Re: First post & three questions

> Date:         Fri, 19 Jan 2001 23:55:55 -0500
> From:         Roger Mills <romilly@...>
>
> Teoh [has] written:
> >Is it possible for other vowels to be semi-vowelized as well? 'cos my
> >conlang has "smooth vowels": smooth /i/ --> [ji], smooth /u/ --> [wu].
> >How would this generalize to other vowels such as [a] or [e]? I pronounce
> >a smooth /a/ almost like [Qa] (Q = velar fricative) except that it's a
> >very weak [Q], almost non-velar-like. How would you classify something
> >like this?>
>
> Seems to me, theoretically it ought to be possible for almost every vowel
> sound to have a corresponding glide-- phonetically. After all, the position
> of the tongue e.g. in [u_vowel] is different from [o_vowel].  Whether any
> language, nat or con, would utilize the very fine distinction between the
> two seems unlikely.
I suppose that is possible for glides based on rounded vowels, where
the 'approximant action' is only performed by the lips. But as you
say, this will most likely be an allophonic variation, not a phonemic
distinction.

(For instance, when I try to pronounce French <oui> /wi/ and <roi>
/Rwa/, I'm sure my tongue never gets as high as a /u/ in the second
word --- more like /O/. But I'm definitely not a native speaker).

But for glides where the tongue creates the approximation, any
difference based on tongue height necessarily disappears. /j/ must be
the glide corresponding to all of /i/, /e/, /E/; /H/ covers /y/, /2/,
/9/; /M\/ takes care of /M/, /7/, /V/.

(X-SAMPA: /M/ is the close back unrounded vowel, IPA reversed-m,
Korean <u>, and /M\/ is a velar approximant).

> Aside from [j] / front V and [w] / rounded back V, and[H] (?)
> founded front vowel, there might be some kind of velar glide after
> barred-i, and perhaps a voiced velar/uvular/glottal after [a]-- cf.
> a lax pronunciation of "aha" for the voiced glottal (ayin in
> Semitic, I think?). Tagalog and many Indonesian langs. get by with
> [j], [w] and [?] as hiatus avoiders-- the [?] corresponds to [a].
X-SAMPA does not list approximants corresponding to close central
rounded or unrounded vowels --- which are IPA barred-u and barred-i,
X-SAMPA /1/ and /}/ --- so I assume they aren't in the IPA either.
That doesn't necessarily mean they haven't been attested in a natlang,
but they can't be common --- and if they occur, it may well be as
conditioned allophones of /j/ and /H/.

To return to Teoh's question, I'm guessing that vowels in absolute
initial position would get an epenthetic /?/ --- so smooth wovels get
something else to avoid the 'hard' /?/. If you don't want open vowels
to acquire glides (/ja-/ or /wA-/, for instance), some sort of
velar/pharyngeal/glottal segment would seem to be the next choice.
Just an /h/, perhaps.

Lars Mathiesen (U of Copenhagen CS Dep) <thorinn@...> (Humour NOT marked)