APA Newsletters
Fall 1999
Volume 99, Number 1

Book Reviews

Previous | Next

Rita Nolan, Cognitive Practices
Blackwell Publishers, 1994. 171 pp.

Reviewed by Jonathan Wilcock

Teaching an introductory course in the philosophy of language is an enduringly difficult task, owing not only to the inaccessibility of much of the material but also to the difficulty for the student in seeing how many of the topics are related. Some may find Cognitive Practices helpful for its broad range of discussions united by a common thread of argument. In the course of arguing against a particular conception of a theory of meaning, Nolan presents criticisms of Fodor's language of thought hypothesis, an interpretation of Wittgenstein, and discusses the work of Paul Grice and Nelson Goodman, amongst many others.

Faced with the diversity of views in the philosophy of language, Nolan amalgamates certain key features of different positions into what she calls "The Standard Theory" as her target for criticism. The main assumptions of the Standard Theory are (1) naturalism, (2) that propositional thought is independent of language, and (3) that the primary function of language is the communication of propositional thoughts by conventional means. On this theory, we are presented with beliefs about the world in perception and the problem of learning a language is that of translating these beliefs into sentences in the public language. Although perhaps views such as these are widely held, the choice of name for her amalgam is unfortunate as it misleadingly suggests that the Standard Theory is the view standardly held in the philosophy of language, when in fact the view that thought is independent of language is controversial.

Nolan's main criticism of the Standard Theory is its inability to account for language acquisition. The account she attributes to holders of the Standard Theory is something she calls the "code metaphor." On this account, a child learns a language by encoding propositional formulae in its language of thought into sentences of the language being learned. Understanding a language, therefore, is a matter of being able to encode, transmit, receive and decode messages. Nolan's main purpose is to argue that this metaphor is severely limited. Following a useful discussion of code conceptions from Locke and Mill through Wittgenstein's "picture theory of language," she suggests an alternative construal of Wittgenstein's views in the Tractatus. Her suggestion is that Wittgenstein claims that on a code conception of language, the text which is to be coded cannot itself be linguistic, and so although it may be grasped it cannot be spoken. Nolan expresses this insight as the claim that if human language is a code, then we can never decipher it, in the sense that we can never write the clean (uncoded) text. Nor are we able to write down the coding rules themselves, for we cannot write down the nonlinguistic things which are correlated with the linguistic elements. She then proposes understanding the Investigations as arguing against the antecedent of the above-outlined conditional thesis of the Tractatus. When the later Wittgenstein criticizes referential semantics, he is implicitly arguing that human languages are not codes. His remarks concerning the indeterminacy of ostension and the private language argument are both seen by Nolan as attempts to show that one cannot identify the things to be coded, nonmental things in the case of the former argument and mental things in the case of the latter, independently of knowing the code, i.e., independently of understanding the language.

Nolan presents Fodor's language of thought hypothesis as a successor to the coding theories of the past. She claims that the language of thought hypothesis survives the success of the private language argument because that is an argument only against the possibility of coding sense data. It would have been welcome for Nolan to have acknowledged Wittgenstein's other criterion of a private language: one in which there are no public criteria for the application of its terms, for Fodor's language of thought is a private language in this sense.

Nolan's objection to Fodor's hypothesis is that if it is right to suppose that in order to learn what the word "ball" means we must already have the concept of a ball, then surely we must also have the concept of the word "ball" in order to learn this. Nolan infers that this would be to make the absurd claim that no one learns even his or her first language. The discussion of Fodor, as perhaps the leading modern proponent of a code conception of language, is not as developed as one might wish. Fodor's claim is that in order to learn a language, L, it is necessary to know some language other than L which has the semantic resources to express the extensions of the terms of L. It seems reasonable to conclude, as Nolan does, that this prior language must also be rich enough to represent the terms of L. It seems unreasonable to suggest that this is tantamount to the claim that no one learns even a first language. Presumably, on Fodor's account of learning L, what is learned is how to match up the various predicates of L, that the learner is already capable of representing, with classes of things, which the learner is also already capable of representing. Nolan's objection commits her to the view that it is not possible to discriminate the terms of a language independently of understanding that language. This view is not so manifestly true that it does not need more than the casual defense Nolan gives of it.

Nolan's general argument that it is impossible to explain language acquisition on the assumption of a coding theory is similar to that which she attributes to Wittgenstein: that it is not possible to successfully correlate the elements of the language with items in the world without already understanding the language. Her main argument is based upon a particular stage of language learning. Two- and three-year-old children, Nolan reports, do not have mastery over the hierarchical relations between various concepts. Such a child will answer the question "What is a robin?" with the answer "a bird" but will answer the question "What is a bird?" with the answer "a robin." Nolan's explanation of this phenomenon is that a child who is learning its first language treats predicates as common names for different things, much as everyone in the Smith family is called "Smith." The child attempts to guess the pattern of correlation between the things she can discriminate on the basis of sense perception and the words. Nolan's suggestion is that when the child says "a bird is a robin" the child is not in fact predicating "robin" of anything but is instead stating an equality between the labels "robin" and "bird," i.e., stating that "robin" and "bird" are two names for the same thing. Nolan concludes that although a child's utterances have the syntactic form of subject-predicate sentences before the mastery of superordination, the child is not at that time capable of making predications. The child's protopredicates become genuine predicates when she recognizes that words are not applied to things only because of their sensory resemblance but also because other words also apply to them. It is these relationships that Nolan supposes the child cannot work out for herself and must be given to her by the speech community.

It seems that what Nolan has in mind is that no matter how much else one knows about the Fs, without the crucial information that all Fs are also called Gs, one will never learn what the extension of the term F is, and this crucial information can only be acquired from speakers of the language. There is a certain lack of clarity in her conclusion that learning a language is an essentially social process, which makes it difficult to assess whether her argument succeeds. She wishes to show that the Standard Theory is wrong to present the process of language learning as one of the formulation and testing of hypotheses. Nolan does not make clear why, when trying to determine which class it is that forms the extension of the term F, one may take into account not only the similarity in appearance of various objects but also the fact that they are all said to be G.

Nolan takes the same empirical evidence to show also that the assumption of the Standard Theory that prelinguistic beings are capable of propositional representation is false. If thought is dependent on language, then certainly a language learner will at some time be incapable of propositional thought before becoming a mature language user. However, even if Nolan is correct to suppose that presuperordinate children have not mastered the English language sufficiently to make subject-predicate judgements in it, it does not immediately follow that they are unable to make propositional judgments in some other language. Her discussion could show at most which point it is in the process of learning a language that human beings become capable of publicly expressing propositional thoughts, not that they are ever incapable of entertaining such thoughts. It may be that she takes herself to have already shown this in her discussion of particular coding theories, yet she frequently writes as though the empirical evidence provides additional support for this claim.

Much of the later part of the book, where Nolan begins to articulate her alternative to the Standard Theory on which learning a language is understood as initiation into a social practice, will be of less value to students than the earlier critical part. An exception is her discussion of Goodman in the final chapter in which she offers a way of making the distinction, based on some of the main elements of her view, between projectible and unprojectible predicates.