I recently sat down to read the wonderful article by the Naima Louali & Gérard Philippson (2004) "Berber expansion into and within north-west Africa: a linguistic distribution", Afrika und Übersee, pp. 105-130.
A wonderful article that discusses some of the chronological issues of the reconstruction of Proto-Berber. The internal coherence between the Berber languages strongly suggests a reconstruction of Proto-Berber no further back than the first millennium BC. At the same time, even if Proto-Berber forms a subbranch of Afro-Asiatic together with Proto-Semitic, their shared ancestor must be many thousands of years before that.
This would suggest that either Proto-Berber somehow 'koinized' at some point, to be a lot less differentiated than one would expect, or that many expected sister languages of Proto-Berber once existed but have now died out (or both!).
Louali & Philippson sensibly suggest that the Libyco-berber inscriptions of North Africa and the inscriptions of the Canary Islands (and perhaps the original indigenous Guanche language) are not forms of Berber, but rather "Para-Berber", that is, sister languages of Proto-Berber.
This would explain quite nicely why this material is so unreadable, but the few things recognisable still clearly look Berber in morphology.
Louali & Philippson also give in an appendix an overview of lexical items clearly reconstructible for Proto-Berber that relate to animal husbandry; However, no actual reconstructions are provided. I thought it would be nice to supply some reconstructions for these lexical items here.
*a-zgər 'bull, ox'
L&P suggest a reconstruction *a-zgăr. And indeed the ă vowel in the Niger Tuareg azgăr generally require such a reconstruction. However, there are, to my knowledge no other *CCăC stems (only *CăCăC), and that vocalism would usually cause Mid Vowel Harmony that shifts the prefix vowel *a to e. The vocalism therefore is rather unusual.
The Zenaga form äzgər seems to point to *a-zgər, and this is in line with what we find in the other Berber languages.
*w/ʔulliʔ 'sheep and/or goats'
The initial consonant gives some problems. Zenaga seems to point to an initial consonant *ʔ, while others sometimes have w and other starts with a vowel u.
*ta-βăl-e pl. *ti-βatt-ăn 'ewe'
Unproblematic reconstruction of a noun with an unusual supletive plural, and an unusual (but Pan-Berber) masculine plural-looking suffix *-ăn, rather than the usual -en, or for nouns with the feminine suffix -e: -iw-en.
Vycichl, who derives the feminine ending -en (without making the sound laws explicit) from feminine suffix -t + plural suffix -ăn suggests this word as evidence for this, and considers *ti-βatt-ăn is an assimilation of *ti-βăl-t-ăn. It is somewhat difficult to understand for this noun because 1. it does not have the feminine suffix -t in the singular, and 2. there's no direct evidence for a lt > tt assimilation. It works better for the noun*ta-ɣaḍ-t, for which see below.
*ta-aɣs-e pl. ti-aɣs-iw-en 'female sheep or goat'
Unproblematic reconstruction.
*ta-ɣaḍ-t pl. ti-ɣaṭṭ-ăn < *ti-ɣaḍ-t-ăn? 'female goat'
Unproblematic reconstruction, and a fairly convincing example of the assimilated *-t-ăn suffix, which blocked it from shifting to *-en.
*a-ɣăyd '(goat) kid'
Unproblematic reconstruction.
*a-akrăr/*a-kărr-əʔ pl. *i-akrar-ăn 'male sheep'
The plural reconstructs fairly unproblematically. The singular, however presents some problems. Besides an etymon *a-akrăr, which is reflected, mostly, by the Tuareg dialects, the northern Berber varieties occasionally point to forms with some kind of -i suffix, which is not easily explained.
*a-kărăw pl. i-kərw-an 'lamb'
The Tamazight and Tashelhiyt forms point to a reconstruction *a-kărăw for the singular. The Tuareg forms a-kərwa(-t) appear to be rather idiosyncratic back formation from the plural stem.
*a-zulăɣ pl. *i-zulaɣ 'he goat'
The *a-zalaɣ plurals found in several Northern dialects are probably back-formations from the plural. The Siwi form zalaq pl. izaluqən is difficult to explain. Moreover the q in Siwi and Awjila azáləq pl. zulíq in this word are irregular
*a-aɣəβ 'milk'
Somewhat unusual reflexes, here and there, but this reconstruction nevertheless seems fairly certain.
*a-dăɣăs/a-dɣəs 'colostrum'
The Tuareg varieties point to a formation with CăCăC pattern, while the Northern Berber varieties and Zenaga point to a reconstruction with a CCəC stem. One might consider that the Northern Berber varieties represent a back formation from the plural *i-dəɣs-an, but most northern dialects don't have a plural, and Tamazight has a different formation. Moreover, 'colostrum' is not very typical concept to have a plural form.
*ăksəʔ 'to herd; pasture'
Unproblematic reconstruction.
*a-mă-ksaʔ 'shepherd'
Unproblematic agentive formation of the previous verb.
Uncannily similar to one of the IE "goat" words, none of which can be reconstructed to PIE and all of which are probably loanwords.
What is ă supposed to be phonetically?
Posted by: David Marjanović | 02/09/2017 at 11:13 PM
Yes, the similarity is probably not a coincidence. There are a couple of agricultural and animal husbandry Wanderwörter shared between Berber and Indo-European.
I did a little bit of work on Guus Kroonen's wonderful Database of Indo-European
Agricultural Terminology.
http://dieat.inss.sc.ku.dk/display/entry.php?lemma_id=905
cf. also Proto-Semitic *gady 'id.', which is presumably related.
http://dieat.inss.sc.ku.dk/display/entry.php?lemma_id=705
http://dieat.inss.sc.ku.dk/display/entry.php?lemma_id=334
Also in metallurgy you find interesting ones:
*a-ẓrəf 'silver' certainly is related to the wanderwort family that 'silver' itself belongs to.
(http://dieat.inss.sc.ku.dk/display/entry.php?lemma_id=5443 I should add more cognates to this one)
Also: *a-ldun~a-buldun 'lead', uzzal 'iron' (cf. Hebrew barzel)
There are probably more, but I haven't found them yet.
===
*ă is normally thought to be a central low vowel [ɐ], but if I'm correct about mid vowel harmony, it's likely that it was closer to [ə]. It ultimately corresponds to Proto-Semitic short *a (a low front or mid short vowel), so does probably come from something like [ɐ] originally.
Posted by: PhoeniX | 02/09/2017 at 11:52 PM
Curiouser and curiouser. Were there Ancient European Farmers on both sides of the Mediterranean, perhaps?
Posted by: David Marjanović | 02/10/2017 at 11:40 AM
I think you should think in that direction yes.
I just remembered another one: *a-agăr 'field', cf. Lat. ager.
It might be a loanword from Latin, but it has undergone Mid Vowel Harmony, and therefore shows up as igr in most Berber languages, quite different from the Latin word.
http://dieat.inss.sc.ku.dk/display/entry.php?lemma_id=68
DIEAT doesn't mention any cognates outside of europe, but I believe the Sumerian word is quite similar as well...
Posted by: PhoeniX | 02/12/2017 at 01:42 PM
David, Phoenix: there are a substantial number of very Indo-European-looking words in Ancient Egyptian too, which intriguingly cannot be traced back to any known branch of Indo-European and some of which as I recall relate to animal husbandry.
So: did these Indo-European loanwords enter Proto-Berber via Ancient Egyptian? Or the other way around? Or did Proto-Berber and Ancient Egyptian each borrow the words separately from some extinct and otherwise unattested Indo-European language of North Africa? Or did Proto-Berber and Ancient Egyptian each borrow their lexemes from different branches of Indo-European?
And in the case of the Ancient Egyptian loanwords, I have often wondered if they might be connected to some of the alleged Indo-European loanwords found in Ancient Hebrew? Sapir wrote a nice article on one such possible Indo-European loanword...
(Just thinking out loud here, in case that wasn't obvious)...I wonder: the Celtic scholar John Koch has argued that the Tartessian language (attested in inscriptions in Southern Spain) was in fact Celtic. Could Tartessian have been the source of (at least some) Indo-European loans into Proto-Berber?
If any reader out there is looking for a thesis topic, I hope my somewhat disjointed thoughts above has given you some ideas...
Posted by: Etienne | 02/12/2017 at 11:55 PM
Dear Etienne:
For many of these words (e.g. 'bean') it's not so much Proto-Indo-European words that are entering Berber (and Basque and Egyptian etc.), but rather a language of the original agricultural settlers of Europe (and, apparently, North-Africa) that were there already before the Proto-Indo-Europeans arrived.
That's certainly the Hypothesis from which Guus Kroonen's database is working. It's elegant because it helps explain why there is such a strong mismatch between what archaeologists generally thing about the settlement of Europe, and what linguists think.
Posted by: PhoeniX | 02/13/2017 at 11:03 AM