In Chapter 2, we saw how Wittgenstein used the builders game to open space for serious consideration of an alternative to the Fregean picture, one that makes meaningfulness prior to meanings and patterned behavior prior to Fregean systematicity. He attacked the central idea that the basic units of meaning are propositions that have the general form: this is how things stand (TLP 4.5). In consolidating his view that this is a bad picture to impose on all language, he later brings out that the expressions “proposition,” “general form of the proposition,” and “logical form” are all proprietary notions that are made for each other (PI §136). So, we can see that Wittgenstein would agree with Dummett that once we introduce certain parts of the picture (propositions and the Fregean notion of the sense of a sentence) we find other parts follow (a truth-conditional semantics and the systematicity of language). Wittgenstein urges that this is wrongly construed as a discovery about the deep structure of any possible language. It is a grammatical mistake. A banal feature of natural language is taken for a deep metaphysical insight into the place of the proposition in language. The exploration of the highly sophisticated picture of language that follows obscures a deeper problem, one that I identified as “the problem of normative similarity.” As we saw, the builders game is used as a foil to the Fregean picture, which is clearly unnecessary to either describing or explaining the linguistic competence of the builders. But what is necessary is some appropriate link between words and objects. Traditionally that link has been identified with reference.

The first problem of normative similarity arises with Wittgenstein’s critical examination of the explanatory role assigned ostensive definition by

denotational theories of meaning. The object denoted by an expression fixes the meaning of that expression by providing the standard for the correct application of the term to other objects in the environment. This theory is associated with a companion picture of the acquisition of language, one that is compelling in its simplicity. Part of the enduring appeal of the denotational theory is precisely its capacity both to explain meaning and to account for the acquisition of language. Adults point to individual objects in the immediate environment of the child and name them. The child, as a result of this demonstration, associates the name with the particular object and subsequent to making this association uses the same name for any other object belonging to the class of which the first originating object was a member. Though she knows much less than the adult, the understanding that the child has of the words she has learned is essentially the same as that of the adult. Yet how can the utterance of a word by one person while standing close to

an object effect an association in the mind of the second person such that the second person can subsequently generalize to the same objects as the first person would? The answer is that the relation established between the word and object is not one of simple association, but of naming. Through ostensive definition, the novice takes the object as the standard guiding subsequent use of the word. The effectiveness of the ostensive definition lies with the cognitive capacity of the novice, in particular her capacity to grasp that the baptismal object (or property of an object) is the exemplar that functions as the standard for correct application of the term uttered. Wittgenstein’s criticism of ostensive definition appeals to the indetermi-

nacy of the demonstrative gesture itself. The range of properties that can be distinguished in the region of the pointing gesture is indefinitely large, and the target property (even if reduced to the range of obvious properties like color, shape, size, kind of object) cannot be individuated solely by pointing, no matter how emphatic. This indeterminacy can be resolved by identifying the target property independently of the pointing gesture, through the use of a sortal expression.1 But this leads to an explanatory regress. Teaching by ostensive definition cannot explain how the novice comes to identify members of a class precisely because it presupposes that the novice already has some relevant classificatory categories, understands the notion of class membership, can take the baptismal object as the paradigm by which other members of a class can be so recognized (that is, identified as the same), and realizes that the teacher is not only uttering a word, but a general name which functions to designate indifferently the members of a class (cf. PI §32). The child both has the concepts of the game she is to learn and can engage in the semantic game of naming. Moreover, all of these abstractly characterized cognitive abilities must be relativized to the particular context and setting within which the ostensive definition is made. The conclusion that Wittgenstein draws is that naming cannot fix meaning, but is itself a semantically sophisticated act that presupposes a great deal of cognitive and

contextual stage-setting: “only someone who already knows how to do something with it can significantly ask a name” (PI §31). Asking for a name is thus a metagame, one presupposing facility with a primary game. This first problem of normative similarity concerns sameness with respect

to categorization, especially object categorization. It is raised in the context of how meaning is originally fixed for the linguistic novice. Naming presupposes a conceptual background that is semantically rich, that already enables the recipient of an ostensive definition to categorize successfully. The success of ostensive definition for initiate learning depends upon the novice already having that very background and conceptual competence that ostensive definition is supposed to engender. One must already be master of a language in order to learn a new language through ostension: