Re: Kash phonology I
|From:||Roger Mills <romilly@...>|
|Date:||Sunday, January 28, 2001, 7:45|
Nik Taylor wrote:
>Roger Mills wrote:
>> 3. "e" mid front close [e] in open syllables; lower open [E] in closed
>> syllables, before /r/, the NC, and stop clusters; so mesa ['mesa] 'one',
>> keli ~ ket ['keli] ~ [kEt] 'six', lero ['lEro] 'day', engol ['ENgol]'hull
>> (boat)', cetre ['tSEtre] 'married', yendren ['jEndrEn] 'to spin'. Open
>> syllable [E] in ende ['EndE] 'and then..., and so...' is an exception.
>Sounds like words like _engol_ are actually syllabified as en-gol, /e/
>having the allophone of closed syllables there. What is the
>justification for considering those to be unitary? Can [mp] occur?>
Well, that's precisely the problem. I guess the waffle-y answer would be
"yes and no". Yes, the NC do trigger the closed-syl. allophone. Perhaps
the NC could be considered lenis or lax, vs. fortis or tense plain stops.
Perhaps in native analysis, the NC (and /r/) are considered "heavy" or some
such. As a matter of fact, the /Cr/ clusters arise, by metathesis, from
original *-rC-, where *-r- was a morpheme of some sort, probably a
pluralizer; perhaps the nasal accretion also had some function, now lost. In
addition, the vowel /e/ was not present in the proto-language AFAIK; most
likely it descends from a central vowel, schwa or barred-i. That may be why
it's the only vowel with significant allophony.
The arguments for NC as units are probably weak, being based on how THEY
syllabify words, and the writing system. Perhaps that's their (my?) blind
I haven't really gone to work on the proto-language, but it had voiced and
voiceless plain stops, *ptk,bdg, (the palatals developed later) as well as
nasalized/cluster *mp,nt,Nk, mb,nd,Ng; and perhaps there were doublets e.g.
*p ~ *mp with or without clear distinctions in meaning (this is a problem in
Austronesian history too) . In this flavor of Kash, the plain voiced stops
have all gone elsewhere-- *b > v, *d > r, *g > k/N/Ng among others, not all
figured out yet. And the voice distinction has been lost in the NC, *mp/mb
> mb etc. (So, in this language, there are no nasal+voiceless stops).