[Neal Whitman (visiting from agoraphilia.blogspot.com), June 24, 2004 at 1:30pm] Trackbacks
When I Say Everyone Can't, I Mean It!

One time back in elementary school, I heard a teacher talking about the logistics of an upcoming field trip, and she said something like this:

  1. Everyone can't fit on the bus.

I was confused. Did she seriously mean to say that not a single one of us could fit on the bus? How was that possible? Oh, wait—she must mean that not everyone could fit on the bus. But even when I'd figured out what she'd really meant, mentally attaching the intended meaning to the actual utterance was like trying to push two magnets together the wrong way.

This happened whenever I heard a sentence with a universal subject (e.g, everyone) and a negated main verb (e.g, can't). The resistance was so strong that for years, I thought the adage "All that glitters isn't gold" meant that by golly, if it glittered, it wasn't gold! Of course, I always thought it would make more real-world sense to say that not everything that glittered was gold, but hey, that wasn't how the saying went, and who was I to try to reinterpret it to suit my own taste?

In formal semantic terms, I was taking the negation to have scope only over the rest of the verb phrase, as illustrated in (2) with the Everyone can't fit example. I balked at allowing the negation to have scope over the whole sentence, as illustrated in (3):

  1. For every person x, x cannot fit on the bus.
    (I.e., No one can fit on the bus.)
    every(x, ~fit_on_bus(x))
  2. It is not the case that everyone can fit on the bus.
    (I.e., Not everyone can fit on the bus.)
    ~(every(x, fit_on_bus(x)))

Through the years, I (in company with many other people with strong opinions about English grammar) remained convinced that anyone who said "Everyone can't" and meant "Not everyone can" was making a mistake, plain and simple, despite the accumulating evidence that for many people, both scopings shown in (2) and (3) were OK.

As a linguist, though, I can't simply dismiss the scoping in (3) as a mistake. If I want to accurately describe how speakers are using a language, I have to respect the fact that Everyone can't is used by many speakers to mean "Not everyone can", and try to uncover the grammar rules that people are following that let them do this. This is not to say I have to like the construction; in writing-instructor mode, I can and do encourage writers to avoid it for the sake of clarity. But in linguist mode, to say that Everyone can't is wrong is just plain irresponsible.

Having recognized the two scopings of (1) as a fact of English, the task is now to write a grammar such that both scopings are generated. In fact, it turns out that it is easy to do this. First consider sentences like (4), similar to (1) except that the everyone is now an object instead of the subject.

  1. I can't talk to everyone.
    (I.e., it is not the case that I can talk to everyone.)

This sentence has two scopings for the negation, shown in (5) and (6), both of which are completely OK for me, and for every other English speaker as far as I know. And once you specify definitions for your negation and quantifiers such as every so that you can generate a sentence like (4) with these two scopings, the two scopings for sentences like (1) are generated automatically.

  1. For every person x, I cannot talk to x.
    (I.e., I can't talk to anyone.)
    every (x, ~(I_can_talk_to(x)))
  2. It is not the case that I can talk to everyone.
    ~(every (x, I_can_talk_to(x)))

Semanticists have known this for years, and the usual thinking about why some people take issue with saying "Everyone can't" when they mean "Not everyone can" is that it's an issue of avoiding ambiguity: Not everyone can is more specific than Everyone can't, so why not just say that if it's what you mean, and reserve Everyone can't for situations when you mean "No one can"? (Or for that matter, avoid Everyone can't entirely, and say "No one can," if that's what you mean.) I'm pretty sure that this is the position that Larry Horn takes in his authoritative A Natural History of Negation, though I'd have to look it up again just to make sure.

This is the position I have defended in recent years in discussions with my parents and my brother Glen. However, Glen recently raised a telling point. In school we learned the usual prescriptive rules about not splitting infinitives, not starting sentences with because, and the other favorites, and in that way learned (or refused to learn) that we shouldn't do these things that we'd been doing for years. But when I was getting confused by Everyone can't fit on the bus, nobody was telling me that the teacher was making a mistake. My rejection of the "Not everyone can" meaning (and his, too) came from the gut, not from a usage manual. So wasn't it possible that in our dialect of English, it really and truly is ungrammatical for Everyone can't to mean "Not everyone can"?

I've been working on this problem for a week or two, now, and I can say that it is much easier to arrange things so that scopings (2), (3), (5), and (6) are generated than it is so that (2), (5), and (6) but not (3) are generated. It can be done, but at a minimum, it will require that there be two definitions for can't, one of which will give you the narrow-scope negations seen in (2) and (5), and the other of which will give you the wide-scope negation seen in (6), and not allow the one in (3). Probably we'll also have to say that subject-everyone has one syntactic category, while object-everyone has another one. Both these measures are ill-motivated (read: hacky) proposals, good for solving the current problem, but without any independent evidence in their favor.

What would be some independent evidence? If we're proposing that the single word can't (or other negation) is ambiguous between the two relevant meanings, then we could expect that there would be other languages out there in which the two meanings are hooked up to two different words. Likewise, if we're claiming that everyone is ambiguous between a subject version and an object version, perhaps there are other languages out there where there are actually two words for the two meanings, instead of one ambiguous one—not just the same word with different case markings, but actual different words. Of course, if such evidence is never found, it doesn't mean that our analysis is wrong, just that there is not much reason to put credence in it. But if evidence like this did turn up, wouldn't that be cool?

UPDATE:

Several readers have noted that "Everyone can/can't fit on the bus" is ambiguous between a distributive reading (each person is/isn't individually able to fit) and a cumulative one (the group of people can/can't fit), which clouds the issue of the scope of the negation. Here is a better-chosen example: Everyone didn't go. I still can get only the "Nobody went" reading (though I can recognize when people other speakers intend the "Some didn't go" reading).

Some readers have also observed that they get only the "not everybody" reading of (4). I admit, this reading is the much-preferred one for me, though I seem to recall situations in which the other reading was appropriate and grammatical. But even if the "not everybody" (i.e. wide-scope negation) reading is the only one available here, it's still strangely different from the obligatory narrow-scope negation that I have to have for (1).