Logical positivism was to a large extent an offshoot of Wittgenstein’s Tractatus. As the Circle understood (and often misunderstood) that book, it demonstrated how ‘consistent empiricism’, as they put it, is possible. It did so by showing, so they thought, that truths of logic and mathematics are tautologies, hence ‘analytic’, true by convention or true in virtue of the meanings of the constituent logical terms, and hence that pure reason alone can arrive at no substantive truths about reality.1 From the Tractatus, members of the Circle derived their conception of the task of philosophy, namely the logico-linguistic analysis of ‘scientific propositions’, and the disclosing of pseudopropositions of ‘metaphysics’. The contribution of philosophy is not to human knowledge, but to the clarification by means of logical analysis of what is known. They accepted the thesis of extensionality, the analysability of all empirical propositions into basic propositions, and the conception of a language as a calculus of signs connected to reality by means of ‘concrete definitions’ (ostensive definitions) of the primitive terms. From discussions with Wittgenstein, transmitted to the Circle by Schlick and Waismann, they derived the principle of verification.
Logical positivism constitutes the third great phase of twentieth-century analytic philosophy, following the pluralist Platonism of early Moore and Russell and the logical atomism of the early Wittgenstein and Russell. As a result of the rise of Nazism, most members of the Vienna Circle and of the affiliated Berlin Society for Scientific Philosophy fled to the USA. Though by the early 1940s orthodox logical positivism had been abandoned, the fundamental principles of the ‘scientific world view’ were retained. The impact of these European emigrés upon American philosophy was colossal. Grafted onto the native stock of pragmatism, their conception of philosophy, of philosophical analysis and the relation of both to science determined the growth of post-war philosophy in the USA. The positivist legacy was, however, transmuted by the greatest of twentieth-century American philosophers, W.V.Quine. More than any other single figure, Quine is responsible for the turn away from the heritage of analytic philosophy, both in its Viennese phase and in its post-war Oxonian phase. As the former was derived from the early Wittgenstein, so the latter was inspired by the later Wittgenstein of the Philosophical Investigations.
Many of the idées reçues of contemporary American philosophy originate in Quine’s writings, and are inimical to Wittgenstein’s later philosophy.2 This alone would make a comparison of Wittgenstein and Quine instructive, and fundamental to the understanding of the development of anglophone philosophy in the last three decades. But there is a further reason. A first glance at the philosophies of Quine and the later Wittgenstein suggests an extensive convergence of views. Given that Quine’s philosophy has been a major factor in the waning of Wittgenstein’s influence, and in the deep change in the conception of philosophy that has occurred, this may seem very surprising. The convergence of the incompatible needs explaining.
Quine and Wittgenstein converge, at least so it seems, over the following points:
(1) The meanings of words are neither ideas in the mind nor objects (Platonic or otherwise) in reality. Both philosophers deny that the concept of meaning can be explained mentalistically, i.e. by reference to mental acts of meaning or intending, or by reference to mental images or ideas. Wittgenstein remarked in 1931 that the concept of meaning is now obsolete save for such expressions as ‘means the same as’ or ‘has no meaning’ (M 258; AW L 30). Quine wrote in 1948 that
The useful ways in which people ordinarily talk about meanings boil down to two: the having of meanings, which is significance, and sameness of meaning, or synonymy. What is called giving the meaning of an utterance is simply the uttering of a synonym, couched ordinarily, in clearer language than the original…. But the explanatory value of special irreducible intermediary entities called meanings is surely illusory.
(OWTIb 11f.)
(2) One of the most famous Wittgensteinian dicta is ‘Don’t ask for the meaning, ask for the use.’ Quine, in one of his relatively rare references to Wittgenstein, quotes it approvingly:
Wittgenstein has stressed that the meaning of a word is to be sought in its use. This is where the empirical semanticist looks: to verbal behaviour. John Dewey was urging this point in 1925. ‘Meaning’, he wrote, ‘…is primarily a property of behaviour.’ And just what property of behaviour might meaning then be? Well, we can take the behaviour, the use, and let the meaning go.
(UPM 46)3
(3) Quine denies the intelligibility of the analytic/synthetic distinction. Wittgenstein does not invoke it (save, very occasionally, to remark ironically that if anything is a candidate for being synthetic a priori, it is mathematical propositions (e.g. RFM 246)).
(4) Both philosophers reject the Vienna Circle’s view that logical truths are true by convention, or true in virtue of meanings. According to Quine, the idea that meanings of words, whether construed as ideas in the mind or as abstract entities, can determine truths or determine us to use words in a certain way is ‘the myth of a museum in which the exhibits are meanings and the words are labels’ (OR 27). According to Wittgenstein, to say, for example, that the truth of ‘p=~~p’ follows from the meaning of negation is to be committed to the mythical Bedeutungskörper (meaning-body)
conception of meaning, which he condemned (PG 54, PLP 234ff.).
(5) Both deny that a natural language is a calculus with determinate rules which fix necessary and sufficient conditions for the application of all meaningful expressions in a language.
(6) Both deny the reducibility of all propositions or sentences to a set of propositions or protocol sentences which are conclusively verifiable by reference to what is immediately given in experience. Hence,
(7) Both repudiate classical foundationalism in epistemology. Quine’s stance is epitomized in the dictum that ‘There is no first philosophy.’ Holism displaces foundationalism, and ‘naturalized epistemology’, drawing upon psychology, neurophysiology and physics, replaces the investigation of the justification of knowledge claims with causal explanations. Wittgenstein’s private language arguments undermine classical foundationalism. It is replaced (in On Certainty) not by naturalized epistemology but by socialized epistemology.
(8) They agree that language learning rests upon training.
Language acquisition presupposes neither thought nor innate knowledge.
(9) They agree that language learning involves ostensive teaching, and that the mere ostensive gesture by itself does not suffice to determine the use of the word in question (RR 44f.; OR 30f., 38f.)
(10) They agree that the way an expression was learnt, the manner of its introduction, as such, is irrelevant to its status and role. Quine argues that the conventional, legislative, introduction of definitions or postulates ‘is a passing trait, significant at the moving front of science but useless in classifying the sentences behind the lines. It is a trait of events and not of sentences’ (CLTa 112). Wittgenstein argues that ‘the way we actually learn its meaning drops out of our future understanding of the symbol’; ‘the history of how we came to know what [the colour-word ‘green’, for example] means is irrelevant’ (LWL 23). ‘The historical fact of the explanation is of no importance’ (LWL 38). There is, he argued, ‘no action at a distance in grammar’, and what fixes the status of a proposition is its use, which may change over time or even from occasion to occasion of its employment.
(11) Both invoke radical translation, the translation of the language of a wholly alien people, as a heuristic device to illuminate the concepts of language, meaning and understanding. Like Quine, Wittgenstein approached philosophical questions in this domain (and others) from ‘an ethnological point of view’. He wrote:
If we look at things from an ethnological point of view, does that mean we are saying that philosophy is ethnological? No, it only means that we are taking up a position right outside so as to be able to see things more objectively.
(CV 37)
Hence he remarked, as Quine would, ‘The common behaviour of mankind is the system of reference by means of which we interpret an unknown language’ (PI §206).
(12) Both recognize a problem of indeterminacy in the use of language and the interpretation of its use. Wittgenstein raises a problem of apparent radical indeterminacy in the applications of rules, since it seems that quite different courses of action can be made out to accord with a rule, given an appropriate interpretation. This leads to the paradox that there is no such thing as correctly or incorrectly following a rule (PI §201). That paradox must be defused, on pain of concluding absurdly that there is no correct or incorrectapplication of rules, and hence no such thing as a correct, meaningful use of language. For Quine, there is a problem of radical indeterminacy of translation (both abroad and at home), and a problem of radical indeterminacy or inscrutability of reference. These too must be defused, on pain of concluding absurdly that all reference to objects is nonsense (OR 48).
(13) At first blush, both approach questions of understanding behaviouristically. Quine holds that
Semantics is vitiated by a pernicious mentalism as long as we regard a man’s semantics as somehow determinate in his mind beyond what might be implicit in his dispositions to overt behaviour. It is the very facts about meaning, not the entities meant, that must be construed in terms of behaviour.
(OR 27)
Wittgenstein wrote: ‘I conceive of understanding, in a sense, behaviouristically…. What is behaviourist in my conception consists only in that I do not distinguish between “outer” and “inner”. Because psychology does not concern me’ (BT 284).
(14) They converge in their conception of truth, repudiating correspondence and coherence theories alike, and, relative to those theories, trivializing truth Wittgenstein adopted a deflationary (Ramseian) account of truth (NB 9, TLP 4.062, PG 123f., PI §136), while Quine treats ‘is true’ as a disquotational device.4
(15) Holism with regard to understanding a language is common to both. Quine remarks: ‘It is of theoretical sentences such as “neutrinos lack mass”, etc. above all that Wittgenstein’s dictum holds true: “Understanding a sentence means understanding a language”‘ (BB 5), and adds in a footnote ‘Perhaps the doctrine of indeterminacy of translation will have little air of paradox for readers familiar with Wittgenstein’s latter-day remarks on meaning’ (WO 76f.).
(16) Both adopt holism with respect to the web of belief. They concur that the web consists of beliefs which are differently related to experience, some exposed to direct verification or falsification, others deeply embedded within the network. Wittgenstein wrote: ‘All testing, all confirmation and disconfirmation of an hypothesis takes place within a system…. The system is the element in which arguments have their life’ (OC §105). Again,
A child learns to believe a host of things. I.e. it learns to act according to these beliefs. Bit by bit there forms a system ofwhat is believed, and in that system some things stand unshakably fast and some are more or less liable to shift. What stands fast does so, not because it is intrinsically obvious or convincing; it is held fast by what lies around it.
(OC §144)
(17) Both agree that we hold mathematical statements immune to falsification. Quine’s ‘maxim of minimum mutilation’ is one of the two guidelines5 of his holistic doctrine of accommodating the falsification of what he calls ‘an observation categorical’6 which is implied by a hypothesis in conjunction with other sentences of the theory. We need not reject the hypothesis, but may instead reject some of the other sentences. However, ‘The maxim constrains us, in our choice of what sentences…to rescind, to safeguard any purely mathematical truth; for mathematics infiltrates all branches of our system of the world, and its disruption would reverberate intolerably’ (TI 11). Similarly, Wittgenstein remarks that we should never allow anything to prove that we are wrong in saying 12×12 =144 (LFM 291). We deposit mathematical propositions ‘in the archives’ (RFM 165), and they are thereby withdrawn from doubt (RFM 363). A proof shows one how one can hold fast to the proposition without running any risk of getting into conflict with experience (RFM 436). The ‘hardness of the logical “must”‘ indicates our refusal to depart from a concept (RFM 238).
(18) Both reject de re necessity. Quine continues the previously quoted remark by saying:
If asked why he spares mathematics, the scientist will perhaps say that its laws are necessarily true; but I think that we have here an explanation, rather, of mathematical necessity itself. It resides in our unstated policy of shielding mathematics by exercising our freedom to reject other beliefs instead.
So too Wittgenstein holds that the apparent inexorability of logic and mathematics is our inexorability in cleaving to them (RFM 37). What appear to be necessities in the world are merely the shadows cast by grammar.
To a large extent, the two philosophers were concerned with similar questions. Both explored all the issues above mentioned in extenso, tracing the threads that connect the conceptual manifold. But despite superficial appearances, the tapestry Wittgenstein wove is profoundly different from Quine’s. The negative points of convergence (roughly (1), (3–7), (10), (14) (18)) are genuine, although the reasons for them are often very different (especially (3), (7), (18)). The positive points, as we shall see, often mask profound disagreement (especially (2), (12), (15–18)). Even where there is a degree of methodological agreement ((11), (13)), the employment of the methodology is altogether distinct. For Wittgenstein’s conception of language, unlike Quine’s, is normative. This disagreement also infects the partial agreement over such points as (8). Similarly, the agreement over ostensive teaching (9) is superficial, since Quine does not conceive of ostensive definition as a rule for the use of a word or of a sample as belonging to the method of representation.
The most significant influence upon Quine was Carnap. He was, as Quine acknowledged, his ‘greatest teacher’. ‘Even where we disagreed’, Quine wrote, ‘he was still setting the theme; the line of my thought was largely determined by problems that I felt his position presented.’7
Quine shared much common ground with Carnap and members of the Circle:
(1) Like them, he was and remained an empiricist, holding that all knowledge is derived from experience. Unlike them, he came to explicate (or, as he put it, ‘to make an analytic tool of’) the concept of experience in neither phenomenalist nor physicalist terms, but rather in terms of stimulations of sense receptors. The common-or-garden concept of experience, he came to think, is ‘ill-suited for use as an instrument of philosophical clarification’ (TT 184f.).
(2) Like the scientifically trained philosophers of the Circle, Quine held that the paradigm of knowledge is scientific knowledge. It is science and scientific theory that yield the best picture of the nature of reality. All understanding is cut to the model of scientific understanding.
(3) The Circle cleaved to the doctrine of the unity of science. Quine held analogously that all knowledge can be unified in a single system, the foundations of which are given by the master science—physics. For ‘every change of any kind involves a change in physical micro-states’, and these are to be explained by physics. Physics gives us the fundamental description of reality, and all deep
(1) explanations of phenomena are physical explanations, for the fundamental laws ofthe universe are physical laws. Explanations in less fundamental sciences, though not reducible to physics, are at best local generalizations supervenient upon physical law.
(4) Although Quine rejected the principle of verification, i.e. that ‘the meaning of a statement is the method of confirming or infirming it’ (TDEb 37), he did not reject verificationism:
The Vienna Circle espoused a verification theory of meaning but did not take it seriously enough. If we recognize with Peirce that the meaning of a sentence turns purely on what would count as evidence for its truth, and if we recognize with Duhem that theoretical sentences have their evidence not as single sentences but only as larger blocks of theory, then the indeterminacy of translation of theoretical sentences is the natural conclusion. And most sentences, apart from observation sentences, are theoretical. This conclusion, conversely, once it is embraced, seals the fate of any general notion of prepositional meaning or, for that matter, state of affairs.
Should the unwelcomeness of the conclusion persuade us to abandon the verification theory of meaning? Certainly not.
(EN 80f.)
(5) Quine shared the Circle’s (general, though not uniform) distaste for ‘abstract entities’ and the nominalist preference for austere ‘desert landscapes’ (OWTIb 4). Although he came to ‘accept’ the existence of classes, functions and numbers, his philosophy is run through with a preference for, though not a commitment to, nominalism. Abstract entities are to be admitted into one’s ontology only in so far as they are required for respectable science and philosophy, and in so far as sharp extensional criteria of identity for them are forthcoming. He is therefore a qualified, economical realist, but an unqualified ‘extensionalist’ (TT 182–4). Among what Quine thought of as illegitimate abstract entities are propositions, which he conceived of as the purported meanings of sentences.8 Meanings, and indeed ‘intensions’ of any kind, were banished from Quine’s landscape as ‘entities’ wrongly posited by sundry theories.
Unlike the Vienna Circle, Quine had a substantial American heritage consisting of (a) pragmatism, derived from Dewey (and perhaps C.I.Lewis, who taught Quine at Harvard), and (b) behaviourism derived from Watson, and behaviourist language theory derived from Skinner. Early and late, he believed that ‘in linguistics one has no choice but to be a behaviourist’. For ‘Each ofus learns his language by observing other people’s verbal behaviour and having his own faltering behaviour observed and reinforced or corrected by others. We depend strictly on overt behaviour in observable situations’ (PTb 38). His behaviourism is the driving force behind his doctrine of the indeterminacy of translation (PTb 37). It is also the driving force behind his rejection of the analytic/ synthetic distinction. Holism alone will not yield that result, as is evident from the fact that Carnap accepted Duhemian holism too, but that did not affect his acceptance of an explicated form of the analytic/synthetic distinction (see n. 10 below).
Sharing some of the basic tenets of Viennese logical empiricism, Quine nevertheless rejected three of its fundamental doctrines in the name of a purified empiricism, a verificationism revamped to the requirements of holism, and behaviourism:
(1) He rejected the intelligibility of the analytic/synthetic distinction, interpreted as a distinction between truths that are grounded in meanings, independently of facts, and truths that are grounded in empirical fact. Hence too, he rejected the pivotal positivist claim that so-called necessary truths are analytic, i.e. true in virtue of the meanings of their constituent expressions, or true by linguistic convention.
(2) He rejected the reductionism that had informed the early phases of Viennese logical positivism, i.e. the claim that all significant empirical sentences are reducible to what is given in immediate experience. This conception had informed the programme of logical construction apparently sanctioned by the Tractatus and pursued (most notably by Carnap in Der logische Aufbau der Welt) in the wake of Russell.
(3) He repudiated sentential verificationism, i.e. the claim that the unit of empirical significance is the sentence which is confirmed or disconfirmed in experience. Instead, Quine, following Duhem, defended a holistic conception of confirmation.9 Our statements about the external world face the tribunal of sense experience not individually but as a corporate body. It is, however, noteworthy that already in The Logical Syntax of Language Carnap too had accepted Duhemian holism with regard to the confirmation or disconfirmation of hypotheses, without renouncing, but rather insisting upon, the validity of the analytic/synthetic distinction.10
These anti-positivist doctrines undermine the Vienna Circle’s conception of philosophy, and not just that of the Circle, but that of analytic philosophy from the 1920s onwards. Of course, it is not truethat analytic philosophy in all its phases was committed either to sentential verificationism or to reductionism. Nor was it necessarily committed to upholding the analytic/synthetic distinction as traditionally conceived or as explicated by Carnap—Wittgenstein was not so committed. He distinguished rather between logical and grammatical truths on the one hand (which are not to be assimilated) and empirical truths on the other (which are not uniform—propositions of the Weltbild, which he discussed in On Certainty, occupying a special position). Nevertheless, a fundamental tenet of analytic philosophy, from its post-Tractatus phase onwards, was that there is a sharp distinction between philosophy and science. Philosophy in the analytic tradition, whether or not it was thought to be a cognitive discipline, was conceived to be a priori and hence discontinuous with, and methodologically distinct from, science.11 Similarly, analytic philosophy in general held that questions of meaning antecede questions of truth, and are separable from empirical questions of fact. If Quine is right, then analytic philosophy was fundamentally mistaken. On Quine’s view, philosophy is continuous with science (NK 126), and ‘philosophy of science is philosophy enough’. In this respect Quine reverts to an older tradition, for example of Herbert Spencer, Samuel Alexander, and (with qualifications and inconsistencies) Russell in the 1910s. Contemporary philosophers who follow Quine have, in this sense, abandoned analytic philosophy. Or, to put the same point differently, if this conception is compatible with what is now to be called ‘analytic philosophy’, then analytic philosophy has become so syncretic as to lose any distinctive marks other than stylistic and thematic, and has severed itself from its roots and trunk in the philosophical developments that run from Moore and Russell, through the early Wittgenstein and the Vienna Circle, Cambridge Analysis, the later Wittgenstein, and Oxford analytic philosophy. Quine’s conception places him in stark opposition to Wittgenstein’s twofold revolution in philosophy (the first heralded by the Tractatus, the second by the Investigations).
Evidently the convergences noted above stand in need of scrutiny. Some of them are indeed shared views. Others are mere apparent convergences, masking fundamental disagreements. In the following I shall draw out some of these differences with respect to the following themes: (1) use; (2) meaning and synonymy; (3) analyticity and necessary truth; (4) ostensive teaching and explanation; (5) revisability of beliefs; (6) understanding, interpreting, translating and indeterminacy. Differences regarding ontology will not be discussed here. It should be stressed that the following discussion does not purport to adjudicate definitively between Wittgenstein and Quine. Although I have not masked my opinion that in the confrontation between the two philosophers it is Wittgenstein whose arguments carry the day, a proper refutation of Quine would require a book in its own right. The purpose of the ensuing discussion is to pinpoint the differences between the two protagonists and the grounds of their disagreements, and to indicate the trajectory of the further arguments that need to be pursued systematically and dialectically.
Quine quotes the Wittgensteinian dictum ‘Don’t ask for the meaning, ask for the use’ with approval, construing ‘use’ as mere behaviour, and concluding ‘Well, we can take the behaviour, the use, and let the meaning go’ (see above, p. 3). But ‘the use’ of an expression, for Wittgenstein, signifies not merely behaviour, but rule-governed behaviour, or more generally, behaviour subject to standards of correctness. The use of a piece in a game, for example a chess piece, is not merely the way in which people move it, but the way they move it when they move it correctly—in accord with the rules for its use. In a passage in which he was addressing behaviourist conceptions of language, Wittgenstein wrote:
If when language is first learnt, speech, as it were is connected up to action—i.e. the levers to the machine—then the question arises, can these connections possibly break down? If they can’t, then I have to accept any action as the right one; on the other hand, if they can, what criterion have I for their having broken down?
(PR 64)
Language learning is indeed rooted in training, and such training is in some ways similar to setting up a causal mechanism by stimulus conditioning. It does not follow that in general ‘the pronouncement of a word is now a stimulus, now a reaction’ (PLP 113f.). Suppose we trained a dog to behave in such-and-such a way in response to the stimulus of a sign ‘p’. Now contrast (a) The sign ‘p’ means the same as the command to do so-and-so, and (b) The dog is so conditioned that the occurrence of the sign ‘p’ brings about so-and-so. The behaviourist account of language reduces the explanation given in (a) to the description of a causal nexus given in (b). But (a) specifies a rule or convention for the use of the sign ‘p’, an explanation within the network of rules of language. Whereas (b) describes a causal mechanism. The truth of (b) is independent of the truth of (a), and the rule is independent of the reactions of the dog. A dog, no matter how well trained, may misbehave. But that what it does is misbehaviour is determined by reference to the stipulated convention of meaning. Otherwise what meaning a sign has would always be a matter of a hypothesis about what reaction it will call forth, and its meaning would not be determinable in advance of the behavioural consequences of its use from occasion to occasion.12
The objection applies to Quine’s behaviouristic conception no less than to Russell, Ogden, and Richards to whom it was addressed. Quine argues, correctly, that a learner has not only to learn a given word (for example ‘red’) phonetically; ‘he has also to see the object; and in addition to this…to capture the relevance of the object’ (OR 29). ‘A child learns his first words and sentences by hearing and using them in the presence of appropriate stimuli’ (EN 81). For the child ‘is being trained by successive reinforcements and extinctions to say “red” on the right occasions and those only’ (RR 42). But what, on a pure behaviourist account, makes a stimulus ‘appropriate’, an object ‘relevant’, or an occasion ‘right’? It is, to be sure, conformity with the use of the rest of the speech community into which he is being acculturated—but, of course, only in so far as their uses are correct, and not misuses.13
The vast majority of the utterances of members of a speech community doubtless employ the expressions of the language correctly, that being presupposed by their being members of a speech community with a shared language. Hence any statistical sampling will collect what are predominantly correct instances of the use of the language. But it will not provide an adequate criterion to distinguish correct uses from misuses (let alone from divergent, metaphorical, poetical or secondary uses). For correct use is not merely a statistical concept.14 The use of an expression is not just the verbal behaviour of users of the expression, but their verbal and other behaviour in so far as it accords with the acknowledged rules for the correct employment of that expression, rules which the users themselves acknowledge in their humdrum explanations of meaning, and of what they mean, and in their recognition of explanations by others of what certain expressions mean. These rules or conventions are not, of course, axioms or postulates of a formal system. Nor are they ‘implicit rules’ postulated by the field linguist. They are not ‘mental entities’. Nor are they mere history, for their role is not exhausted in the original teaching of the expressions.15 Far from being ‘explanatorily idle’ as Quine suggested in his criticism of Carnap (TCa 98f., CLTa 112f.), they are explanatorily indispensable, since they determine the difference between correct and incorrect use, as well as the difference between sense and nonsense.16 They are exhibited in explanations of meaning, which are as accessible to observations of behaviour as are descriptive uses of declarative sentences.
These explanations include answers to questions such as ‘What is a gavagai?’ (and Wittgenstein’s field linguist will fairly rapidly master the native technique of asking such simple questions). Such answers may take the form of ostensive definitions—many by reference to paradigmatic samples which are to be used as standards for the correct application of the definiendum.17 They may take the form of synonyms (precise or rough and ready), or of exemplification (‘Running is doing this’, ‘Hitting is this’), or of a series of examples (with a similarity rider) which are to be taken as a rule, or of paraphrase or contrastive paraphrase. (It can be presumed that the native will be willing to teach Wittgenstein’s field linguist, no less than he is willing to teach his own children.) The normative (i.e. rule-governed) use of words in sentences and the norms that are being complied with by speakers’ applications of words are perfectly accessible—as accessible as the difference between showing how to use a measure and a judgement of the length of an object. The field linguist can come to identify the native judgements of lengths, say, by observing their measuring activities, and, hesitantly no doubt and presuming upon native tolerance, by participating in the measuring practices. He will come to identify what the natives call ‘such-and-such a length’ (a foot, or a span), i.e. what their standard of measurement is, no less than he will come to identify their judgements that something is so-and-so many spans long. It is behaviour and participatory practice, and not something arcane and mysterious, that give us access to standards of measurement (and analogously to standards of correct use of terms) no less than to judgements of measurement (to correct applications of terms thus explained).
Quine and Wittgenstein agree that the genesis of an ability is irrelevant to its later characterization—how and whether one learnt the use of an expression does not matter inasmuch as it is true that there is no ‘action at a distance’ in grammar. But Wittgenstein insists, and Quine denies, that rules, thus understood, play a constant role in the use of language—as standards of correct use, cited in explanations, appealed to in criticisms of use and in clarification of disagreements (to determine whether the disagreement is one in judgement or in definition), and employed in teaching. The relevance of teaching is not causal or genetic, but rather immanent: ‘what matters is what is given in the explanation’ (LW L 38, my italic). What is thus given is a rule, a standard, against which to judge the correctness of the application of an expression from case to case, and by reference to which we can generally differentiate between disagreements in judgements and disagreements in definitions.
It is explanations of meaning that constitute standards for the correct use of their explananda, and what counts as a correct application of an expression is exhibited in the practice of its application (and the critical reactions, as well as the uncomprehending questions, that are forthcoming when an expression is misused). For communication by means of language to be possible, Wittgenstein argued, there must be agreement not only in judgements (as Quine holds), but also in definitions or explanations of meaning—in standards of correct use (see PI §242). There is an internal relation between an explanation of meaning (definition, or a rule for the use of an expression) and applications of that expression, and understanding an expression is grasping that relation, i.e. grasping what counts as applying the expression correctly. For applying an expression in accord with its explanation is one criterion of understanding. Another is explaining it correctly in context—for someone who cannot say what he means by the use of an expression in some way (by paraphrase, contrastive paraphrase, exemplification, ostension, etc.) will be said to be speaking without understanding what he is saying. And if what he means by it deviates significantly from what it means, he will be said to be misusing it. A third criterion of understanding is reacting appropriately in context to the use of the expression, and what counts as ‘appropriate’ is partly determined by what the expression means, as given by an acceptable explanation of its meaning.
It may well be that, as Quine claims, the child’s early training in the use of language involves primarily one-word sentences, but surely not only observation sentences. Expressive sentences will be at least as important—for example ‘Hurts!’, ‘Good!’, as will ersatz imperatives—for example ‘Want!’, ‘Drink!’, ‘Apple!’. And assent or dissent will be exhibited in responses to requests or demands no less than in responses to questions. However, he must rapidly progress beyond this to terms, and not by constructing analytical hypotheses (the child is no theorist or linguist), but by learning their use, mastering the technique of their application, including their combinatorial possibilities and impossibilities with other expressions. This is learnt not by theory construction, but by guided practice, subject to correction of error-which is not the same as conditioning and reinforcement. For what he learns includes, among other things, how to justify and give reasons for what he does by reference to the standards of correctness he learns, how to criticize and correct misuses, including his own. Once the child has learnt to ask ‘What is that?’, ‘What is this called?’ and ‘What does “such-and-such” mean?’, he has passed the stage of ostensive training and moved on to the stage of being taught, by ostensive and other explanations, the use—the meaning—of words. He must learn, in rudimentary form no doubt, the differences, from case to case, between sense and nonsense. And nonsensical or ungrammatical forms of combination which he employs can be, and often are, corrected by parents and teachers.
It is evident that although Quine and Wittgenstein agree that in a sense all the field linguist and child have to go on in learning the language is behaviour, that agreement masks profound disagreement. I shall defer for a moment consideration of the differences between Quine’s field linguist and Wittgenstein’s (see below, pp. 26–30). While Quine presents the child as being conditioned in the use of language, this conditioning being aided by the existence of innate responsive similarities and by induction, which is ‘animal expectation or habit formation’ (NK 125), Wittgenstein conceives of language learning as not just a matter of conditioned response. Although it rests on shared reactive propensities and discriminatory capacities, and begins with mere training, what are to be learnt are the techniques of a normative practice.18 Those rule-governed techniques are learnt by engaging in the practice, subject to correction, guided by example and explanation.
From the point of view of a normative conception of meaning such as Wittgenstein defends, a behaviouristic conception like Quine’s is simply no conception of meaning at all, not even an ersatz one.19 Indeed, it is no conception of language, for a language stripped of normativity is no more language than chess stripped of its rules is a game.
Quine denies, rightly, that ‘meanings’ are ‘entities’. He claims that at best we can talk of expressions having a meaning, i.e. being significant, and of different expressions as having the same (or different) meaning. But we can speak of sameness of meaning, or synonymy, only if there are clear criteria of identity for meanings. He argues that none are forthcoming, since the concept of synonymy can only be explained by reference to equally problematic intensional notions like necessity, self-contradictoriness, definition, semantic rule, immunity to falsification by experience (unassailability come what may), and apriority. It is, however, important to note that he does not take the concept of synonymy to be incoherent. ‘The explicitly conventional introduction of novel notation for purposes of sheer abbreviation’ is perfectly licit.
Here the definiendum becomes synonymous with the definiens simply because it has been created expressly for the purpose of being synonymous with the definiens. Here we have a really transparent case of synonymy created by definition; would that all species of synonymy were as intelligible.
(FLPVc 26, my italics)
It is unclear whether we are to conclude that in such transparently intelligible cases, in which synonymy yields perspicuous criteria of identity, meanings are ‘entities’.
If stipulation can produce synonyms, then there is such a thing as two expressions having the same meaning (rather than being merely ‘stimulus synonymous’). If so, why cannot there be unstipulated synonyms in use, as manifest in the explanations that competent speakers give of the use of terms (which is precisely what lexicographers typically catalogue)? Maybe there are none, but at any rate, we understand what would count as a pair of synonymous expressions. Grice and Strawson compare Quine’s position here to a man who claims to understand what it is for two things to fit together if they are specially made to fit together, but denies that it is intelligible that things not so made should fit together. So far from that being unintelligible, they further argue, synonymy by explicit convention is only intelligible if synonymy by usage is presupposed. There cannot be law where there is no custom, or rules where there are no practices.20 To be able to stipulate that a novel expression is to mean the same as a previous one, one must already have a conception of synonymy. It may be that natural language so evolves as largely to exclude the kind of redundancy that is involved in the common existence of exact synonyms, but that is surely something to investigate, not to dismiss. If it be so, we may find it useful (as lexicographers do) to consider synonymy a matter of degree-context-and purpose-relative. But if so it be, that is a fact, not a defect.
Wittgenstein has no qualms about talking of the meaning of expressions. Meanings are indeed not ‘entities’. To know the meaning of ‘A’, like to know the length of X, the age of Y or the price of Z, is not to be acquainted with an entity, but to know the answer to the question ‘What does “A” mean?’ (‘What is the length of X, Y’s age, or Z’s price?’). The ‘what’ here is an interrogative pronoun, not a relative one. To say that ‘A’ has the same meaning as ‘B’ is not to say that there is some third thing they both mean, but rather that ‘A’ means (the same as) ‘B’, that they are used in the same way, that an explanation of what ‘A’ means will also serve as an explanation of what ‘B’ means, and indeed that citing ‘A’ will serve as an answer to the question ‘What does “B” mean?’ The meaning of an expression is determined by its use; it is given by what are accepted as explanations of meaning; it is what we understand when we understand or know what an expression means. And that is exhibited in the criteria of understanding. Expressions are synonymous if the explanation of what the one means will also serve as a correct explanation of what the other means. To be sure, expressions are typically more or less synonymous, or synonymous in some contexts and not in others or for some purposes and not others—the matter of synonymy is indeed often context-dependent and purpose-relative:
The question whether ‘He can continue [the series 2, 4, 6, 8…]’ means the same as ‘He knows the formula [An=2n]’ can be answered in several different ways: We can say ‘They don’t mean the same, i.e., they are not in general used as synonyms as, e.g., the phrases “I am well” and “I am in good health”‘; or we may say ‘Under certain circumstances “He can continue …” means he knows the formula.’
(BB 114f.)
Synonymy is not an all or nothing affair. For some purposes of describing spatial relations, ‘on’ and ‘on top of mean the same. ‘The book is on the table’ means the same as ‘The book is on top of the table.’ But ‘Hillary is on Everest’ does not mean the same as ‘Hillary is on top of Everest.’ The criterion of adequacy for a dictionary definition (specification of synonymy) is that the definiens should be standardly substitutable for the definiendum, but such specifications do not and need not indefeasibly license substitution. The demand for absolute, context-free and purpose-independent standards of synonymy is as absurd as the demand for completeness of definition or determinacy of sense (the exclusion not of vagueness, but of the very possibility of vagueness), prominent in Frege and the Tractatus.21
Quine takes so-called ‘analytic truths’ to be true in exactly the same way as empirical propositions, and does not see them as having any different role from any other propositions embedded in the web of belief. Like Carnap, who never abandoned his conviction that, at least in a constructed language, one can sharply differentiate analytic truths from empirical ones, Quine never raises the question of the role of such truths as ‘Red is darker than pink’, ‘Bachelors are unmarried’, ‘Either it is raining or it is not raining.’ Truth is truth, and there’s an end to the matter; and no one would deny that such statements are true.
From Wittgenstein’s point of view, this is like saying that knowing is knowing, no matter if it is knowing that grass is green, that green is a colour, or that nothing can be red and green all over; or that believing is believing, no matter whether what is believed is that it will rain tomorrow, that 2+2=4, that Goldbach’s conjecture is true, that one should not steal, that one’s name is N.N., that the world has existed for many years. It is not that ‘true’, ‘know’ or ‘believe’ are ambiguous (as are ‘bank’ or ‘port’—ambiguity being coincidental, and unlikely to be preserved through translation into another language, save per accidens), but rather that we need to investigate, from case to case, what it is for one kind of proposition (for example ‘2+2=4’) to be true as opposed to another (for example ‘Grass is green’, ‘Kindness is a virtue’), what counts as knowing one sort of proposition rather than another, etc.22
Like Carnap, Quine takes it that analytic truths, if there were any, would be type-sentences, every token of which is analytic. Indeed, he assumes, wrongly, that Carnap and the Vienna Circle were committed to the view that if a sentence is analytic, its status cannot be changed—whereas Carnap’s view was that an analytic truth cannot be falsified by experience, but that we can ‘abandon’ it, cease to count it as such. However, to abandon it is to change the meaning of its constituent terms.
Wittgenstein, unlike the Vienna Circle, did not explain analytic truths by reference to type-sentences which are either (instances of) laws of logic or reducible to a law of logic by the substitution of synonyms for constituent expressions in accord with definitions. Nor did he clarify the nature of so-called necessary truths by arguing that they are consequences of the meanings (definitions) of their constituent expressions. Indeed, Wittgenstein does not invoke the category of analytic truths in his later work. This may be due partly to a distaste for received jargon, partly to radical disagreement with the construal of such truths by the Vienna Circle and others, and partly to the fact that the concept of analyticity employed by his predecessors and contemporaries, no matter whether Kant, Frege or Carnap, does not cut along the distinction or distinctions that most concerned him, and hence, in his view, does not serve to explain or elucidate what it is for a proposition to be a ‘necessary truth’. The Circle’s account assimilated disparate linguistic phenomena, namely logical truths, mathematical truths and analytic truths as traditionally conceived. Further, it proved powerless to illuminate such ‘meta-physically necessary propositions’ as ‘Red is darker than pink’, ‘Red is more like pink than like blue’, ‘There is no transparent white.’ Instead, Wittgenstein distinguished between logical propositions, mathematical propositions and so-called metaphysical truths, the first being senseless but internally related to inference rules, the second being rules for the transformation of empirical propositions about quantities and magnitudes of things, and the last being rules for the use of their constituent expressions in the misleading guise of descriptions.
Whether a sentence expresses what we so misleadingly call ‘a necessary truth’ is a matter of what it is being used for, hence a feature of the use of token sentences. Two tokens of the same type-sentence may be differently used, now to express a ‘necessary truth’, now to express an empirical proposition. ‘War is war’, for example, is rarely used as an instance of the law of identity, and ‘What will be, will be’ is not typically used to express a theorem of tense logic. ‘This is red’ may be used to make an empirical statement about the carpet, or used as a ‘grammatical proposition’ (‘This (colour) is red’), which can indeed be taken as a ‘necessary truth’, although, like ‘Red is a colour’, it is in effect a rule for the use of the word ‘red’. ‘Acids turn litmus paper red’ was once used to define acids, i.e. as a grammatical proposition, but is no longer so used. Since criteria and symptoms in science often fluctuate, a proposition of physics may in one context be taken as an empirical law, and in another as a definition-depending on how it is employed in an argument. What Wittgenstein was adamant about was that no proposition could be used simultaneously to state an empirical truth and to express a grammatical rule, any more than a ruler can be used simultaneously as a measure and as an object measured (‘measures’ is irreflexive).
For Wittgenstein, the crucial question is: ‘What is the use of so-called “necessary” or “analytic” truths?’ We say that the following are all true: ‘2×2=4’, ‘p v~p’ ‘Red is a colour’, ‘Nothing can be red and green all over’, etc. But what is their point? What information are we conveying to anyone? What go under the name of necessary truths are expressed by the use of a mixed bag of kinds of sentences, and Wittgenstein does not impose uniformity upon them, but rather explains why we think of them as ‘necessary’ and what is meant by calling them so. He does not try to explain what ‘makes them true’— a dubious question, since they are unconditionally true (not made true by anything). A fortiori he does not claim that they are made true by a convention. In the sense in which ‘The sun is hot’ is made true by the sun’s being hot, ‘Red is a colour’ or ‘Either it is hot or it is not hot’ are not made true by anything—although precisely because red is a colour, one may say that A’s being red makes it coloured. Unlike the Vienna Circle, he never argued that any necessary truths are ‘true in virtue of meanings’, but condemned such a view as a mythology of meaning bodies. Unlike Quine, he did not hold that the truth of statements (by which Quine meant sentences) depends upon both language and extra-linguistic fact— it is not sentences that are truth-bearers, any more than it is sentences that are supported by evidence, believed or doubted, feared or suspected, but rather what is said by their use. What it is that is said by the use of a sentence depends upon language, but whether what is thus said is true or false does not (save in the case of empirical assertions about language). Unlike Quine, he did not hold that what we call ‘necessary truths’ are simply those which we ‘shield’ from empirical disconfirmation by exercising our freedom to reject other beliefs instead (TI 11), although that is second cousin to the truth (see below).
Truths of logic, he held, are vacuous (senseless, i.e. limiting cases of propositions with a sense). Despite the fact that they all say the same, namely nothing, they nevertheless differ. For they are internally related to rules of inference, and different tautologies may be related to different rules of inference. Inference rules are in turn definitive of what we call ‘thinking’, ‘arguing’ and ‘reasoning’. Mathematical truths are rules which belong to a vast system of interconnected rules, the essential point and purpose of which is the transformation of empirical propositions about the magnitudes or quantities of things, etc. Analytic truths are rules in the guise of descriptions: ‘Bachelors are unmarried’ is a grammatical proposition, an explanation of the meaning of the word ‘bachelor’, given in the material mode. It is a rule that licenses the inference from ‘A is a bachelor’ to ‘A is unmarried.’ Non-analytic necessary truths are similarly grammatical propositions, even though they are not transformable into logical truths by substitution of synonyms. For example, ‘Red is darker than pink’ is a rule licensing the inference from ‘A is red and B is pink’ to ‘A is darker than B.’ Where Quine argued that ‘(x)(x=x)’ can be said to depend for its truth upon the self-identity of everything (CLTa 106), Wittgenstein held that there is no finer example of a useless proposition than ‘A thing is identical with itself’, it being comparable to ‘Every coloured patch fits into its surrounding’ (PI §216). The proposition ‘a=a’ is a degenerate identity statement which says nothing (LFM 27, 283). ‘An object is different from itself is nonsense, and so too is its negation. Although the law of identity seems to have fundamental significance, the proposition that this ‘law’ is nonsense has taken over its significance (BT 412).
Necessary truths are indeed unassailable. They persist unalterably, independently of all that happens—as the construction of a machine on paper does not break when the machine itself succumbs to external forces (RFM 74). Nothing is allowed to falsify them, but their ‘necessity’ is not explained merely by the fact that we refuse to abandon them—that indeed would not distinguish so-called necessary truths from truths of our world-picture, such as ‘The world has existed for many years’, ‘I was born of parents’, ‘I have never been to the stars.’ What is marked by the ‘must’ of ‘If it is red, then it must be coloured’, ‘If there are ten Xs in each of ten rows, then there must be a hundred’, ‘If it is red, then it must be darker than pink’ is the normative role of such propositions as ‘Red is a colour’, ‘Red is darker than pink’, ‘10×10=100’—they are rules, ‘norms of representation’ or ‘norms of description’. ‘Red is a colour’ does not ‘owe its truth’ to red’s being a colour in the sense in which ‘Some dogs are white’ owes its truth to the fact that some dogs are white (or to some dogs’ being white). Its being true consists in its being an expression of a rule for the use of its constituent expressions ‘red’ and ‘colour’, as the truth of the proposition ‘The chess king moves one square at a time’ consists in its being the expression of a rule of chess. If we know that A is red and B is pink, we are entitled to infer without further observation that A is darker than B; if we know that there are ten Xs in each of ten rows, then we can infer without counting that there are a hundred Xs in all. If B turns out to be darker than A, then it was not pink, or A was not red, or one or the other has changed colour. If there are more or less than a hundred Xs, then there was a miscount, or some were added or removed. What we hold rigid is not a truth about the world, but a rule for describing how things are in the world.
It is true that we can transform an empirical proposition into a rule or norm of representation by resolving to hold it rigid. (But ‘The world has existed for many years’, which we could not abandon without destroying the web of our beliefs, is nevertheless not a rule, since its role is not to determine concepts or inference rules.) It was an empirical discovery that acids are proton donors, but this proposition was transformed into a rule: a scientist no longer calls something ‘an acid’ unless it is a proton donor, and if it is a proton donor, then it is to be called ‘an acid’, even if it has no effect on litmus paper. The proposition that acids are proton donors, like ‘25×25=625’, has been ‘withdrawn from being checked by experience, but now serves as a paradigm for judging experience’ (see RFM 325). Though unassailable, so-called necessary truths are not immutable—we can, other things being equal, change them if we so please (with provisos concerning logic (see pp. 24f.), and appropriate qualifications when it comes to expressions that are so deeply embedded in our form of life as to be unalterable by us). But if we change them, we also change the meanings of their constituent expressions—here Carnap was right. If we abandon the proposition that red is a colour, we thereby change the meanings of ‘red’ and ‘colour’; if we drop the law of double negation, we change the meaning of negation.
The above characterization of the disagreement between Wittgenstein and Quine in the matter of analyticity and necessary truth makes it possible to deal briefly with an otherwise large and ramifying topic, the nature and role of ostensive teaching and definition.23 The depth of the difference between a causalist viewpoint and a normative one is strikingly evident here.
Quine takes ostension to be a matter of conditioning and induction (OR 31), i.e. learning to associate a given stimulus with an utterance. It depends upon a snared innate standard of similarity (NK 123). In the case of what he calls ‘direct ostension’, ‘the term which is being ostensively explained is true of something that contains the ostended point [i.e. the point where the line of the pointing finger first meets an opaque surface]’ (OR 39).24 Wittgenstein similarly argues that ostension presupposes shared behavioural dispositions (for example, to look in the direction of the pointing hand) and discriminatory capacities. But unlike Quine, he distinguishes ostensive training (which he is willing to take behaviouristically) from ostensive definition or explanation. Of course, an ostensive definition sets up a connection between a word and a ‘thing’ (namely a sample). But ‘the connection doesn’t consist in the hearing of words now having this effect, since the effect may actually be caused by the making of the convention. And it is the connection and not the effect which determines the meaning’ (PG 190). An ostensive definition (the connection between word and sample) is an explanation of what a word means, and the explanation ‘is not an empirical proposition and not a causal explanation, but a rule, a convention’ (PG 68) for the use of the explanandum, a standard for its correct application—as is evident in cases in which the ostensive gesture, the utterance ‘This’, and the sample ostended can replace the definiendum in a sentence. Where a sample is employed, the sample is not an object of which the concept being explained is predicated, but rather belongs to the method of representation. It is the standard for the application of the term, not an instance of its application.
Quine takes everything within the web of belief to be capable in principle of being relinquished, including logic and mathematics— even though we are least willing to relinquish these in the face of recalcitrant experience. He argued that
The totality of our so-called knowledge or beliefs, from the most casual matters of geography or history to the profoundest laws of atomic physics or even pure mathematics and logic, is a man-made fabric which impinges on experience only along the edges…. Any statement can be held true come what may, if we make drastic enough adjustments elsewhere in the system. Even a statement very close to the periphery can be held true in the face of recalcitrant experience by pleading hallucination or by amending certain statements of the kind called logical laws. Conversely, by the same token, no statement is immune to revision.25
(TDEb 42f.)
Similarly, he later claimed ‘In science all is tentative, all admits of revision—right down…to the law of the excluded middle’ (SLSa 232), ‘mathematics…is best looked upon as an integral part of science, on a par with the physics, economics, etc., in which mathematics is said to receive its applications’ (ibid., 231), and ‘Logic is in principle no less open to revision than quantum mechanics or the theory of relativity…. If revisions are seldom proposed that cut so deep as to touch logic, there is clear enough reason for that: the maxim of minimum mutilation’ (PL 100). His invocation of the principle of minimum mutilation is wholly pragmatic, and does not rest on any discernment of a difference in function of mathematical and logical truths from any other truths (‘truth is truth’). Castigating Carnap for putting grammar and logic on the same footing (qua analogues of formation and transformation rules in a formal deductive system), Quine wrote:
We do better to abandon this analogy and think in terms rather of how a child actually acquires his language and all those truths and beliefs, of whatever kind, that he acquires along with it. The truths or beliefs thus acquired are not limited to logical truths, nor to mathematical truths, nor even to analytic truths, if we suppose some sense made of this last term. Among these truths and beliefs the logical truths are to be distinguished only by the fact…that all other sentences with the same grammatical structure are true too.
(PL 101)
Wittgenstein agreed that we can envisage a language without the law of double negation. Nevertheless, fundamental propositions of logic, such as the law of non-contradiction ‘~(p & ~p)’, or the tautology ‘p & (p?q)?q’, are renounceable only at the cost of renouncing all thought and reasoning. For these tautologies are internally related to inference rules which are constitutive of what we call ‘reasoning’, ‘arguing’, ‘thinking’. And he takes propositions of mathematics as concept-forming rules, characteristically licensing inferences among empirical propositions. Moreover, he denies that even humdrum empirical propositions such as ‘The world has existed for a long time’, i.e. certain propositions of the Weltbild, can be revised or rejected. For their repudiation would tear apart the whole web of belief. It is these, and not the propositions of mathematics and logic, that are so deeply embedded in the web of belief that they cannot be revised, even though they are not ‘necessary truths’.
On the other hand, propositions of logic are misconstrued as being akin to propositions of the Weltbild, i.e. so deeply embedded in the web of belief as to be impossible to extricate without total mutilation. Rather, they are the correlates of the inference rules that constitute the connecting links between the nodes of the web. It is the logical relations between beliefs that make for the difference between a web of beliefs and a collection of beliefs, for to believe that all As are F is ipso facto to believe that this A is F, as it is to believe that there are no As which are not Fs. The ‘abandonment’ of the law of noncontradiction would not be, as Quine suggests, ‘inconvenient’. Nor would it simply mean that we would score a poor ratio of successes over failures in our predictions. It would mean that the web of belief collapsed into a knotted tangle of incoherence. The role of the fundamental laws of logic is toto caelo different from that of the beliefs they connect within the web.26 Indeed, one cannot be said to believe them as we believe empirical propositions—to believe that either it is raining or it is not raining is not to have any belief about the weather, and to believe the principle of bivalence is simply to determine the concept of a proposition as that which can be either true or false.
Quine’s thesis of indeterminacy of translation is rooted in empiricist qualms about the under-determination of theory by evidence. Wittgenstein’s explicit paradox of rule following is, he argued, rooted in a misconception which turns on the under-determination of a function by a fragment of its extension. This paradox is defused by the consequences of realizing that the relation between a rule and its extension is not akin to the relation between an empirical hypothesis and its evidence, since the relation is internal. A rule is not an explanatory hypothesis which explains the acts that constitute conformity with it. The instruction ‘Observe a man’s behaviour in the course of the day, and infer which of his acts were intentionally performed in conformity with rules given to him’ is as absurd as ‘Here is a husband: now tell me who is his wife.’27 That a given activity (a game of chess, for example) is conducted according to such-and-such rules may indeed be a hypothesis or conjecture (of an observer who has not learnt the game), but it is quite wrong to suppose that there is no ‘fact of the matter’ as to how chess is to be played. It would doubtless be exceedingly difficult to pick up the rules from mere observation of moves alone, independently of observations of the discussions and explanations of the game, but then no one has to—rather we receive instruction and practice in playing the game.
Both Quine and Wittgenstein consider that reflection upon radical translation may be philosophically illuminating, and both approach radical translation behaviouristically—but each in a different sense. On Quine’s official view, the problem set the field linguist is to map ‘surface irritations’ onto dispositions to verbal behaviour. What is to be studied is the relation between the ‘meagre input’ of ‘certain patterns of irradiation in assorted frequencies, for instance’, and the ‘torrential output’ (EN 83) of intricately structured talk of things (WO 26).28 It is less than obvious that Quine cleaves to his rigorous behaviourism here, since behaviourism requires that behaviour be viewed as ‘bare bodily movement’, and speech as the emission of sounds, from which bare basis a translation is held to be derivable. The field linguist’s point of access, according to Quine, is the one-word observation sentence, to which assent and dissent are allegedly identifiable inductively. But assent and dissent are intensional (as well as intentional) notions—a person assents not to a sentence, but to what is said by the use of a sentence, i.e. to an assertion that thing are thus-and-so, and assents to what he understands inasmuch as he believes it to be true.29 The identification of assent and dissent therefore presupposes viewing the observed behaviour not as mere bodily movement, but intentionalistically30—and it is not obvious that Quine’s austere behaviourism entitles him to this intentionalist stance.
Wittgenstein’s ‘behaviourist’ approach to radical translation is unconnected with Watsonian or Skinnerian behaviourism. What is behaviourist about his conception of understanding is only that the distinction between the ‘outer’ and the ‘inner’ is irrelevant for him, since understanding is not a mental state, but akin to a capacity. The nature of the capacity, and the degree to which it is possessed, is to be seen in a person’s behaviour, including his linguistic behaviour. Wittgenstein recognizes ab initio that the ‘common behaviour of mankind’ by reference to which we interpret an unknown language is behaviour intentionalistically conceived. When an explorer comes to a foreign land, he wrote, he can come to understand the native language
only through its connections with the rest of the life of the natives. What we call ‘instructions’, for example, or ‘orders’, ‘questions’, ‘answer’, ‘describing’, etc. is all bound up with very specific human actions and an order is only distinguishable as an order by means of the circumstances preceding or following //accompanying it//.
(MS 165, 97f.)
Hence, too, ‘If a lion could talk, we could not understand him’ (PI, p. 223), not because his growls are unclear, but because his behavioural repertoire is so profoundly different from human behaviour, human expression, gesture and mien, and the forms of possible interaction we can engage in (even with a tame lion) are so limited. Our human ‘form (or forms) of life’ is not shared with lions. But ‘speaking a language is part of a form of life’. ‘It is a feature of our language that it springs up //it grows// out of the foundations of forms of life’ (MS 119, 148). ‘Instead of the unanalysable, specific, indefinable: the fact that we act in such-and-such ways, e.g. punish certain actions, establish the state of affairs thus-and-so, give orders, render accounts, describe colours, take an interest in others’ feelings. What has to be accepted, the given—it might be said—are facts of living //forms of life’ (RPPI §630, with an MS variant).
According to Quine all understanding is translating. Understanding utterances of another in one’s own language involves homophonic (and sometimes heterophonic) translation. To understand a language or conceptual scheme, to determine its ontological imports, is always to translate it into another language. ‘It makes no sense to say what the objects of a theory are, beyond saying how to interpret or reinterpret that theory in another’ (OR 50). ‘Commonly of course the background theory will simply be a containing theory, and in this case no question of a manual of translation arises. But this is after all just a degenerate case of translation still—the case where the rule of translation is the homophonic one’ (OR 55). For it only makes sense to ask what the references of terms are relative to a background language. Further, ‘questions of the reference of the background language make sense in turn only relative to a further background language’ (OR 49). But in practice ‘we end the regress of coordinate systems by something like pointing. And in practice we end the regress of background languages, in discussions of reference by acquiescing in our mother tongue and taking its words at face value’ (OR 49).
But understanding is not the same as translating or interpreting. The former is akin to an ability, while the latter are typically activities one engages in (although there is a use of ‘interpret’ which is synonymous with one use of ‘understand’, as in ‘He interpreted the order to mean…’, i.e. he took it (understood it) to mean). Nor can Quine licitly argue that all understanding involves translating or interpreting. Translating is a matter of rendering the utterances of one language in another. Interpreting is a matter of clarifying utterances by means of more perspicuous paraphrases, especially in cases where an utterance admits of divergent readings (legal statutes, poetry)—it is this interpretation as opposed to that one. Interpretation therefore presupposes understanding—where more than one way of understanding is on the cards, and interpretation weeds out the worse from the better way of understanding. If the speaker is still available, one is likely not to interpret his ambivalent utterance, but to ask him to explain what he meant—and he does not have to interpret his own words for himself. In cases where an utterance in one’s own language is not understood at all, one neither translates it nor interprets it, but rather, one explains it. ‘Homophonic translation’ is no more translating than photographing a painting is a kind of painting.
Understanding utterances of one’s own language is not exhibited by homophonic disquotation—this being neither necessary nor sufficient for understanding. A child exhibits understanding of the request ‘Shut the door!’ by shutting the door, not by engaging sotto voce in homophonic translation antecedently to shutting the door.31 Someone who has mastered the device of disquotation may exhibit this skill without manifesting any understanding at all. The fact that misunderstanding is rectified by interpretation and lack of understanding (of a foreign tongue) by translation does not show that understanding ordinarily involves either.
Wittgenstein argues that ‘any interpretation [of the expression of a rule in our own language] still hangs in the air along with what it interprets, and cannot give it any support’ (PI §198). Not all understanding can consist in assigning interpretations. How I understand something is shown not only by the interpretation I give of it if asked, but in what I do in response—which shows what I call ‘such-and-such’. In the case of an order, how I understand it is shown by what I do in compliance with it. Here ‘He has interpreted it to mean …’ just means ‘He has understood it to mean…’, not ‘He has interpreted it to mean…and now he has acted on that interpretation.’ For, if all understanding required an interpretation, this would indeed generate a regress, since he would now have to interpret the interpretation he gave. Moreover, it would follow that what was understood was not the order given, but only the interpretation of it (PG 47). An interpretation is given in signs, so the idea that every sentence stands in need of an interpretation amounts to claiming that no sentence can be understood without a rider. But this is absurd since the rider would need an interpretation. We do sometimes interpret signs. But when asked what time it is, we do not; we react. We react, and our understanding is manifest in what we do (see PG 47). That a symbol could sometimes be further interpreted does not show that one does further interpret it. There is an internal relation between an order and what counts as compliance with it, as there is an internal relation between an assertion and what makes it true— and what one understands by an order or assertion is to be seen in one’s behaviour, which manifests one’s grasp thereof.
To be sure, Wittgenstein never considered Quine’s theses of indeterminacy of translation and of inscrutability of reference. Nevertheless, some of his remarks and general strategies can be brought to bear upon the matter. In the first place, he would reject Quine’s behaviourist methodology. For Quine, what is ‘given’ to the field linguist is surface irradiations and responses. In strict consistency, the latter should be characterized in terms of bare bodily movements and emission of sounds (a limitation which, as we have seen, Quine fails to recognize). For Wittgenstein’s field linguist, what is given is human forms of life, to be characterized intentionalistically. For Quine, the primary leverage to be employed by the linguist is prompting assent or dissent by one-word observation sentences in circumstances of appropriate stimulus. For Wittgenstein’s linguist, it is participation in the alien form of life and practices, engaging in discourse aided by gesture and facial expression (and not merely prompting Yes/No answers from the native), requesting, ordering, thanking, expressing pleasure and dissatisfaction, warning and heeding warnings, commiserating with suffering, and so on.
Three associated Quinean presuppositions might be questioned from a Wittgensteinian perspective. First, the assumption that there is no role in the process of translation for explanations of meaning (construed normatively) given by the native, in particular none for ostensive definition by reference to samples and their use. ‘Someone coming into a strange country will sometimes learn the language of the inhabitants from ostensive definitions that they give him’ (PI §32). That ‘he will often have to guess the meaning of these definitions; and will guess sometimes right, sometimes wrong’ (ibid.) does not mean that there is no fact of the matter regarding correct understanding of them. For what counts as understanding such an explanation is manifest in correct application, which is internally related to the explanation.
Secondly, Quine pays no attention to the grammar (and grammatical form) of expressions that are being translated (this is an aspect of his disregard for any distinction between nonsense and falsehood). His claim that the term ‘gavagai’ may indifferently signify ‘rabbit’, ‘rabbit stage’, ‘undetached rabbit part’ or ‘rabbithood’ is wrong. For the grammar of these expressions, their combinatorial possibilities in language, is wholly different. If the linguist succeeds in translating ‘Hungry!’ (a fairly early achievement, one would think),32 then if gavagai (or a gavagai) is said to be hungry, he can be sure that ‘gavagai’ does not mean rabbithood or undetached rabbit part. A defender of Quine might respond that the native utterance might signify not ‘This rabbit is hungry’, but ‘This undetached rabbit part is a part of a hungry animal.’ It might—if it possessed the appropriate grammatical multiplicity. But if an expression might signify ‘is a part of an F animal’, then to be sure, it cannot, in another utterance, do service as the copula if such there be. An expression signifying a rabbit-stage can only be interchangeable in translation with one signifying a rabbit if the grammar of phase-sortals is indistinguishable from the grammar of their corresponding sortal—which it is patently not. The supposition that all grammatical categories are permutable in different translations compatible with making sense rests on no argument but only on Quine’s bold assertion. One would like to see translations of a page of humdrum English prose into German in accord with such divergent ‘translation manuals’ which severally preserved intelligibility.
Finally, the use of language is embedded in the stream of human life. It is a part of the endlessly differentiated pattern of human behaviour. The thought that there can be two or more equally acceptable translation manuals for a given language, and no fact of the matter in choosing between them, rested for Quine foursquare on the translatability (in terms of stimulus synonymy) of observation sentences (on the basis of identification of assent and dissent), the alleged indeterminacy of translation of standing sentences, the underdetermination of theory by evidence, and the inscrutability of reference of terms in general. But the thought that the network of standing sentences is capable of divergent interpretation consistent with translation of observation sentences (including (pace Quine) expressive utterances and sentences containing indexicals), and consistent with the intelligibility of the associated human behaviour is misconceived. Learning a language is no more learning a theory than is learning any other normative practice, for example learning how to play a game. There are behavioural criteria for understanding words, i.e. for having mastered the techniques of their use, no less than there are behavioural criteria for understanding the moves of pieces in a game. It is striking (and no coincidence) that attempts by Quine’s followers to defend his theses of indeterminacy of translation and inscrutability of reference take as examples not the natural languages of mankind, but one fragment or another of mathematics or logic which admits of sundry permutations or alternative projections into some other part thereof without affecting truth. It is evident that such examples do not exemplify radical translation at all, let alone indeterminacy of translation.
If understanding is not a matter of translating, and if ‘homophonic translation’ is no translation, then to be sure, radical translation does not begin at home. It is, trivially, understanding that begins at home. Does one not understand one’s own utterances? Is there no fact of the matter about what one is referring to when one uses words? A person normally knows what he means when he says ‘N.N. is in the next room’, knows whom he means, and can say whom he means if asked. Quine argues that the question of what our words refer to is meaningless save in relation to ‘a background language’ (OR 49). From Wittgenstein’s perspective, taken one way, this is right; taken another, it is wrong. ‘The meaning of a word is its use in the language’ (PI §43), and a word has a meaning only as part of a language. Moreover, ‘It is only in a language that I can mean something by something’ (PI, p.18n.). To put this hyperbolically, as Wittgenstein does (PI §199), ‘To understand a sentence means to understand a language.’ For the sentence is the minimal unit for making a move in a language-game. It is comparable to a move in chess—and a move is only a move in the context of a game. Hence one might say that what a word refers to is a question that can only be raised and answered in relation to its use in a sentence of the language to which it belongs. But this does not make the question of its reference relative—as the question of the reference of an indexical in a sentence is relative to the context of its utterance. What Quine means, however, is quite different from this, and has no such justification. It is false that ‘If questions of reference of the sort we are considering make sense only relative to a background language, then evidently questions of reference for the background language make sense in turn only relative to a further background language’ (OR 49). For all questions of reference arise only, and receive their answer only, with respect to the use of words in sentences of a language. It is misconceived to suppose that a metalinguistic question such as ‘What does “rabbit” (as employed in an antecedent utterance) mean?’ involves regress to a different language from the (English) utterance in which the word ‘rabbit’ occurred. And it is equally misconceived to suppose that one cannot ask for an explanation of what a word signifies save by metalinguistic ascent—’What is a rabbit?’ will do just as well. The supposition that there is a regress of different languages is as gratuitous as the relativity thesis. Quine’s manner of extricating himself from the absurdity is ‘That in practice we end the regress of background languages, in discussions of reference, by acquiescing in our mother tongue and taking its words at face value’ (ibid.). The truth of the matter is that there is no regress, and the question of inscrutability of reference does not arise, precisely because we use our mother tongue, having mastered the technique of its use, and we normally take its words ‘at face value’, since they are not normally used metaphorically or in a secondary sense, and we know, and can explain, what they mean. But that is not a conclusion Quine would wish to arrive at, or one to which his argument entitles him.
We began this discussion with a survey of apparent convergences between Quine and Wittgenstein. Closer scrutiny, however, reveals the two philosophers to be as proximate, and as distant, as members of the far Right and the far Left in the horseshoe-shaped French National Assembly—one must travel through the whole spectrum of opinion to reach the one viewpoint from the other. The one is (to use Isaiah Berlin’s Archilochean typology) an exemplary ‘hedgehog’, a methodological monist, a defender of scientism in philosophy, a naturalizing epistemologist and propounder of an ontology guided by physics and canonical notation. The other is a paradigmatic ‘fox’, a methodological pluralist appalled at the misguided idea that the only forms of knowledge and understanding are scientific, who viewed scientific method in philosophy as the source of misconceived metaphysics, who socialized epistemology without naturalizing it, and held that the canonical notation of mathematical logic had completely deformed the thinking of philosophers.
If Quine is right, then philosophy is an extension of science, and philosophical understanding is homogeneous with understanding the phenomena of nature as well as of mathematics and logic. It is part of the vast man-made web of belief with which we confront experience, differing from the rest only in its generality. The philosophical enterprise is part of the human endeavour to achieve knowledge of the world. If Wittgenstein is right, then philosophy is sui generis. It is a quest for understanding, not for knowledge. What it aims to understand is the structure of our familiar conceptual scheme, which is presupposed by all our knowledge of the world, and is partly constituted by logic and mathematics, which are a priori, and fundamentally distinct from science. It attains such understanding not by theory construction, hypotheses and explanation, but by description of the way we use words, and such arrangement of the rules for the use of expressions that enables us to see where entanglement in these rules leads us astray and generates the idiosyncratic problems of philosophy, which are categorially distinct from the cognitive problems of science. The resolution of these problems does not add to the sum of human knowledge about the world. What it produces is understanding, clarity about our own thought, and, to use a phrase from the Tractattts (TLP 4.1213) made famous by Quine, a correct logical point of view. Then we can see the world, and ourselves, and our place within the world aright (see TLP 6.54).
*This chapter is a much-shortened version of chapter 7 of my forthcoming book Wittgenstein’s Place in Twentieth Century Analytical Philosophy (Blackwell, Oxford). I am indebted to Dr H.-J. Glock, Professor O.Hanfling, Dr J.Hyman, Dr D.Isaacson and Dr Anat Matar for their comments on that longer version.
1 In fact the Tractates argued that logic is transcendental, that all the truths of logic flow from the essential (bipolar) nature of the proposition as such, that is, reflect the logical properties of the world, and that mathematical propositions are, technically speaking, nonsense.
2 I shall not be concerned here with the Tractatus.
3 Quine’s references to Wittgenstein are few and sometimes, as here, betray little understanding. Wittgenstein did not suggest, as Dewey did, that meaning is a property of behaviour (see below, pp. 11–16). Dewey’s conception of meaning was behaviouristic, ‘use’ being construed as behavioural effect. Congenial though this is to Quine, it is far removed from Wittgenstein’s normative conception of use. Elsewhere (OR 27) Quine suggests that Dewey’s claim that language presupposes the existence of an organized social group from which speakers have acquired their speech habits is a rejection of the possibility of a private language in Wittgenstein’s sense. This is mistaken, since Wittgenstein is not concerned with the social genesis of a language. A private language in his sense (a language the individual words of which refer to the speaker’s immediate private sensations, which can be known only by him) might, if it were possible at all, be thought to be acquired only in social interaction—as Augustine intimated (PI §1). Similarly, Quine suggests that Wittgenstein’s conception of philosophy as dissolving philosophical problems by showing that there were none really there is satisfied by Carnapian explication (WO 260). This is the converse of the truth. If one wants to know how birds can fly, it avails little to be told how to build an aeroplane (I owe the simile to Avishai Margalit). Carnapian explication does not dissolve philosophical problems, but sidesteps them by banishing the words that give rise to them. Wittgenstein, by contrast, put those problem generating words and the contexts in which they generate problems under the microscope. He aimed to dissolve the philosophical problems by showing how entanglement in the grammar of those very words that a Carnapian explication banishes gives rise to the philosophical problem that bewilders us, and his solvent is the description of the use of the problematic expression, of its place in the grammatical network of related expressions, and of its grammatical differences from superficially similar expressions.
4 Ramsey’s account is not disquotational. In his view, truth is ascribed primarily to propositions, not to sentences. Hence he claims not that ‘“p” is true’=‘p’, but rather that ‘It is true that p’=‘p’ (‘Facts and Propositions’, repr. in D.H.Mellor (ed.), F.P.Ramsey: Foundations— Essays in Philosophy, Logic, Mathematics and Economics (Routledge and Kegan Paul, London, 1978), pp. 44f.). Wittgenstein, although he asserted in the Investigations that “‘p is true’”=‘p’ ( PI §136), had argued in the Grammar that ‘the quotation marks in the sentence “‘p’ is true” are simply superfluous’, since ‘“p” is true’ can only be understood if one understands the sign ‘p’ as a prepositional sign, not if ‘p’ is simply the name of a particular ink mark (PG 124). Like Ramsey, he had no qualms about prepositional quantification, agreeing with him that ‘What he says is true’=‘Things are as he says’ (PG 123).
5 The other is the maxim of simplicity of theory.
6 A statement of the form ‘Whenever p, q’, which is compounded of observation sentences. It specifies a generality to the effect that the circumstances described in the one observation sentence are invariably accompanied by those described in the other (PTb 10).
7 W.V.Quine, ‘Homage to Carnap’, in R.Creath (ed.), Dear Carnap, Dear Van (University of California Press, Berkeley, 1990), p. 464.
8 This is rather surprising, since most philosophers who cast propositions in the role of truth-bearers do not make the mistake of characterizing them as meanings of sentences. What is true (or false) is also what is believed, assumed or claimed to be true, but it makes no sense to believe, assume or claim the meaning of a sentence to be true. What is believed may be implausible, exaggerated or inaccurate, but the meaning of a sentence cannot be any of these. (See A.R.White, Truth (Macmillan, London, 1970), ch. 1.)
9 But it is noteworthy that he stretched Duhem’s holism far beyond anything which Duhem would have countenanced.
10 ‘The test applies, at bottom’, Carnap wrote, ‘not to a single hypothesis but to the whole of system of physics as a system of hypotheses (Duhem, Poincaré)’ (The Logical Syntax of Language (Routledge and Kegan Paul, London, 1937), p. 318). Daniel Isaacson has pointed out that analyticity, according to Carnap, is relative to pragmatic constraints on theory. We can relinquish any kind of statement in the face of experience, but to relinquish L-valid truths is different from relinquishing empirical truths. The former, but not the latter, involve change of meaning. The one involves admitting falsehoods, the other change of concepts (D.Isaacson, ‘Carnap, Quine and Logical Truth’, in D.Bell and W.Vossenkuhl (eds), Subjectivity and Science (Akademie Verlag, Berlin, 1993), pp. 114–16.
11 Even in its pre-Tractatus, Russellian phase, analytic philosophy, though construed as cognitive and continuous with science, was committed to, indeed limited to, reductive and constructive analysis—and this too is repudiated by Quine.
12 For more detailed discussion of Wittgenstein’s objections to behaviourist accounts of meaning, see P.M.S.Hacker, Wittgenstein: Meaning and Mind, Volume 3 of An Analytical Commentary on the Philosophical Investigations (Blackwell, Oxford, 1990), in an essay entitled ‘Behaviour and Behaviourism’, §§2–3, from which the above remarks are derived.
13 See S.Shanker, ‘The Conflict between Wittgenstein and Quine on the Nature of Language and Cognition and its Implications for Constraint Theory’, in this volume, pp. 212–51.
14 If it were, then, inter alia, there would be no deferring to experts to explain the use of technical and quasi-technical terms (appeal to socio-linguistic surveys would suffice).
15 For detailed exposition of Wittgenstein’s normative (rule-governed) conception of language, see G.P.Baker and P.M.S.Hacker, Wittgenstein: Understanding and Meaning, Volume 1 of an Analytical Commentary on the Philosophical Investigations (Blackwell, Oxford, 1980), passim and Wittgenstein: Rules, Grammar and Necessity, Volume 2 of an Analytical Commentary on the Philosophical Investigations (Blackwell, Oxford, 1985), passim. For criticism of theoretical linguists’ failure to apprehend correctly the normative character of language and speech, see G.P.Baker and P.M.S.Hacker, Language, Sense and Nonsense (Blackwell, Oxford, 1984), chs 7–10. For criticism of Kripke’s reduction of rule-following practices of language use to social regularities of behaviour, see G.P.Baker and P.M.S.Hacker, Scepticism, Rules and Language (Blackwell, Oxford, 1984), ch. 2.
16 For lucid criticism of Quine’s view that speaking a language is not a normative practice and that invoking rules in philosophical elucidation of language and its features is explanatorily idle, see H.-J.Glock, ‘Wittgenstein vs. Quine on Logical Necessity’, in S.Teghrarian, Wittgenstein and Contemporary Philosophy (Thoemmes Press, Bristol, 1994), pp. 211–20.
17 Quine makes room for ostensive instruction, but interprets it causally rather than normatively, thus failing to distinguish ostensive training from ostensive teaching, ostensive definition and explanation of meaning (see below, p. 23).
18 It should be noted that not everything that is not normative (rule-governed) is conditioning. Innumerable purposive activities, skills and techniques, for example how to whistle tunes or tell jokes, are neither normative or theory-construction nor a matter of stimulus/response conditioning. They are typically open-ended and ‘plastic’—adaptable to indefinitely many circumstances.
19 Quine’s concept of stimulus meaning is allegedly an ersatz behaviourist concept of meaning, trimmed to the demands of rigorous science. The affirmative stimulus meaning of an observation sentence (for a speaker) is the class of all stimulations that would prompt his assent—stimulations taken as the impact of radiation, etc. on his sense receptors. This, he claims, ‘is a reasonable notion of meaning’ for such observation sentences as ‘Rabbit’ or ‘The tide is out’ (WO 44). But it is a notion of meaning that has broken all connection with what we understand by ‘meaning’. This will not disturb Quine, but may give pause to those who are less cavalier about our workaday concepts. (1) It violates the grammar of ‘meaning’: for some stimulus meanings are larger than others (since some classes are larger than others), some stimulus meanings include members which are exclusively sound waves (for example the stimulus meaning of ‘Noise!’), and some stimulus meanings consist exclusively of painful stimuli (for example ‘Hurts’, ‘Stings’, ‘Burns’). But the meaning of a oneword sentence cannot intelligibly be said to be larger than that of another, the meaning of the exclamation ‘Noise!’ (or of the sentence ‘There is a noise’) cannot be said to include sound waves among its members, and the meaning of ‘Hurts!’ or ‘Stings!’ does not include members that are painful or pleasurable stimuli—since the meaning of an expression is not a class of anything. On the other hand, the meanings of some sentences are hard to grasp, difficult to explain, impossible to render precisely in French—but classes of stimulations that prompt assent are neither easy nor difficult to grasp, cannot—in the relevant sense—be explained (since there is nothing in the semantic dimension to explain), and there is no rendering classes of stimuli in French. (2) It provides no standard by reference to which the use of an expression can be said to be correct or incorrect. The class of stimuli (construed in terms of surface irritations) that prompt one’s assent to ‘Gavagai’, let alone those that prompt another’s assent, is not only inaccessible (since, scientists apart, few speakers know anything about the character of surface irritations and their description) but also no standard of correct use. (3) It bears no connection to understanding an expression. For to understand an expression is to have mastered the technique of its use, and that is a normative skill, not a conditioned response.
If this is correct, it is far from obvious why Quine’s notion should be characterized as a concept of meaning (even an ersatz one) at all. Saccharine is ersatz sugar, but something that is neither sweet nor water soluble is not.
20 H.P.Grice and P.F.Strawson, ‘In Defense of a Dogma’, repr. in H.P. Grice, Studies m the Way of Words (Harvard University Press, Cambridge, Mass, and London, 1989), p. 207.
21 For more extensive discussion, see G.P.Baker and P.M.S.Hacker, Language, Sense and Nonsense (Blackwell, Oxford, 1984), pp. 211–18.
22 For a detailed discussion of Wittgenstein’s strategy in this matter, see G.P.Baker and P.M.S.Hacker, Wittgenstein: Rules, Grammar and Necessity, Volume 2 of an Analytical Commentary on the Philosophical Investigations (Blackwell, Oxford, 1985), in the chapter entitled ‘Grammar and Necessity’, §4, ‘The Psychology of the A Priori’.
23 For detailed discussion, see G.P.Baker and P.M.S.Hacker, Wittgenstein: Understanding and Meaning, Volume 1 of An Analytical Commentary on the Philosophical Investigations (Blackwell, Oxford, 1980), in a chapter entitled ‘Ostensive Definition and its Ramifications’.
24 Yet it is noteworthy that one can ostensively define directions of the compass. And one can ostensively define smells and sounds by reference to samples, even though one does not, strictly speaking, point at an object (see ‘Ostensive Definition and its Ramifications’, §2).
25 Even the Law of Non-Contradiction has the same status as all else. It is just that ‘without it we would be left making mutually contrary predictions indiscriminately, thus scoring a poor ratio of successes over failures’ (see Quine’s ‘Comment on Quinton’, in POQ 309).
26 See A.M.Quinton, The Nature of Things (Routledge and Kegan Paul, London, 1973), pp. 216f., and his ‘Doing without Meaning’, in R.Barrett and R.Gibson (eds), Perspectives on Quine (Blackwell, Oxford, 1990) p. 307, and Glock, ‘Wittgenstein vs. Quine on Logical Necessity’, pp. 210–11.
27 For detailed argument, see G.P.Baker and P.M.S.Hacker, Scepticism, Rules and Language (Blackwell, Oxford, 1984), pp. 92f.
28 It is curious that Quine should think the input ‘meagre’. What would it be like if it were richer? Even more ‘irradiations’, incessant noise and flashing of lights? In Word and Object, he wrote:
We have been reflecting in a general way on how surface irritations generate, through language, one’s knowledge of the world…. The voluminous and intricately structured talk that comes out bears little evident correspondence to the past and present barrage
of non-verbal stimulation; yet it is to such stimulation that we must look for whatever empirical content there must be. (WO 26)
This is equally curious. If the ‘input’ is to be described in terms of surface irritations, then the ‘output’ should be described in terms of bare physical movements and the generation of sound waves. If the output is to be described in terms of structured talk (and human action), then the input should be described in terms of what is perceived, the visible and audible, etc. environment, including the voluminous and intricately structured talk of our fellow human beings.
29 Assent to a sentence, according to Quine, is passing a verdict on its truth, which may be mistaken. The subject is held to believe what is uttered (UPM 48). The observation sentences which are the ‘entering wedge in the learning of language’ are vehicles of scientific evidence, verbalizing the predictions which check scientific theories (PTb 4f.). Consequently the concept of assent which he deploys is intimately interwoven with epistemic and intensional concepts. Invoking the principle of charity as a pragmatic guideline for translation makes this evident.
30 The primacy of behaviour viewed intentionalistically is a leitmotif of G. H. von Wright’s extensive writings on the explanation of human action, from Explanation and Understanding (Routledge and Kegan Paul, London, 1971) onwards.
31 For detailed discussion, see G.P.Baker and P.M.S.Hacker, Language, Sense and Nonsense (Blackwell, Oxford, 1984), ch. 8.
32 And one regarding which it would be difficult to argue, given the associated behaviour, that there is no fact of the matter about the translation of the term, no less than about the one-word sentence.