Language Log

They may be midgets

Yes, indeed, as Geoff Pullum reports, there really was an NPR interview this Easter morning in which James Cochrane, author of Between You and I: A Little Book of Bad English, really did claim that the modal verb form might was being replaced by may and "has practically disappeared from the language".  Geoff supposes that Cochrane (a "mendacious pontificating old windbag" and "an utter fraud", as Geoff characterizes him) gets away with peddling this twaddle because of the abysmal level of public awareness about language; people like him are "convinced you'll believe absolutely anything, so they have little motive to stick to even a vague semblance of truth."

I think this credits Cochrane (and others of his kind) with more knowledge about his audience than he actually possesses and with more calculation in his pontificating than he's probably capable of.  I think he's earnest enought, but he's also ignorant, lazy, and self-important — ignorant of the facts about English, too lazy to do his homework (it's not like no one has ever thought about these phenomena before), and so self-important that he takes himself to be the measure of all things linguistic and just relies on his gut feelings about the state of the language.  As a result, what he says is wildly hyperbolic — he hugely overestimates the scope and significance of the innovations he reports on (and probably their recency as well) — and misses essentially everything of interest about the might/may phenomenon.

But enough about Cochrane.  Let's talk facts.

A good place to start is the old reliable Merriam-Webster's Dictionary of English Usage (1989).  Section 2 of the may entry concerns may vs. might, citing references back to 1966 for "the puzzling use of may where might would be expected", notably "in describing hypothetical conditions, and in a context normally calling for the past tense."

The first MWDEU case is epistemic may (the may of possibility, not the "root" may of permission) in the apodosis of a counterfactual conditional: If he'd have released the ball a second earlier..., he may have had a touchdown (note the innovative 'd = would in the protasis, as well as the innovative may in the apodosis).  Here a past modal form -- would, could, should, might -- is usually called for.  The may is innovative, but not particularly troublesome, since in the hypothetical context neither root may nor present epistemic may (the may of I may vomit and It may rain tonight) makes sense.

(The Cambridge Grammar of the English Language notes another context in which past modal forms are usually called for -- in backshifted reports, as when the utterance It may rain before we get home is reported as I thought it might rain before we got home -- and in which may has been encroaching on the territory of might: I thought it may rain before we got home.  CGEL (p. 203) observes, "Conservative usage manuals tend to disapprove of [this] usage, but it is becoming increasingly common, and should probably be recognised as a variant within Standard English.")

The second MWDEU case is in reports of possibility in the past, as in Born in Buffalo, N.Y., he may have gone to Princeton... but he made his reputation as a railroader.  Here there is the potential for the may to be taken as a present epistemic ('It may (well) be the case that he went to Princeton') rather than a past epistemic ('It might have been the case that he went to Princeton (but he didn't)').  This is described as "confusing" by MWDEU.

There are still other cases where may might give the reader or listener pause.  For instance, with negation in its syntactic scope, may could be either root may or a present epistemic: I may not eat the peanuts 'I am not allowed to eat the peanuts, I must not eat the peanuts' OR 'I might (well) not eat the peanuts'.

I've been aware of the "confusing" cases at least since 1958-61, when I worked on a newspaper and did a fair amount of copy editing.  The innovative uses of may sometimes result in ambiguities that can be troublesome (at least to readers and hearers who aren't so accustomed to the innovative uses), and I dutifully clarified things then (by replacing may with might, or sometimes with must), and since then I have often advised my students to do the same.  But I have to admit that lots of people seem to have no difficulty in working out, in context, the intentions of the writer or speaker, so maybe my clarificatory work was just fussiness.  People cope with potential ambiguity all the time, after all.

Back to MWDEU, which admits to puzzlement: "No one has a satisfying explanation for why these substitutions [of may for "expected" might] occur, and we are as stumped as anyone else."  I think they're giving up too easily.  To start with, present epistemic may and might are in widespread alternation: I may/might vomit, It may/might rain tonight, etc.  The semantic or pragmatic difference is subtle (by the way, there's plenty of literature on it), so that it's always open for people to see the difference as primarily stylistic and to opt for what they see as the informal variant, may -- sometimes preferring it in situations where both variants are possible, and then extending it (variably) to situations (like hypotheticals and backshifted reports) where particular constructions used to call for might only.

MWDEU entertains the germ of this proposal when it cites a brief footnote in the Quirk et al. Comprehensive Grammar of the English Language (1985), to the effect that the spread of may might be related to "some speakers' not perceiving any [semantic] difference between" You may be right and You might be right.

In any case, the result is a situation where even innovators still have use for might, especially as a strongly tentative epistemic, even though might has lost some ground to may.  (And of course there are still all us old folks using might like crazy and keeping the Google numbers up.)

This message has been brought to you by They Might Be Giants, the writers of the advice song "No":

(Not that we're so disapproving here at Language Log Central.  We're really very nice people.  Unless you spell badly.)

zwicky at-sign csli period stanford period edu