Defects

One of the striking differences between linguists/philosophers of language on the one hand, and philosophers who write about language on the other, is that the regular philosophers use remarkably few examples of defective sentences. For various reasons (sometimes because they aren’t relevant) there will be few examples of syntactically well formed sentences. But there often won’t even be many examples of false sentences.

For example, and I’m just using this because it’s an illustration, in Judith Thomson’s Carus lectures at the APA last week, there were dozens and dozens of good sentences involving evaluative terms given as examples, and used to motivate some theories about how widely some of these terms could be used, and no examples whatsoever of defective sentences. This doesn’t seem particularly unusual outside philosophy of language. So my example below, *Jack and Jill ought to be dropping by soon, had to be added because it wasn’t there.

Sometimes this means philosophers miss nice arguments for their own positions. But often it means they don’t check that all the things that should be defective on their accounts really are defective, or that all the things they claim are non-defective are really non-defective. For instance, I hypothesised below (following Thomson) that there’s an epistemic sense of ‘ought’, meaning something like probably. But it’s very hard to see how to paraphrase (1) using ‘ought’. Certainly not as (2) or (3).

(1) Oswald probably shot Kennedy.
(2) *Oswald ought to have shot Kennedy.
(3) *Oswald ought shot Kennedy.

Maybe there is an epistemic sense of ‘ought’ that can’t be used to paraphrase (1), or maybe I’m just missing an obvious paraphrase, but it would, I think be nice to know just what’s going on here.

The opposite problem came up in the session on the metaphysics of fiction. Amie Thomasson, following among others Peter van Inwagen, has been arguing that reflection on ‘critical’ statements about fiction, like (4), shows us we’re committed to believing in fictional characters.

(4) Some of the characters in the Odyssey are drawn with more detail than any character in O Brother Where Art Thou.

There’s a lot to be said on this topic, especially on the issue of just outlandish this kind of ontological commitment might be. (Thomasson has done quite a bit of work on this, arguing that it’s no more outlandish than most of our everyday real-life ontological commitments, such as our commitment to the existence of socially constructed abstracta like marriages and mortgages, and those look like interesting arguments to me whether or not they are ultimately right. At the conference Mark Richard made some good points suggesting the analogies might not be as close as Thomasson has argued. Some smart young metaphysician should sort all this out I think!) But I want to stress a different point here.

Thomasson, like van Inwagen, doesn’t think that we get ontological commitments from sentences like (5).

(5) Scylla was a six-headed monster.

In particular, we don’t get ontologically committed to six-headed monsters, of which I’m pretty sure there are none. Nevertheless, it’s true that Scylla was a six-headed monster. Why no commitment? Because when we assent to (5), what we really mean is something like In the Odyssey, Scylla was a six-headed monster. There’s a difference between claims which we accept because they are true in the fiction, like (5), and claims we accept because they are true about fictions, like (4). The latter are meant to be interpreted literally, so they carry ontological commitment. To use a more familiar illustration, (6) is meant to be a literal truth, while (7) is only something we take to be true in the fiction.

(6) Sherlock Holmes was created by Arthur Conan Doyle.
(7) Sherlock Holmes was a clever detective.

Now if these two uses of ‘Sherlock Holmes’ are meant to be quite different, one might well expect that we couldn’t easily conjoin these sentences. And I think that’s sort of right, though not all informants agree. I think, that is, that (8) is at least questionable.

(8) ?Sherlock Holmes, Arthur Conan Doyle’s most famous creation, was a clever detective.

But I’m not sure this generalises. I’m not sure, that is, that English recognises the smooth distinction between critical claims (like (4)) and fictional claims (like (5)) that Thomasson and van Inwagen need. So (9) for instance sounds perfectly fine to me, whereas it should sound at best like a cheap pun if Thomasson and van Inwagen are correct.

(9) Odysseus is the craftiest character in the Western Canon.

The point is that (9) is saying that Odysseus the guy is crafty, indeed a crafty character, not that Odysseus the character in the poem is a crafty construction. We’re attributing a certain kind of guile here to Odysseues, not to Homer. But (9) also says that Odysseus is in the Canon, and hence presumably is fictional. The overall form of (9) is the same as (10), which is unambiguously critical by these lights. (And I guess false, but that’s no importance here.)

(10) Odysseus is the most influential character in the Western Canon.

The kind of influence here is the influence that Robinson Crusoe, the fictional character, has in abundance (think of how many Crusoe references you’ve heard around the traps) and not the influence that Robinson Crusoe, the washed up and mostly lonely sailor, lacks. So is (9) meant to be literally true, or something that should be paraphrased away? I’ve got no idea, and I suspect until and unless there’s a good answer to this, the argument that (4) should be paraphrased away just like (5) will look more plausible than perhaps it ought.

I don’t mean to be too critical here. I’m rather partial to the van Inwagen line on fictional characters, and I’m pleased that Thomasson has done all this good work on making that line more plausible. But I think there’s still some underinvestigated assumptions being used to prop up the theory, and they wouldn’t be so underinvestigated if philosophers were all in the habit of looking for defective sentences, rather than just leaving this to the philosophers of language.