I think y'all are talking past each other - and neither being particularly
fair about it, IMESHO.

Any 'lect, be it dia-, regio-, ideo-, or other, may certainly be subjected
to independent phonemic analysis; indeed, you could say that those are the
only things you *can* analyze directly.  As far as I can tell, Daniel was
doing this for his 'lect, while Ray was arguing that Daniel's choice of
symbols was inappropriate for a phonemic analysis of "English" the global
language.  Since Daniel wasn't proposing a new phonemic analysis of the
global language, I suppose they're both right.

When I said that phoneme symbology is arbitary, I didn't mean to criticize
Daniel's choices or trigger a defensive response.  I certainly agree that
the choice is not *entirely* arbitrary, and the closer to phonetic reality
the better; on that basis I must reject Daniel's modest proposal of /x/,
/y/, and /z/ for the three "E"-phonemes in his 'lect.  The only point I was
trying (and apparently failing) to make was  that features qua features are
not themselves phonemes.  That is,  if aspiration is phonemically
significant, the result is not a new phoneme /_h/, but rather a doubling of
the number of voiceless stop phonemes (or whatever aspiration applies to).
In general, while phonemic analysis identifies the phonemes, it can only
hint at what actually distinguishes them.  For instance, voice is not the
key discriminant between /p/ and /b/ in most varieties of English.  And - to
go back to a moment to the earlier thread on Pinyin spelling - an analysis
of Mandarin that used those same two symbols would not be "wrong", although
perhaps more misleading about the phonetic reality than the use of those
symbols for English is.

Likewise, the use of /a:/ and /a/ for the vowels of "part" and "pat" in
Danglish does not introduce /:/ as a phoneme, nor does it mandate the use of
that symbol within other phonemes that happen to be phonetically long but
whose quantity is not contrastive (/e:/).  Using /e:/ is certainly
reasonable, and I never intended to claim otherwise.  But I disagree that
using /e/ instead would be misleading, despite the phonetic reality, simply
because phonemic symbology is not feature-oriented.

So, to sum up, if I may . . . I think we can all agree that (1) "pat" and
"part" have two different vowels in most varieties of English, and (2) the
traditional (diacrhonic, diaspatial :)) phonemic analysis of pan-global
English represents the distinction as one of quality and rhoticity.  I
further see no reason for any of us to dispute Daniel's word (3) that the
distinction is realized in his 'lect solely by length.  In those
circumstances, his choice of symbols is perfectly reasonable for a phonemic
analysis of his 'lect, and as that's all he was offering, the thread can now
descend into violent agreement. :)

Mark J. Reed <[log in to unmask]>