Richard Wordingham mentioned:
> 2. (p7) Voiced aspirates without voiceless aspirates exist in Kelabit
(spoken in Sarawak).
========
Arnaud's reply:
>Strange
Yes it is....
I haven't read your paper, so don't know if you've cited a ref. for this--
but Robert Blust's doctoral diss. (Hawaii, mid 70s) and a longish and
complex conference paper based on it, were devoted to the question. As I
recall, he claimed it was often due to vowel deletion in the environment PAN
*...[vd.stop]___["Laryngeal"]... and a few other less well-defined cases. I
don't recall why voiceless stops weren't affected, if indeed any reason was
given.
The so-called AN "Laryngeals" (I. Dyen's term, actually) were at that time
rather ill-defined and mysterious, based mainly on poorly knownTaiwanese
languages where the reflexes seemed to be very mixed; with improved data
from Taiwan over the years, they're now assumed with pretty good reason to
have been 1.*q (uvular stop, which > /?/ in Philippine langs., with which
Kelabit probably subgroups, or often > /h/, as in Malay), 2. *h (then
written *S, > /h/ in Phil., mostly lost everywhere else), and a recently
proposed 3. *? (glottal stop, still a bit uncertain).
The main problem with Blust's theory was that it required reconstructing *q
or *h in places where they hadn't previously been posited, leading to new
trisyllabic forms in what was believed to be a mainly bisyllabic
proto-language, and for which there was very little evidence, aside from
Kelabit. I haven't kept up with the debate, but I suspect the jury may still
be deliberating. And I don't see the proposed trisyllables cited very often
in current work.
Whether the voiced aspirates of Kelabit reveal anything relevant to the IE
field, I'm not competent to say.