metalinguistic negation and the presuppositions of belief-attributions

January 5, 2007 in Language, Philosophy, Pragmatics

I hardly know anything about metalinguistic negation, except that it occurs in the following case:

Ron is asked whether he thinks Sue quit smoking. He responds: "no, of course I don’t think that Sue quit smoking (since I never believed that she was a smoker, to begin with)." Ron’s response challenges the questioner’s way of putting the question: the question "do you believe that Sue quit smoking?" appears to presuppose that Ron believes that sue was a smoker, and Ron’s response echoes the presupposition by way of rejecting it. The negation involved here is metalinguistic, as it is when we attempt to characterize Ron’s doxastic state by uttering (1).

(1) Ron doesn’t believe that Sue quit smoking.

Normally, in uttering (1), we would presuppose that Ron believes that Sue was a smoker. But since the negation involved in our utterance of (1) is metalinguistic, we don’t presuppose that Ron believes that Sue was a smoker. "Of course Ron doesn’t believe that Sue quit smoking," we might say, "since Ron didn’t believe that she was a smoker to begin with." The negation here appears to concern a way of putting things. In this case, we can’t felicitously utter (1) unless we presuppose that Ron believes that Sue was a smoker.

It is tempting to draw the following lesson from the case:

(L) For any belief-attribution B (negated or otherwise) about some individual A, if B’s clausal complement carries the presupposition that P, then an utterance of B will normally presuppose that A believes that P, except in cases of metalinguistic negation. 

That is to say, in the belief-attribution (1) about Ron, since (1)’s clausal complement carries the presupposition that Sue was a smoker, an utterance of (1) normally presupposes that Ron believes that Sue was a smoker. In the case we described it doesn’t, since that case involves metalinguistic negation.

I am very dubious about (L). Consider the following case.

Some researchers are investigating a group of people’s responses to the
following question: "do you believe that Sue quit smoking?" Every
subject responds "no" to the question. Some challenge the presupposition borne by the question (that Sue was a smoker), some don’t. But the researchers don’t know about any of that; the data they have about the subjects’ responses is given in the form of a checklist. Next to each subject’s name, there is a checkmark if the subject does profess to believe that Sue quit smoking, and no checkmark otherwise. Let’s assume that the list they are given after the latest round of testing contains ten names and no checkmarks.

Let’s stipulate that the researchers don’t presuppose
that Sue has smoked, nor do they presuppose that any of their subjects
believes that Sue has smoked. How do we imagine the researchers characterizing their results in the case? It seems perfectly ok for them to
characterize their results by uttering (2) and (3):

(2) None of our subjects believes
that Sue quit smoking.

(3) Ron (a subject in our test) doesn’t believe that Sue quit smoking.

(2) and (3) appear to be felicitous, even though the researchers don’t presuppose that any of their subjects believes that Sue has smoked. This is evidence that neither an utterance of (2) nor (3) carries the presupposition that anyone believes that Sue was a smoker.

Additionally, there is no metalinguistic
negation here, as far as I can tell. The researchers are not challenging a certain way of putting things. Notice in the earlier case involving (1), we were pushed to say there was metalinguistic negation by the fact that if (1) involved genuine negation, it would carry the presupposition that Ron believes that Sue has smoked. Since (1) doesn’t appear to carry this presupposition in the case as we described it, we inferred that (1) doesn’t involve genuine negation; rather, it involves metalinguistic negation. We can’t apply this reasoning to the case involving (2) or (3), since neither (2) nor (3) appears to carry the presupposition that anyone believes that Sue was a smoker.

So (L) is wrong.

ADDENDUM: here’s what appears to be a pretty reliable test for distinguishing metalinguistic from genuine negation in cases like this. If the negation involved in an utterance of ~P is metalinguistic, then for some pragmatic presupposition Q carried by the non-metalinguistic-negating reading of ~P, it will normally be felicitous to say something like:

~P; ~Q!

~P, because ~Q

Of course ~P, since ~Q

~Q, so ~P

We see this pretty clearly in the case involving (1). (4) is clearly felicitous in this case:

(4) Ron doesn’t believe that Sue quit smoking, because Ron doesn’t believe that Sue has smoked

In the case involving (3), however, (4) doesn’t seem to be felicitous, since, as we stipulated, whether or not Ron believes that Sue has smoked is not part of the common ground. By the test given above, the negation involved in the case involving (3) is not metalinguistic.