On Oct 18, 2004, at 7:46 AM, Roger Mills wrote:
> Those dialects that distinguish the vowels of _Mary, marry, merry_ may
> be
> able to assign the phoneme differently, provided a derivative form
> exists where the non-neutralized variant shows up.

Some of us have a different problem :-)

"Mary" /me:r\i/ = [me@r\i]
"grass" /gr\&s/ = [gr\e@s]

I.e., the allophonic merger of /&/ in many environments with /e:/
before /l/ and /r/.
That's why i've always thought of /e@/ as its own phoneme (and included
it in Rokbeigalmki).

-Stephen (Steg)
  "`od yom shel hhol be'eretz haqodesh"
  (another profane day in the holy land)
      ~ hadag nahhash