Matthew Yglesias and quantities

Matthew Yglesias had two posts yesterday that dealt, directly or indirectly, with the quantities denoted by various determiners.

First, Yglesias's article in the American Prospect (linked from this post) took a look at the Bush administration's claims regarding connections between al-Qaeda and Iraq through the lens of Gricean implicature. It's nice to see the work of linguists applied in the real world. It would be even nicer if those considering the importance of language looked at how people use it in the real world.

Here's Yglesias's example of Gricean implicature:

If I tell you, "they're not all in the meeting yet" when, in fact, no one is in the meeting, I haven't lied to you about anything. If no one is there, then, indeed, they're not all there. Nevertheless, any reasonable listener will have understood me to mean that some, but not all, of the expected attendees are then. Again, if I say, "some people are in the room" when only one person is in the room, I'm not speaking falsely, I'm simply speaking uncooperatively. You'll infer that more than one person is in the room although, strictly speaking, I said no such thing.

I disagree. If you say, "Some people are in the room" and it turns out there's only one person, you're lying. Now, "Some person is in the room," is a different story. As I pointed out in the comments, "Morphologically plural NPs don't always refer to multiple individuals (e.g. 'every dog' when there's only one dog in the discourse), but 'some X' when X is plural requires at least two individuals fulfilling the predicate of the sentence for the statement to be true."

Another commenter pointed out that, in logic, "some" is taken to mean "at least one." I can live with that. But conversational English is not equivalent to logic. If you tell me that some people are in the room, there's gotta be at least two for you to be telling the truth.

Second, in his discussion of how "a majority of the US public wound up believing Iraq was behind the attacks," Yglesias elided a statement of Jeff Jarvis. "Few if any people in power said that Iraq was behind the attacks" (the original can be found here) became "Few ... people in power said that Iraq was behind the attacks." It's a small change, to be sure. But Yglesias's commenters raked him over the coals for it, claiming that the difference in meaning is substantial.

The commenters' criticisms are, in my mind, unfair. I think Yglesias is wrong, but for a different reason.

First, as to why the commenters are wrong. The difference between "few" and "few, if any" is not one that affects the truth conditions of the sentence in question. The commenters' point is that by getting rid of the "if any," Yglesias eliminated the possibility, hinted at by Jarvis, that no people in power connected Iraq and 9/11.

The problem is that "few" already includes the possibility that no people in power made that connection. Suppose I'm a teacher and someone in my class how many students passed a recent test. Suppose I respond by saying, "Few students passed the exam." Suppose no students, in fact, passed the exam. Did I lie? No, I didn't. I misled the class, suggesting that at least some students (at least two... remember what I said above?) passed. But it's a matter of Gricean implicature. We know because that implicature can be cancelled. "Few students passed the exam. In fact, none of you did," is not a contradiction. Less than felicitous communication, sure, but not a contradiction. Truth condtionally, "few" and "few, if any" are identical. It's fine for Yglesias to remove the "if any" to ease reading.

Yglesias errs in assuming that "few" and "a few" are equivalent. As we've just seen, sentences of the form "few X Y" (where X is a noun and Y is a verb) can be true without any Xs actually Y-ing. "A few" is different. If Jarvis had said, "A few people in power said that Iraq was behind the attacks," Yglesias would have been right in assuming that there existed some administration officials who connected Iraq to the 9/11 attacks. But Jarvis used "few," not "a few." It's a small difference, even smaller than the removal of "if any."

In short, Yglesias took the implicature of Jarvis's statement as truth. It's an entirely reasonable step to take, one that everyone does without thinking. But if you're concerned with logical truth, as Yglesias already showed himself to be, it's a step you just can't take.

For more on this, see my paper on the meaning of few and a few.