Published online by Cambridge University Press: 01 December 2003
The goal of this study is to reintegrate the theory of generative grammar into the cognitive sciences. Generative grammar was right to focus on the child's acquisition of language as its central problem, leading to the hypothesis of an innate Universal Grammar. However, generative grammar was mistaken in assuming that the syntactic component is the sole course of combinatoriality, and that everything else is “interpretive.” The proper approach is a parallel architecture, in which phonology, syntax, and semantics are autonomous generative systems linked by interface components. The parallel architecture leads to an integration within linguistics, and to a far better integration with the rest of cognitive neuroscience. It fits naturally into the larger architecture of the mind/brain and permits a properly mentalistic theory of semantics. It results in a view of linguistic performance in which the rules of grammar are directly involved in processing. Finally, it leads to a natural account of the incremental evolution of the language capacity.
1. Essentially the same” is a matter of perspective. When we are talking about “English speakers” as a whole we can treat them all as essentially the same. But if we’re talking about dialect differences, dialect contact, or language change, we can just as easily switch to treating different speakers as having (slightly) different linguistic systems in their heads. And of course when we’re talking about language acquisition, we take it for granted that the young child has a different system than the adults.
2. To some extent Chomsky's point has been lost on the larger cognitive neuroscience community. For instance, the widely cited connectionist parser of Elman (1990) is a variation of a finite-state Markov device, and is subject to some of the same objections raised by Chomsky in 1957. See Marcus (2001) and Pinker (1999) for extensive discussion.
3. For example: “The deep structure that expresses the meaning is common to all languages, so it is claimed [by the Port-Royal grammarians – who of course did not use the term “deep structure”], being a simple reflection of the forms of thought” (Chomsky 1966).
4. Some opponents of Chomskyan generative grammar (for instance some Cognitive Grammarians) have rightly objected to syntactocentrism, but proposed instead that all properties of language are derivable from meaning. I take this to be equally misguided, for reasons that should be evident as we proceed.
5. A standard mark of recursivity is a constituent occurring within another constituent of the same type. For instance, a clause can appear within another clause: The man who comes from New York is tall; and a noun phrase can appear within a noun phrase: the king of the Cannibal Islands. In phonology this sort of situation does not occur nearly so freely: In particular, a syllable cannot occur within another syllable.
6. Interestingly, Chomsky (1965) brings up an example like (4) and analyzes the prosody as a fact of performance: Speakers don't pronounce the sentence in accordance with its syntactic structure. This is about the only way he can analyze it, given that he does not have independent principles of intonational constituency at his disposal. Contemporary theory allows us to say (correctly, I believe) that (4) is well-formed both syntactically and prosodically, with a well-formed but non-isomorphic correspondence between the two structures.
7. It is important to distinguish two interpretations of “syntactic” here. In the broader sense, every combinatorial system has a syntax: mathematics, computer languages, music, and even phonology and semantics. In the narrower sense of technical linguistics, “syntactic” denotes the organization of units such as NPs, VPs, and prepositions. I am reserving “syntactic” for this narrower sense and using “combinatorial” for the broader sense.
8. For the semantics I have used the Conceptual Structure notation of Jackendoff (1983; 1990); readers invested in other frameworks should feel free to substitute their own notations.
9. Stratificational Grammar (Lamb 1966) also proposed a thoroughgoing organization into independent generative components linked by interfaces.
10. The lexicon, a large collection of learned arbitrary associations between very particular bits of structure, also has parallels in other domains of memory. For instance, it is an arbitrary fact that the sound kæt means “feline animal,” and the child must learn it from the environment. Now consider the association between the appearance of foods and their tastes. It is similarly arbitrary (from the point of view of the organism) that something that looks like a cucumber tastes the way it does, and organisms learn probably hundreds or thousands of such associations. (There are even ambiguous looking foods: Think of mashed potatoes and vanilla ice cream.)