Page 561 - Encyclopedia of Philosophy of Language

namely, blue. What is the relation between its doing that and its meaning what it does? Perhaps, for it to do that just is for it to mean what it does; in saying that it names the color blue, one states exactly what it means. (What extra fact should one mention? What fact not determined by that one is determined by what 'blue' means?)
But if that is so, as it seems to be, then Wittgenstein's next question arises. What has all that—either what it means or what it names—got to do with the stan- dards for its correct use or, inter alia, with what it would be true of? Here Wittgenstein uses his notion of a language game to make the point. (A language game is defined by its rules. By contrast with actual words, it is thus explicit which rules set its standards of correctness for words and responses that are moves in it.) A language game is an object of comparison: we may sometimes view some properties of our words as modeled in one or another such game. There is no such thing as 'the language game we are playing' in speaking given words (see Sects. 81,130). That a word, e.g., 'blue,' names what it does is compatible with its figuring in an indefinite variety of language games, with indefinitely many different, and sometimes con- flicting, standards of correctness. So its naming what it does does not settle what it could truly apply to.
Now all of these remarks about what naming does not do apply intact to meaning. That 'blue' means what it does does not fix what it would be true of. Take the sky. Is it blue? If you said so, in the wrong surroundings, someone might be very disappointed when he looked at it for the first time from close up, in an airplane. Sometimes it counts as blue—in some surroundings, for some purposes—and it is then true to say that it is blue. Sometimes it does not so count, and it is false to say so. What changes from one case to another is not what 'blue' names, nor what it means. Throughout it speaks of the color blue, and does so by and in meaning what it does.
By now, the above point has been widely taken, even by those who, not long ago, still thought that meanings of words could be stated by specifying what they were true of. The most usual way of assimilating the point is to echo J. L. Austin in saying that an English sentence is precisely not what can be true or false (it doesn't say anything), then positing some other item (a thought, a proposition) that may be. The point then becomes: an English sentence expresses different thoughts on different speakings, an English predicate different concepts, etc. But a thought has just the sort of truth condition a sentence was orig- inally supposed to.
2.2 Semantic Properties and Occasion-sensitivity
It is just here that Wittgenstein's second guiding idea—the avoidance of 'shadows' between us and the world—leads to major innovations. The principle is: do not try to solve the problem (accounting for vari-
ation in what words with fixed meaning say) by post- ulating new bearers of semantics, since the same problem will arise for them. They, too, will be gov- erned by different standards of correctness in different surroundings. (We cannot cancel out the sign.) If we are debarred from this solution, what solution remains? The answer is to treat semantic properties as normal properties, on a par with ground-level ones. Consider the property of being blue. Having it, we have just seen, is, or may be, an occasion-sensitive affair. Sometimes the sky counts as having it, some- times it does not. Now shift to the property of being true, or true of the sky. If that behaves normally, then having it is an occasion-sensitive affair too. Whatever the semantic item—say, a sentence—if that item may ever count as having that property, then the basic state of affairs is: in some circumstances it would, in others it would not. More generally, the having of a semantics (some set of semantic properties) is an occasion-sensitive affair: which semantics an item counts as having varies with the occasions for count- ing it as having, or lacking, any.
If an item may count, in different surroundings, as having different semantics, then something more than just the item's occasion-independent nature is required for fixing which semantics it counts as having in any given surroundings. The surroundings must help, of course. But, Wittgenstein points out, there being a result depends also on there being such a thing as the reasonable way of understanding the item, or assign- ing it a semantics, in those surroundings. Facts as to what is reasonable depend, in some way, on facts about us, beginning with the simple fact that we are very often capable of seeing what is reasonable. With- out the right background of facts about users, or treaters, or evaluators, of an item, there would also be no facts as to the semantics it counted, in given surroundings, as having.
If the above sort of sensitivity to occasions or sur- roundings is intrinsic to semantic properties, thus a feature of any item that has a semantics, then the Fregean ideal language is ruled out. Any language would be, necessarily, dependent on the reactions of us, or its users, for the semantic facts about it, par- ticularly those about its applications, being what they are. Languages cannot, in principle, be self-propelled, as they would be on Frege's view.
Wittgenstein argues for the pervasiveness of sem- antic occasion-sensitivity in two ways: first in his dis- cussion of rules and what they require (see especially Sects. 84-7); and second in the private language dis- cussion (roughly Sects. 243-72). The first is a direct argument, relying on facts as to the conceivable ways of specifying which rule a given rule is. Any speci- fication leaves at least conceivable doubts as to whether in this case, this, or perhaps rather that, would be in compliance with the rule. ('Can't we imagine a rule determining the application of a rule,
Wittgenstein, Ludwig
539