Re: retroflex consonants

From: "Jan van Steenbergen" <ijzeren_jan@...>

> I have been under that assumption too, but then I was told by a native
speaker
> that Polish _sz_ and _z._ are alveolars and not retroflexes.
> Czech, OTOH, is supposed to have retroflexes, I think.
I've read conflicting descriptions. The book I have has <cz>, <sz>,
<z.>/<rz> has palatoalveolars /tS/, /S/, /Z/ -- but somewhere on the web I
read they're retroflexes. Since you have contrasting and fully phonemic
alveolopalatals <c'>/<ci>, <s'>/<si>, <z'>/<zi> (IPA c-curl and z-curl are
involved), I tend to think retroflexes would set them off more. Maybe it
depends on the locale and the speaker.

I know Russian <sh> and <zh> (but not <ch> and <shch> which are always
palatized) are realized either retroflex or near-retroflex, once again
because of the phonemicity of softness vs. hardness. Don't know about Czech,
but I know <r-hacek> is a palatal trill (or something like that), reflected
as <rz> in Polish but pronounced as the much less difficult /Z/ (or /z`/?).

ObConlang: Tech has retroflexes in most dialects; they result from backing
of Proto-Tech palatals <c'>, <3'>, <c'`> (ejective/glottalized), <s'>, <z'>,
<n'>, <r'>, <l'> and <y> before <a>/<@> and (in this case with a secondary
articulation of /w/) <u>/<o>. Backed <y> is a retroflex continuant, like
English <r>. The same remained palatal before <i>/<e>, where IPA c-curl and
z-curl are found for the affricates.

The inspiration for these: Northwest Caucasian, of course!