Re: Vowels?

John Cowan:
> What's really amazing about this is what a variety of sounds can
> all come across as /r/, even if with a foreign or cross-dialect
> accent.  How is it that the difference between an alveolopalatal
> approximant and a uvular trill can be heard as mere sub-phonological
> surface noise?  Someone who rendered /T/ as [X] would experience
> no such tolerance.
This is discussed in Ladefoged & Maddieson 1996 _The sounds of the
world's languages_ in the chapter on 'rhotics'. I forget what their
final conclusions are, but at any rate, your amazement is widely
shared.

ObConlang: Livagian /r/ can be realized only as a postalveolar
approximant or as bunched/molar R (one of its several phonological
similarities to English). A tap would be realization of /l/ or
/d/, depending on environment. A uvular trill or fricative would
be a realization of /kh/ or /gh/. An alveolar trill would be a
realization of r+gh+r or r+kh+r. A labiodental approximant would
be heard as a defective realization of /v/.

--And.