August 16, 2006

Majority/minority complex

Time to comment on the results of last week's Language Log poll concerning the words majority and minority. Unlike the majority of folks who commented on the poll, I'm not going to try to explain the results, and I'm certainly not going to tell you which choice is correct. This is Language Log, not ... well, not someplace else.

I don't know what got me thinking about all this in the first place, but a couple of weeks ago I came to what I thought was an odd conclusion about my own native English-speaking intuitions: that I would fill in the blanks this way.

The poll shows that a majority of people are against the war.

The poll shows that a minority of people is against the war.

(FWIW: at least two commenters agree with me, but at least two commenters don't. Go figure.)

My intuitions are a little more subtle than this, though. In the majority case, I'm very certain, and substituting is for are sounds wrong to me. In the minority case, on the other hand, I'm significantly less certain, and substituting are for is sounds OK -- though is is preferable enough for me to make the choice.

As of 3:50pm PDT today, 71% of you (930 of 1316) chose are in the case of majority, vs. 29% or 386 of 1316 who chose is. In the case of minority, 63% of you (711 of 1129) chose are, vs. 37% or 418 of 1129 who chose is. Given comments like this one ("That's what I learned, and I'm sticking to it."), I'm thinking that the is answers in both cases are somewhat overrepresented. In any event, there's clearly a strong preference for are in both cases, and that preference is stronger in the case of majority, which loosely matches the difference in my intuitions in the two cases.

Some quick-and-dirty googling on several variations reveals an even more interesting picture. I did two types of searches, one with a wildcard in object-of-of position and another with that same wildcard plus another one in prenominal position (to catch examples such as a significant majority of ...), and I grouped the results together. I didn't do anything fancier than that, and so these numbers are not as reliable as one would like, and I was too lazy to try any other verbs besides is/are. Anyway, here are the results when the determiner is a alongside the results when the determiner is the:

search string% ghits, x = a% ghits, x = the
x-maj-of-are total76.38%72.18%
x-maj-of-is total23.62%27.82%
x-min-of-are total66.43%54.84%
x-min-of-is total33.57%45.16%

If you leave out the of-phrase and just search for a/the (*) majority/minority are/is, things get really interesting. The percentages get a little closer to each other in the case of the (*) majority are/is, and they flip almost completely in the case of a (*) majority are/is. The minority cases flip, too, but the a case looks like a flip of the previous the case (closer to 50-50) and vice-versa (closer to 60-40).

search string% ghits, x = a% ghits, x = the
x-maj-are total29.12%62.55%
x-maj-is total70.88%37.45%
x-min-are total45.52%37.49%
x-min-is total54.48%62.51%

The flipping observed here is predicted by the theory in this first comment (and others): that it's the (presence and) proximity of the plural object of of that establishes a preference for are in the former set of cases with the of-phrase. (Why flipping fails in that one case beats me, though.) You might think that a (as opposed to the) would tend to pull things toward is, but this is only true in one case (the completely flipped case vs. the nonflipped case, so there may be something else going on here). I have to admit, I'm baffled.

Many commenters are also baffled, but by something else: why anyone in their right mind would choose are in either of these cases (which, as I've just summarized, most people do in most cases, so everyone seems to be out of their mind). For example, Emily writes:

I'm stunned (stunned!) that there could be any disagreement on this topic at all. Both are singular nouns! One can have "a majority."

"A lot" is an exception. It may have once denoted exactly one "lot" of people, but since that noun became old-fashioned (hie ye hence, the lot of ye!!), it's become a phrase like "many." (You can tell because people keep trying to write "alot," the same way they often write "alright.")

I think what's confounding people is not the semantic "manyness" of the words but the proximity of the "of people" modifier. This is a classic SAT error we teach all our students to avoid. ("They stuff it with words to confuse you!" we tell them. "don't be seduced!")

Because really, which would you prefer: a language where verbs agree with any damn nouns they're next to, or one with heat-seeking verbs that see through entire clauses to lock unshakably onto their subjects?

There's the proximity-of-the-plural-of-object argument again. While this indeed seems to be part of what's going on, there has to be more to it: even without the of-phrase, there are still plenty of folks who insist on using are.

The "a lot" example is (I think) a reference to this immediately preceding comment:

Suppose you replaced 'a majority' with 'a lot'. I'd expect all those who apply this singular rule to suddenly retract and go for the plural form of 'be'. I mean, "A lot of people is opposed to the war" sounds utterly aweful (sic), yet it is just as singular an entity as 'a majority'.

This is an interesting tactic: bringing up an example about which virtually nobody disagrees to talk about an example about which there's significant disagreement. This reminds me of the arguments that are sometimes used to convince people that you shouldn't say "Me and so-and-so did such-and-such". You would never say "Me did such-and-such", right? It's obviously "I did such-and-such", so it must be "So-and-so and I did such-and-such". It makes a certain amount of sense, but it's clearly not very relevant to the facts: there's disagreement in one case but not in the other, and what we should be trying to figure out is why there is this difference in the two cases.

Well, I've about exhausted myself with this one. I'll bet that a significant minority of you will comment. (How's that for avoiding the issue?)

Posted by Eric Bakovic at August 16, 2006 06:57 PM