July 23, 2006

Will Shortz sets impossible puzzle on NPR

Will Shortz's word puzzle for last week (on NPR's Weekend Edition Sunday) was to find a name from classical mythology that was, in spelling, a concatenation of English pronouns. And the problem was probably impossible. What's for sure is that the answer he gave was not correct.

The answer was supposed to be Theseus, the name being a concatenation of these and us. The latter for is indeed the accusative form of the pronoun lexeme we. But the former, these, is not a pronoun.

I'll be using the terminology of The Cambridge Grammar in explaining why, but this isn't some sort of terminological quibble: there really are pronouns in English, and they share certain very clear, sharp properties. And the word these is not one of them, no matter which system of terminology you prefer. The thing is that there are syntactic differences in behavior between the two classes of words.

These is the plural inflected form of the lexeme this, which is a determinative, specifically a member of the demonstrative subclass of determinatives (another subclass is the articles, the definite article the and the indefinite article in its two shapes an and a).

The lexeme this is one of the special determinatives (like some and most, but not the articles) that are permitted to function as simultaneously determiner and head of a noun phrase. (There are some subtleties to the argument; but see page 422 of The Cambridge Grammar for some crucial discussion.) So that's why a form of this can occur on its own where a noun phrase can occur (which is perhaps the root of the confusion): This is typical has a subject noun phrase consisting of only one word, which is both determiner and head of the subject noun phrase.

Here, very briefly, are four lines of evidence for arguing that these is a determinative, not a pronoun. The last one is particularly telling, I think.

1.  Third-person pronouns do not co-occur in a noun phrase with a common noun the way determinatives do: the book is a (singular) noun phrase (and the is a determinative); *it book is not a noun phrase (and it is not a determinative). [It must be noted here that the pronouns we and you are special in that they have additional uses as determinatives, in phrases like we linguists or you boys; see page 374 of The Cambridge Grammar for discussion. But this is special to just those two lexemes. None of the 3rd person pronouns have second lives as determinatives in Standard English (notice, these is 3rd person); and none of the singular ones do; and none of the reflexive ones like yourself do; and so on. Don't be misled by How 'bout them apples? or them bones, them bones gonna walk around; those are from non-standard dialects where the shape of some items is different, and those dialects have a determinative with the form of them and the meaning of those. Tricky, isn't it?]

2.  Pronouns do not in general allow modification by preceding quantifiers: *every it, *all they, etc., are not grammatical noun phrases.

3.  Pronouns occur before the particle in verb-particle idioms, not after it: Don't even bring it up in conversation (with the pronoun before the particle up) is grammatical but *Don't even bring up it in conversation (with the pronoun following) is not.

4.  A particularly salient point is that pronouns occur in what are known as confirmatory tags: compare Susan is clever, isn't she? (grammatical) with *Susan is clever, isn't Susan? (not grammatical).

Now, by all four of the tests these facts make available, these is a determinative, not a pronoun!

First, these books is a grammatical noun phrase (confirmation of determinative status).

Second, all this and all these are grammatical (disconfirmation of pronoun status).

Third, Don't even bring up these in conversation is grammatical (disconfirmation of pronoun status).

Fourth, *These pastries are delicious, aren't these? is not grammatical (disconfirmation of pronoun status).

There is lots of other evidence that could be brought to bear, but this will do for a start. Sadly, Will Shortz did not consult Language Log's capable staff before setting his puzzle, even though our rates for non-profit organizations like National Public Radio are so reasonable.

This is not the first time, my friend Aaron Kaplan points out to me. The New York Times crossword puzzle for last December 30 had the clue "Lord's Prayer adjective," and the answer is 3 letters long. The answer is supposed to be thy, of course. But thy is not an adjective. It is the genitive form of a (now archaic) pronoun. It can be used as a determiner, just as any other genitive noun phrase like the children's can. Old-fashioned traditional grammars insist that anything that can occur before the noun in a noun phrase (or anything that can modify a noun) is an "adjective", but that policy gives you a class of "adjectives" infected with a diverse array of members that have almost nothing syntactic or semantic in common (I>London has to be an adjective because of the phrase London fog, and so on). As Aaron notes, the error here may be the fault of the original puzzle author. But Shortz is ultimately in charge, and is paid to be. This man needs a linguist on call.

Why do people neglect the informational resources that are available to them? I do not know. Shortz could have just called the main switchboard at Language Log Plaza and ask to be put through to the Lexical Categorization Department in the Grammar Division. (That's who Jon Stewart should have called before telling the College of William and Mary that terror is not a noun. And who Microsoft should have called to learn whether it was even remotely plausible to try and stipulate that trademarks should never be used in the possessive or the plural.)

I know, some of you will say that this earlier post, written when I was young, perhaps suggests a certain lack of sympathy with the whole puzzle genre, a mild prejudice against the very idea of puzzles that inclines me to be mean to Will Shortz. But no, I am perfectly capable of maintaining a level head on this issue. Not everything that Will Shortz bases his sometimes ingenious puzzles on is mistaken. But he is clearly drawing his grammar information — just as nearly everyone else does — from a superficial grasp of what was printed in school grammars in the 19th century. The subject has moved on. There is stuff you need to consider if you are going to talk about grammar or invent puzzle clues that make reference to grammar.

Others among you will say (as some have already said) that unless Will Shortz had got an accessible copy of The Cambridge Grammar, he had only published dictionaries to go on, and they all say that these can be a pronoun. True, they all do. But the point of view that I take does not make grammar depend on authority. It depends on evidence. Part of the tragedy of the present state of English grammatical studies is that the published resources aren't in line with the known evidence. In particular, all published dictionaries are simply wrong about the categories to which they assign quite a few alleged pronouns, prepositions, adverbs, adjectives, and conjunctions. All of them. It's a crisis.

And it isn't Will Shortz's fault, of course. I don't expect him or NPR to take the slightest bit of notice of what I've said here. But this is Language Log. You get the actual truth, and a glimpse of the evidence that backs it up. Plus a glimpse of the extent to which what is commonly believed about English grammar is at many points demonstrably incorrect.

Posted by Geoffrey K. Pullum at July 23, 2006 01:35 PM