What was meant by what was said
- Language, Sense and Nonsense by G.P. Baker and P.M.S. Hacker
Blackwell, 397 pp, £22.50, April 1984, ISBN 0 631 13519 7
The picture on the dust-jacket of Language, Sense and Nonsense is a 17th-century allegory by Laurent de la Hire. It shows Grammar as a lady seriously engaged in watering some rather spindly potted plants. In her left hand she holds what looks like a very long tape-measure, bearing the words vox litterata et articulata debito modo pronuntiata. Presumably this tape-measure is for checking inch by inch the growth of her diminutive and sickly-looking horticultural specimens. For Grammar in the 17th century, that is fair enough. But in the 20th century the allegory would need to be painted rather differently. Grammar would not need a watering-can at all. The spindly plants would give way to a luxuriance of hothouse foliage, and the tape-measure would have to be calibrated to measure the proliferation of nonsense. That is approximately the canvas sketched in by artists Baker and Hacker.
The ‘nonsense’ of their title is the nonsense propagated in profusion nowadays by grammatical theorists and philosophers of language. Most of us will doubtless have heard of grammar. But what exactly, we may well ask, is ‘philosophy of language’? According to one view, it is ‘the foundation of all other philosophy’ (Dummett). But does it stand in the same relation to linguistics as philosophy of science to science? Or in the rather different relation which aesthetics bears to art? Does the philosopher qua philosopher have any message to pass on to those non-philosophers who are categorised professionally as ‘linguists’? If so, what is it? Although these questions are not directly addressed by Baker and Hacker, their book comes closer to providing intelligible answers than anything hitherto published. For this reason, if for no other, it is a publication which will command the attention of all academics whose interests are either philosophical or linguistic, or both. But what it says will also concern a much wider audience who have watched, either with fascination or with trepidation, the rising star of Modern Linguistics in the firmament of Arts and Sciences.
At the heart of philosophy of language is the problem of ‘meaning’. Or that is how the matter is commonly put by philosophers themselves. But unlike other investigators of meaning, the philosopher of language does not propose to base his semantic theorising upon a systematic study of how words are actually used in speech communities, or of their recorded semantic histories, or of what their users say they mean. He proposes, rather, to reflect upon the notion of ‘meaning’ and reason about the general conditions which would have to be met by any theory of meaning applicable to languages such as English. By a ‘theory of meaning’ he means, usually, an account of how the meanings of sentences are derived from the meanings of their constituent words. To this end, he searches for combinatorial rules which will determine how word-meanings are to be ‘put together’ to yield sentence-meanings. The complete theory of meaning for a given language will thus reveal ‘how the language works’ insofar as it shows how the language is derivable from the meanings of individual words and of the combinatorial procedures involved. Whether what the philosopher of language reflects on and calls ‘meaning’ is what the layman would count as meaning is another question, however – and quite an important one.
The authors shrewdly identify four main theses which conceal the nonsense endemic in much contemporary theorising about language, and give it a superficial plausibility. These are: 1. the doctrine of the separation of the ‘sense’ of a sentence from its ‘force’; 2. the explication of linguistic meaning in terms of truth-conditions; 3. the dependence of thought and speech on ‘tacitly known’ sets of rules; 4. the mystery of our apparent capacity to understand sentences we have never encountered before. Systematically they set about demonstrating how questionable each of the above theses is when seriously scrutinised. They expose the credibility gap between the facts which lend it support and the theoretical task it is pressed into performing. Their aim in all this is not to effect dramatic Pauline conversions upon committed defenders of the current orthodoxies, but rather to influence ‘prospective recruits who have not yet taken the Queen’s shilling’.
Baker and Hacker begin, quite rightly, by refusing to treat the core problems commonly said to be central to philosophy of language either as compelling conundrums which have suddenly dropped unbidden out of this morning’s clear blue sky, or as quasi-permanent linguistic questions which have surfaced at different times and in different guises throughout the Western tradition. Their opening chapter makes the point that even in the history of ideas we cannot hope to escape the consequences of the humdrum Shakespearian truth that ‘what great ones do, the less will prattle of.’ The great ones in the present instance were the natural sciences; and what came to be called ‘philosophy of language’ has to be seen against the historical backdrop provided by the most important intellectual revolution in European history. That was the upheaval when, from the Renaissance onwards, philosophy yielded to the natural sciences as the model of what rational inquiry should be. Baker and Hacker preface their historical critque by pointing to what ensued from the acceptance of the corpuscular theory of perception. Henceforth nothing at all could simply be assumed to be what it appeared to be: on the contrary, the odds were that ‘scientific’ investigation would discover it to be something seemingly quite different. And this profound doubt about whether things ‘really’ are what they appear to be spills over from the world of physical objects to flood the world of language. It would be rash, and almost certainly wrong – so the modern theorist urges – to take words and sentences ‘at their face value’. To understand them, we must subject them to a penetrating analysis which will lay bare the ‘deep’ mechanisms of language. Only then shall we be able to gain a ‘scientific’ appreciation of how mankind’s linguistic devices for conveying thoughts from one mind to another ‘really work’.
As in the natural sciences, the analysis will be no good if it does not reveal an orderly system. For science is in the business of reducing chaos to order. As regards language, the philosophical challenge is to show that the superficial Babel of the many tongues of men, and the incongruities, inconsistencies and inadequacies readily evident in any one of them, mask an underlying simplicity, uniformity and rationality. Faced with this immense (but self-imposed) undertaking, the modern philosopher of language has equipped himself with a number of intellectual tools, of which the sharpest appears to be a radical distinction between two kinds or levels of meaning. He distinguishes between the ‘sense’ of what is said and its ‘force’. This allows him to maintain that, for example, ‘Frenchmen eat snails’ and ‘Do Frenchmen eat snails?’ mean the same as far as their ‘sense’ is concerned, while typically differing in meaning as far as their ‘force’ is concerned. Although the declarative sentence is usually employed with the ‘force’ of a statement, while the interrogative sentence is usually employed with the ‘force’ of a question, both statement and question have something important in common – namely, that both are about ‘the same thing’, the eating of snails by Frenchmen. Now this abstraction which we call ‘the same thing’ does not appear to be tied to any particular language. For the question as to whether Frenchmen eat snails can as well be raised in English as in French or, for that matter, Portuguese, Chinese or German.
Thus far the distinction between ‘sense’ and ‘force’ sounds like a matter of plain common sense. For no one would wish to deny that we can make statements, ask questions or issue instructions about the same thing, whether in one or in several languages. Indeed, if we could not do that, then language would not be a very useful means of communication. Therefore what strikes the layman first of all about the philosopher’s distinction between ‘sense’ and ‘force’ is not any doubt about its validity, but rather why such a very ordinary distinction is of such philosophical importance. The answer to this is that ‘force’ appears to be a matter of what, as language-users, we do with words (communicating with our family, friends and strangers, seeking information, discussing issues of interest, expressing our likes, dislikes, fears etc); whereas ‘sense’ is a matter of what words do for us. It is rather like the difference between what you choose to spend your money on, as distinct from the value of the money you have in your pocket to spend. In short, the distinction between ‘sense’ and ‘force’ allows the philosopher to bypass the tricky empirical questions which beset deciding when and how in real-life situations we actually decide – if we ever do – exactly what was meant by what was said in any particular instance. Instead, the philosopher is left free to concentrate on the ‘sense’ of the words used. For this element of meaning is allegedly invariant, remaining unscathed throughout all the diversities, complexities and mishaps of ordinary communication, and unaffected by translation to boot.
The distinction between ‘sense’ and ‘force’ is thus crucial in the following respect. If that distinction is not valid, the enterprise calling itself ‘philosophy of language’ is not in business at all: for there will be no market of abstract, invariant meanings for the philosopher of language to trade in. And if that is so, to make any serious contribution to the study of language the philosopher risks having to get out of his armchair and do some fieldwork, belatedly joining the linguists, anthropologists, psychologists, sociologists and others in the various kinds of systematic observation, documentation and experimentation which have, throughout the present century, gradually built up the corpus of what is known about human linguistic activity. On the other hand, if the ‘sense/force’ distinction is indeed valid, the philosopher can remain comfortably ensconced in the armchair of theory, secure in the assurance that no new facts which research could possibly discover about language use can affect his semantic pronouncements one jot, since the meanings he is concerned with qua philosopher are simply not affected by the rough and tumble of verbal communication.
Armchair comfort of this kind must be under serious threat if Baker and Hacker are right. In two chapters devoted to the ‘sense/force’ distinction, they argue that insofar as the distinction is linguistically tenable it just will not do the job the philosopher of language tries to make it do. Pressed beyond its natural limits, it simply becomes incoherent. The key issue here is how generally the distinction can be applied. For if only some sentences or utterances are subject to it, then it cannot provide a theoretical foundation of sufficient breadth and depth to support a universal philosophical semantics. It will be of no avail for the philosopher to insist: ‘In every sentence (or utterance) I distinguish “sense” from “force”.’ For it is far from clear how that stipulation makes sense at all. Saying to someone ‘Please shut the door’ does not prima facie give rise to any need for distinguishing what the words mean, as opposed to what would be meant in saying them. That sublety has no evident basis in everyday experience, for a very obvious reason. We do not count anyone as understanding the words ‘Please shut the door’ unless he or she grasps what would count as shutting the door in response to that request to do so. This simple fact of life affords no anchorage at all for a ‘sense/force’ distinction, look at it how we may. Understanding what is said is of a piece with understanding the words used.
Language, Sense and Nonsense offers an impressive range of reasons for doubting whether the philosophical buttresses commonly used to shore up the ‘sense/force’ distinction rest on any firmer linguistic bedrock than the distinction itself. The authors question the claims frequently made to the effect that declarative and interrogative transforms of a sentence have the same ‘descriptive content’; or involve ‘paraphrases’ of the same underlying item; or relate to the ways in which a given statement would be reported in ‘indirect speech’. It is to their credit that they do not at any point deny or take issue with the deliverances of our everyday understanding of relationships such as that between ‘Frenchmen eat snails’ and ‘Do Frenchmen eat snails?’ What they query, rather, are the foundations of the vast theoretical edifice which philosophers of language and generative grammarians have built thereon. We do, certainly, recognise that certain pairs of sentences are ‘about the same thing’. But this is a far cry from the assumption that the two sentences must therefore share some common element, and even further from the invention of abstract ‘levels of representation’ of the sentences where the common element can be invisibly located. It is rather like assuming that our everyday understanding that double yellow lines prohibit parking, whereas a white capital letter ‘P’ on a blue ground sanctions parking, automatically makes sense of the notion that the yellow lines and the ‘P’ sign share a common component.
Even if ‘sense’ could satisfactorily be divorced from ‘force’, how is ‘sense’ itself to be explained? One philosophically attractive proposal is the idea that an indisputable bridge linking different forms of words is the truth or falsity of what is said. Truth is what binds the answer ‘Frenchmen eat snails’ to the question ‘Do Frenchmen eat snails?’ in all cases where, as a matter of fact, Frenchmen do eat snails. Truth, it is often held, knows no linguistic frontiers. This, too, is a view apparently rooted in everyday experience. Would it not be absurd to deny that, say, an international group of journalists could accurately report – each one in his or her native language – some event that all had witnessed (for example, an attempt on the life of the President of the United States of America)? From this it would seem to follow that quite a number of sentences or sequences of sentences in English, French, Italian, Chinese etc must mean ‘the same thing’, in the final analysis. Otherwise, how could truth be international? Reuters and similar news agencies would be permanently discredited, let alone the variant reports which next day might appear in the Times and the Daily Telegraph. It might seem, then, that the theorist’s task of stating the meaning of a sentence like ‘Frenchmen eat snails’ is adequately performed by stating the conditions for the truth of the assertion ‘Frenchmen eat snails.’ These will also be the conditions for the truth of any correct translation of ‘Frenchmen eat snails’ in any other language.
Here again we seem to glimpse a viable basis on which to construct a universal philosophical semantics. But once again, Baker and Hacker argue, the prospect flatters only to deceive. ‘Truth-conditional semantics,’ they say, ‘is a powerful myth in the guise of a scientific theory.’ They trace the origins of the myth back to attempts to elucidate the logical constants within systems of formal logic. They examine various pivotal developments which occurred in the gradual extension of the application of the concept of truth-conditions to sentences of ordinary English, and expose the resultant confusions. One of these is a bizarre blindness to the fact that metalinguistic statements of the form ‘The sentence “Frenchmen eat snails” means that Frenchmen eat snails’ are nonsense. Furthermore, the notion that truth-conditional semantics provides an explanation for how a fluent speaker of English could understand what ‘Frenchmen eat snails’ means is a non-starter. For there is simply nothing which constitutes a satisfactory account of what the criteria would be for grasping the truth-conditions of this or any other English sentence.
It is clear, in any case, that if truth-conditional semantics is to explain meaning successfully for all forms of sentences, it needs to be underpinned by the ‘sense/force’ distinction. This is because it is nonsense to talk of stating the truth-conditions for ‘Do Frenchmen eat snails?’ Questions, unlike assertions, are neither true nor false. If, however, ‘Do Frenchmen eat snails?’ has the same ‘sense’ as ‘Frenchmen eat snails,’ then the truth-conditions for the latter will do the job for both sentences, since the two differ merely in ‘force’. Thus the doctrine of a common ‘sense’ is essential to the truth-conditional programme. Without it, non-declarative sentences of all kinds are left stranded in semantic limbo, their meanings unaccounted for. It follows that if the case earlier presented against the validity of the ‘sense/force’ distinction is sound, the truth-conditionalist stands no chance whatsoever of providing a complete semantic theory of any language other than one artificially restricted to fact-stating as its sole function. And this patently will not suffice for the languages we use for everyday communication.
The third probe of Baker and Hacker’s investigation jabs deep into the currently popular thesis that mastery of a language must be a matter of the individual’s possessing a ‘tacitly known’ set of rules. Here the culpable myth-mongers include Frege, the younger Wittgenstein, Davidson and Chomsky, all guilty of originating, passing on or compounding fundamental misconceptions about the nature of rules, rule-following and normative explanation (although Wittgenstein later recanted and made honourable amends). In linguistics the confusion began with Saussure and his muddled psychologism. The Saussurean attempt to ‘localise’ la langue in a particular segment of his metaphorical ‘speech circuit’, and his misleading analogies which equate a synchronic linguistic state with the cross-cut of the stem of a plant or the positional state of a game of chess, all engender an obfuscation which later theorists did nothing to dispel. It culminated half a century later in the Chomskyan absurdity of speakers who succeed, thanks to being born with an innately pre-programmed ability to do so, in ‘cognising’ linguistic rules they could not possibly be aware of. How a rule, or the mental representation of a rule, could causally determine a behavioural consequence in speech performance, rather than normatively determine the correctness of what was said, is one of many deep generativist mysteries.
Finally, Baker and Hacker deal with the ‘generative theory of understanding’. This is a theory which purports to explain how it is we can understand sentences we have never heard or read before (such as, perhaps, ‘Plastic Frenchmen eat plastic snails’). It relies heavily on the previously examined mythology of rules and their ‘mental representations’. The theory offers a compositional semantics, which builts up the meanings of sentences out of the meanings of constituent words, together with the meanings of their combinations. The notion that there is something deeply perplexing to be explained about understanding a previously unencountered sentence Baker and Hacker trace, once again, back to the Tractatus and to Frege. It is itself a puzzle generated by certain limited and misguided linguistic presuppositions, which do not carry over to everyday verbal discourse. Consequently it is hardly surprising to find that the philosopher of language finds himself here in a cleft stick. For if he grants that any combination of words makes sense which might conceivably, under certain circumstances, be used to say something intelligible, then most of the classic examples of nonsense – ‘Quadruplicity drinks procrastination,’ and others – must be granted to make sense after all. But if, on the other hand, he grants that nonsensical combinations of words can occasionally be used to make intelligible remarks (‘Look at that dead rainbow’), he is driven to abandon the claim that understanding previously unencountered sentences is just a question of computing the sense of the whole from the meanings of the parts.
At no point does Language, Sense and Nonsense commit the strategic error, made by many data-oriented linguists, of trying to discredit linguistic theories on the ground that they do not fit the linguistic facts (or enough of the linguistic facts). Such attacks are easily countered and the outcome is stalemate. The right strategy is the one which Baker and Hacker pursue in this book, and more relentlessly than any critics before them. They show why it is doubtful whether the linguistic theories currently on offer are even coherent, let alone whether they fit the linguistic facts. This is far more devastating. At the same time, Language, Sense and Nonsense provides enough historical commentary to show why presumably intelligent theorists came to perpetrate such astonishing howlers. The errors, time and again, arose from over-ambitious attempts to extend and adapt promising ideas without due regard for the logical underpinnings and explanatory contexts which made them promising in the first place.
Some readers may not be convinced by everything Language, Sense and Nonsense has to say about rules or about descriptions. Nor, doubtless, will the authors’ interpretations of Frege go scot free. But the reservations which might be made on these topics are in the final analysis marginal, neither blunting the edge of their criticism, nor weakening its thrust.
It is not possible to summarise the wealth of detail with which Language, Sense and Nonsense documents the case against the theses it attacks. One is left with three key questions. The first is whether Baker and Hacker have correctly identified the main sources of linguistic nonsense in the theoretical incoherence which they attack. The answer to this must be that, from an academic point of view, they undoubtedly have. There remains, nevertheless, an interesting question or series of questions concerning the psychological and social sources of that infection, and its psychological and social breeding grounds. The second question is whether the authors give a fair account of those general intellectual positions against which their assault is mounted. Again, the answer must be that, for all their irreverent asides and trenchant sarcasm, they indeed do. The third question is what remains of philosophy of language if their diagnosis is correct. The answer, it might seem, is ‘not a great deal.’ Their concluding chapter suggests that ‘nothing can be salvaged from modern theories of meaning’ – for the basic reason that the problems these theories address ‘are all bogus’. Nevertheless, it might be possible to transfer the title ‘philosophy of language’ to some more humble but less futile enterprise which sought, for example, to classify the concepts we use in describing and analysing language. In any event, if philosophy of language is not to be abandoned as worthless, ‘nothing less than a fresh start is needed, a total reorientation of thinking about languages.’ That conclusion might apply equally to the sister enterprise of theoretical linguistics. Language, Sense and Nonsense seems likely to cut at least as large a philosophical swathe as was cut nearly fifty years ago by the famous book whose title it echoes: A.J. Ayer’s Language, Truth and Logic. It could be the most important – and is likely to be the most controversial-challenge of the present decade to academic idées reçues about language.