[tied] Re: Unreality of One-Vowel Systems (was: Bader's article on

--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...>
wrote:

> The main thing is to establish exactly what we're talking
> about. From a _phonetic_ point of view Sanskrit has the
> vowels [i], [i:], [u], [u:], [&] and [a:] (plus [e:] and
> [o:] after monophthongization of /ai/ and /au/).
>
> It's possible to apply an abstract phonological analysis
> which reduces all of these to consonants (/y/, /w/) and a
> single vowel /a/ ([a:] = /aa/ and, to quote Pa:n.ini, [&] =
> /a/).

Except that I believe Piotr quoted some examples where there were two
possible surface forms and no obvious rule to choose between them.
The abstract analysis almost works, but not quite.

Incidentally, why do we have _urdi_ 'of a man' but _vr.ta_ 'chosen' and
_vr.ks.a_ 'tree'?

What stops us applying the same analysis to, say, Classical Arabic?

Richard.