Saturday, June 04, 2005, 18:42 GMT
I think what's happening is this:
/"w@tdi se/
syllabicization: "w@t.di se
vowel length determination: "w@t.di se (no changes)
variation of phonemes' realization based on stress: "wVt.di se
variation of phonemes' realization in intervocalic locations: "wVt.di se (no changes)
voicing assimilation: "wVd.di se
removing release of stops in certain places: "wVd_}.di se
diphthongizing certain tense vowels in places: "wVd_}.di seI
["wVd_}.di seI]
The above order of transformations is not exact at all, and could be quite different, but you get the picture. This is why the vowel in "what" ends up being short rather than long, and [4] is *not* used, at least in my dialect, as voicing assimilation appears to happen after the steps that determine such.
/"w@tdi se/
syllabicization: "w@t.di se
vowel length determination: "w@t.di se (no changes)
variation of phonemes' realization based on stress: "wVt.di se
variation of phonemes' realization in intervocalic locations: "wVt.di se (no changes)
voicing assimilation: "wVd.di se
removing release of stops in certain places: "wVd_}.di se
diphthongizing certain tense vowels in places: "wVd_}.di seI
["wVd_}.di seI]
The above order of transformations is not exact at all, and could be quite different, but you get the picture. This is why the vowel in "what" ends up being short rather than long, and [4] is *not* used, at least in my dialect, as voicing assimilation appears to happen after the steps that determine such.