True might

« previous post | next post »

"Heath Ledger Might Be Dead, but the Heath Ledger Scandal Dept. Is Still Taking Calls."

The gossip-sheet author who wrote that headline wasn't trying to cast doubt on the actor's demise. He was just using a common rhetorical device: granting a point with might or may, before stating the counterpoint in the next clause.

It seems to me that might was added to that first clause not because of any question about whether the clause is true, but rather as a way of signaling doubt about its logical connection to the point made in the second clause. In fact, you could eliminate the modal without really changing the force of the argument: "Heath Ledger is dead, but …"

Some other recent journalistic examples:

Latin might be dead, but it continues to twitch.
[H]is hair might be gone, but not his loopy humor.
The sandbags and floodwaters might be gone, but the paperwork has just begun.
The rain and wild weather that plagued the weekend might be gone, but the delays remain Monday at LaGuardia.
The price of gas might be rising, but that hasn't completely stopped boaters from enjoying life on the water.

The break may be over, but the Padres still appear to be broken.
London may be costly, but it's cheap compared with Moscow and Tokyo.
Their memory may be failing. But people with Alzheimer’s can still sense when someone is talking down to them.
Ernie Banks may be 77, but he hasn't given up on one of his lifelong dreams.
SDCC may be over, but the news continues to roll in.

Given that Ernie Banks is incontestably 77, why write "Ernie Banks may be 77, but he hasn't given up on one of his lifelong dreams", rather than "Ernie Banks is 77, but …"? If the writer were telling us about Ernie's age, rather than using his age as a rhetorical counterpoint for a discussion of his dreams, then "Ernie Banks may be 77" would imply, counterfactually, that there's some doubt about the matter.

Of course, the same rhetorical structure "possible(P) but Q" can also be used when the truth of P is subjective, uncertain or controversial:

Carbon might be recaptured, but cash won't be.
Pickup lines might be out, but flirting is still very in!
This might be true, but Secretary Kopelousos says finding the money to pay for other transportation projects won't be easy.
It might be apocryphal but the story is attaining mythical status just the same.
Their intent might be honorable, but the impact would be devastating.

Your notebook may be old, but that doesn't mean its days are numbered.
Rumors that Jobs is ill may be overblown, but he's seen as so central to Apple's success that investors are jittery.
I may be naïve but I’m going to believe that Favre does want to get this resolved in a way that minimizes the embarrassment to Green Bay.
It may be bravado but former Blairites are warning that if the cabinet does not act before the party conference, there will be a revolt from below.
DIY may be fun, but it's a challenge if you don't know what you're doing.

So what's going on here? I'll leave it to the linguists who work on formal pragmatics to work it out exactly, because my breakfast blogging hour is over. Me, I'm just defending myself against an embarrassing comment.

Yesterday, in discussing the role of dictionaries in determining whether something "is a word", I quoted a sportswriter who used the "possible(P) but Q" rhetorical device as follows:

Granted, "disappreciation" might not be an actual word, but it was what Lito Sheppard came up with to characterize the Eagles' handling of him yesterday, and, syntax aside, his point was clear.

I assumed without discussion (and in fact without thought) that the Les Bowen was taking it for ganted that "disappreciation" isn't a word. But in the comments, DonBoy retorted "Hey, the original author just said that it might not be a word".

DonBoy, you might be right, but I don't think it matters.


  1. James said,

    July 29, 2008 @ 8:58 am

    I think my favorite LL entries are the ones that take a perfectly ordinary construction and make it seem puzzling.

    It never occurred to me before that there was anything to explain here, but now that I see it, I have a theory.

    I think the 'might' in "might… but…" is the same as or closely related to the 'may' in "Be that as it may" or "That's as may be". And my sense is that it is deontic, not alethic or epistemic. The speaker is allowing, though it's the kind of allowing that takes place in conversation and debate and has an epistemic aroma. Compare: "I allow that your notebook is old, but that doesn't mean its days are numbered." The speaker allows that P, but insists that Q. Similarly, in "Be that as it may" or "That's as may be", the speaker allows, stipulates as it were, so as to go on to something else.

    That's my theory.

  2. Rob Gunningham said,

    July 29, 2008 @ 9:51 am

    Are might and may equally utilizable in most of the examples you give?

  3. Mark Liberman said,

    July 29, 2008 @ 10:22 am

    Rob Gunningham: Are might and may equally utilizable in most of the examples you give?

    I think so. I also think that they work with "maybe", "perhaps", etc., but I couldn't find any examples of that pattern in a quick search through Google News. Searching Google Books, though, I find this in Rush Limbaugh's The way things ought to be:

    Well, maybe Switzerland is cleaner, but it is merely a tourist museum. The world has never seen anything like the United States of America.

    And this from Marlowe's Jew of Malta (Machiavelli is speaking):

    To some perhaps my name is odious,
    But such as love me, guard me from their tongues.

  4. Mark Liberman said,

    July 29, 2008 @ 10:36 am

    James: I think the 'might' in "might… but…" is […] deontic, not alethic or epistemic. The speaker is allowing, though it's the kind of allowing that takes place in conversation and debate and has an epistemic aroma.

    This analysis seems right to me. It also explains why "possible(P) but Q" suggests that you'd expect P to imply ~Q, in a way that's stronger than the normal contrastive effect of but.

  5. Marijne said,

    July 29, 2008 @ 11:16 am

    Heath Ledger Might Be Dead, but the Heath Ledger Scandal Dept. Is Still Taking Calls.

    It seems like the "might" sets up what follows as an invariant, i.e. whether Heath Ledger is dead or not, his scandal department is always taking calls.

  6. Morten Jonsson said,

    July 29, 2008 @ 12:24 pm

    "If" is often used the same way, as in "If the king's word was law, a man's home was still his castle." That always confuses me for a moment, since I'm expecting an if-then construction. In this case, I might think the second part of the sentence was conditional on the first–given the premise that the king's word was law, a man's home was his castle; without that premise the case might be different. But "if" actually means "although." It's an old and perfectly respectable idiom, but it gets me every time. Could it be that the "if-then" construction has become common only fairly recently, and before that, no one was likely to be confused?

  7. Russell said,

    July 29, 2008 @ 12:26 pm

    Though I know Paul Kay is a sometimes-contributor to LL, I'll mention in any case that this particular construction was noted as special (but given no analysis) in the 1988 Fillmore-Kay-O'Connor paper on "let alone" in Language.

    It was given a slightly more detailed treatment by Paul in a 1990 Linguistics and Philosophy paper titled "Even." However, Paul's judgment is that only "may" can do this, not "might" or "could." As a result he argues against a conceptual-metaphor analysis (of the type that Eve Sweetser might advance) in which the "may" is operating on a interactional level (i.e., "I may be forced to admit into common ground that…"). It seems like James has the exact same intuition (as Sweetser would have), that this involves a discourse-level deontic meaning. The fact that you can get (some) other modals and epistemic adverbs indicates that Paul's initial analysis was a bit too strong.

  8. Russell said,

    July 29, 2008 @ 12:30 pm

    I can't believe I missed it! I should have started with "Paul Kay may be a sometimes-contributor to LL, but…". Or, better, "I may (well) know that Paul Kay…"

  9. Ryan Denzer-King said,

    July 29, 2008 @ 12:52 pm

    I think Marijne's onto something with invariance. The statement isn't exactly saying Heath Ledger IS dead; it's saying that whether he's dead or not, the scandal dept is still taking calls.

    This seems to me to be a case of analogy. That second group of examples you give seems interpretable as the regular use of "might": "it may be the case that…" The second group also reads in a way that is almost identical to the first group…except that the first group pretty clearly means that it IS the case that, not it might be. Given the similarities, it seems reasonable to think that the "is" usage arose from analogy to the "might" usages.

  10. kip said,

    July 29, 2008 @ 12:55 pm

    Ditto to what Marijne said- I think it might signal that the first clause might be irrelevant to the second clause (even though you might expect it to be very relevant).

    I'm not sure why but this reminds me of a George Carlin bit about the phrase "happened to be X" as opposed to "was X", as in "A man, who happened to be black, was arrested by police on sixth street." As if there was some question as to whether or not the man was black. But my memory is failing to do the routine justice.

  11. Kenny Easwaran said,

    July 29, 2008 @ 2:55 pm

    When I read this post I was immediately struck by a thought that this "may" or "might" could be dual to the "must" used in "The maid was away all day, so the butler must have done it". Use a must for a conclusion, and a might for a premise that's being conceded or granted.

  12. DonBoy said,

    July 29, 2008 @ 8:04 pm

    I see now that the heavy sarcasm in my head didn't make it into the comment box properly. Really, I know what this "might" means; I've just always found it a little weird so I was highlighting it.

    [(myl) And a good thing you did! Thanks!]

  13. Rick S said,

    July 29, 2008 @ 9:25 pm

    This might seem Machiavellian, but weakly conceding or even merely acknowledging a counterargument in a preliminary clause with "might/may…but" is a way to tag it as irrelevant, so that by the time you make your assertion in the main clause it seems as if the counterargument has already been addressed. It can be used as a tactic to switch topics abruptly, terminating a line of argument in which you're struggling, or to fend off a counterargument before it's even begun, if your opponent doesn't notice you're being slick.

  14. Rick said,

    July 29, 2008 @ 9:33 pm

    Another ditto to Marjine's concept of invariance. The concessive element of the first clause adds a tone of wry irony: "HL's death has so failed to silence the scandalmongers, it may be regarded as less fact than hypothesis."

  15. Rick said,

    July 29, 2008 @ 9:36 pm

    sorry, "Marijne!"

  16. Brian said,

    July 29, 2008 @ 9:48 pm

    "I see now that the heavy sarcasm in my head didn't make it into the comment box properly."

    I'm glad it didn't. If it had, this article might not have been written. And it was a good one.

  17. Russell said,

    July 29, 2008 @ 11:41 pm

    I'm reminded of an observation I heard someone make a while ago, namely that "it's true that…" is more likely than you might expect to be in a concessive context (although it's true that…, it's true that…but…). Excluding cases like "it IS (in fact) true that…, so nyah nyah." (you can try a not-very-rigorous experiment by presenting unsuspecting subjects with "it's true that the population of the US exceeds 300 million," in a flat-ish intonation, and then asking them to continue the discourse.)

    And "it may be true that…" is even more likely to have that function.

  18. Joel Shaver said,

    July 31, 2008 @ 4:18 am

    Alan Partington's chapter 'Metaphors, motifs and similes across discourse types' in Corpus-Based Approaches to Metaphor and Metonymy (2006, Stefanowitsch and Gries, eds.) discusses this construction in search results from a corpus of the works of P.G. Wodehouse. The examples Partington finds in Wodehouse all seem to be figurative, with some classed as similes and others as metaphors. For instance:
    'Most of the time I might have been eating sawdust for all the good it did me.' is a simile, with a hedge at the end ('for…'), and
    'so thickly did it bristle with obstacles that it might have been a mile of No Man's Land.' has a hedge at the beginning ('so… that…'). In contrast, 'He might be a child in wordly [sic] matters […] but if the King did not know the difference between home-grown domestic and frozen imported foreign […]' and 'He might be a pain in the neck to the family, but he did know how to stop a dog fight.' are classed as metaphors, because this 'might be', like your examples, seems to be meant as a concessive identity statement. Partington's paraphrase of the first metaphor example fits with James' theory: 'admittedly, he is a child in worldly affairs'.

  19. Mossy said,

    July 31, 2008 @ 9:21 am

    More, please! I sometimes help translators from English deal with modal phrases, and this is a wonderful translation problem. It seems to me that the sentences could be rephrased: Even though/although X, all the same Y. Even though London is costly, Moscow and Tokyo are all the same/still more expensive. Even though Heath Ledger is dead, all the same the Heath Ledger Scandal Dept. is still taking calls. But I still can't get the difference between that and "Heath Ledger might be dead, but…"
    Thanks in advance

  20. Andrew said,

    August 2, 2008 @ 7:21 pm

    In light of this discussion about "may", and "might", I found s John McCain campaign ad:
    rather odd – It describes Obama in mocking pseudo-messianic tones as "the One" and concludes "Obama may be the one, but is he ready to lead?" Is there a category for "P may be true, but…" where the creator of the statement expects to convince the audience that P is ridiculous?

RSS feed for comments on this post