Re: Need some help with terms: was "rhotic miscellany"

Roger Mills wrote:

>>So I am no longer certain that my R is not, in fact, retroflex.  But I
>>got the impression by reading this list that the retroflex approximant
>>was a rare sound; in fact, the CXS for it is a trigraph, [r\`].  Which
>>doesn't jibe with it being the standard American rhotic.
Well, the standard American rhotic is a pretty rare sound, as far as
most languages are concerned, but on the other hand I'm not aware of
many languages that have an actual alveolar approximant either. Danish
does, but it's written "d", so it's not clear why the IPA symbol for
this sound is [ɹ], CXS [r\], instead of some d-like symbol, unless they
had the American /r/ in mind (and Danish "d" sounds nothing like
American "r"). The retroflex approximant symbol with the tail was a
later addition to the IPA, so it's possible that the use of [r\] for the
American /r/ sound instead of [r\`] is a convention based on tradition
like the use of [V] (turned v) for the "short u" sound in "bud" (which
in many dialects is more like [6] "turned a").

> I'll have to check out an IPA sound-site.  There is a retroflexed /r./ in
> Tamil and relatives-- the few times I've heard it, it's very hard to place,
> hard to reproduce; sounds more like a _d_ of some sort....
I've only heard it from Routledge's _Colloquial Tamil_ CD's, where it's
written as "ʐ" (i.e., "z" with a retroflex tail), but it definitely
sounds like an approximant rather than a fricative. It sounds pretty
close to the Jarda /r/, although I'd never heard Tamil spoken when I was
designing Jarda. Not exactly the typical American /r/, but close. The
exact shape of the tongue seems to make a big difference with these
sounds, but there's no easy way to indicate this with IPA symbols, and
without X-rays there isn't any easy way to tell what's really going on
with the shape of the tongue in any case. My vague impression is that
the Jarda /r/ might be a bit palatalized, and the American /r/ more
velarized and labialized.

In any case, my pronunciation of American /r/ seems to depend on its
position in the word, and it's hard to tell if there's much if any
retroflexion going on with it. But it does seem to be labialized (as I
say it, not necessarily in other dialects) -- not nearly as much as /w/,
but enough to color the sound. In any case, the Jarda /r/ is definitely
nón-labialized, which is one reason it sounds different from American /r/.

Hmm, actually I think only intervocalic /r/ is labialized, but not
syllable-final /r/. I'm not sure about that, but there does seem to be a
difference.