Wolfgang Schwarz :: Semantics Without Use

Semantics Without Use

Posted on Saturday, 03 Jun 2006.

I've just noticed that I don't understand those who do not base semantics on use, so I'm asking you for hints or pointers.

Here, very roughly, is the position I don't understand:

Speakers of a language have tacit knowledge of its syntax and semantics. Take Karl. As a competent speaker of German, he tacitly knows that, say, "Berlin" denotes Berlin, "pleite" denotes (or expresses) the property of being broke, and "x ist y" is true iff the thing denoted by x has the property denoted by y. Thus he knows that "Berlin ist pleite" is true iff (or expresses the proposition that) Berlin is broke. That explains why he comes to believe that Berlin is broke upon hearing trustworthy people utter "Berlin ist pleite", and that's why he himself utters "Berlin ist pleite" to tell people that Berlin is broke. The object of semantics is this tacit knowledge of speakers. It has nothing intrinsically to do with use, conventions and the like.

I hope this sounds familiar. I think it's a pretty common position, so I'm a little worried that I don't understand it.

What I don't understand is the content of Karl's alleged semantic knowledge: what do "denote", "true", "express" mean here? Let's use "true". What does Karl believe when he believes that "Berlin ist pleite" is true iff Berlin is broke? Some options:

"true" means "true-in-German": Karl believes that "Berlin ist pleite" is true-in-German iff Berlin is broke.
"true" means "true-in-my-idiolect": Karl believes that "Berlin ist pleite" is true-in-his-idiolect iff Berlin is broke, where Karl's idiolect is defined by his tacitly known semantics.
"true" means "true-in-my-community": Karl believes that "Berlin ist pleite" is true in his idiolect and in the idiolect of his fellow Germans.
"true" doesn't mean true-relative-to-anything here. It means true simpliciter. Karl knows that "Berlin ist pleite" is true simpliciter iff Berlin is broke.
"true" expresses a primitive property here: Karl knows that "Berlin ist pleite" has the primitive property T iff Berlin is broke.
"true" means something like "is correctly used by the linguistic conventions of my community": Karl knows that an utterance of "Berlin ist pleite" conforms to the linguistic conventions of his community iff Berlin is pleite.
"true" here means something else altogether.

(6) is ruled out by the claim that Karl's tacit semantic knowledge has nothing intrinsically to do with use, conventions and the like.

(4) collapses into (5) or (7): maybe propositions can be true simpliciter, but to speak of strings of symbols being true simpliciter makes no sense to me. In other words, whatever that is supposed to mean, either this "true simpliciter" can be explained as something else I can understand (like, "beginning with the letter O") -- option (7) -- or it means something inexplicable -- option (5).

(2) makes the content of Karl's semantic knowledge dangerously self-referential: the content of his belief that "Berlin" denotes Berlin is that he has the belief that "Berlin" denotes Berlin, that is, that he has a belief whose content is that he has a belief whose content is... Even if there is a coherent notion of content in here, Karl's knowledge will presumably be trivial and a priori, making it incomprehensible why it takes people years to learn a semantic theory.

(3) also leads nowhere: what does it mean for something to be true in the idiolect of other members of my community, if truth in an idiolect is defined as truth in the idiolect of other members of the community?

(1) means that truth-in-German cannot be explained in terms of individual speakers' tacit semantics of German, because truth-in-German plays a central role in the clauses of that semantics. But what is truth-in-German if not something that is determined by the use and conventions within the German-speaking community or by the tacit semantics of individual speakers? This road is a dead-end.

Since I can think of no other credible proposal for (7), that leaves (5): "true" here stands for some primitive property T. Since the same resoning goes through for "denoting" and "expressing", these too stand for primitive properties D and E of symbols, or primitive relations between symbols and other things.

I find that already very hard to understand. But it gets worse. First, I don't understand at all how those alleged facts about T, D and E could be learned. What a child observes are various patterns of use: people say "es regnet" when it's raining, etc. How could it infer primitive facts about T, D and E (which at any rate have nothing to do with patterns of use) from these observations? Learning a language would seem be be a matter of making lots of wildly irrational inferences.

Second, the above story now goes something like this:

As a competent speaker of German, Karl tacitly knows that "Berlin" D Berlin, "pleite" D (or E) being broke, and that "x ist y" is T iff the thing D by x has the property D (or E) by y. Thus he knows that "Berlin ist pleite" is T iff Berlin is broke. That explains why he comes to believe that Berlin is broke upon hearing trustworthy people utter "Berlin ist pleite", and that's why he himself utters "Berlin ist pleite" to tell other people that Berlin is broke.

But what kind of explanation is this? Why does Karl infer that Berlin is broke upon hearing trustworthy people utter "Berlin ist pleite", just because he believes that this sentence has the primitive property T iff Berlin is broke? The only explanation for this strange inference I can think of is that Karl has learned inductiveley that when trustworthy people say things, the conditions under which their utterances are T are usually satisfied. If this is the real explanation, it is at best very misleading to call the above story the explanation.

Moreover, since standards of use are no part of the concept T, Karl's inductive knowledge is really contingent. If he lived in a community of people speaking Inverted German (with all truth values reversed), he would have learned just the opposite: that when trustworthy people say things, the conditions under which their utterances are T are usually not satisfied. He would also have learned that when he utters sentences that are T, the others often come to believe that the T-conditions of those sentences are not satisfied. So if he wants to tell somebody that things are so-and-so, he utters a sentence of which he believes that is not-T iff so-and-so. After a while of getting used to this, Karl will be a speaker of Inverted German whose tacit semantics is German. Indeed, since there is no conceptual connection between T and conventional conditions of use, I see no reason why ordinary Germans shouldn't have the tacit semantics of Inverted German, and use something like Karl's inductive knowledge to interpret utterances and to decide what to say. If the conditions for truth-in-German are something like the conditions for truth-in-the-idiolects-of-ordinary-Germans, this would mean that, surprisingly, "Berlin ist pleite" is false in German iff Berlin is broke. But surely that's not a live possibility.

So (5) also doesn't work. Have I missed an obvious alternative?

Comments

# Andrew McGonigal on 06 June 2006, 12:03

I'm not sure I understand what inverted German is, if its tacit semantics can mirror those of German. Say we have a true sentence, P, and a false sentence, Q. Their conjunction in German, P&Q is false, given the semantic rule for conjunction. Now, in inverted German, if P is mapped onto a false equivalent P*, and Q to a true equivalent Q*, then the conjunction P*&Q* ought also to be false if the tacit semantics for conjunction remain the same. But the idea of Inverted German was introduced by saying that all truth values were reversed, so shouldn't the Inverted German conjunction be true, since the German conjunction is false?

Is it just the atomic sentences that get reversed (and if so, how are those identified?) Or am I missing something here?

# wo on 06 June 2006, 20:18

Yes, conjunction is also remapped: in inverted German, A & B is true iff either A is true or B is true. In general, whenever a truth-functional connective o maps (v_1, ..., v_n) to v in German, o maps (!v_1, ..., !v_n) to !v in inverted German.

I didn't want to suggest that the (tacit) semantics for German and inverted German are somehow indistinguishable. Rather, my point was that nothing in the position I sketched (the position that explains semantics in terms of tacit semantic theories) excludes that all speakers of German actually use as their tacit semantics a semantics for inverted German. It doesn't matter for that whether the two semantics somehow "mirror" each other.

# Andrew McGonigal on 07 June 2006, 12:31

Why can't the T simply play the role of a derivational device (and so not mark a 'primitive property' in any more than the thin, logical sense acceptable to deflationists)? Max Kobel suggests this kind of view. Say we have e.g. the fragment of Inverted German semantics

"Berlin" denotes Berlin
"pleite" denotes the property of being broke and
"x ist y" is T iff the thing denoted by x LACKS the property denoted by y

(so that the 'T' is in effect playing the role of deflationist falsehood). Then the semantics module derives

"Berlin is pleite" is T iff it is not the case that Berlin is broke

and then applies a separate T-to-meaning rule:

From a derived T-sentence of the form:

(*) "S" is T iff it is not the case that P

derive (+ make consciously accessible, etc)

"S" means-in-our-language that P

So basically, we hypothesize that derivations of falsity-conditions rather than truth conditions explain the semantic knowledge of Inverted Earth speakers. But in both cases, appeal to the T-property is only really helping us get from a finite set of axioms/rules to some semantic knowledge. Now Karl never has to appeal to a special kind of inductive knowledge to infer that Berlin is broke upon hearing trustworthy speakers of his language utter "Berlin is pleite". He knows what "Berlin ist pleite" means, he knows what it is to be trustworthy (inter alia, to say what one means and mean what one says!) so he can just put those two pieces of information together to work out that (probably) Berlin is broke.

Now, we have two possibilities for the rules that govern the semantic module. One is that it carries out truth-derivations, and the other falsity-derivations. We appeal to empirical considerations (simplicity of derivation, coherence with other cognitive systems, etc) to argue that it is a better empirical hypothesis that the former possibility obtains. (For example, it typically takes less processing to see that an object has a property, than that it lacks it).

When we come to make explicit claims about truth and falsehood in contexts outside of their instrumental derivational role, we appeal to the schemas:

"S" is true iff "S" means-in-our-language that P, and P
"S" is false iff "S" means-in-our-language that P, and not-P

Given the premise - derivable both on the 'German' and the 'Inverted German' semantics - that "Berlin ist pleite" means-in-our-language that Berlin is broke, we can derive the intuitive falsity conditions.

Does that work?

# wo on 07 June 2006, 18:55

Thanks, Andrew. I think I see how you could use "true" (or "T") as mere instrumental derivation device, but that doesn't really answer my concern. I wasn't very clear on that: I'm not particularly concerned with the meaning of "true" in people's semantic theory. What I don't understand is in general the alleged content of our semantic knowledge.

In the framework you offer, the problem mainly arises with "means": When Karl derives from his semantic knowledge that "S" means (or means-in-our-language) that P, what is the content of this belief? That by the conventions of our language, "S" should be uttered only if one believes that P? This I would understand. But it's supposed to not mean that. What then? 1. that "S" means-in-German that P? 2. that "S" means-in-my-idiolect that P? 3. that "S" means-in-my-community that P, where this is defined in terms of meaning-in-idiolects (rather than conventions of use)? 4. that "S" means-simpliciter that P? None of these makes sense to me.

# Andrew McGonigal on 07 June 2006, 22:07

Thanks; that's a bit clearer. This reminds me of Dummett's insistence that a theory of sense that is constituted by a theory of truth-conditions has to be augmented by a theory of force. The theory of sense really only fixes the content of the sentence-radical, and doesn't determine the meaning of the natural language sentence. That only gets into the picture once your theory of force tells you that e.g. an indicative sentence that (includes a sentence radical that) means that P should be uttered only if one believes that P. So the theory of force adds the kind of connection to use that you seem to be wanting.

I think that what somebody who was tempted by the tacit knowledge view would be likely to say is that rules like:

(A) If an indicative sentence S means that P then S should be uttered only if one believes that P

are tacitly known, and encoded in the pragmatics module. The outcome of the semantics module and the pragmatics module working together will yield the kind of 'knowledge of the meaning' of sentences (in a broad sense of meaning) that links them to use, conventions etc. And maybe nothing could be a semantics module properly so-called unless it was wired up in the right way to a pragmatics module. So in one sense tacit semantic knowledge does have something 'intrinsically' to do with use, conventions, etc. It's just that in understanding how a particular set of cogs work, we abstract away from the theory of force, etc, pro tem.

# wo on 09 June 2006, 09:45

OK, many thanks, I think I see how that could work. Though I believe it's quite misleading then to use semantic vocabulary like "means" and "denotes" in describing the tacit semantics. If I understood it correctly, the idea would be that we have a cognitive module that just associates sentences with meanings or truth-conditions. It doesn't tell us that "snow is white" means in English that snow is white, nor that it means that in my idiolect, or that it means that simpliciter, or that it should be used in a certain way, or whatever. The semantic module by itself doesn't give us any knowledge at all: there are no *facts* that obtain according to the semantic module. It offers just a table that associates sentences with truth conditions (or whatever).

Real semantic knowledge then is a result of applying pragmatic rules to the semantic module, rules like

(A) If the semantic module associates an indicative sentence S with P then S should be uttered only if one believes that P.

Here, "should" probably doesn't express an individual norm for the subject in question; otherwise my semantic knowledge wouldn't help me at all to interpret other people's utterances. So (A) should rather say that if S is associated with P then there is a convention in my community to utter S only if one believes that P.

I suspect that this picture doesn't really differ from my view that semantic knowledge consists in something like knowledge of conventional conditions of use. It merely adds the speculative psychological hypothesis that this knowledge is the result of two different modules, one containing no information at all, but just a table, and the other a general rule that derives the semantic knowledge out of this table.

Wolfgang Schwarz

Semantics Without Use

Comments

Add a comment