David Peterson wrote: > Seriously, though, I've noticed something about English schwa. > Specifically, some- > times it comes out sounding like a schwa, [@], and sometimes more like > a [I]. It > seems like these are close to becoming either lexicalized, or > allophones in a bizarre > vowel harmony system. Anyone who works future English thought about > this? > To make it explicit, here are some examples: > > (1) Is "explicit" [EksplIsIt] or [EksplIs@t]? > (2) Is "America" [VmErIk@] or [VmEr@k@]? > (3) Is "impossible" [ImpAsIbl=] or [ImpAs@bl=]? I've noticed this too, although for me, they seem to be still lexically determined, but hard to distinguish. Also, for me, the unstressed [I] is more of a centralized [i-] (and hard to distinguish from /@/), while when stressed it's more front, as [I]. There *are* a few minimal pairs, tho (standard disclaimer, this is my idiolect, yours may vary, etc.) Roses [rozi-z] Rose's [roz@z] Rose is [rozi-z] Rose's rose is red [roz@s rozi-z rEd] woman [wUm@n] women [wImi-n] <-- the variance in the first sylable is far more salient for me than the second -man [m@n] -men [mi-n] <-- very hard pair to distinguish I think there's some that vary freely. Like, I think (but am not positive) I use both [@'mIr)i-k@] ([Ir] is a diphthong for me) and [@'mIr)@k@], in fast speech, probably even [@'mIrk@] The initial and final schwas definitely stay, however. Initial and final schwas seem to be more resistant, in my speech, to dropping than medial schwas.