Date: Tue, 19 Mar 1996 22:31:34 EST


Subject: Re: may/might distinction

I've used minimal pairs like

(1) He {might/could} have won but he didn't.

(2) #He may have won but he didn't.

(3) It was possible for him to win but he didn't.

(4) #It's possible that he won but he didn't.

for many years in my semantics and pragmatics classes to illustrate the

difference between logical and epistemic possibility. This distinction, which

I talked about in my (1972) thesis, borrowing from a Lauri Karttunen paper of

1971 and work by philosophers like Saul Kripke and Ian Hacking, has to do with

which sorts of possible worlds you look at when you try to figure out the

truth conditions of a sentence. (1) and (3) entail something like 'There's a

possible world consistent with the physical (etc.) facts of the actual world in

which he won', i.e. if certain contingent facts had worked out differently, he

would have won. (2) and (4) involve not logical (metaphysical) possibility

but epistemic possibility, i.e. that there's a possible world consistent with

what I know in which he won. But in saying he DIDN'T win, I'm saying there's

no world consistent with what I know in which he won.

Well, anyway, the punch line (as nobody who read this far will be

startled to learn) is that some time ago--maybe in the mid-1980's--I began

realizing that more and more of my students were not sharing my judgments. I

think they DID get the distinction; they just didn't preserve the mapping

between the semantics and the modal auxiliaries. I share everyone's frustra-

tion over the loss, especially since it makes the distinction harder to

demonstrate. But then I think it's an occupational disease of linguists to

regret all neutralizations, especially those wiping out distinctions we make

ourselves. (I do, however, recognize that not everyone shares my grief

concerning those careless dialects in which Mary, marry, and merry are not