Unscrambling chicken and egg

This is an admirable, sometimes infuriating, always enjoyable book that will interest all students of the theory of language, but the questions it raises are of far wider philosophical interest. Roy Harris's topic is not simply linguistic communication, still less "communication theory", but human communication in general. In a non-technical, learned exposition, he attacks a large number of the more comfortable orthodoxies of old and new theories of language, both philosophical and linguistic. The protagonist of his debate is what he calls "integrationism", the antagonist "segregationism".

A way of regarding segregationism is how to exorcise the traditional fear that on a natural picture of communication, communication may not be possible at all. Perhaps we can only try to express it if we do not really take it seriously, and most of us have taken it almost for granted that there is something wrong headed about it. But that is too easy. For suppose communication occurs when A conveys her thought to B via some medium of encoding in such a way that B can recognise it for what it is - the thought she had about so and so. ('A' could as well stand for artist, 'B' for beholder, or for writer and reader, etc.) Then might not A persistently construe the signs in the medium quite differently from how B does? People can be doggedly at cross-purposes on particular occasions - so why not always?

One segregationist device might be to secure the overall viable sign system first. "The basic issue," Harris says, "is one of ontological primacy. For the segregationist, communication presupposes signs, and sign systems exist apart from and prior to the communicational purposes to which they may or may not be put. For the integrationist, on the contrary, signs presuppose communication: they are its products not its prerequisites."

The segregationist strategy is most plausible for linguistic signs and "virtually all theories of modern linguistics, at least from Ferdinand de Saussure onwards, are segregational theories". They tend thereby to segregate in another way by privileging all forms of linguistic communication above any others.

From this standpoint, despite all the obvious differences, the Saussurian doctrine that langue precedes parole , Chomskian thoughts that a "deep" syntax permits semantics, Jerry Fodor's view that all communication presupposes the common code of a "language of thought", and the central doctrines of Jacques Derrida, are all in their different ways "segregationist".

Then what is put in place by a prior system of signs is the subsequent possibility of ascriptions of true or false beliefs to sign-users. Put thus, it is an astonishing claim. It suggests three questions not placed equally in the foreground of Harris's account. It would certainly imply, among other issues, a radically Cartesian division between code users and other, mostly non-human, animals. Such a belief is indeed all about us, and surprisingly Harris takes up the issue only as it concerns human non-sign-users. Second, how could any intelligent being have a system of signs unless it first possessed what has come to be called a "theory of mind" so that false beliefs may be ascribed to others? Third, how on a segregationist theory could sign systems come about in the first place? These questions have often been set aside together as if they were a distraction from good theory.

Harris's integrationist response is that the condition for communication is communication. This neither involves a paradox nor a vicious regress since we may progressively push back the presupposed conditions for any particular fully contextualised occasion of communication to what are in effect a thinner and thinner set of biomechanical, macrosocial and circumstantial conditions: together these constitute the necessary communicational infrastructure. He intends these categories to be exhaustive. The last of these implies that integrationism places at the core of theory what segregationism places at its periphery. As Harris put it in his earlier book on Saussure and Wittgenstein, "communication is the reaching of agreement I in particular interactional episodes". What matters is how a speaker may in particular integrate the various types of conditions. This suggests the need for a clear account of how highly sophisticated forms of communication can be related to less sophisticated forms, together with how different varieties of these conditions may interact. Wisely, perhaps, Harris here merely points a way. His point is that we will not find it in the segregated code.

He makes two further epistemological claims about how signs emerge from communication. Their effect is to re-shape the initial picture quite radically. The first is that it is "self-communication" that "shapes and informs our understanding of interpersonal communication, and not vice versa". The second is that "temporal integration needs to be seen as the sine qua non for all sign making". The former may seem contentious, but it matters what he does not mean by it. "Self communication" should neither be seen as a degenerate case of talking to others, nor be regarded as a "private language" of the kind Wittgenstein taught us to eschew.

Harris's theory will not convince everyone. I am for instance inclined to agree with him concerning the orthodox distinction between "use" and "mention" by quotation (which manifestly breaks down in a thoroughly integrationist way in the case of rude words), but I am less convinced that the related doctrine that distinguishes words that can be counted as types rather than tokens need carry the segregationist load he suggests.

There is much more to do in the spirit of his discussion on specifically non-linguistic forms of communication, but Harris's book should reinvigorate such issues, which have been too much marginalised by the orthodoxies he criticises.

Andrew Harrison is reader in philosophy, University of Bristol.