/kw/~/k/

[Dennis R. Preston]

Y'all have really got me goin' on the /kw/~/k/ stuff. Of course Arnold is
right for those dialects of US English which have a rounded vowel in such
words as "quart" and "quarter." That is, there is very good phonetic
motivation for a coalescence between the /w/ and a following +round
segment. (I am venturing only into phonetic ground here; I don't think
anybody has raised any phonological issues, although they always lurk).

How about all those folk for whom the open-o (the vowel of "caught" for
right-thinking people) has merged with the vowel of "cot"? That mereger,
for most speakers, has gone in "cot" (not "caught") direction. Since "cot"
in US English is definitely not round, any /w/ deletion among such speakers
should yield "quart" - "cot" and "quarter"-"carter" homophones (not the
"quart" "court" one discussed so far).

(I completely ignore such fossils as me and the few others who have
reported a long-O versus open-O distinction before /r/, the
"hoarse"-"horse" distinction.)

Since there is this instability of open-O before /r/ (with regard to
long-O) and the enormous conflation of open-O with short-O (i.e., the vowel
of "cot"), I suspect there is a great deal more dialectal variety going on
here than we might all suspect. IN the South, for example, there are very
important social and geographical lines surrounding those areas which have
obligatory open-O after /w/ (so that my Louisville "water" with short-O is
obligatoprly pronounced /wOHtr/ (where "OH" = open-O). (Again, the
phonological repercussions of this are interesting but not commented on
here.)

Furthermore, Peter Jennings was cited a /w/-dropper, but, although open-O
and short-O are merged in all non East Coast varieties of Canadian English,
it's not clear that they merge at the same fronter, nonrounded short-O
position that most US varieties do. If Tom Brokaw (Dakotas) deleted /w/,
however, one would think that he was an open-O short-O merger and,
therfore, unlikely to have "quarter-carter" homophones, but maybe open-O is
preserved in western US varieties after /w/, allowing and underlying /w/ to
feed a rule which "preserves" open-O which then allows this licensed open-O
to reach back and delete the empowering /w/, masking the entire operation.

dInIs

Dennis R. Preston
Department of Linguistics and Languages
Michigan State University
East Lansing MI 48824-1027 USA
preston at pilot.msu.edu
Office: (517)353-0740
Fax: (517)432-2736