> >Even if we accept typologically normal vowel patterns /a i u/
> >and /a i u o e/ for PIE, is there any other known language
> >where this sort of restriction occurs?
>
> That's a kind of unfair question. There is no language like IE anymore
> than there is another language just like Urdu.
Not at all unfair, since we were using language typology to help us
understand which collection of vowels was most believable. We were already
saying "are other known languages like this?" My point is that we were
looking only at the phonemes, not at the peculiar restrictions on their
occurrence. And if language typology has nothing to say about these
restrictions, perhaps it has nothing to say about them as phonemes either.
If - if- we have to admit the PIE is bizarrely odd, even unique in its use
of the vowels, how can we rely on typology to help us understand the vowels
as phonemes?
Peter