February 07, 2004

The giant logical spider web of grammar

Sir Winston Churchill is reported to have said once about his electoral opponent, Labour party leader Clement Atlee, "He's a modest man, with much to be modest about."

Friday afternoon's colloquium in my department at the University of California, Santa Cruz, was by a modest man whose modesty is completely unjustified. Dick Oehrle doesn't do anything much in the way of self promotion. He just produces wonderfully rich research papers, mildly unorthodox but not intellectually isolated, often very technical -- quite a bit of logic and mathematics. His fascinating talk on Friday was built from a detailed examination of the extraordinary surprises of the English tense and modality system and a proposal about the precise description of the meanings of the tenses. "Fasten your seatbelts," said Dick quietly just before launching into the main part of the talk.

An encouraging event. Don't beam me up yet, Scotty; there is intelligent life in my subject down here.

Since you weren't there, I'll give you just one example, out of dozens, of the sort of strange factual stuff to be described. It comes in four steps.

1. The form knew is the preterite (or "past") tense form of know. So if you think you know but you don't, the way I will describe the situation tomorrow, when it's in the past, is to say that you thought you knew but you didn't.

2. In the same way, the form could is the preterite of can. So if you think you can but you can't, the way I will describe the situation tomorrow, when it's in the past, is to say that you thought you could but you couldn't.

3. You can't generally use a preterite with an explicit indication of future time; it sounds completely nuts:

??I knew the answer tomorrow.

There's really no way to make that sound sensible without a fairly strained invocation of time travel or dreaming something. It does not sound normal.

4. So now look at this sentence, with the preterite form could:

I could do the job tomorrow.

No time travel there. It's a perfectly normal thing to say. So why on earth would that be? Do we have to postulate two different words with the spelling could, suspiciously similar in their connection to ability but differing just in their compatibility with future time reference?

The point here is not to come up with some particular way of brushing aside this as merely a case of this, or merely necessitating that, and easily handled in some ad hoc way. The point is, rather, that the grammarian confronts dozens, scores, hundreds of facts of similarly unexpected character, all apparently related in some vast abstract web of grammatical and semantic connectedness whose nature is not known, so that any ad hoc descriptive move you make is likely to be undone by its deleterious consequences for the solution of one of the other puzzles (either grammatical or semantic).

Plenty is known about the relevant generalizations. There is a careful description of all of this domain of English, due to Rodney Huddleston, in The Cambridge Grammar of the English Language (see chapter 3), which Dick's talk referred to several times. It fits togther proposed solutions to most of the descriptive puzzles (though possibly not quite the right solutions; you can never quite get all the data to fit in perfectly). Dick thinks all previous accounts place too much reliance on syntactic and morphological classification and have not devoted enough effort to work on the semantics of tense.

What I liked was not any particular technical proposal in the talk, but the way Dick sees what the project is. What we're doing here is like exploring a giant logical spider web of abstract relationships, in many dimensions, in darkness, with no access to evidence about it other than to tally the reports of people who have run into tiny bits of it. That's what makes it so interesting (and such a long-term prospect) to investigate the structure of even a well-studied language like English. That's what Dick Oehrle understands about the work that grammarians do.

Posted by Geoffrey K. Pullum at February 7, 2004 02:55 PM