Chomsky wipes slate clean

Stephen Pulman asks if necessity is the mother of complex linguistics

This is a collection of three essays based on lectures given by Noam Chomsky at the University of Siena in 1999, prefaced by a lengthy introduction by the editors, the linguists Adriana Belletti and Luigi Rizzi, and supplemented by an edited transcript of an interview with Chomsky. While most of the book concentrates on linguistics, the final - chilling - essay concerns the role of western intellectuals, propaganda and US foreign policy.

The editors' introduction gives a historical survey of Chomskyan linguistics. Two basic puzzles about language motivate his research programme. First, the "creative use of language": native speakers can produce and understand an infinite number of new utterances that are appropriate to, though not dependent on, their circumstances. Second, the "logical problem of language acquisition": the problem of how children learn the rich and complex system needed to support this creativity so quickly and uniformly, despite the variety of circumstances in which they do so.

His answer to the first question is that linguistic structures are built by recursive rules, that is, rules that are finitely stateable but that can apply and reapply to yield a potentially infinite number of distinct structures. Recursive structure is found in other human symbolic systems (mathematics, music and so on) but apparently not in non-human activities and thus seems to be unique to our species. His answer to the second question is that humans are genetically predisposed to learn language: the language faculty is a "mental organ" that grows automatically given only a modest amount of exposure to the relevant input.

However, the nuances of this second answer have changed in Chomsky's recent work. Until recently, the assumption was that properties of language that were uniform across languages and apparently otherwise unmotivated were a direct reflection of the genetically determined structure of the "language organ". An "explanatorily adequate" linguistic theory would explain these properties of the language system by showing how they arose from the structure of the language organ, triggered by exposure to relevant linguistic data. The hypothesis that Chomsky is currently exploring in the "minimalist program" is that such structure instead may arise from mental constraints external to the language faculty proper. If the task of language is to relate thought to speech, to connect the output of the "conceptual-intentional" system to the articulatory organs, then some of its structure may be determined by the properties of these systems and the need to connect them as efficiently and economically as possible. In fact, the language faculty may represent a "perfect" solution to this problem.

This may sound like a functional explanation, but it is of a radically different kind than the many (unsuccessful) attempts to explain the nature of language in terms of external properties or communicative functions.

What Chomsky is exploring is the hypothesis that the language faculty is an optimal solution to a problem imposed by the structure of the two (or more) systems that it connects. The intricate and complex nature of linguistic structure may arise, not directly from some genetically specified template, but from a kind of intrinsic necessity, in the way that the beautifully varied and regular structure of a snowflake arises from a small set of relatively simple physical properties.

This is a powerful and compelling idea, but it is a radical shift in perspective, and most of the long interview is devoted to explaining and defending it. The typical pattern of explanation in generative linguistics used to rely on the apparent arbitrariness of properties of language, in the sense that it was usually easy to imagine other ways, logically speaking, in which the same effects could be achieved. This arbitrariness could be explained if it were a direct reflection of the genetic substrate of the language faculty. Now, the typical question for a minimalist is instead: here is some apparently uniform property of language that does not seem to be motivated by the need to relate deep semantic structure with speech, and may even seem dysfunctional in this respect. Can we nevertheless see it at some more abstract level as part of a non-arbitrary, even optimal solution to this mapping problem? Chomsky is in effect tearing up many of the "results" of the past 20 or so years and beginning again, reconstructing some of them within this new framework.

A recurrent thread in Chomsky's work concerns the nature of explanation in the physical and cognitive sciences. In two of the chapters here he pursues interesting parallels between the study of language and the history of the physical sciences, particularly chemistry. For Galileo explanation was mechanistic: we can understand physical phenomena to the extent that we could envisage constructing a mechanical device to simulate them. Clearly, mental phenomena such as language do not fall within such a mode of explanation. With Newton's invocation of "action at a distance", a force he was not able to give a Galilean account of, the mechanistic mode of explanation became untenable. It had to be replaced by theories that would admit a limited number of such unexplained properties but try to explain as much as possible with them, in the belief that at some point the properties themselves would be properly understood. From this point of view, mental properties become no less mysterious than some physical ones: they are associated with physical matter but not reducible to mechanistic explanation.

For the early chemists, this stance was expressed by Joseph Black, advocating that "chemical affinity be received as a first principle, which we cannot explain any more than Newton could explain gravitation... let us defer accounting for the laws of affinity, till we have established such a body of doctrine as (Newton) has established concerning the laws of gravitation". The "laws of affinity" were finally understood only in the 20th century with the development of physics to the point where the chemical behaviour of elements could be described in terms of atomic structure.

For Chomsky, this is an appropriate attitude for linguists and other cognitive scientists to hold today. Mental properties such as the structure of the language faculty are to be studied by the methods of the natural sciences - they are part of nature - but linguists cannot be required to achieve standards of explanation that would be unrealistic for the physical sciences. The first 50 years of experimental chemistry resulted in a "body of doctrine" that is very different from what would be regarded as current orthodoxy. It is highly unlikely that the results of the first 50 years (almost) of generative linguistics will appear any less quaint to future generations of cognitive scientists, and it is certain that the study of the language faculty is no less difficult than the study of chemistry.

One may criticise many aspects of the actual practice of some Chomskyan linguists: baroque, theory-internal analyses; cavalier treatment of data; wilful refusal to spell things out in enough formal detail for the consequences of claims to be clear. Nevertheless, to read Chomsky himself is always to be gripped by his breadth of vision, his intellectual depth, and his refusal to take any assumption, however hallowed, for granted.

Chomsky makes linguistics intellectually interesting, he makes it exciting, and he makes it a discipline that matters to the wider scientific community.

Stephen Pulman is professor of general linguistics, University of Oxford.