5/29/2009

puzzles from the later Quine on meaning and synonymy

In the Quine volume of the Library of Living Philosophers, Quine says the following in his "Reply to Alston":

"It would be reasonable to refer to those conditions [="the conditions under which a sentence may be uttered"] collectively as the meaning of the sentence."

But "the synonymy relation gains no support from this notion of meaning. The reason is that, on this notion of meaning, no two sentences can have the same meaning; for no two sentences are wholly alike in their conditions of utterance." (1986, p.73)

That last claim strikes me as implausible; is that just me and my un-Quinean prejudices? Is there a decent argument for Quine's claim that two sentences never have the same conditions of utterance?

That's my main question. But I should mention that Quine gives his own very terse argument: "A sentence can be uttered only to the exclusion of all other sentences, and// hence only under conditions not totally shared, if we grant determinism" (73-74). But that strikes me as a (for lack of a better word) weird argument, for at least two reasons. 1. The fact that sentence A is uttered instead of sentence B at a given time and place does not mean that B could not have been uttered (note that the definition of 'meaning' is the conditions under which a sentence MAY be uttered, not IS (ACTUALLY) uttered. 2. I would've thought that any reasonable notion of 'conditions of utterance' would not require the conditions to be specified up to the level of detail of full physical theory; that is, there could be physically different instantiations of the same 'conditions of utterance'. And it strikes me as very strange to require determinism at that linguistic level (even if we want to be hardcore determinists about physics): sometimes I just keep my mouth shut, even if there's some utterance that would have been fully appropriate for those conditions.

11 comments:

Jonathan Livengood said...

Consider the following game. Player one, the proposer, will select two sentences (suggesting that they are synonymous). Player two, the opponent, attempts to specify conditions of utterance satisfied by only one of the two sentences. If the opponent can always do so, then Quine is correct.

Since Quine is talking about utterances, it seems appropriate to take phonemic characteristics as fair game for the opponent in setting conditions of utterance. But if that is allowed, then surely there will always be phonemic differences between any two distinct sentences?

As an aside, I think this is a *ridiculous* attempt to cash out "meaning." (One reason to think it ridiculous is that it admits the above dodge, which relies on purely syntactic differences between sentences.)

What do you think?

Protagoras said...

It occurs that Quine's skepticism about modality might have made him disinclined to allow your first reason, with all of its questionable "may" and "could have been" content. Of course, I think it's absurd to try to get along without modality, but if the goal is to understand Quine's position, rather than arrive at the truth about meaning, Quine's hostility to modality is almost certainly going to be very important.

P.D. Magnus said...

Along the lines of Jonathan's suggestion, suppose that Phi and Psi are would-be synonyms. Now imagine the circumstance under which a madman threatens to kill you unless you say exactly 'Phi.' Under these circumstances, Phi would be prima facie appropriate but Psi would not.

This provides further evidence that any philosophical problem can be clarified by a thought experiment involving a murderous madman.

Rafal Urbaniak said...

Interesting. Thanks, Greg.

My first reaction is, Quine confuses the meaning-normative sense of "the conditions in which a sentence can be uttered" (something like: two sentences have the same conditions in which they can be uttered iff any competent speaker when faced with someone stating one of them, would agree that the person could've equally well stated the other one). In this sense, it seems true that you can identify the conditions under which a sentence can be uttered with its meaning, but it is false that any two distinct sentence will have different conditions, just because there are various physically possible situations where they're affirmed.

In another sense, in which the conditions in which a sentence can be uttered are just all physical conditions at the moment of utterance, it is quite plausible that distinct sentences will have different conditions (for quite trivial reasons). But it does seem slightly insane to claim that THIS is the meaning of a sentence.

Think about it from a computational perspective: if the meaning of a sentence were the set of its conditions in the second sense, then to know the meaning of all sentences of a language will require one to know all distinctions between possible physical situations in which they can be uttered. But this is hardly what we're doing when we understand sentences, and the amount of work to compute all the relevant distinctions in order to understand a sentence would be enormous (I don't even think, doable, in usual circumstances).

Rafal Urbaniak said...

P.D., the murderous madman example might be read as indicating that the notion of `bein appropriate' is too coarse-grained to be identified with (meaning-relevant) utterability conditions.

Say phi and psi are would-be synonyms that differ on their first letter only. Suppose the murderous madman says:

1) I will kill you unless you say exactly phi.

2) I will kill you if you don't tell me how phi and psi differ in meaning.

Now imagine, you utter phi, but to the second question you reply: look, psi's first letter is different.

I reckon, any sensible murderous madman would be then consistent if they killed you for failing to satisfy 2).

P.D. Magnus said...

Rafal: Quine thinks he has offered independent arguments for his account of meaning - that meaning of X is just the conditions under which it's appropriate to utter X. My original madman scenario shows that any two words differ in meaning (given Quine's account).

Your madman scenario is different. That madman is clearly labouring under an inadequate conception of what meaning is. He'd kill me not because I was wrong, but because he hadn't learned the Quinean lesson about meaning.

This provides further evidence that not every thought experiment involving a murderous madman clarifies a philosophical problem.

Rafal Urbaniak said...

"Quine thinks he has offered independent arguments for his account of meaning - that meaning of X is just the conditions under which it's appropriate to utter X."

P.D. Right, and my impression is that this might seem compelling only when we already have a certain sense of what's appropriate. But in this narrower sense, change in the shape of the letters by itself doesn't carry any change in meaning (say the murderer tells you to write something in capital letters only: would that mean that the same phrase written in capital letters have different meaning??).


"My original madman scenario shows that any two words differ in meaning (given Quine's account).

Your madman scenario is different. That madman is clearly labouring under an inadequate conception of what meaning is. He'd kill me not because I was wrong, but because he hadn't learned the Quinean lesson about meaning."

Your scenario works, given Quinean approach, I guess. My point was, I'm not sure there's any lesson about meaning to be learned. If Quine was getting at capturing how "meaning" is used, his account is in wide disagreement with how we think about meaning: "TABLE" does not have a different meaning from "table", even if in certain contexts it might be appropriate to write one but not the other. Of course, you can disagree and accept the Quinean approach. But then, it's likely that you're not analysing "meaning", you're completely redefining the word. And fun as it might me, the notion of meaning which entails that two different tokens have different meanings is hardly useful.

P.D. Magnus said...

Rafal writes: 'Of course, you can... accept the Quinean approach. But then, it's likely that you're not analysing "meaning", you're completely redefining the word.'

Quine would agree with this, wouldn't he? There's a sense in which he does away with meanings (the sense in which you describe) but another sense in which he reconstrues meaning (conditions of appropriate use are the closest thing to meaning that's left in his account). That's why he says that "[i]t would be reasonable to refer to those conditions ... as the meaning of the sentence," rather than saying that those conditions just are what ordinary people have in mind when they say meaning. (He'd object to the idea that ordinary people have anything at all in mind!) Whether we use the label 'meaning' for these conditions or eliminate the label is, for Quine, just a decision about how we are going to use words.

Rafal Urbaniak said...
This comment has been removed by the author.
Rafal Urbaniak said...

I think you're right P.D. But again, I'm not sure how useful considerations of this sort are, where "meaning" is construed that way.

I mean, yes, we can redefine "cow" to mean what we mean by "bat" and then "cows can fly" comes out true. The question is, what philosophical puzzle does this strategy help to clarify or solve?

Anonymous said...

I agree with your point.
I can define the "same" situation
and therefore the same meaning
in any waya I like. Thus, last year at some conference you made a
statement x, and at this year's conference made the same statement -- or even one similar (as I deteremine)--I could say that the meaning is exactly the same. I determine what same and different is. Thus, it seems to me that Quine's distinction has no claim to much solidity.