Back Vowel Phonemes
Aaron E. Drews
aaron at LING.ED.AC.UK
Mon Sep 18 16:40:45 UTC 2000
TERRY IRONS wrote:
>
> dog, lost, horror, on, upon, water, wash
>
> My query is, what vowel would you consider to be the underlying phoneme
> in each case, ) or a?
I would agree with dInIs when he said:
> If you are assuming a Labovian paradigm, then the "historical" class of the
> vowel gets first consideration.
Since I have a merged vowel for this region, I've had to learn to draw
on the orthography to help me decipher what would be the "proper"
phonemic category. So, <aw>, <au>, etc. are the ) vowel evidenced by
some folks who still pronounce a diphthong. The /a/ vowel is
represented by <a> and <o>, with a couple of exceptions. The first five
on your list fall into this /a/ category.
When <a> or <o> are preceded by <w> (or even /w/), the rounding of the
dubya supposedly rounds and raises the following vowel. If I remember
correctly, these vowels are ) phonemically, although since the vowel is
in a predictable position, I would question that. This category fits
the last two on your list.
The <o> and <a> also become ) before nasal+obstuent clusters and before
voiceless anterior fricatives in some varieties of American (the same
environment as the _can_~_can't_ distinction in RP). The raising and
rounding of the vowel in these two environments isn't entirely
consistent, so the predictable realization is more a rule of thumb
rather than a hard-and-fast phonological rule. For that reason, it is
generally assumed that the underlyer is /)/.
I think I'm regurgitating a lot of Wells, vol. 1 and bits of vol. 3.
Those might be able to help you categorize the vowels.
Of course, this is all in an ideal, static situation.
--Aaron
________________________________________________________________________
Aaron E. Drews The University of Edinburgh
aaron at ling.ed.ac.uk Departments of English Language and
http://www.ling.ed.ac.uk/~aaron Theoretical & Applied Linguistics
Bide lang an fa fair Awa whair nae man has gaun afore
More information about the Ads-l
mailing list