Re: Minyeva Babel Text

Jesse Bangs wrote:
(re Garrett Jones' Minyeva):

>> I did look at your phonetics page, and I'm baffled by your "vowel" /r/.
>> If you're describing the sound found in "bird" in rhotic American
>> dialects, it's a retroflex approximant, and not a vocoid at all.
However,
>> if it's a round mid-central vowel, as you said, then it's not like
>> American /r=/.
I don't agree here.  American (rhotic) /r/ is realized as a retroflex
approximant in initial, intervocalic and postconsonantal position (rose,
berry, brick).  In the env. V__C it is realized as retroflexion of the
vowel-- bird, part [b3^d, p_hA^t]-- (replaced in non-rhotic dialects  with
length [b3:d, p_ha:t], and the odd [3^] > [3j] in older NYC and Deep South.)
In rhotic dialects the entire phonetic sequence [V^] can be called a vocoid,
in that it functions as the syllablic nucleus.

It is true that there is sometimes rounding in the vowel-r of "bird, her"
etc., just as there is-- for some speakers-- in consonantal [r], but it's
strictly sub-phonemic.

(Final -Vr is a separate problem, [V^] in __#, but [Vr] when a vowel
follows, even in the non-rhotic dialects.)

So IMO it would be entirely possible to have a phonemic vowel [3^], or
rounded [3\^],  in a language, and it could just as well be symbolized /r/--
provided there is not also a consonantal /r/.  This is the case, IIRC, in
Chinese.  (A lot of learners and conlangers, I think, consider Engl. [3^] a
difficult and ugly sound; personally, I like it, probably because my dialect
has it.)

You could have a language with strict CVCV structure, but with contrasts
like [paka] vs. [pa^ka].  Surely you'd want to posit phonemic "retroflexed
vowels" in that case??? Perhaps /paka/ vs. /päka/ (or any other handy
symbol) rather than /paka/ vs. /parka/ (violates CVCV).

SPECULATION:
One _could_ posit a series of phonemic "retroflexed vowels" for English /I^,
E^, A^, O^, 3^/ --not sure about /U^/-- that occur only in the env. (C)__C
(e.g. mono-morphemic beard, Baird, bard, board, bird).  But it would be
obtuse to do so, not only because it multiplies entities, but also because
it creates a problem with  the occurence of the same vocalic nuclei  in
(C)__# (beer, bear, bar, as well as cases like care/cared/caring, etc.)
where the  [r] can reappear.  The situation with VrC is quite comparable to
other diphthongal vowels [aj, aw] as well as the sequence /...VlC.../
(where the /l/ also vocalizes/drops in some envs./dialects e.g. palm
[p_hA:m] or [p_hQ:m] , help [hEwp]), and such considerations (phonetic
similarity, similar patterning with the other approximants/"semivowels" /w j
l/, permissible morpheme structure etc) make it more sensible (in Engl.) to
phonemicize the retroflexed vowels as sequences of V-r.

The status of nasal vowels in French is comparable-- there are minimal pairs
of [V~] : [V] where it is impossible to identify the conditioning nasal, so
that the nasal vowels must be called "phonemic"-- presse /prEs/ vs. prince
/prE~s/; but other cases where nasal/non-nasal alternate-- mon cher vs mon
ami, parfum vs. parfumerie.  Unfortunately I don't recall how classical
phonemics handled this problem, but it would have been one of the difficult
areas of that theory.