May 21, 2005

Unsavory Propositions

"There, now," she said.
I glanced at her to see if this utterance was deliberate. We had often embarked on prolonged discussions of "there, now" as children, an exclamation I had deemed redundant and, worse, senseless.
"What does it mean?" I would say.
"It means 'there, now'" was all that Martha would reply.
--Cathleen Schine, The Evolution of Jane, p. 36

As you may know, Herman Cappelen and Ernie Lepore's newish book Insensitive Semantics argues for a position called Semantic Minimalism: that the only context-sensitive terms are certain obvious ones, like 'I', 'you', 'he', 'she', 'here', 'now', tenses, and some others. An utterance that doesn't contain any of these terms semantically expresses the same proposition, no matter what context it is used in. So Cappelen and Lepore say things like "when Nina says 'John is ready', she expresses the proposition that John is ready."

This position is hard to one-up, but Martha has done it. On her style of explanation, even the obvious context-sensitive terms mean just what they say. And I think she wins. Because the relevant question here is, exactly what are the things that utterances semantically express? And I don't think that there's any reason to think that the answer is 'propositions', if we take different utterances of "I was here three minutes ago" to express different propositions.

Cappelen and Lepore argue for Speech Act Pluralism--that a single utterance performs many speech acts. So if I say "Matt Weiner is in his office" I say that Matt Weiner is in Curtin Hall, and that Matt Weiner is in his office, and that the author of the blog Opiniatrety is in his office, and a whole mess of other things besides. In this I think they're absolutely correct.

But why should we single out one of these speech acts to say, "This is the proposition semantically expressed"? What makes one of these speech reports special?

Cappelen and Lepore's answer, as I understand it, is that people can communicate across very different contexts and in contexts in which there is great ignorance--when the speaker doesn't know anything about the circumstances, beliefs. The proposition that is semantically expressed is what can be exactly communicated. Thus "the proposition semantically expressed is that content the speaker can expect the audience to grasp (and expect the speaker to expect them to grasp) even if they have mistaken or incomplete communication-relevant information" (p. 184), and mutatis mutandis audiences can expect the same of the speaker, and non-participants can grasp the content.

It seems to me as though the proposition semantically expressed is supposed to be a minimal fail-safe content. If you don't grasp this, you don't grasp what has been said. Somewhere (I can't find it right now [UPDATE: see bottom of post]) I believe C & L say something like, if you're reporting someone who said "She is tall," and you don't know who they were talking about, you have to say "I don't know exactly what they said, but they said that some woman was tall." Whereas if you're reporting someone who said "John is ready," and you don't know what activity was salient, you can still say "They said John was ready," and you're guaranteed to have expressed something the speaker expressed--even if you don't know what John was ready for.

But in fact sometimes we can communicate even without knowing what proposition is semantically expressed, as C & L have it. Take their own expression of the possible gaps in communication (5stC is the context of their utterances):

Take you, our reader; we have no idea who you are; we know next to nothing about your beliefs... (you're not here in 5stC with us).

What does the bolded sentence semantically express? According to C & L's theory, it's a proposition of the form <Matt Weiner, in 5stC with, C & L>. (To keep it simple, I'll assume that the utterance in question is the token of that sentence in my copy of Insensitive Semantics, which I never let anyone else see; so 'you' uniquely picks out me.) But C & L, the speakers, don't know that it expresses that proposition. And I know that they don't know that--in fact, they as good as told me that they don't know it. So here's a case where communication is possible even though the hearer can't expect the speaker to grasp the proposition that's semantically expressed.

(I picked on the bolded sentence because it's the only one in this passage where 'you' isn't in an intensional setting. "We have no idea who you are" is an interesting case--obviously it's meant to mean something like "We have no idea who's reading this sentence" rather than "We have no idea who Matt Weiner is, in the sense that we wouldn't recognize his face on the side of a milk carton"--though that's also true. I think here 'you' has to be treated as one of Nunberg's descriptive indexicals; 'you' contributes the property 'being the reader of this sentence'.)

So the 'proposition semantically expressed' as C & L have defined it doesn't necessarily play the role they want it to play. What does play that role--that of (as I interpret it) the minimal fail-safe content? The meaning of the words. I can grasp, and can expect C & L to grasp, that they said "you're not here in 5stC with us," where 'you' is the term that picks out the addressee of the context, 'us' is the term that picks out the speakers of the context, etc.

That is to say--if we are looking for some rock-bottom level of content that is communicated whenever there is understanding, the sort of thing expressed by "'There, now' means 'there, now'" is better suited than the sort of thing expressed by "'Nina, in saying 'John is ready', says that John is ready."

Of course, on that view what is semantically expressed by an utterance isn't what ordinarily gets called a proposition. And I have thus proved the thesis of this post. I haven't proved what I'd really like to--that propositions, as usually taken in philosophy, are no damn use whatsoever--but maybe that'll wait until I turn this into a real paper.

[UPDATE: The passage I was thinking of above is on pp. 93-4, where C & L consider how someone might report an overheard utterance of "That's a nice one" if she doesn't know what was demonstrated. They say that one of four things will happen: She'll say "I don't know what he said," she'll investigate to find out what was demonstrated, she'll quote directly "He said, 'That's a nice one,'" she'll say something like "He said that some demonstrated object was nice." This is supposed to generalize to every expression in the Basic Set of context-sensitive expressions, and to contrast with "John is tall" and "John is ready" and the like.

But note: If I see "I am better than you" written on the wall, and someone asks me what is written there, I can say "Somebody wrote that they were better than me. I don't know who."]

Posted by Matt Weiner at May 21, 2005 05:47 PM

OT - I'd appreciate your take on whether the statement "They're trying to see it to that more children get killed" is an ascription of intention to cause children to get killed.

Apparently post titles don't have the quasi-mystical powers I had imagined.

Posted by: washerdreyer at May 23, 2005 09:27 PM

While I suppose I might want to agree with you that propositions are no use for understanding language (how often to linguists talk about propositions rather than sentences or utterances?) I'm not convinced by your examples. It sounds like if your/their interpretation of the examples is correct, then when I say "Hesperus is Hesperus" I also say that Hesperus is Phosphorus and many other things of that sort, which seems a little distasteful to me.

Nice to meet you this week!

Posted by: Kenny Easwaran at May 30, 2005 11:57 AM