I was struck by Lameen’s recent post on Jabal al-Lughat about a language of South Sudan:
The least well documented subgroup of West Nilotic is the Burun group, spoken around the borders between Sudan, South Sudan, and Ethiopia. The largest language in this subgroup is Mabaan, spoken in South Sudan, for which there exists at least one dictionary (available without bibliographic information on Roger Blench’s site), and several very interesting articles by Torben Andersen. But we are no longer in the era where a non-field linguist could be content to look at printed sources alone; there is a fair amount of Mabaan content on YouTube, including a channel by a BA-trained linguist and first language speaker of Mabaan, Thomas Anour: Learn Maban, African Language with Thomas Anour. (Like and subscribe, or whatever it is you’re supposed to do on YouTube to encourage creators.) Between these, that makes enough material to observe an interesting phonological difference.
In Mabaan as described by Torben Andersen and in the aforementioned anonymous dictionary, /h/ seems to show up only in interjections or loans, and /ħ/ is not mentioned at all. The variety spoken by Thomas Anour, however, features a number of words with initial [ħ] (occasionally varying with [h]). A single cognate in a North Burun language, Mayak, suggest that this is the reflex in his variety of *r, which otherwise becomes a semivowel in Mabaan; more would be desirable. […]
Edit (12/12/2024): The Elenchus comparativus (von Hurter, 1800) records, s.v. “souris” (mouse), <hén> for “Abugonos Burun” vs. <rine> for “J. Kurmuk”. This is the only word in the list transcribed with initial h – and the only word on the list corresponding to any of the ones above – but seems sufficient to suggest that this pronunciation is indeed old. Among words with *r, one notes Abugonos <yonga> “meat” and <ímaghi> “blood” (Kurmuk <rin>), which do not support the hypothesis of *r > ħ, but, given the imprecise transcription, do not disprove it either. My thanks to Shuichiro Nakao for sending me a link to this exceptionally early source.
Isn’t that great? Of course there’s material on YouTube for all sorts of languages, but it still surprises and pleases me that a linguist can use it for this kind of historical analysis. (If you’ve developed an interest in Mabaan pharyngeals, there’s a follow-up post.)
DM has actually pointed out on Lameen’s site that r -> h happened in Brazilian Portuguese, too.
I was thinking, that although [ʕ] strikes Europeans as a quintessentially exotic sound, this is a bit parochial of us. Egyptian seems to have changed proto-Afro-Asiatic *d to /ʕ/. Even. So much for the principle of least effort.* Maybe.
[Western Oti-Volta has non-initial proto-Oti-Volta *d -> /j/, and Hausa has proto-Chadic non-initial *r -> /j/, but those are pretty small potatoes in comparison. Pre-proto-Oti-Volta may have had non-initial *j -> *ʔ, though, which is a bit more enterprising …]
* Not taken seriously for some time now, on account of not actually working.
And of course Spanish has managed /j/ -> /x/.
Does Mabaan actually permit full words to begin with a vowel?
In other words, could one of these initials be there just to repair the phonotactics?
Gurma has done that, which is why POV initial *s seems to have become /ŋ/: the actual change was *s -> zero, and the the /ŋ/ was added because Gurma full words can’t be vowel-initial.
Is there any evidence of vowel pharyngealisation as an earlier feature of the Burun group?
(I’m wondering if the initial ʕ as opposed to ʔ could go back to an older vowel contrast. Western Oti-Volta languages with contrastive glottal vowels don’t permit them word-initially, presumably because such words formerly began with /ʔ/ and there has been dissimilation.)
Well, there are sound changes that make things easier for the speakers, and there are sound changes that make things easier for the listeners. These two often go in opposite directions.
Notably, if a language has an independent /ʔ/ and doesn’t permit vowel-initial words, that doesn’t automatically mean it adds initial /ʔ/ for repair. Arapaho adds initial /h/ instead.
That, on the other hand, does strike me as *j dropping out and the hiatus being filled with *ʔ. Same for the ʕ-ʔ “merger” of most kinds of Hebrew.