Insensitive Semantics, chapter 5

Today’s discussion was led by Brian Hedden, an MIT grad student. His handout is available here, and it covers chapter 5 of the book. Structurally, the chapter runs parallel to the discussion in chapter 3, though content-wise, it raises somewhat different issues.

In the beginning of the books, C&L suggest that there are really only two arguments that theorists advance in favor of the conclusion that a certain sentence is context-sensitive in its interpretation: context-shifting arguments and incompleteness arguments. Chapter 3 sought to show that context-shifting arguments, if they’re any good at all, threaten to prove conclusions that are too strong by the lights of people who propound the arguments. The arguments threaten to show that every sentence is context-sensitive, and many theorists who put forth these arguments are want to stop short of that conclusion. We’ve talked about this argument recently. C&L want to employ the same strategy here, arguing that if incompleteness arguments work sometimes, they work for all sentences.

C&L start out with their reconstruction of the form of argument. It has two stages and a conclusion. Here’s their statement.

1. A solicitation of an intuition to the effect that the proposition semantically expressed by an utterance of a sentence S (according to Semantic Minimalism) is incomplete, i.e., it is not the kind of thing that can take a truth value
2. A solicitation of an intuition to the effect that utterances of S have a truth value, e.g., that they can express propositions, and hence, do have truth conditions, and so, can take a truth value
3. (Conclusion:) Something unaccounted for by Semantic Minimalism must be added in the context of the utterance in order for a complete proposition to be semantically expressed

C&L then argue that we can have the intuition the argument trades on in stage 1 for any sentence, no matter how spelled out. Hence, any sentence, no matter how spelled out, is incomplete in the sense of the argument and requires contextual supplementation. Once again, that’s the view (more or less) that they call radical contextualism, and any argument that leads to it serves as a reductio of its premises, at least according to C&L.

The key question for the moderate contextualist who wants to advance this kind of incompleteness argument, then, is whether she can draw a principled distinction between the initial cases where it seems pretty convincing to many people, and the rest of the language where she wants to resist its application. As far as I can tell—and this came out in the discussion today—that question interacts with how exactly the argument goes.

As Brian points out quite correctly on the handout, the first stage of the argument is somewhat odd. Here it is again.

A solicitation of an intuition to the effect that the proposition semantically expressed by an utterance of a sentence S (according to Semantic Minimalism) is incomplete, i.e., it is not the kind of thing that can take a truth value.

And this requires a bit of comment, because you might have thought that in this context, a proposition is, by definition, something that can take a truth-value. Hence, we’re not going to have any intuitions to the effect that there are propositions that don’t take truth-values.

In fact, the situation is even odder. If you look towards the end of the chapter and consider the objections and replies C&L consider on behalf of their opponents, they say this (focus just on how they characterize their own reconstruction of the argument):

Our reply is simply to point out that an incompleteness claim, as we have construed it, is a metaphysical claim. It is not a psychological claim and it is not a claim about what speakers usually do. It is a claim about what propositions exist. It is, for example, the claim that there is no such thing as the proposition that Jane has had enough. The thing picked out by the italicized part of the previous sentence is not a complete, genuine, real, etc. proposition, according to those who invoke Incompleteness Arguments. (p.65)

Here’s why I think that this is weird. Suppose for a minute that we’re in a possible worlds framework where we identify propositions with sets of worlds. In that case, you might think that everybody can agree on which propositions there are: take the set of worlds W. The set of propositions is the powerset of W. If you’re going to have an argument about which propositions there are or are not, you’re going to do high metaphysics, worrying about abstract objects and such. You’re not going to go out and have yourself intuitions about the sentence Jane has had enough or it’s raining.

Here’s a different way of characterizing the first stage of the incompleteness argument. You’re asked to have an intuition that, if you just go by the semantic values of the words in the sentence you can see on the page or hear in speech, you don’t determine something that can be true or false when evaluated with respect to a situation. A different way of putting things: suppose someone describes a situation to you and asks whether a sentence like it’s raining or Jane has had enough is true. The intuition you’re supposed to have is that you’d just be stumped (to use Ephraim’s phrase).

On this way of reading it, the incompleteness argument isn’t about which propositions there are or aren’t. The argument is about whether the compositionally determined semantic value of the sentence, so long as we restrict ourselves to the actual words in the sentence, is a proposition. And the main evidence for that conclusion is the (psychological) feeling that when confronted with a sentence like it’s raining, you just can’t evaluate it as true or false with respect to any world.

(NB: this is a different alternative reading from the one that Brian gave on the handout. It may be that Brian’s reconstruction is more friendly to C&L, and thus is worth pursuing.)

Here’s why I’ve been belaboring this point. Like I said at the outset, the name of the game is to draw a distinction between pairs like these:

1. Jane is ready.
2. Jane is ready for the exam.

C&L’s opponent, the moderate contextualist, wants to say that (1) is incomplete while (2) is complete. C&L then say that by their best lights about what incompleteness amounts to, (2) is just as incomplete as (1). Here are their lights: to say that a sentence is incomplete is to say that the sentence does not settle all questions of the form “would the sentence be true if…?”

Here are a couple of responses: just because a sentence doesn’t answer all of these questions doesn’t mean that the sentence doesn’t express a proposition. It may well be the case that a sentence divides the set of worlds into three parts: the ones that are compatible with the sentence, the ones that are incompatible with it, and ones that the sentence is just plain silent on. You might be drawn to this picture if you’re thinking that propositions are used to model a thinker’s mental state, and if you also believe that a thinker’s mental state isn’t committed about all possibilities. Perhaps we should rather believe that a thinker’s mental state is only committed on the worlds that the thinker has or can reasonably consider: possibilities that are very far removed just never occurred to her, and she doesn’t have a view on those. So to say that a sentence is incomplete in the sense of not expressing a proposition because it doesn’t settle all questions of the form “would the sentence be true if…?” presupposes a particular view about propositions, one that I think is pretty implausible.

Second (related) response: thinkers can intuitively assess (2) as true or false relative to some situations, whereas they’re stumped with (1). That’s obviously a psychological response, and it’s about the felt need for something that we might as well call “completion” that is present when confronted with (1) but not, or not as much, when confronted with (2). So on the reconstruction of incompleteness arguments that I just went through, the psychological response is quite a propos.