Why I reconstruct *β and not *v

Glen Gorden asked me a while ago, why I insisted on using *β for the Proto-Berber sound that corresponds to Touareg h, Aujila v and Ghadamès β, and not just the more easily comprehensible *v.

Motivated by his latest post on the transcription hell of Ancient Egyptian, I felt it was necessary to elaborate a bit more on it.

It mainly has to do with my current understanding of the Proto-Berber consonant system and in what place it fits in. In African linguistics v is commonly used as the symbol for the voiced fricative while β is used for the labial approximant. My current reconstruction presents a consonant system within which I deem more likely to have had a labial approximant than a voiced fricative, and my transcription system attempts to reflect this.

As it stands, I think the phonology works something as in the following table:

There is some indication that *f may have been a *p originally, but it's not yet convincing. The voiced counterpart to *f is *b. The dental obstruent series is two-fold. It has a stop and fricative variant.*γ was probably originally *q, historical data confirms this, but with the comparative method you cannot reconstruct *q, since it has become *γ in every single Berber language.

The glottal *h may or may not have existed. This is currently an issue that needs to be figured out. I suspect that the Palatal series is not original, though I still have to find solid proof that these are the result of a palatalizing context. But there already seem to be some indications which I might elaborate on in the future.

Taking that into account we can further reduce the Proto-Berber Phonology to the following:

Note the exciting gap of the 'glottalised b', not unlike the gap that we find in Proto-Indo-European. The lack of a velar nasal shouldn't surprise anyone, this is quite typical in languages of the world.

So I don't use v to transcribe Proto-Berber *β, because it would suggest that it is the fricative counterpart to *b. While to me it seems that it is probably the labial approximant instead.

Comments

[Octavià Alexandre]

Hi, Marijn. I find your article interesting, especially because I've been trying to sort out the Wikipedia article on Proto-Berber. Not being myself a specialist on Berber, I'd suggest you could also contribute to it.

I'd also like to comment some points. For example, what you call "emphatic" are usually regarded as pharyngealized dˤ~tˤtˤ, zˤ~zˤzˤ (see for example Kossmann & Stroomer, 1997, "Berber Phonology"). Hence I can't see how the pair γ~qq would fit in this category.

[Octavià Alexandre]

You might also be interested to read this this article by Arnaud Fournet.

[PhoeniX]

I'll look into contributing to the Wikipedia article.

I sometimes feel that it is maybe difficult to say much on Proto-Berber in form of an article like that because so little has been written on it.

Kossmann 1999 is the best and most complete work on Proto-Berber right now, but does not include *ʔ yet as it was only discovered later by Kossmann 2001 and Taine-Cheikh 2004.

I wouldn't even dare touching upon the subject of the Proto-Berber vowel system in an encyclopaedic work on Proto-Berber, our knowledge is much too limited to give a good overview. We need *a, *e, *i and *u. I wonder sometimes if *e isn't an allophone of *a (same reflex in the état d'annexion).

For the short vowels I lean towards a reconstruction of *ă and *ə, but there could have been two short high vowels. More research needs to be done.

You bring up a fair point about the way I interpret *γ~*qq.

There's several reasons why I think it should be seen as pharyngealized consonants.
As you can see, there is voicing distinction in non-emphatic there isn't in the emphatic ones. *γ~*qq doesn't have voicing distinction, this raises a suspicion that it might be an Emphatic consonant.

*q functions as an emphatic of *k in Proto-Semitic. I feel that it mirrors this.

Now comes the actual question on whether it is pharyngealized. The answer is no, q is not an actual pharyngealized consonant.

But, I have yet to find a languages that does have a pharyngealized k. Of course, this typological appeal loses its strength since virtually all languages, and definitely all language that I'm familiar with, that have pharyngealized consonants are from the same language family (Semitic, and by extension Afro-asiatic)

But I still feel that a kˁ is very difficult to pronounce, and at least causes some form of backing of the velar articulation to something more uvular. As soon as the point of articulation would completely shift to uvular, there would be no reason to uphold the pharyngeal co-articulation, as the consonant would be sufficiently distinguished in another way.

Another approach that more or less leads to the same point is assuming that emphatic consonants in Proto-Berber were velarized rather than pharyngealized, which they are quite often in Berber languages.

A velar consonant cannot be velarized, so to provide an emphatic k with extra 'backing' or 'vowel lowering' timbre like all other emphatic consonant do, the emphatic k is made a uvular consonant.

I agree that you are free to disagree here, but I think the lack of voicing distinction especially is very telling that it used to be the emphatic velar.

[PhoeniX]

I've deleted part of your second comment which wasn't relevant to the discussion and somewhat ad hominem and at best troll bait.

I read his article, I think he did a good job finding cognates that tie Proto-Afro-Asiatic gutturals. There might be something there, but he is too sloppy to make a convincing point.

First of all, Berber does have a guttural consonant (*ʔ) which Fournet simply does away with as an insertion by Zénaga. It is clear that this consonant is more widespread, and was part of the Proto-Language. Whether this guttural comes from Proto-Afro-Asiatic guttrals, I don't know. But it will have to be looked at rather than ignored.

At some point, Fournet tries to claim that Proto-Berber *z that has become h in several Touareg languages actually had it the other way around.

This is simply wrong, and cannot be the case. First of all, Tahaggart Touareg has a h from two origins, first of all from Proto-Berber *β second of all from Proto-Berber *z. This is completely ignored.

You would have to assume that a language shift has taken place in all Berber languages from h to z except for a few, not even all, Touareg languages. In other words, at a time that the Touareg languages were already split up, which can't have been longer ago than 1000 years ago. If it can't have been longer than 1000 years ago, we have a problem. We would have noticed in Latin transcription of Berber tribe names that h in those transcriptions always corresponds to z in Modern languages except for Tahaggart touareg. This is not the case.

So the z~h correspondence cannot be upheld.

In general his article shows a lack of familiarity in proto-Berber. His example at the top of Page 5 from Kabyle is simply wrong. These words can be reconstructed as *tariruwt and *irrew, where the š is a normal Phonotactical result of a y in front of t (Kossmann 1999:220)(The sequence uw is avoided, though usually this seems to shift to iw rather than uy).

His example of the word donkey in page 7, where he seems to reconstruct ʕ on basis of the ž is unfortunate, as it is clearly the result of the sequence *zy (Kossmann 1999:231).

Fournet tries to say that Tahaggart h is original, and not a shift of z to h, on Page 7 he reconstruct *izi with Tahaggart ehi with an original *z and on Page 9 he reconstruct *habar with an *h on basis of Tahaggart, where earlier this wasn't enough convincing proof!

In other words, he is not applying his own sound-law consistently. Messy.

I could go on for much longer, but I also won't be able to disprove everything. But I think it's clear that Fournet lacks the knowledge of Proto-Berber and the rigor of a traditional comparative linguist to make a convincing point on this subject.

[Octavià Alexandre]

Thank for your review of Fournet's article. It's also clear he hasn't read Kossmann's book (it's on sale on Amazon.de) in despite of quoting it in the bibliography (I'm affraid this is typical of him). Probably I'll buy myself a copy.

[Glen Gordon]

"In African linguistics v is commonly used as the symbol for the voiced fricative while β is used for the labial approximant."

In other words, you're basing orthography on the phonetic level, which is always cumbersome. If there are any languages that make a phonemic contrast between /v/ and /β/, they're statistically insignificant and so too is the acoustic and articulatory differences. So you're quibbling about the phonetic level to justify an orthography that should be grounded on the *phonemic* level.

The use of "f" to write a bilabial fricative in Japanese romanji hasn't caused an apocalypse, nor in Etruscan linguistics. It's obviously doable.

Whenever possible, a healthy rule is to use plain Roman characters for unmarked sounds in languages employing the Roman alphabet in order to improve general readability for not only specialists but non-specialists.

One should reserve extraneous symbols for the representation of either marked phonemes or sounds whose qualities are still under reasonable debate.