Re: the sound [a]

Quoting "Mark P. Line" <mark@...>:

> Nik Taylor said:
> > Andreas Johansson wrote:
> >> Assuming that having [i] and [u] in non-final position would be
> >> perceived as
> >> odd, wrong or accented, certainly.
> >>
> >> I notice that your hypothetical example language appears to use fairly
> >> long
> >> words or alot of coda consonants.
> >
> > Eh, so make it 2:1, then.  Or even 1.5:1, the ratio isn't important.
> > :-)  I just meant, whichever is more common, or can be analyzed as the
> > default (that is, it's simpler to say "use [i] in case X, [e] otherwise"
> > than to say "use [e] in case X, Y, or Z, [i] otherwise", should be
> > considered the phoneme.
>
> So, when we're studying the basic color terms of a language, we believe
> that we can ascertain what a speaker's perceptual prototypes are.
>
> But in the case of phonemes, we can't...?
>
> Maybe we could just ask native speakers which sound they hear.
Again, if [e] and [o] are recognized as correct pronunciations of the phonemes
in question, while [i] and [u] are rejected as wrong, there would seem to be
little doubt that /e/ and /o/ is the more sensible analysis.

I however see no reason to believe that native speakers would necessarily be so
forthcoming. Languages with small vocalic sets often (always?) allow much
variation in the precise phonetic realization of their vowel phonemes, and it's
very possible the speakers of our hypothetical language would accept both [e]
and [i] for the front vowel, both [o] and [u] for the back one.

And if we're really lucky, we get answers along the lines of "Well, I use [e]
and [o], but my uncle use [i] and [u], except when he's drunk, which is about
23% if the time".

                                                               Andreas