Friday, September 28, 2007


Language in Context: Stanley's motivations

I've been reading the introduction to Jason Stanley's Language in Context, which I got my hands on this week, and it gives a nicer summary of the considerations motivating his project than I think we've had previously. I'm going to be reading the essays over the next while, and I thought I'd try to blog along the way. Here I'm going to be unambitious, and simply try to get straight on the motivations for the project Jason tries to carry out over the course of the papers.

The main thesis seems to be that semantics serves up the intuitive truth-conditions of utterances in a systematic way, even given the ubiquity of context-sensitivity in natural language. There seem to be two main kinds of considerations pushing Jason towards this view:

a. Only then can we avail ourselves as semanticists of the resources afforded by speakers' judgments concerning the intuitive truth-conditions of various utterances
b. Only then can we account for our ability to effortlessly acquire information about the world from hearing utterances of novel sentences (even when those sentences are context-sensitive)

In this introduction, Jason seems to focus on three principle kinds of opponent to his project. Firstly, there are those who think that successful communication doesn't require that one grasp the same proposition as was expressed by another's utterance, but only one sufficiently similar (Jason attributes this thesis to Heck and Bezuidenhout). From his perspective, Jason just thinks this rests on an unsupported pessimism - part of his project is precisely to explain why grasping the proposition another's utterance expressed isn't nearly as demanding as people have supposed.

The second kind of opponent, represented by the authors of Insensitive Semantics, agrees with Jason that semantics should offer up something truth-conditional, but thinks that asking for the intuitive truth-conditions is demanding too much. Instead the semanticist can only offer the minimal proposition - an 'informationally impoverished monstrosity' as Jason describes it (12) - and has to leave all of the real work - explaining how we so effortlessly gain information about the world from each other's utterancs, which presumably involves entertaining something other than the minimal proposition - to pragmatics. And, as Cappelen and Lepore very explicitly stress, if my memory serves me correctly, this view cannot avail itself of the resources afforded by speaker-judgments about the intuitive truth-conditions of utterances. As far as Jason is concerned, that's like trying to do syntax whilst ignoring native speakers' judgments of grammaticality and ungrammaticality; it's tying both your arms behind your back before the big fight.

Lastly, we have those who think that semantics can only serve up things which fall short of determining truth-conditions - again, the real work is done at the level of pragmatics. Jason thinks that this just renders the whole process of interpreting an utterance of a sentence in a language one understands altogether too unconstrained and mysterious. We might be pushed to such a view should it prove that no project like Jason's can be successful, but it's not a fruitful starting point for understanding communication.

Does this seem like a reasonable summary of the kinds of pressures Jason takes himself to be responding to?

Labels: , , , ,

I haven't read that stuff, but it sounds like the sort of things that make me uncomfortable with the Cappelen and Lepore stuff, and also make me generally like what Jason has to say about this stuff.

But in general, some of the motivation for me is that this project is very ambitious (trying to do so much with just semantics), and if you don't try for it, then you can't tell whether or not it will succeed.
Although I haven't read the introduction to the new collection, your summary of the challenges seems right to me.

I'm not sure I'm addressing Kenny's point with this, but in general, I think it should be mentioned, as I know you're aware, that Jason almost never discusses these issues in the abstract (one exception, partly, is his and King's 'Semantics, Pragmatics and the Role of Semantic Content'). Most of the time, we get detailed studies of particular problem cases for which some of his opponents have argued that the intuitive truth-conditions (relative to the context) are not due to the compositional semantic interpretation.

In my opinion, this is what gives Jason's position real strength.
Hi Aidan,

You say, "The second kind of opponent, represented by the authors of Insensitive Semantics, agrees with Jason that semantics should offer up something truth-conditional, but thinks that asking for the intuitive truth-conditions is demanding too much."

It strikes me that whether what you say is true depends on how you construe 'intuitive truth conditions'. You might mean something like ordinary speaker judgements about truth conditions. But it strikes me that Cappelen and Lepore can capture this, just not in a way that Stanley would like. C&L might respond that they are capturing the intuitive truth conditions of a sentence like (1) in the following way.

(1) John's ready.

Ordinary speakers (in the right contexts) will report that sentence (1) is true in those worlds where John is ready.
Well, let's suppose that's works for the case in question (Let me cancel any implicature to the effect that I don't think it will work - I just don't feel up to trying to think it through this evening).

Aren't Jason and other just going to suggest that this seems to go pretty awry in cases of quantifier domain restriction and others (e.g. 'every beer is in the fridge', and other overly familiar cases)?

So I guess what my first response is to worry about the scope for this response on C&L's behalf, even if it works for constructions like (1).
[My computer crashed and I'm not sure if my last comment went through. This says it better anyways.]


I don't think that the problem with C&L's proposal is that they can't supply, or don't appeal to, any intuitive truth condition.
After all, even in 'Every beer is in the fridge', C&L will say, people in disparate contexts will intuitively report the speaker as having said that any beer is in the fridge. As far as there is an intuitive proposition ascribed, there is an intuitive truth condition ascribed. That is, intuitively, the proposition that every beer is in the fridge is true iff every beer is in the fridge.

Stanley seems to take other evidence as relevant to determining the intuitive truth conditions, like speakers judgments about truth or falsity in counterfactual situations.
So, it seems to me to be that Stanley and C&L disagree about what should count as being the intuitive truth conditions, not on whether intuitive truth conditions should be consulted.
Ok, good, I see the point. My quick response above is indeed much too quick. Let me think about it, and take a look at what C&L say on this issue again.
Hi Bryan and Aidan,

Waters are muddled in the case of C&L, I think.

They tend to talk mostly about "intuitions about speech act content". What they deny (p. 53) is what they call the Mistaken Assumption, which they think drive all contextualist arguments, namely that a theory of semantic content is adequate iff it accounts for all or most of the intuitions speakers have about "speech act content".

On the other hand, a central claim of their theory of speech act pluralism is that the proposition semantically expressed (PSE), or the minimal proposition as it has come to be called, "does not exhaust the speech act content" (p. 145). A single utterance asserts many propositions, one of them being the PSE.

Furthermore, the PSE is the proposition which gives truth conditions. This is seen by the fact that they accept (p. 155) the disquotational schema for utterances: An utterance of 'a is F' expresses the proposition that a is F and is true iff a is F.

When they reply to Carston's objections (p. 184), they assert that the PSE is the content which speakers and audiences know can be transmitted through indirect quotation or reproduction to receivers in different contexts. So, they do seem to hold, as Bryan says, that intuitions can always pick up on the PSE.

However, given the above, we should expect C&L to hold that intuitions about truth conditions are merely a species of intuitions about speech act content.

In other words, in the presence of speech act pluralism, it would be tendentious to generalise Bryan's remark that "As far as there is an intuitive proposition ascribed, there is an intuitive truth condition ascribed." It depends on whether the proposition being intuitively ascribed is the PSE or some other proposition belonging to the set of propositions asserted by an utterance.

Or what do you think?
Hi Andreas,

What you say sounds entirely right. All that I meant by my remark that ascribing an intuitive proposition ascribes an intuitive truth condition was that a truth condition could be extracted from the proposition by intuitive principles. I was thinking along the following lines. (1) You get the intuitive proposition expressed. Then, (2) you extract a truth condition via the disquotation schema (which is intuitive). Finally, (3) this deliviers what I was calling an intuitive truth condition (by preservation of "intuitively" under logical consequence as it were).

I didn't mean to suggest that the truth condition was intuitive in the more robust sense that Stanley seems to be testing for. That is, I didn't mean that it delivers a proposition which agrees with our intuitions what happens about counterfactual situations.
Hi! I'm currently writing a paper on Cappelen & Lepore's "Insensitive semantics" and was therefore very happy to see this conversation. What do you think about this idea: Consider instructions that you give to a high-school class. You give them homework. Their task is 1)to go out and count the "red items" on the school yard 2)to determine how many of the pupils in the other class are "ready". The first task is comprehensible, but the second is not. Isn't this as difference that C&L's theory cannot account for?
Hey, I'll post a reply when I get a chance. But as an advertisement, I'll have several posts on Cappelen and Lepore's tests for context sensitivity up in the next couple of weeks. Stay tuned.
Thanks for the comment Stellan. A quick thought (this isn't my own reactions to your suggestion - rather it's an educated guess at what C&L might say in response):

The context of utterance hasn't been fleshed out enough in these examples for the instruction to count the ready kids to be comprehensible (to us, reading the dialogue in our contexts). Now, I take it your point was precisely that this marks a contrast between that instruction and the instruction to count the red items in the playground - I guess that your suggesting that the latter doesn't require that kind of supplementary information about the extralinguistic context in order to be comprehensible. But I think C&L would appeal again to the kinds of arguments offered by radical contextualists here. If we asked some kids to count the red things, and they included watermelons or Bezuidenhout's red apples, have they complied with our instruction or not? What about something painted red on one side, but blue on the other? Or something striped red and blue?
Thanks for your interesting response to my comment, Adrian. Your comments are very important to me, since I am writing a paper on C&L's book.

I think that you are right, that is what C&L would say. But as you notice, there is a difference between the instruction concerning the instruction "count the red items in the playground" and "count the ready kids in the other class". The difference is that the former is, as an initial instruction, comprehensible, whereas the latter one is incomprehensible. When
carrying out the task, there will maybe be some diffirences in what they choose to regard as red (like Bezuidenhout and others have pointed out), but this does not change the intuition that the kids will find the former instruction

Now my point is that this indicates
that slope from radical contextualism to moderate contextualism is not as slippery as
C&L claim.

C&L claim that there is no difference between "A is ready" and "A is red" concerning context's impact on meaning. But as eg. Kent Bach argues, there seems to be a difference; "A is ready" is conceptually incomplete, but "As is red" is not.
I'm entirely sympathetic - I'm no fan of the supposed slide of MC into RC either. But let me play devil's advocate a bit more.

Suppose they say: Whadya mean the latter instruction is incomprehensible? You understand it perfectly well. It's an instruction to count the kids in the other class who are *ready*. And you know what state you should bring about in order to have complied with that instruction; you've got to have counted the kids in the other class who are ready. Where's the lack of comprehension?

You might say that you don't have any idea how to go about achieving that that until it is specified what the kids are to be ready for. But it's the same with the red case; we don't really know how to go about counting the red things until it is specified in what way they are to count as red. Moreover, absent further clues, we don't have much better a grasp of what it would take to count the kids in the other class who are ready for an exam than we do about how to count the kids who are ready, even though the former is (according to MC) supposedly complete in a way the latter isn't.
Hi Aidan and Stellan,

I think Aidan is write in his response to Stellan's comment. Two quick points:

(A) My first comment is that I think the example of something we can count needs to be changed. I know how to count the kids on the playground, but I have no idea how to count the *red things on the playground* for reasons independent of the ones Aidan puts in the mouths of C&L. Lots of people say things like 'red things' doesn't give us a criterion of identity, and I think something like this is right. Take any red surface, there are lots of red subsurfaces which I don't know whether to count.

(B) My other comment is that I don't know what to make of the fact that I don't know how to count the ready things. I can't count the stars b/c some of them are really far away, I don't know how to count the electrons in the room b/c they're itty bitty. But I don't think of that as giving me information on the semantics of 'star' or 'electron'. Maybe it's just hard to figure out which things are ready and which things aren't. Why should I take this to be a deep semantic point about the semantics of ready things?
Thank you Bryan for the very interesting and challenging comment. I will try to respond and explain my ideas in more detail. The example is just a sketch,
so I'm very happy to receive your comments - they will help me improving it.

A) My point is not that there aren't unclear cases where we do not know if a given item is red or not. There are. My point is just that an instruction where you are asked to count or collect (or whatever) "red things" is far more clear than if you are asked to count "ready people".
If you don't agree with this, we have a clash of intuitions. Since this is intuitions of
how competent speakers would react, this is an empirical question and an experiment would be
very interesting to carry out.

B) My point is not that you need to be able to know how to verify a sentence in order to understand it. I admit that the example might indicate that view, but it's not my
intention. My point is that you do not understand a predicate expression if you do
not know what it is for the predicate to be true of an object. Hence, you do not understand
"star" or "electron" if you do not know what an object is like if it is a star or an electron.
And you do not understand "red" or "ready" if you aren't capable of grasping the conditions under which these predicates are true of objects. (This is just a classical
truth-conditional view of meaning.)

I have some problems with really grasping what you mean when you say that it just might
be problematic to find out "which things are ready and which aren't". I agree that it is difficult to understand the sentence "John is ready" without further specification
of what he is ready for (that is my point). But in a given context, this is not difficult to
understand (or to find out). But as I said, I might just miss you point here.

I have also difficulties with interpreting your last sentence. Maybe it's a slip of the pen
or maybe there is some deeper conflict between our viewpoints.
I don't really understand what you mean by "semantics of ready things". From my viewpoint, it's reasonable to speak of the semantics of expressions, not of

Thank you again, and I hope to hear more from you, Bryan.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?