On Tue, 3 Jun 2003, Richard Wordingham wrote:
>
> I suppose 'ninyire' should be sound, but the PIEML reference demolishes
> the claim to monovocalicness claim by citing the contrasting prefixes ura-
> and vra-. Such contrasts demonstrate that [u] ~ [v] was a phonemic
> contrast, not an allophonic variation. I suppose the critical point is
> that prosodic contrasts are fundamental, while the phonemic contrasts
> are incidental.
I have no objection to this; ura- : vra- was one of my own counterexamples
when I published the analysis (back in 1974), making it one that at least
demanded a "silent consonant" to be put right. However, *if* /vra-/ had
been realized [ura-], which is indeed imaginable, *that* language would
have been monovocalic. That is how close Sanskrit gets to hitting the
mark.
There is also a foreign word like toyam 'water' which runs counter to the
syllabic structure of avyas gen. of avi- 'sheep'; in a synchronic analysis
they cannot both have /avya/. In this case however one can just prefer a
synchronic analysis in which toyam has not yet been integrated in the
language.
The relevance for IE is that, in the discussion of IE, it is the sheer
possibility of reducing the sounds [e, a, o, e:, a:, o:] to a single vowel
phoneme /e/ that is rejected on typological grounds. And for these
elements Sanskrit does in fact make do with /a/ and /aa/ only.
Jens