[tied] Re: ka and k^a

In languages with few vocalic phonemes, each phoneme tend to consist
of two or more allophonemes. In a three-wovel system, it is probably
not of any value to distinguish between e/a/o and i/a/u. Because the
system would probably consist of these allophoneme combinations:
e/i, u/o, a/ä.

In a two-phoneme system you could likely have this allophoneme
combination: @/e/i, a/o.

In Abkhasian the letters a and e does not denote two disting sounds,
but two groups of allophonemes, of which each allophoneme is a sound
that in other languages has status as a separate phoneme.

--- In cybalist@yahoogroups.com, "Rob" <magwich78@...> wrote:
> --- In cybalist@yahoogroups.com, "Patrick Ryan" <proto-
language@...>
> wrote:
>
> > ***
> > Patrick:
> >
> > 1) There is nothing "unrealistic" about *e, *a, *o.
>
> Do you know of an existing language with such a vowel inventory?
>
> I cannot (yet) say that /e a o/ is unrealistic, but /i a u/ is
*more*
> probable, since languages with only three vowels tend to maximize
the
> distinctions between them.
>
> > 2) Whether more three-vowel languages have *i, *a, *u, or *e,
*a,
> > *o does not determine what pre-PIE had.
>
> True, but it does affect the likelihood for different scenarios of
> (pre-)IE.
>
> > 3) Since Miguel remains silent, let me ask you: what
possible
> > practical difference does it make to prefer *e, *a, and *o to
your
> > *i, *a, and *u?
> >
> > ***
>
> The difference of less typological realism (notice I do not
> say "unrealism") to more.
>
> - Rob