February 16, 2004

That Thing You Do

I'm prepping a post which is going revolve around slicing and dicing of linguistic data to yield metaphysical conclusions. So I think it's time for some methodology: why I don't think you can always slice and dice linguistic to yield fine metaphysical conclusions. (I haven't got far enough in that post to figure out whether I'm going to need this Get-Out-of-Jail-Free card, but I want to have it ready.)

It's going to turn on this question: What are the things that we do? Are they repeatables or specific individual events? If I wash my car and you wash your car, have we literally done the same thing or have we done things that are numerically distinct but share a common description?

Jennifer Hornsby has often claimed that in ordinary language, things done are always act-types rather than particular action-events. That certainly seems to be true of, say, "Anything you can do I can do better." Witness the following obviously valid inference:

(1) Anything you can do I can do better [... than you]. You can bake a pie. So I can bake a pie better than you.

First of all, it's not even clear what it would be to say that you can do a particular pie-baking, as opposed to baking a pie in general. Even if that were solved, the conclusion isn't that I can accomplish the particular pie-baking events that you can accomplish, but better; it's that I am better than you at the repeatable action of pie-baking.

But let's take "Every little thing she does is magic." What would an inference look like from this?

(2) Every little thing she does is magic. This morning she read the Times. Her reading of the Times was a little thing. So her reading of the Times was magic.

The things referred to here are clearly (by my lights) particular-events. The claim isn't that any action-type she participated in is magic--not that reading the Times is in itself magic. If the things quantified over in (2) were the same as the things quantified over in (2), the conclusion would have to be that reading the Times is magic. But it's her reading the Times that is magic--not even any particular way of reading the Times that anyone else could partake in.

Nor do I think it can be said that we have two senses of "do" here, taking different objects. Consider this sequence:

(3) What she did on her first entrance as Olivia astounded the audience. It was magic. No one else could have done it better.

This seems pretty unexceptionable to me if you give it an ordinary amount of attention. And it looks like both occurences of it have to be anaphoric to "what she did...." But if (1) and (2) involve different construals of 'do', which sense does "did" have here? To make sense of the second sentence, "did" would have to be construed as in (2); to make sense of the third, as in (1).

Yet I think we can comprehend (3) pretty easily. Dialogue might continue:

(4) What did she do on her first entrance? She straightened her tunic.

From (4) it follows that her [particular] straightening her tunic was magic, and that no one else could have straightened their tunic [action type] better than she did.

Why is this sort of ambiguity permissible? In ordinary language, we rarely have the occasion to distinguish between things done as action types and as particular actions. For instance, if I say "Alice did something that made her mother happy," it's rarely material whether Alice's mother was happy because of the fact that Alice called on her birthday or whether it was the event of Alice's calling that made her mother happy. One can most likely find some action-type that will fit the bill for any particular action that fits the bill.

So here's a case where I think the fine points of linguistics don't help us with the fine points of metaphysics. Fundamentally, it doesn't matter whether the direct object of "do" is really a particular event or really an action type--because we usually don't need to think about the difference. You can probably back people into corners about this--maybe you can ask the speaker in (3) "So there's a magic thing that other people can't do as well?"--but a metaphysically imprecise language is usually enough to get us through the day. And why would it have to be any different?

[UPDATE: I should point out that there's a pretty good chance that someone has done some work on the linguistics of "do" that answers all these questions.]

Posted by Matt Weiner at February 16, 2004 07:32 PM