Looking in for the meaning without

What are the foundations of language? The answer varies greatly according to what you think "language" is. Ray Jackendoff, professor of linguistics at Brandeis University, distinguishes linguistic foundations of five different kinds: psychological, biological, "architectural", semantic and conceptual. Something very odd strikes one straight away about this basic list. It does not include any social foundation. If you take it that the existence of language depends rather crucially on the existence of human society, such an exclusion will seem perverse. Jackendoff's perfunctory apologia for the exclusion marks him out, in terms of linguistic theory, as a segregationist, one who believes in a strict separation between the academic study of languages as internally coherent codes or systems and the study of how people actually use them for practical purposes. For segregationists, languages are constituted ab initio as autonomous objects of theoretical inquiry, and may be analysed independently of the ongoing human activities in which they are manifested. Jackendoff does not deny that society is the matrix of language, but claims that you cannot use a language socially unless you have the cognitive capacity to do so. From this it is held to follow that linguistics need not bother much with speech acts or society. The non sequitur is gross. But it conveniently lets segregationists off the hook of finding it necessary to ask what communication is, what communication needs society or individuals have and how these might shape the development of the kind of communication system that most of us (in Anglo-American culture) have come to call "a language" (such as English, French and so on).

Having dismissed society and social intercourse, segregationists such as Jackendoff are free to fix their linguistic gaze exclusively on what (allegedly) goes on inside the head, where "cognition" takes place and the relevant systems of knowledge are supposedly stored in some form of cerebral hardware. Not the head of any real individual, of course; but the hypothetical head of some ideal speaker of the language. For linguists who regard this idealisation as not only inadequate but pernicious into the bargain, that is like expecting to find the foundations of politics concealed somewhere inside the anonymous head of the ideal politician, rather than emerging from the study of political events.

Foundations of Language , Jackendoff reminds us, was the title of a journal that flourished in the 1960s and 1970s, and in which he recalls publishing his first paper. He has forgotten the wise observation in the first editorial of that journal (1965) that pointed out that there are "as many linguistic theories of language as there are trends in contemporary linguistics". Reading Jackendoff, one would think there was only one such trend that ever deserved to be taken seriously; namely the particular version of segregationism developed at the Massachusetts Institute of Technology. This is what explains Jackendoff's narrow view of the foundations of language, and also his foundational equation of language with speech.

Non-vocal forms of language, such as writing and the sign systems of the deaf, are barely visible in Jackendoff's phonocentric perspective. They must be merely parasitic on vocal forms, or else byproducts of the same "universal grammar". But the biological apparatus necessary for speech will not get you far if the task before you is reading a book. For that you need eyes, not ears. Furthermore, the organisation of Jackendoff's text on the printed page obeys principles of spatial arrangement that have no counterpart in phonology. Jackendoff seems blind to the foundational implications of these facts, and blissfully unaware that he is presenting direct counter-evidence to his thesis. He has written a book without realising what he is doing.

Jackendoff's opening chapter on "the complexity of linguistic structure" plunges us deep into the familiar algebraic formulas and tree diagrams of Chomskyite grammar. Then we are promised an "evaluation" and "rearticulation" of the academic "agenda" set out in the first chapter of Noam Chomsky's Aspects of the Theory of Syntax (1965). But what Jackendoff's reformist zeal offers is too little and too late. Chomsky's chapter said nothing about the biology of language. It discussed the criteria for an idealised grammar and raised the associated issue of the "psychological reality" of grammars, first introduced into modern linguistics in 1916 by Saussure's Cours de Linguistique Génerale .

Jackendoff evidently believes the first question to be addressed is whether mental processes can be "reduced" to brain processes. His question-begging solution is to invent a domain of his own, baptised "the f-mind". It is in our f-mind, he assures us, that "linguistic structure" resides. Saussure, sensibly in the view of many, regarded the mind/brain issue as a philosophical red herring and did not pursue it. For it has no more to do with speech than with tennis or cookery or any other intelligent human activity. All intelligent activities have some biomechanical basis in neurophysiology. But it does not follow that focusing on the biomechanical basis leads to a better understanding of the activity. On the contrary, it may obscure understanding. The reason is simple. All activities cannot be reduced to "f-activities". They may have a social foundation that is not "instantiated in the nervous system" and cannot be described in biomechanical terms.

Part two is devoted to "architectural" foundations. By "architecture", Jackendoff means, roughly, syntagmatic organisation; and his beef against Chomsky here is that Chomsky's conception is "syntactocentric". Instead, Jackendoff advocates thinking of language as having "multiple parallel sources of combinatoriality". Shorn of all the hieroglyphs and jargon, what this comes down to is a major concession to the view long held by non-generativists that linguistic analogy is open-ended, which means there is no predicting what will eventually turn up as acceptable and comprehensible. (Should anyone doubt this, I suggest looking at the unanticipated explosion of Netspeak for countless examples.) Part three examines "semantic and conceptual foundations". The general public, according to Jackendoff, regards "meaning" as "the central issue in the study of language" and here, he says, the Chomskyite programme has let everyone down. For it concentrated on fiddling little grammatical minutiae and remained indifferent to the fact that "what people really want to know about is still meaning". Why not go the whole hog and admit that what people really want is the truth? Jackendoff discusses truth briefly and seems to hanker after what used to be called a "correspondence theory" but cannot quite see how to get one without, as he puts it, "psychologising the world". Thus for Jackendoff, language and everything to do with it end up "trapped inside the brain" and the discussion of the trap is so riddled with computer metaphors that it is difficult to tell where neurophysiological explanation begins or ends.

Does Jackendoff eventually succeed in breaking free from the Chomskyite shackles? To report that he does would be overstating the case. For Jackendoff ultimately cannot bring himself to relinquish what he calls one "surely uncontroversial postulate". (Perhaps this is a misprint for "incontrovertible"?) It is the postulate that: "People find sentences (and other entities) meaningful because of something going on in their brains."

Alas, no. Jackendoff's postulate is miles wide of the mark. We social creatures find things meaningful because of what we notice going on in that outside world not "trapped inside the brain" and of our own pragmatic resources in dealing with it. It is the day-by-day requirement to integrate the two that produces, for each and every one of us, the recognition of what is meaningful and what is not.

The publisher's hype tells us that this work is "the most fundamental reformulation of linguistic theory in a generation". If that is so, one wonders where this generation has lived. Perhaps cloistered behind the intellectual walls of certain American universities. Some books are out of date even before they are published. It is not that Jackendoff is belatedly trying to shut the stable door on language long after one particular theoretical horse has bolted. The generativist horse never bolted: it was found dead in its stall. Jackendoff's book is an attempt to resuscitate a corpse.

Roy Harris is editor, Language and Communication , and is emeritus professor of general linguistics, University of Oxford.