Necessary coupling

The "connection" referred to in Roy Harris's The Language Connection is between philosophy and linguistics - or, as the publisher's blurb appealingly puts it, between "philosophers and linguistis in the West". Harris, in truculent mood, but with considerable elegance and wit, has put together a powerful critique of the way these disciplines have established a tradition of using language for talking about language (ie "metalinguistically", "reflexively"). No hostages are taken. Linguistic metalanguage is seen as "meretricious", "self-serving", "tunnel vision", "implausible metaphysical speculations" with "irreparable flaws" which "throw no light at all on our lay linguistic experience" and which force speech into a "metalinguistic sausage-machine". The Language Connection reminds me a bit of the drama critic who, asked to talk about a particularly emotional and challenging play, could only say: "at the end, everyone dies".

At the end of this book, everyone is certainly dead (apart from Harris, who writes of both philosophers and linguists in the third person throughout). The whole basis of metalinguistic discourse in both disciplines has been found wanting. Ever since it was first formulated in ancient Greece, Harris argues, language study has used a conceptual framework which is confused, distorting and illusory. The criticism applies equally to each discipline, because essentially the same metalinguistic framework is used by both - notwithstanding the various differences of focus which have arisen over the centuries, such as the linguist's sentence and the philosopher's proposition.

Neither discipline has any desire to eliminate the distinctions which separate them, or to clarify the foundations of their subjects, Harris asserts, because they each have a vested interest in maintaining their separate academic identities - for each, reflexivity is a "carefully doctored version designed for their own disciplinary purposes". The two disciplines are maintaining the status quo for purely self-perpetuating reasons. "They claim a specific field of expertise and develop a methodology of their own. They institute a technical terminology which serves simultaneously to exclude outsiders unfamiliar with the field and to ensure that only those questions which can be couched in its terms are recognised as valid. They set up hierarchies of employment and monitor their own 'professional' qualifications." Students "are expected to write essays which deploy this academic jargon". They learn to play the relevant game.

What is so wrong with linguistic metalanguage that it deserves such a pasting? Essentially, everything. Harris argues that linguists and philosophers have created a totally spurious world of linguistic objects (the various units of linguistic description). They have named and classified these units, believing that their inventions are real and have universal validity. He examines seven "doctrines" (so called because they have been formally addressed in the teaching tradition) which make up what he calls a "segregational" view of language - a view of language as linear and segmentable into identifiable units, such as word and sentence (that "controversial one-dimensional monstrosity"). If these units cannot be satisfactorily identified, he argues, then everything collapses.

The view relies on the distinction between use ("John is monosyllabic" meaning "John speaks in monosyllables") and mention ("John is monosyllabic" meaning "The name John has one syllable"), and between types and tokens ("there are four tokens of the type John in this sentence"). It involves the notion of parts of speech, sentences/propositions, telementation (the transferring of thoughts between minds through the use of words), fixed codes (different users attach the same meaning to the same linguistic form), and plain representation (the binary correlation between words and things). These notions, Harris argues, cannot provide a coherent basis for language study. For instance, there is the long-standing problem of the definition of parts of speech; if they are viewed as divisions along a formal linear continuum, they will not necessarily correspond to what is meaningful; conversely, a semantic definition raises questions of how "meaning" can be divided into parts.

The "moral of my story", Harris says, is that "philosophers and linguists need to rethink the whole problem of discussing language right from scratch". We cannot tinker with it, because the whole framework is flawed. Either we accept it (the "vested interests" option) or reject it. Harris is in no doubt: we must reject it. We need a comprehensive rationale, a more meaningful approach, an "adequate theory of language". But say we agree, what form could it conceivably take? About this, Harris is largely silent. This will not come as any surprise to readers who begin at the beginning. They are warned about it by Ray Monk in his preface.

At various points, Harris hints that there may be no alternative: sorting out the muddle "does not require anything as ambitious as the construction of a new error-free metalanguage (a project about which I am sceptical)". Or again: "even the severest critics of the traditional parts-of-speech doctrine found themselves at a loss to provide any more convenient basis for comparing one language with another". Or again, after an enjoyable critique of A. J. Ayer: "it seems that linguistic enquiry in western culture is locked into a metalinguistic framework which is highly resistant to basic change, even though its presuppositions are extremely dubious and easily shown to be so." Harris is well aware that he has to use the same metalanguage himself, though he tries to defend himself against the obvious criticism by saying that his "intention was never to call a halt to the metalinguistic games we play . . . but to prevent . . . metalinguistic illusions". If we understand these illusions, it "does not take us as far along the road as we might like towards an adequate theory of language. But at least it is a first step in the right direction".

So, does Harris give us any clues about the direction we should be taking? There are some hints. He accepts that reflexivity is central to linguistic communication, as long as it does not become "parasitic upon a more basic non-reflexive function of language, something extra to the primary purposes words serve". "In both disciplines, we see the adoption (of doctrines) . . . because that is what answers to the demands of the theorist rather than because it corresponds to the observable practice(s) of the linguistic community" (his italics). So what are these "primary purposes" and "observable practices"? Context seems to be the key. There must be no uncontextualised theorising. Harris seems to be arguing towards a social or psychological processing view of some kind, in which the individual takes centre stage. Language must not be reduced to "anonymous, interchangeable speakers and hearers". The notion of speech "as a machinery producing a series of autonomous verbal objects" must be replaced by one in which it is seen "as a creative interactive function of individuals".

So that is the way Harris is inviting us to go - towards a world of contexts, processes, functions. But these are notorious areas for vagueness and obscurity, as we have learned from the proliferation of functionalist models of language in recent decades, and some formal criteria are needed if they are to be kept under control. Harris's replacement notions are surely going to be just as problematic as those he wishes us to abandon.

Which is the way with models. Indeed, the one thing I miss in this book is a serious discussion of the notion of models. The term model eventually turns up in passing on page 145. Yet models, even very faulty ones, can illuminate as well as obscure. Harris has focused only on the obscuring features of the traditional metalinguistic model, and seems to deny that it could ever be illuminating. Yet it would not be difficult to scan the ranks of assembled linguists (I cannot speak for philosophers) and demonstrate ways in which the traditional metalanguage has helped clarify the nature of language, or solve various applied problems. Coincidentally, on the day I was writing this review, the latest issue of Language arrived on my desk, with some 50 or so reports on all kinds of linguistic topics. Is the traditional metalanguage so distorting that none of these works can illuminate our understanding of language? I think not. Harris's book reminds me of the optimist and pessimist who look at a glass containing some water: one (Harris) sees it as half-empty, the other (your reviewer) as half-full. More, Harris wants us to throw the glass away completely, and drink from something else - though exactly what is not clear.

The Language Connection succeeds well in its aim to make us reflect seriously about matters which have too long been taken for granted. It is the most convincing attack on intellectual laziness and conservatism in the linguistic domain that I have read in a long time, and it left me feeling healthily disturbed, as after a good workout in the gym. It was therefore quite a relief to turn to a nice straightforward issue: the origin of language. The publisher comments about the Key Issues series, to which The Origin of Language (edited by Harris) belongs, that it "makes available the contemporary reactions that met important books and debates on their first appearance". That sounds like an excellent idea, and this well-chosen selection of items on the origins of language, originally published between 1851 and 1892, certainly does that. We have substantial pieces from R. C. Trench, Max Muller (twice), F. W. Farrar, E. B. Tylor, Charles Darwin, George Darwin, W. D. Whitney, R. L. Garner, and two anonymous authors. It is fascinating to see the way the set of issues surrounding this topic was presented, and to see the vigour with which cases were argued. The topic excites widespread interest more than a century on.

I am not so sure whether more than 300 pages of debate will attract a continuous read. Skimming or selective reading may well be the order of the day. For student readers, Harris does not provide much help. There is no index, so it is impossible to trace individual themes or examples. And the context of individual items is not explained: an editorial paragraph saying who the author was, why he was writing, and what the gist of his piece was, would have been invaluable. Odd for Harris to leave us so decontextualised. As it is, of the present selection, some items have no sources mentioned (Trench, Muller, Farrar, Charles Darwin); and we are not told who several authors are (Trench, Farrar, George Darwin, Garner). Yet, in those days of creationism versus evolution, it is important, surely, to know, for example, that Trench was archbishop of Dublin, Farrar dean of Canterbury. The very first line of the opening piece cries out for such editorial contextualisation: it begins, "But the truer answer . . .". Truer than what, we shall never know.

Students of language today, who get little more than a short paragraph on each of the old theories (pooh-pooh, ding-dong, etc) will find the discussion in these pages a revelation. I would have liked a fuller introduction. The one provided deals only with the evolutionary climate of the time, and stops short (apart from a brief mention) of bringing these arguments into the present day, where new perspectives have emerged within linguistic anthropology (fossil studies of early hominids), psychology (ontogenetic studies of the human infant), and zoosemiotics (the analysis of animal communication). Maybe this is the fault of the series - "each text has a new editorial introduction to supply the necessary historical background" - but it is nonetheless a fault. There is more to the history of ideas, as Harris par excellence knows, than digging up and contextualising the intellectual past. There needs to be a renewal of connection with the intellectual present, and that this volume does not do. Such books on the past need to get back to the future.

David Crystal is honorary professor of linguistics, University of Wales, Bangor.