cross-post from Linguist List

[Arnold M. Zwicky]

On Jul 16, 2005, at 6:43 AM, Mark A. Mandel wrote:

> This is getting more and more interesting. First we get "their"
> split out
> from its (to most of us) homophones, then a different two-way split
> with
> "they're" as the odd man out (from Wilson, not surprisingly... that
> is, I'm
> not surprised by now that Wilson has interesting data to offer).
>
> Let's see how this matches up with the merry/marry/Mary isoglosses.

well, i have -- more accurately, sometimes have -- a merry/Mary
distinction (and a distinct marry as well) , and i have it here too,
distributed in the same way as in wilson's variety, at least when the
words are accented (as they are in citation forms): their/there with
[E], they're with [e].

i *think* that adverbial there always has some accent, so it always
has [E].  but expletive there is often unaccented, so there are more
possibilities.

in unaccented occurrences, their/there can have, for me, either [Er]
or a "reduced" vowel, syllabic [r], but i believe that this reduction
is much more frequent for (expletive) there than for their.  (so that
their is distinguished from there in my phonology, but statistically
rather than categorically.)

now, on to they're.  for many american speakers (of whom i am
definitely one), personal pronouns in combination with a reduced
("contracted") auxiliary very frequently have their nuclei laxed/
monothongized; this is almost categorical for unaccented occurrences,
but can happen in accented occurrences as well.  so, I'm is very
frequently just [am] (rhymes with Tom) for me.  and they're very
frequently has [Er] rather than [er].

the result is that there are contexts in which the three are
homophones (with [E]) for me, but i nevertheless distinguish all
three (at least statistically) in other contexts.

arnold (zwicky at csli.stanford.edu)