The next episode of my serialized reflections on double modals (the series began here) concerns problems of adjunct placement. Notice, first, that modals like might allow not to come after them (the negation of You may have any home-brewed beer is You may not have any home-brewed beer), while regular verbs don't (the negation of We make home-brewed beer these days isn't *We make not any home-brewed beer these days). That gives us the following contrast between can tomatoes (with the regular transitive verb can meaning "put into cans") and might can tomatoes (the modal verb can meaning "be able to"):
| . | regular transitive verb can | modal verb can |
| WITHOUT NOT: | We can tomatoes in the summer. | We can eat tomatoes in the summer. |
| WITH NOT: | *We can not tomatoes any more. | We can not eat tomatoes any more. |
We can exhibit a similar contrast between the regular transitive verb will meaning "leave as a bequest") and the modal verb will (with meanings involving future time and volition):
| . | regular transitive verb will | modal verb will |
| WITHOUT NOT: | He willed his estate to charity. | He will leave his estate to charity. |
| WITH NOT: | *He willed not his estate to charity. | He will not leave his estate to charity. |
So, other things being equal (in the absence of special constraints), one would expect a dialect that allows Senator Fred Thompson's might could get done (meaning "possibly could get done"), if could were finite, to also allow I might could not get done (meaning "possibly could not get done"). Do people report that as well?
Moreover, since we do not find notafter adverbs in adjunct function and right before modals — that is, we do not find *He maybe not can get there in Standard English — my speculation that might is simply an adverb in some dialects would predict that in those dialects you cannot say [?*]That might not could get done (to mean "possibly could not get done"). Do people speaking his dialect find that ungrammatical? If they accept it, that would be some evidence supporting the double-modal analysis.
Because of something a student in my class said today, I have realized something else: Marianna Di Paolo's analysis, where sequences like might could are single lexical items, also predicts very firmly that you could not get [?*]That might not could get done — it would involve an occurrence of the independent word not right in the middle of a lexical item!
And because of something Brett Reynolds wrote to me today, I have noticed something else. Adverbs with meanings like "probably" or "possibly" usually occur in quite a few positions: clause-initial, VP-initial, clause-final... and after the tensed auxiliary. This suggests that if might were an adverb there is no reason not to expect alongside I might could get that done by Friday a variant with a flipped order with the adverb after the modal: I could might get that done by Friday. I don't know why, but somehow that doesn't feel plausible.
I don't know the answers to the questions I've been raising; I'm not biased against or in favor of any hypothesis. I really don't know what the syntactic facts about the supposed double modal dialects are. And I think most people who mention them don't; they just have at best a small collection of examples they're heard — anecdotal evidence rather than a systematic basis for a syntactic analysis.
Maybe I could work on the construction here in Scotland: there are Scots dialects with formations like might could (there is at least one speaker in my class). But as far as analysis is concerned, right now I am feeling that (a) my adverb hypothesis about might is basically junk, very unlikely to be true; but (b) every other hypothesis I can think of looks like it has a heck of a hard time too. Nothing seems to fit.
But of course, I need to read more. I'm very ill-educated on this topic. So the next thing I'm going to do is to read the unpublished paper by Chris Barker (New York University) and Cynthia Kilpatrick (University of California, San Diego) that Chris has kindly sent to me. It is about double modals, and Cynthia is a native speaker of a dialect that apparently has them. This should be interesting. In due course, if I learn things, I will report on them here, of course, at your linguistic one-stop shopping location, Language Log.
Posted by Geoffrey K. Pullum at November 27, 2007 09:24 AM