Re: THEORY: vowel harmony [was CHAT: Another NatLang i like]

At 5:22 pm -0500 28/6/99, Nik Taylor wrote:
>"Raymond A. Brown" wrote:
>> In Turkish, e.g. many suffixes have a vowel harmony whereby the vowel in
>> the suffix may be /u/, /y/, /|/ or /u/ (where /|/ is the high, back
>> unrounded vowel written as an undotted i in Turkish).  But the vowels
>> certainly have separate phonemic status in the language.
>
>Well, Turkish isn't what I was talking about.  What I meant was if the
>vowel forms cause allophones in the consonants, as with, say,
>roundedness, then why call the vowel phones phonemes, but the consonant
>phones allophones?
But aint't that what the do in Turkish!

>For that matter, in Turkish, do all the vowels of a word have to be the
>same backness?  Can you have roots that combine front and back vowels?
Only in borrowings.

>If not, then perhaps it could be analyzed as a word-level feature on the
>whole word, either [+front] or [+back] word.
There are analyses that do take that sort of approach, e.g. the prosodic
analysis of the Firthians.  But then they're not taking the "classical
phonematist" approach.

>For that matter, in my
>vowel-harmony conlang, I considered an orthography wherein sets of
>vowel-harmony variants (for instance, if it were Turkish, /u/ and /y/)
>would be indicated by the same letter, and then at the beginning of the
>word, there would be a symbol that would indicate [+front] or [+back]
>(or rather, in the case of that conlang, IIRC, [+high], [+mid], [+low]
I can't see why one could not do that - though possibly in the case of
Turkish one might need characters to indicate round/unround prosodies as
well as front/back.

Ray.