Wolfgang Schwarz

Non-Indexical Context-Dependence

I wonder about the best treatment of the following kind of context-dependence, and its relation to analyticity and apriority.

1)	Mozart's piano sonatas are difficult; France is hexagonal; there is no more beer; it is impossible to travel from Berlin to London in less than 3 hours; Tourists from China are always friendly.

Whether such a sentence is true in a given context depends on the contextually determined domains of quantification, standards of difficulty, of precision, etc.

Compare ordinary indexical sentences:

2)	I am hungry; the meeting starts now.

Intutively, and very roughly, to know whether a (1)-sentence is true in a context, you need the context to figure out what is meant by the words, whereas to figure out whether a (2)-sentence is true, you only need to figure out whether what the sentence means is actually the case. Of course, in a sense, you also need contextual information to figure out what is meant by the words in a (2)-sentence: who "I" refers to etc. But isn't that a different sense?

The case of (1) seems closer to ambiguity, as in

3)	the soup is hot; Fred went to the bank; Jones plays the piano.

Here, again, you need contextual information to figure out what is meant. Sometimes this is handled by assuming that there really are (at least) two words "hot" and "bank" (and "playing"?) in English, which just happen to be spelled alike. The contextual information needed for disambiguation is then part of the information you need to figure out what sentence is actually uttered.

I don't think this is a satisfactory account of all (3)-cases, but at any rate, (1)-cases can't be handled like that. There aren't indefinitely many different words in English that just happen all to be spelled "impossible" or "there is". There aren't indefinitely my rules for the use of quantifiers in the way there might be two distinct rules for the use of "bank".

So (1)-cases appear to be somewhere intermediate between (2) and (3).

Here's the relation to analyticity and apriority. I would like to say that in a suitable context, these (or their tokens) are analytic and a priori:

4)	if L is a law of nature, then it is impossible to do something that is incompatible with L; if professional pianists need more than a week of practice to play a short sonata, it is difficult.

But only in suitable contexts. In other contexts they can be false. Yet this is odd: if as a competent speaker of English you need information about the present context to determine whether a sentence (or token) is true, then surely it isn't analytic or a priori!

(If information about semantic values is excluded, even "water = H2O" will turn out analytic and a priori.)

A related problem concerns truth at a world considered as actual. Suppose we evaluate a sentence S at a a possible situation C considered as actual. What if S is a sentence of type (1) and C contains information relevant for the domains, standards of precision, etc.? To decide whether S is true at C considered as actual, do we hold fixed the relevant features of the actual context, or do we interpret S by the features of C? Either option looks bad. On the one hand, the whole point of considering worlds as actual is that we don't hold fixed features of the actual context, not even the semantic values determined by those features; on the other hand, I would have thought that what is a priori in one context can be a posteriori in another.

It seems that we need an intermediate step between disambiguated sentences and propositions:

sounds/marks 
+          => disambiguated sentence
context       +             => determinate content
              context          +          => proposition/truth value    
                               context

Then I could say that apriority, analyticity etc.\ are not features of (disambiguated) sentences, but of those 'determinate contents'.

But I don't like that at all. I'd rather think the rules of our language simply tell us which sounds are true in which contexts. All I can say to motivate the complicated three-step assignment are vague intuitions about content, analyticity, etc. (Moreover, at least for analyticity, it seems obvious to me that it is a property of sentences, not contents.)

So I'm more inclined to lump all these steps together, treat (1), (2) and (3) alike, and claim that whether a sentence is analytic or a priori never varies from context to context. But I'm afraid I'd then have to say that there is no context in which it is analytic, a priori or necessary that whatever is hot is not cold (or even that whatever is hot is hot), and that sounds a little crazy.

Is there any good literature on this?