is more likely a loan from Wolof or Pulaar.

Interestingly, most of the problem cases I've noticed in this quick skim are related to agricultural terminology. I wonder if that has anything to do with the particular interest of such terms for archeologists motivating a more intense search for cognates.

More on Nile Valley Berber [?]

I finally got around to borrowing Bechhaus-Gerst's Sprachwandel durch Sprachkontakt am Beispiel des Nubischen in Niltal. It's tough going because I don't really speak German, but she briefly suggests (p. 37) that the C-Group Culture of 2200 BC-1500 BC in lower Nubia, known as Temehu to the Egyptians, were Berbers (referencing Behrens 1984/5), and that Nobiin-speaking Nubians came in about 1500 BC and replaced them. This would explain the possible Berber loanwords in Nobiin, notably aman "water". Apparently, the archeology shows a change of cultures and of body types around 1500 BC, and ancient Egyptian paintings first begin depicting their southern neighbours as black around this period, while the Egyptian loanwords in Nobiin seem to date to the New Kingdom or later.

The identification of the Temehu with the Berbers is not based on linguistic evidence, as far as I know, and the small inventory of possible Berber loans in Nubian is neither conclusively established nor necessarily dates from as early as 1500 BC. So I don't know how much confidence to put in this scenario. However, it points to an interesting avenue for studies of Berber to explore. A lot of evidence suggests that Afroasiatic originated further east than North Africa, so it would make sense for there to have been Berber speakers in the Nile Valley - that could even be where Berber spread from in the first place. I previously discussed this issue in The Berbers of Southern Egypt.

The book is interesting for other reasons, incidentally - if her scenario for the development of Kenzi/Dongolawi is correct, it has borrowed an astonishing amount of grammatical material from Nobiin.

References:
Behrens, P. 1984/5. "Wanderungsbewegungen und Sprache der frühen saharanischen Viehzüchter", SUGIA 6:135-216.

Coptic adjectives

A little follow-up on the previous post, based mainly on Reintges' Coptic Egyptian (Sahidic Dialect): A Learner's Grammar:

In Coptic, predication of properties is handled exactly as for nouns, including the use of an determiner with the adjective:

hen-noc gar ne neu-polytia
indef.pl-great for are their-labours.
For their labours are great.

In attribution, the structure is Determiner - A - n - B, where A can be the noun and B the adjective, or vice versa:

ou-kohi n-soouhs: a-small n convent
t-parthenos n-sabê: the-virgin n prudent

To express the material of which something is made, you use the same structure, except that only B can be the material:

t-kloole n-ouein: the-cloud n light "the cloud of light"

Note that this is separate from the attributive construction:

ntof pe-iôt pahôm "He, our father Pahom"

So can adjectives be distinguished as a separate word class, when they behave so much like nouns? The answer is yes: an adjective is an item that can occupy either A or B in the attributive structure without a change in referential meaning. (See Coptic Grammatical Categories, Shisha-Halevy, p. 53.) If you reverse the constituents of a genitive or material construction, you change the referential meaning: "a vessel of wood" vs. "vessel wood (ie wood for vessels.)" If you do so for an adjective-noun attributive construction, the referential meaning stays the same: ou-noc n-polis or ou-polis n-noc both refer to the same entity, "a big city". So for this case, Dixon's hypothesis scrapes through.

Triliterals in strange places

In a grammar I was looking at lately, I came across the following sentences:

"Nouns may be verbalized, or verbs nominalized, simply by bringing the stem into a suitable rhythmic form... Most of the rhythmic patterns call for a tri-consonantal stem. If a stem is di-consonantal in its primary form, a consonant (usually the glottal stop) is added to give it the proper structure... Often in the course of forming derivatives, stems that are too long are forced into one or the other of the regular patterns. They are cut down by the loss of quantity or of vowels or consonants as may be necessary."

Was this a Semitic language, or perhaps some less well known Afro-Asiatic cousin? No: this was Sierra Miwok, the pre-conquest language spoken by the Native Americans of central California inland from the Bay. (See map.) The "rhythmic patterns" only involve changes in quantity and CV>VC metathesis, not insertion of specific vowels as in Semitic, but the parallel is striking. Here are a few examples:

leppa- "to finish", with a CVCVCC pattern imposed, becomes lepa''- (gaining a glottal stop).
ṯolookošu- "three", with a CVCCV pattern imposed, becomes ṯolko- (losing the š).

Compare Arabic:
'ab- "father", with a 'aCCaaC plural template imposed, becomes 'aabaa'- "fathers", gaining a glottal stop (historically a semivowel, but never mind that)
`ankabuut- "spider", with a CaCaaCiC plural template imposed, becomes `anaakib- "spiders", losing the t.

Reference:
Freeland, L. S. 1951. Language of the Sierra Miwok. Baltimore: Waverley Press.

Is Omotic Afroasiatic?

Omotic, a small group of non-Cushitic, non-Semitic languages spoken in the highlands of Ethiopia, has always been the odd one out in Afroasiatic; by anyone's tree it is the first to have split off, and the noted Chadicist Paul Newman expressed scepticism about its membership in the family. I know little about Omotic, or Cushitic for that matter, but after reading a few sketch grammars in Omotic Language Studies , I found it very difficult to imagine these languages as Afro-Asiatic; with Berber or Hausa or Beja or Semitic the cognates are instantly visible, but none of the most familiar grammatical morphemes or lexical items seemed to be present. However, a paper I just came across by Rolf Theil is the first I've seen to present an argument against the hypothesis, and a pretty good one at that. There are parts I would question - for example, the suggestion that pronouns are unreliable (they are conspicuously unreliable in regions where extensive politeness systems have developed, like East and Southeast Asia, but I didn't think highland Ethiopia fell in that category) - but the overall argumentation seems good. In particular, the attempt to show that a roughly equal number of similarities can be observed between Omotic and families other than Afro-Asiatic is on the right track - if Omotic were to have more similarities with Afro-Asiatic than with any other family, then merely pointing out problems with some of those similarities would be inadequate. I'll be interested to see the reactions of people better acquainted with the family.

On another note, I passed my upgrade presentation yesterday - yay!

Beja and beyond

Some interesting news this week from the Beja, an ethnic group of the Red Sea coast of Sudan and Egypt. It's unclear whether this rebellion is representative of the Beja's general feelings or just a figleaf for Eritrean intervention (or both), but it's a story to watch - and an excuse to bring up a cool language.

Beja is Afro-Asiatic* - either part of Cushitic or a separate branch, depending on who you ask - and happens to be among the most obviously similar languages to Semitic and to Berber. The noun morphology is already fairly suggestive:

	Beja definite article	Arabic noun endings	Kabyle obligatory prefix
Masculine nominative singular	u:-	-u	w-
Masculine accusative singular	o-	-a	a-
Feminine nominative singular	tu:-	-atu	t-
Feminine accusative singular	to-	-ata	ta-

And the pronominal object suffixes add credence:

	Beja	Arabic	Kabyle
me	-i, -o	-ni:	-iyi
you	-ok	-ka	-ik
us	-on	-na:	-aγ
you (pl.)	-okn	-kum	-kən

(Beja, apparently, has no third person suffixes.) However, what really clinches it is the verbal system. Beja has two principal classes of verbs: one that often takes prefixes, and one that usually just takes suffixes. In Semitic, the prefixes are used for the imperfect, and the suffixes developed from a stative (still to be seen in Akkadian) into a perfect; Berber mostly retains the prefixes, whereas only minor traces of the suffixes remain. The prefixes are especially telling:

	Beja	Arabic	Kabyle
I	a-	'a-	-γ
you (m.)	ti- -a	ta-	t- -ḍ
you (f.)	ti- -i	ta- -i:	t- -ḍ
he	i-	ya-	i-
she	ti-	ta-	t-
we	n-	na-	n-
you (pl.)	ti- -na	ta- -u:na	t- -m
they	i- -na	ya- -u:na	-n

while the suffixes are best exemplified in Beja in the conditional mood:

	Beja	Arabic	Dahalo general non-past (Cushitic)
I	-i	-tu	-o
you (m.)	-tia	-ta	-to
you (f.)	-tii	-ti	-to
he	-i	-a	-:i
she	-ti	-at	-to
we	-ni	-na:	-no
you (pl.)	-tina	-tum	-ten
they	-ina	-u:	-en, -ammi

Just for good measure, in the prefix verbs you also have a feature found in Akkadian (among other Semitic languages) and Berber but lost in Arabic: a present tense formed by doubling the middle radical (in Berber and Akkadian) or adding n before the middle radical (in Beja). Compare:

• Beja aktim ("I arrived") > akanti:m ("I arrive")
  
• Akkadian almad ("I learned") > alammad ("I am learning")*
  
• Tamasheq əlmədǎγ ("I learn", irrealis) > lammǎdǎγ ("I am learning", realis)
  

It's really remarkable, considering all this, that Afro-Asiatic research isn't more advanced. There are two etymological dictionaries out there, admittedly - Ehret's and Orel and Stolbova's - but, though valuable, they frequently disagree with each other, and neither has attained general acceptance.

* Some people think Afro-Asiatic is not proved. I can't think why. Omotic's membership is not entirely clear, but all the rest is just plain obvious.

* Previously misquoted forms corrected, thanks to Matthew Loran.