I just returned from a very stimulating workshop on philosophy and linguistics at the University of Michigan. There was a lot of good discussion throughout the weekend (and much merriment to boot). One of the highlights was Sunday morning’s discussion of Michela Ippolito’s very clever paper on “only”#[1]. During that discussion, we came across what I think is a new discovery.
One of the puzzles about “only” is that the presupposition/implicature/entailment/whatever that the prejacent is true can be suspended:
Only Kim can pass the test, and maybe even she can’t.
It is striking that straightforward cancellation is impossible, but that with epistemic modal riders like “maybe” (and, as Mandy Simons noted in the discussion, the almost obligatory appearance of “even”), suspension seems possible.
Now, the cool thing was that Hans Kamp said that this felt to him to be quite similar to the famous Sobel sequences in the counterfactual literature:
If John had come to the party, it would have been fun.
But of course, if John and Bill had come to the party, it would have been a disaster.
One of the characteristics of the Sobel sequence is that it can’t be reversed:#[2]
If John and Bill had come to the party, it would have been a disaster.
??But of course, if John had come to the party, it would have been fun.
So, the new discovery about “only” is that suspension cannot be run in reverse:
??Maybe even Kim can’t pass the test, but only she can.
I suspect that this fact will play a non-negligible role in the eventual discovery of the truth about “only”.
[1]: Michela has asked me not to link to the draft of her paper. If you’d like to know more about her paper, send her email at michela@bu.edu.
[2]: I first learned this from Irene Heim and it plays a central role in my paper “Counterfactual in a Dynamic Sequence”. Hans said that he thought this had been part of the folklore for a while, but I am not sure about that.
What I see in this is the same problem as I mentioned in my working paper, which I still have to re-write. The accessibility relation in this case is expressed by maybe, and what is accessed is a situation where Kim cannot pass the test, i.e., ¬π is true, while in another situation π is true. To avoid the contradiction both π and ¬π are True, the utterer of the sentence separates them in different containers, in this case, different situations (whether in the same possible world or not).
November 1st, 2005, at 4:13 pm #An initial observation would be that the kernals are linked in a fuzzy manner. “Only Kim can pass the test.” implies an indicative mood, but linked to ” Maybe even [Kim] can’t [sic][pass the test]”, mixes the mood.
First, the use of “and”, rather than “but” or “or”, should be addressed, as a conjunction carries it’s own implied inference between kernals. “Only Kim can pass the test, or maybe (even) she cannot”… still fails to address the change in mood and aspect.
Only Kim can pass the test, but maybe she will not.
Only Kim could pass the test, but maybe even she cannot.
Only Kim is able to pass the test, and maybe even she could not.
Only Kim could have passed the test, and maybe even she has not/wlll not/cannot.
By asserting that the subject “is able” to do something she has yet to do, the aspect is imperfective for the predicate. Yet the next kernal qualifies the same verb with a negation only upon a change of aspect, which requires a temporal change as well (unless both kernals are subjunctive).
Only Kim ought to pass this test, and maybe even she cannot.
November 2nd, 2005, at 10:01 pm #