Linguistics demythologised

Michael Dummett

  • The Language Myth by Roy Harris
    Duckworth, 212 pp, £18.00, August 1981, ISBN 0 7156 1528 9

This book, a follow-up to the same author’s The Language Makers, published in 1980, is a wholesale onslaught on ‘orthodox modern linguistics’. It is, and is meant to be, provocative and stimulating, and will prompt fruitful debate in an area still made murky by difficult conceptual problems, but surely rightly seen as critical in our present state of knowledge of ourselves and of the world.

The ‘language myth’ of the title is defined in Chapter One to consist of two fallacies: that words stand for ideas, speech thus being a means of conveying ideas from one mind to another; and that a language consists of a fixed coding of ideas into sounds or written signs. These are somewhat inelegantly labelled by Harris the ‘telementational’ and the ‘determinacy’ fallacies respectively. His characterisation of the determinacy fallacy rapidly introduces a second strand: the combination of words into sentences. The fallacy involves attributing to the speaker of a language a knowledge of the principles governing the possible forms of its sentences: that is, the permissible combinations of words in sentences capable of expressing thoughts.

This initial characterisation may rightly be felt to do extremely rough justice to the conception Harris wishes to criticise, in two respects. First, it is tendentiously entitled a ‘myth’, and its two components ‘fallacies’, before any attempt has been made to show-that it is in any way in error. Nor is the characterisation immediately followed by such an attempt: indeed, the entire book contains no head-on refutation of the alleged myth. Rather, Harris’s plan is to draw out, in the course of five chapters, the consequences of various versions of the ‘myth’ in the writings of philosophers and, more particularly, linguists, and to raise objections to them and point out difficulties to which they lead: only in the sixth and last chapter is a sketch given of a linguistic theory not infected by the myth. If the reader is to be persuaded that the myth really is a myth, and its components genuinely fallacies, he must acknowledge that the difficulties raised by Harris are truly insurmountable, that the consequences drawn by him truly follow, and that the versions of the myth he surveys exhaust all possibilities, or, at least, that any other conceivable version would be subject to analogous objections. Harris makes no pretence of proving these contentions. The work is tacitly left – in the manner of mathematics textbooks – as an exercise for the reader, but as one he is not expressly encouraged to undertake.

In the second place, the initial characterisation is excessively vague. Just what is sufficient to convict anyone of the crime of thinking that words stand for ideas? Frege is surely the founder of modern semantic theory, and to him contemporary philosophy of language and contemporary linguistics are both deeply indebted: should we nevertheless see him as having committed this crime? He certainly denied that words have any essential connection with the contents of the mind, and held that, when an expression can be held to designate anything, what it designates will normally be something in the external world about which we intend to speak, except, of course, when our intention is to speak of our inner experience or something else not part of the external world; and perhaps this is enough to secure his acquittal. But he also held that each unitary expression of a language expresses an objective sense grasped by the speakers of the language; perhaps this may, after all, render him guilty. There is really no saying: the ‘telementational fallacy’ has been characterised so imprecisely that we have no sure way of telling when it has been committed.

Furthermore, the appearance of sentences in the characterisation is singularly abrupt. What is supposed to be the relation, according to believers in the myth, between ideas, for which words stand, and thoughts, conveyed by means of sentences? Do thoughts consist, for them, of ideas combined in certain ways? If so, what must they take the relation to be between the principles governing combinations of ideas and those governing combinations of words? Surely not any set of ideas can combine to form a thought. Surely, also, the same ideas can be differently combined to form different thoughts, as with the thoughts expressed by Miles imperatorem interfecit and Militem imperator interfecit. But surely also the principles governing the combinations of words in one language may notoriously differ from those in another, as shown, e.g., by the fact that Du musst nicht schreiben does not mean the same as ‘You must not write.’ Just what theory, then, does the determinacy fallacy embody? And, once more, what constitutes committing it?

Harris’s stigmatisation of these two conceptions as fallacies appears extremely bold, because our natural reaction is that something along these lines must be right. Words surely do have meanings: in listening to an Italian weather report, may I not be stumped by not knowing what nebbia means, and enlightened by looking it up or asking a friend? In some sense, speech serves to convey thoughts; in some sense, a variety of principles governs the construction of sentences. To attain clarity, what is called for is a discrimination between the allowable senses in which these propositions hold, and those in which Harris means to disallow them: and then the two alleged fallacies will stand out sharply, so that we may proceed to investigate whether they are in fact fallacious, and, if so, why. But just this is what he fails to provide: the fallacies are left so vaguely formulated that we have no clear idea what we must do if we are to avoid them to Harris’s satisfaction.

In Chapter Two, Harris announces his principal theme: namely, that the ‘orthodox tradition of modern linguistic theory’ enshrines a version of the ‘language myth’, and is therefore irremediably flawed. This particular version is founded upon the idea that competence in a specific language consists in possessing a body of knowledge, and that the task of the linguist is, accordingly, to render this knowledge explicit. This is indeed an idea of enormous importance, with which many linguists would concur; many philosophers of language, likewise, either share this idea or hold, more cautiously, that competence in a language could in principle be attained by explicit acquisition of a body of knowledge, and that the only, or at least the best, way of characterising how a language functions is to set out the content of the knowledge that would serve this purpose. It is therefore disappointing that Harris’s refutation of the idea is so perfunctory. He maintains that, when we speak of someone as ‘knowing French’, the use of the verb ‘to know’ has no genuine epistemic significance, but ‘is merely a stand-in, called upon to do duty because there is no specific superordinate verb which subsumes the common verbs to speak, to read, to write and to understand’.

Mere assertion will not convince. We know, from common experience and from pathological cases, that there is a gap between an active and a passive mastery of a language. Yet they can hardly be unconnected: and the connection surely does not lie only in the fact that a speaker is guided by feed-back, being (unless deaf) simultaneously a hearer of his own words. The difficulty is not one of finding a single verb to embrace four more specific ones denoting well-understood activities: if we could give a plausible account of the character of any one of the four component abilities – say the ability of a competent hearer to grasp what a speaker intends by his utterances – we should have gone far towards an explanation of how language works. To explain it in terms of what the hearer knows is an attempt at such an account: he understands what the speaker says because he has various items of knowledge, which the linguist or linguistic philosopher aims at formulating, concerning the words the speaker uses and the ways he puts them together; more accurately, his understanding simply consists in his hearing what is said, and hearing it as articulated into words in the appropriate manner, against the background of the relevant knowledge. Such an account undoubtedly generates problems about the character of such knowledge. Perhaps it will not work at all: but that needs to be shown, and it needs to be acknowledged that the problem it attempts to solve is a problem, indeed the central problem concerning language. In this passage, at least, Harris reveals no consciousness that any explanation is called for of what understanding an utterance may be; he does not argue that some rival account will explain it better, but apparently assumes that we know what understanding is as well as we know what swimming is.

Harris writes as if mastery of a language were a matter of possessing four diverse, if connected, practical abilities; and, although we do use phrases like ‘he knows how to swim,’ he is quite likely right if he thinks that such abilities do not, in general, have a genuine epistemic component. But he fails to observe that the cases are disanalogous. To learn to swim is to acquire a technique for doing something whose nature we already grasp: we know what it is to swim, but do not yet know how to do it. To learn French is not to learn a technique for doing that of which we already know what it is to do it: someone who does not know French does not know what it is to speak French; he could, for example, be fooled by people speaking nonsense words with a French intonation. One can grasp what swimming is without grasping how it is done. But there is no saying what it is to speak French without saying how it is done, just as there is no saying what it is to play draughts without saying how it is done: that is why one cannot even try to speak French without having learned it, as a man who had never learned to swim might well try to do so. This fact may not justify us in supposing that to know a language is to be in a genuinely epistemic state: but one can hardly reject this contention without acknowledging, and attempting to account for, the fact which tempts us to advance it.

What is it, then, that Harris would desire of a ‘demythologised’ or ‘integrational’ linguistics? Such a linguistics would be, he says, ‘an investigation of the renewal of language as a continuously creative process’. By this he means that it should not content itself with attempting to describe a language only as spoken at a particular time, as Saussure proposed, but should describe it as incorporating the possibility of ‘what is traditionally classified as belonging to “language change” ’. This last phrase is characteristic of the somewhat impressionistic style in which Harris often writes. What appears as a linguistic change from one standpoint will not count, from another standpoint, as involving any change in the language, but, say, as a widespread change of belief or of social convention: but there is no warrant to imply that there is no such thing as linguistic change by putting the phrase ‘language change’ in scare quotes, since, if there is not, then most of Europe and the Americas and much of India are still speaking Proto-Indo-European. Harris still intends a description of a language as it is at a certain time, but he wants the description to provide for the possibility of linguistic innovation without predicting what it will be.

Harris gives, in his last chapter, several entertaining examples of such innovation, beginning with the description of Mr Heath by the Times as ‘a doorstep loser’, which, as he remarks, involved no allusion to any propensity on Mr Heath’s part to mislay doorsteps. He refers disparagingly to explanations of such innovatory uses by ‘orthodox linguistics’ which invoke a distinction between linguistic knowledge proper and pragmatic or other knowledge. We have, however, to ask whether a theory of the kind he desires would distinguish between innovatory and non-innovatory uses of language. If it did, it is hard to see how it would differ very sharply from the ‘orthodox’ approach he so unfavourably contrasts with it. If it did not, then it would surely be unable to recognise anything effected by the innovatory uses as involving change; and so, although it would have set the boundary between change and stasis at a different place, it would not have presented language as capable of evolving at all.

Harris wants a reformed, ‘integrational’ linguistics to take ‘as its point of departure the individual linguistic act in its communicational setting’: ‘language,’ he says, ‘cannot be studied in isolation from the investigation of “rationality” ’; when both are studied together, ‘linguistic behaviour is... placed on a par with all other forms of voluntary human action.’ It is unquestionably true that, in linguistic interchange, we are constantly guided by the same kind of evaluation of other people’s motives and intentions as when we are concerned with non-linguistic behaviour; we rely on this to grasp the point of what other people say, why they take it as relevant, or perhaps are deliberately changing the subject, whether they are saying something by way of concession, or illustration, or corroboration, etc – in short, what they are driving at. For this reason, Martians who failed to recognise us as rational agents would not attain true comprehension of human speech. It in no way follows that we cannot separate out what is specific to the mastery of a particular language from what is essential to the understanding of all voluntary human action, including the use of language. Someone who does not know English very well may not understand the word ‘doorstep’or the word ‘loser’; or he may as yet be unaware of the rather loose manner in which, in English, it is permissible to use nouns as quasi-adjectives to qualify other nouns. But, if he knows all three, his failure to understand ‘doorstep loser’ will not testify to a defect in his knowledge of English; nor will his English teacher at any stage need to instruct him in principles that would enable him to understand the phrase, as opposed to background information about the practice of canvassing.

Harris tells us much of what an integrational linguistics will do: he says nothing about how it will go about doing it. In science, the proof of the criticism is in the positive theory. Berkeley subjected the calculus, as it was in the mathematics of his day, to what we now recognise as trenchant and fully justified criticism: but, as he made no positive proposals for setting it on a sure foundation, his criticism cut no ice with mathematicians. If Harris or a disciple can produce the beginnings of a linguistic theory which would accomplish what he thinks it should, his criticisms of the contemporary ‘orthodox’ variety will then appear profound and far-sighted, and my complaints that they are defective in demonstrative cogency niggling and myopic. If not, however, his observations are unlikely to prove much hindrance to the survival of the ‘orthodox’ tradition.