I'm just back from a fun conference on Moral Contextualism in Aberdeen. Some of my photos of the conference are now online. (I can assure readers that there is a perfectly rational explanation of those ones with the cutlery.)
Highlights from the conference included Berit Brogaard's attempts to convince us that the truth-value of one and the same proposition can vary with the speaker. (Her presentation is available online.) She appealed to Recanati's views on direct speech reports (whereby the quoted sentence is not merely mentioned but used) to argue that such reports, although they create shifed contexts, do not change the parameters of the circumstance of evaluation.
While I agreed that the view seems to have advantages over Macfarlane-style relativism, one thing that made me suspicious about it was that one surely wants to deliver the same result for unquoted sentences as for quoted sentences. The following pair, for instance, sounds very odd:
1. At t, John said 'Murder is wrong' and he was mistaken.
2. The sentence John uttered at t was true.
But it wasn't clear how the account could deliver this uniformity; since it appeals to special features of quoted speech to deliver the desired result in case 1, it wasn't clear how to get it for cases like 2.
John Hawthorne suggested in his talk that 'ought' claims might be subject to contextual variation in semantic value for the same reasons as are the 'can' claims which they (supposedly) imply. This, I thought, might sit quite well with the view (suggested in Lewis) that there is contextual variation in the semantic value of modal utterances due to contextual restriction of the quantifiers over worlds which they involve.
I was particularly interested, therefore, to hear Ralph Wedgwood developing a version of Angelika Kratzer's possible-world semantics for 'ought' claims in the next-but-one session. Kratzer's basic idea here (subject to many bells and whistles, of course) is that context supplies a set of propositions which are held fixed across all relevant worlds, and an ordering on those worlds. 'It ought to be that p' is then true iff p is true in all the worlds ranked as unsurpassed.
Tomorrow I'm off to Southampton for the Joint Session, so more paper reports should follow soon.
Subscribe to:
Post Comments (Atom)
5 comments:
Hi Carrie,
Thanks for the objection! It was bugging me all the way back to the states (In fact, I missed all the movies on the planes :-). Anyway, I agree that there is a problem here. In the past I have been willing to admit that the Kaplan-style semantics I am endorsing is committed to absolute truth for utterances and sentences in contexts. So, in the past I would say that utterances and sentences-in-contexts are true simpliciter. I now think that is a mistake. John Hawthorne gave me the following annoying counter-example (u is any utterance):
U is true
U means that p
So, p
If utterances are true simpliciter, then the argument is invalid (for U might have been made 3 years ago). So, let U be my assertion 3 years ago of ‘I am hungry’. That utterance means that I am hungry (assuming Kaplan-style semantics). But it doesn’t follow that Brit is hungry (now).
So, the argument is invalid. Bad news! For the argument is, of course, valid (I lost a lot of sleep over that one).
So, I now deny the notion of truth simpliciter. We are already required to give up the notion of truth simpliciter for propositions within standard semantics. For proposition truth is, as a minimum, relative to possible worlds. In other words, propositions aren’t just true or false; they are true or false relative to circumstances of evaluation. What’s new on my part is that I now want to give up truth simpliciter for sentences-in-contexts and utterances as well.
That blocks Hawthorne’s and your objections in the following way. U isn’t just true. ‘U is true’ is short for ‘U is true at the circumstance of evaluation determined by the context of use. So, ‘U is true’ means ‘U is true at the circumstance of evaluation determined by my current context of use’. Hawthorne’s argument then comes out valid, which is as it should be.
Now, back to your objection, Carrie. Am I committed to the following?
1. At t, John said 'Murder is wrong' and he was mistaken.
2. The sentence John uttered at t was true.
I would say no. Suppose (for the sake of argument) that my view is that murder is o.k. Sentence (1) is then true at the circumstance of evaluation determined by my context of use (my circumstances, like Macfarlane’s, have judges in them). But I am not committed to the truth of sentence (2). For by my standards murder is o.k. (we are assuming :-)
But Carrie would now say: “but wait Brit. You have to accept sentence 2. For you just told us earlier that the sentence John uttered at t was true back then when he uttered it.”
Reply: I am committed to ‘the sentence John uttered at t was true at the circumstance of evaluation determined by John’s context of use’. But I am not committed to ‘the sentence John uttered at t was true at the circumstance of evaluation determined by MY context of use’ (for simplicity, I am ignoring tense here, since tense operators don’t shift the judge parameter).
My view still has an advantage over MacFarlane’s, for we don’t need contexts of assessment (everything follows given contexts of use, circumstance-shifters and Recanati’s account of direct speech reports)
I hope I can watch some movies now. But I’d better hurry up in case Carrie replies :-)
Hi Brit,
Thanks for the comment. Sorry about the movies!
I remember thinking at the time that something like this was probably your best move. (In fact if I remember rightly, when I started asking my question in the session, I was just assuming this was what you believed. Which is probably one reason why my question was not maximally clear at the time!)
But what I wondered about this combination of views was: does it still have room for (you to express a commitment to) the kind of faultlessness that is so often what motivates the move away from vanilla invariantism? As you say, you shouldn’t take yourself to be committed to ‘the sentence John uttered at t was true at the circumstance of evaluation determined by MY context of use’; indeed, you should say: ‘The sentence John uttered at t was false at the circumstance of evaluation determined by my context of use’. But then, in what sense was he faultless according to you? He was faultless according to the circumstance of evaluation determined by his context of use, but so what? You should reason to yourself: ‘To be faultless according to some circumstance of evaluation different from my own current circumstance of evaluation is not in general a way to be faultless.’ (Cf: ‘To be true according to some world different from my own is not in general a way to be true.’) Of course, I’m assuming here that what it takes for something to be ‘faultless’ without qualification is for it to be faultless according to the context of evaluation which is currently operative. (Maybe there’s wiggle room there, but it seems like it would need some work.)
It could be that you have some motivation for non-invariantism other than faultlessness, but if not, the motivation seems hard to state given the view you end up with. I was trying to structure my point as follows: you need faultlessness to get the anti-invariantism off the ground, but you need disagreement in order for the view not to collapse into contextualism. But it’s hard to have both: as you make an effort to respect one, you seem to lose hold of the other.
Thanks, Carrie! I think you raise a lot of important issues here. Let me begin by reviewing a bit. I think faultless disagreement is a problem for both the standard contextualist and the perspectivalist/relativist. Consider:
John: Murder is wrong
Mary: No, it’s not.
Given naïve contextualism, the exchange cashes out to:
John: Murder is wrong by my standards
Mary: No, it’s not wrong by my standards
It’s obvious what the problem is here (A: I am hungry. B: No, I’m not). Given sophisticated standard contextualism, the exchange cashes out to:
John: Murder is wrong by our standards
Mary: No, murder isn’t wrong by our standards
Here it seems that John and Mary are disagreeing about what their standards are, not about whether murder is wrong (assuming that any propositions are expressed all when the speakers have different standards)
Perspectivalism/relativism does a bit better: John and Mary assign different truth-values to one and the same proposition, viz. the proposition that murder is wrong. The problem for perspectivalism/relativism is that the proposition that murder is wrong and its negation do not contradict unless they are evaluated at the same circumstance (but they are not evaluated at the same circumstance, for John and Mary have different standards; hence, different circumstances). MacFarlane has a nice article on this online.
Back to the main issue you raised, Carrie. The problem for perspectivalism, as I see it, is this:
Suppose John says: murder is wrong. Call this sentence-in-context/utterance (A).
The perspectivalist agrees with the relativist that ‘murder is wrong’, as uttered by John, expresses the proposition that murder is wrong. The moral standards are included in the circumstances of evaluation rather than in the proposition. But unlike the relativist, the perspectivalist wants to say that the proposition expressed is true iff it is true at a circumstance that includes the speaker’s (= john’s) standards (the genuine relativist will say that there is no fact of the matter b/c we need to specify the evaluator first). Let’s say that John’s standards prohibit murder. (A) is then true.
If Mary then says ‘John said “murder is wrong” and he was wrong’, then on my account (inspired by Recanati) the quoted sentence expresses a proposition that is evaluated at a circumstance that includes Mary’s standards. Let’s suppose Mary likes murder. Then what Mary said is true.
But, and this is what I think is the core of your objection Carrie (correct me if I am wrong),
when I (as a semanticist) evaluate (A), I am in the same situation as Mary. So, why shouldn’t we get the result that it’s my (the semanticist’s) standards that count (rather than John’s). Suppose, for instance, that I (the semanticist) am on Mary’s side (I like murder). Then we should have concluded earlier that John’s utterance (= A) is false – not that it is true.
I think that was why I started rambling about object-language/meta-language issues at the conference :-)
And I am going to do it again. Perhaps we can simply say that object-language occurrences of both S and ‘S is true’ are true iff S is true at the circumstance of evaluation determined by the speaker who uttered S or ‘S is true’. That seems to avoid your objection.
The occurrence of ‘true’ in the meta-language is the usual one (given Kaplan-style semantics). So, in the meta-language, S is true iff S is true at the circumstance of evaluation determined by the context of the speaker who uttered S (and not the context of the semanticist).
Hope this will work :-)
"Perhaps we can simply say that object-language occurrences of both S and ‘S is true’ are true iff S is true at the circumstance of evaluation determined by the speaker who uttered S or ‘S is true’. That seems to avoid your objection."
The trouble, I thought, was that you, i.e. the perspectivalist semanticist assessing John, want to be able to say that John's utterance, S, was true. But you have to say this by uttering 'S was true', thereby making yourself, not John, the speaker ... And since you are now the speaker, 'S was true' is not true. No?
You are right, Carrie. And I definitely need to think more about this. But it is exactly for the reason you point that that I want to distinguish between object-language occurrences and meta-language occurrences of 'is true'. Unlike proposition truth, utterance truth in Kaplanean semantics is absolute (i.e., not relative to circumstances of evaluation). But I think there are several good reasons to think that object-language occurrences of 'is true' are not absolute:
First, object-language occurrences of 'is true' can be tensed (witness: what John said/uttered was true but it isn't true anymore). Meta-language occurrences of 'is true', on the other hand, are not tensed (you can make it tensed but it won't help you for if something is true absolutely or simpliciter, then any added tenses will be redundant).
Second, Schiffer, Hofweber and others have argued that 'it is true that p' and 'p is true' are just pleonastic paraphrases of 'p'. If, however, meta-language occurrences of 'is true' behave in the same way, then "Utterance U is true. So, U" should be valid. But it is not in Kaplanean semantics (when formulated in the meta-language). For utterance truth is absolute. So U may have been made 3 years ago or in some non-actual world.
So, I think there is already good reason to distinguish between object-language occurrences and meta-language occurrences of 'is true'.
To make a long story short, I can agree with you that I, i.e. the semanticist assessing John, want to be able to say that John's utterance, S, IS true. But that's o.k. for I am now speaking in the meta-language, not in the object-language.
Even if I am wrong, there is definitely some interesting issues here to sort out concerning the differences between meta-language occurrences and object-language occurrences of 'is true'.
I am still struggling with:
U is true
U means that p
So, P
For on Kaplan-semantics it comes out invalid!
Post a Comment