may/might confusion

[Laurence Horn]

At 1:17 PM -0500 1/17/02, RonButters at AOL.COM wrote:
>In a message dated 1/16/2002 7:49:19 PM, sagehen at WESTELCOM.COM writes:
>
><< AG Ashcroft suffers from this increasingly common complaint, causing him to
>say, absurdly, that " had not the other passengers" aboard the plane
>prevented his doing so, the would-be shoe-bomber "MAY have succeeded in
>bringing down the plane".....
>To which I say, had anyone not familiar with this story been relying on
>Ashcroft's description for their understanding, they MIGHT well have been
>pretty confused about the outcome.
>A. Murie >>
>
>There has been a lot written about the merger of MAY and MIGHT, but for many
>people this seems to be a done deal. I know, it is hard to understand how a
>distinction that seems so natural and clear to ME simply does not register
>with most educated Americans under the age of 45 or so (and even older in the
>case of such folks as Ashcroft). I don't even remember, as a youth, having
>had anyone point out to me that there was such a distinction--it seems to me
>to have always been a part of my linguistic "competence." So my gut reaction
>is that this is not something like the prescriptivist SHALL/WILL distinction,
>which has been preserved largely only in textbooks (but cf. "Shall I
>castigate the evil doers?" vs. "Will I castigate the evil doers?"); there is
>really a linguistic change in progress here, and it is inevitable and well on
>its way to completion. MAY is replacing MIGHT in virtually all environments,
>though MIGHT still exists for such speakers as a free variant in some
>environments.

I agree with almost everything Ron says about this merger, except that
  (i) while there is the age correlation he mentions, it's not
absolute.  (Actually Ron does say "most", which is probably right.)
I check this with my students annually when I teach semantics, and
there's always a (non-age-correlated) split among them.  The half (or
so) who retain the distinction between epistemic 'may' vs.
metalinguistic 'might' are as incredulous about the other half (or
so) as positive "anymore" speakers are at the fact that not everyone
is one.
(ii) I don't think it's AS significant a variable as invariant "be",
as Ron opines below, but it's one I'm more interested in.

>
>I think this is a linguistic change that is at least as important as the
>extension of invariant BE, though it is scarcely mentioned in the
>sociolinguistic literature.