Strawson Meaning and Truth
Strawson Meaning and Truth
Strawson Meaning and Truth
P. F. STRAWSON
What is it for anything t o have a meaning at all, in the way, o r in the sense, in which words o r sentences o r signals have meaning? What is it for a particular sentence to have the meaning o r meanings it does have? What is it for a particular phrase, o r a particular word, to have the meaning o r meanings it does have? These are obviously connected questions. Any account we give of meaning in general (in the relevant sense) must square with the account we give of what it is for particular expressions to have particular meanings; and we must acknowledge, as two complementary truths, first, that the meaning of a sentence in general depends, in some systematic way, o n the meanings of the words that make it u p and, second, that for a word to have a particular meaning is a matter of its making a particular systematic contribution to the meanings of the sentences in which it occurs. I am not going t o undertake to try to answer these so obviously connected questions. That is not a task for one lecture; or for one man. I want rather to discuss a certain conflict, o r apparent conflict, more o r less dimly discernible in current approaches to these questions. For the sake of a label, we might call it the conflict between the theorists of communication-intention and the theorists of formal semantics. According to the former, it is impossible to give an adequate account of the
concept of meaning without reference to the possession by speakers of audience-directed intentions of a certain complex kind. The particular meanings of words and sentences are, no doubt, largely a matter of rule and convention; but the general nature of such rules and conventions can be ultimately understood only by reference to the concept of communication-intention. The opposed view, at least in its negative aspect, is that this doctrine simply gets things the wrong way round or the wrong way up, or mistakes the contingent for the essential. Of course we may expect a certain regularity of relationship between what people intend to communicate by uttering certain sentences and what those sentences conventionally mean. But the system of semantic and syntactical rules, in the mastery of which knowledge of a language consists-the rules which determine the meanings of sentences-is not a system of rules for communicating at all. The rules can be exploited for this purpose; but this is incidental to their essential character. It would be perfectly possible for someone to understand a language completely-to have a perfect linguistic competence-without having even the implicit thought of the function of communication; provided, of course, that the language in question did not contain words explicitly referring to this function. A struggle o n what seems to be such a
central issue in philosophy should have something of a Homeric quality; and a Homeric struggle calls for gods and heroes. I can at least, though tentatively, name some living captains and benevolent shades: on the one side, say, Grice, Austin, and the later Wittgenstein; on the other, Chomsky, Frege, and the earlier Wittgenstein. First, then, as to the theorists of communication-intention. The simplest, and most readily intelligible, though not the only way of joining their ranks is to present your general theory of meaning in two stages: first, present and elucidate a primitive concept of communication (or communication-intention) in terms which do not presuppose the concept of linguistic meaning; then show that the latter concept can be, and is to be, explained in terms of the former.1 For any theorist who follows this path, the fundamental concept in the theory of meaning is that of a speaker's, or, generally, an utterer's, meaning something by an audiencedirected utterance on a particular occasion. An utterance is something produced or executed by an utterer; it need not be vocal; it could be a gesture or a drawing or the moving or disposing of objects in a certain way. What an utterer means by his utterance is incidentally specified in specifying the complex intention with which he produces the utterance. The analysis of the kind of intention in question is too complex to be given in detail here, so I shall confine myself to incomplete description. An utterer might have, as one of his intentions in executing his utterance, that of bringing his audience to think that he, the utterer, believes some proposition, say the proposition that p; and he might intend this intention to be wholly overt, to be clearly recognized by the audience. Or again he might have the intention of bringing his audience to think that he, the utterer, wants his audience to perform some action, say a; and he might intend this intention of his to be wholly overt, to be clearly recognized by the audience. Then, provided certain other conditions on utterer's intention are fulfilled, the utterer may be said, in the relevant sense, to mean something by his utterance: specifically, to mean thatp, in the declarative mode, in the first case and to mean, in the imperative mode,
that the audience is to perform action a in the second case. Grice, for one, has given us reason to think that, with sufficient care, and far greater refinement than I have indicated, it is possible to expound such a concept of communication-intention or, as he calls it, utterer's meaning, which is proof against objection and which does not presuppose the notion of linguistic meaning. Now a word about how the analyis of linguistic meaning in terms of utterer's meaning is supposed to proceed. Here again I shall not go into details. The details would be very complex. But the fundamental idea is comparatively simple. We are accustomed, and reasonably, to think of linguistic meaning in terms of rules and conventions, semantic and syntactic. And when we consider the enormous elaboration of these rules and conventions-their capacity, as the modern linguists stress, to generate an infinite number of sentences in a given language-we may feel infinitely removed from the sort of primitive communication situation which we naturally think of when trying to understand the notion of utterer's meaning in terms which clearly do not presuppose linguistic meaning. But rules or conventions govern human practices and purposive human activities. So we should ask what purposive activities are governed by these conventions. What are these rules rules for doing? And the very simple thought I spoke of which underlies the suggested type of analysis is that these rules are, precisely, rules for communicating, rules by the observance of which the utterer may achieve his purpose, fulfil his communication-intention; and that this is their essential character. That is, it is not just a fortunate fact that these rules allow of use for this purpose; rather, the very nature of the rules concerned can be understood only if they are seen as rules whereby this purpose can be achieved. This simple thought may seem too simple, and in several ways. For it is clear that we can, and do, communicate very complicated things by the use of language; and if we are to think of language as, fundamentally, a system of rules for facilitating the achievement of our communication-intentions, and if
THE P H I L O S O P H Y OF LANGUAGE
the analysis is not to be circular, must we not credit ourselves with extremely complicated communication-intentions (or at least desires) independently of having at our disposal the linguistic means of fulfilling those desires? And is not this absurd? I think this is absurd. But the program of analysis does not require it. All that the analysis requires is that we can explain the notion of conventions of communication in terms of the notion of preconventional communication at a rather basic level. Given that we can do this, then there is more than one way in which we can start pulling ourselves up by our own linguistic boot-straps. And it looks as if we can explain the notion of conventions of communication in terms of the notion of pre-conventional communication at a rather basic level. We can, for example, tell ourselves a story of the analytic-genetic variety. Suppose an utterer achieves a pre-conventional communication success with a given audience by means of an utterance, say x. H e has a complex intention, vis-a-vis the audience of the sort which counts as a communicationintention and succeeds in fulfilling that intention by uttering x . Let us suppose that the primary intention was such that the utterer meant that p by uttering x; and, since, by hypothesis, he achieved a communicationsuccess, he was so understood by his audience. Now if the same communication-problem presents itself later to the same utterer in relation to the same audience, the fact, known to both of them, that the utterer meant that p by uttering x before, gives the utterer a reason for uttering x again and the audience a reason for interpreting the utterance in the same way as before. (The reason which each has is the knowledge that the other has the knowledge which he has.) So it is easy to see how the utterance of x could become established as between this utterer and this audience as a means of meaning that p. Because it has worked, it becomes established; and then it works because it is established. And it is easy to see how this story could be told so as to involve not just a group of two, but a wider group. So we can have a movement from an utterer pre-
conventionally meaning that p by an utterance of x to the utterance-type x conventionally meaning that p within a group and thence back to utterer-members of the group meaning that p by a token of the type, but now in accordance with the conventions. Now of course this explanation of conventional meaning in terms of utterer's meaning is not enough by itself. For it only covers the case, or only obviously covers the case, of utterance-types without structure-i.e. of utterance-types of which the meaning is not systematically derived from the meanings of their parts. But it is characteristic of linguistic utterance-types to have structure. The meaning of a sentence is a syntactic function of the meanings of its parts and their arrangement. But there is no reason in principle why a preconventional utterance should not have a certain complexity-a kind of complexity which allowed an utterer, having achieved one communication-success, to achieve another by repeating one part of the utterance while varying the other part, what he means on the second occasion having something in common with, and something which differentiates .it from, what he meant on the first occasion. And if he does thus achieve a second success, the way is open for a rudimentary system of utterance-types to become established, i.e., to become conventional within a group. A system of conventions can be modified to meet needs which we can scarcely imagine existing before the system existed. And its modification and enrichment may in turn create the possibility of thoughts such as we cannot understand what it would be for one to have, without supposing such modification and enrichment to have taken place. In this way we can picture a kind of alternating development. Primitive communication-intentions and successes give rise to the emergence of a limited conventional meaning-system, which makes possible its own enrichment and development which in turn makes possible the enlargement of thought and of communication-needs to a point at which there is once more pressure on the existing resources of language which is in turn responsive to such pressure. . . . And of course there is an element of mystery in this;
but so there is in human intellectual and social creativity anyway. All the foregoing is by way of the roughest possible sketch of some salient features of a communication-intention theory of meaning and of a hint as to how it might meet the obvious objection that certain communicationintentions presuppose the existence of language. It has all been said before, and with far greater refinement. But it will serve, I hope, as a sufficient basis for the confrontation of views that I wish to arrange. Now, then, for the at least apparently opposed view, which I have so far characterized only in its negative aspect. Of course the holders of this view share some ground with their opponents. Both agree that the meanings of the sentences of a language are largely determined by the semantic and syntactic rules or conventions of that language. Both agree that the members of any group or community of people who share knowledge of a language-who have a common linguistic competence-have at their disposal a more or less powerful instrument or means of communicating, and thereby of modifying each other's beliefs or attitudes or influencing each other's actions. Both agree that these means are regularly used in a quite conventional way, that what people intend to communicate by what they say is regularly related to the conventional meanings of the sentences they utter. Where they differ is as to the relations between the meaning-determining rules of the language, on the one hand, and the function of communication, on the other: one party insists, and the other (apparently) refuses to allow, that the general nature of those rules can be understood only by reference to this function. The refusal naturally prompts a question, viz. What is the general character of those rules which must in some sense have been mastered by anyone who speaks and understands a given language? The rejected answer grounds their general character in the social function of communicating, for example, beliefs or wishes or instructions. If this answer is rejected, another must be offered. So we ask again: What is the general character of these meaning-determining rules?
It seems to me that there is only one type of answer that has ever been seriously advanced or developed, or needs to be seriously considered, as providing a possible alternative to the thesis of the communication theorist. This is an answer which rests on the motion of truthconditions. The thought that the sense of a sentence is determined by its truth-conditions is to be found in Frege and in the early Wittgenstein, and we find it again in many subsequent writers. I take, as an example, a recent article by Professor Davidson. Davidson is rightly concerned with the point that an adequate account of the meaning rules for a language L will show how the meanings of sentences depend on the meanings of words in L; and a theory of meaning for L will do this, he says, if it contains a recursive definition of truth-in-L. The "obvious connection," he says, between such a definition of truth and the concept of meaning is this: "the definition works by giving the necessary and sufficient conditions for the truth of every sentence, and to give truth-conditions is a way of giving the meaning of a sentence. To know the semantic concept of truth for a language is to know what it is for a sentence-any sentence-to be true, and this amounts, in one good sense we can give to the phrdse, to understanding the lang~age."~ Davidson, in the article I quote from, has a limited concern. But the concern finds its place inside a more general idea; and the general idea, plainly enough, is that the syntactic and semantic rules together determine the meanings of all the sentences of a language and do this by means, precisely, of determining their truth-conditions. Now if we are to get at the root of the matter, to isolate the crucial issue, it seems to me important to set aside, at least initially, one class of objections to the adequacy of such a conception of meaning. I say one class of objections; but it is a class which admits of subdivisions. Thus it may be pointed out that there are some kinds of sentences-e.g., imperatives, optatives, and interrogatives-to which the notion of truth-conditions seems inappropriate, in that the conventional utterance of such sentences does not result in the
saying of anything true or false. Or again it may be pointed out that even sentences to which the notion of truth-conditions does seem appropriate may contain expressions which certainly make a difference to their conventional meaning, but not the sort of difference which can be explained in terms of their truth-conditions. Compare the sentence "Fortunately, Socrates is dead" with the sentence "Unfortunately, Socrates is dead." Compare a sentence of the form "p and q" with the corresponding sentence of the form "p but q." It is clear that the meanings of the members of each pair of sentences differ; it is far from clear that their truth-conditions differ. And there are not just one or two expressions which give rise to this problem, but many such expressions. Obviously both a comprehensive general theory of meaning and a comprehensive semantic theory for a particular language must be equipped to deal with these points. Yet they may reasonably be regarded as peripheral points. For it is a truth implicitly acknowledged by communication theorists themselves3 that in almost all the things we should count as sentences there is a substantial central core of meaning which is explicable either in terms of truth-conditions or in terms of some related notion quite simply derivable from that of a truth-condition, for example the notion, as we might call it, of a compliancecondition in the case of an imperative sentence or a fulfillment-condition in the case of an optative. If we suppose, therefore, that an account can be given of the notion of a truthcondition itself, an account which is indeed independent of reference to communicationintention, then we may reasonably think that the greater part of the task of a general theory of meaning has been accomplished without such reference. And by the same token, on the same supposition, we may think that the greater part of the particular theory of meaning of a particular language L can also be given, free of any such, even implicit, reference; for it can be given by systematically setting out the syntactic and semantical rules which determine truth-conditions for sentences of L.
Of course, as already admitted, something will have to be added to complete our general theory and to complete our particular theories. Thus for a particular theory an account will have to be added of the transformations that yield sentences with compliance-conditions or fulfillment-conditions out of sentences with truth-conditions; and the general theory will have to say what sort of thing, semantically speaking, such a derived sentence in general is. But this, though yielding a large harvest in sentences, is in itself a relatively small addition to either particular or general theory. Again, other additions will be necessary in connection with the other objections I mentioned. But, heartened by his hypothesized success into confidence, the theorist may reckon on dealing with some of these additions without essential reference to communication-intention; and, heartened by his hypothesized success into generosity, he may be happy to concede rights in some small outlying portion of the de facto territory of theoretical semantics to the theorist of communication-intention, instead of confining the latter entirely to some less appetizing territory called theoretical pragmatics. I hope it is now clear what the central issue is. It consists in nothing other than the simpleseeming question whether the notion of truthconditions can itself be explained or understood without reference to the function of communication. One minor clarification is called for before I turn to examine the question directly. I have freely used the phrase "the truth-conditions of sentences" and I have spoken of these truth-conditions as determined by the semantical and syntactical rules of the language to which the sentences belong. In such a context we naturally understand the word "sentence" in the sense of a 'type-sentence'. (By a sentence in the sense of a type I mean the sense in which there is just one English sentence, say, "I am feeling shivery," or just one English sentence, say, "She had her sixteenth birthday yesterday," which one and the same sentence may be uttered on countless different occasions by different people and with different references or applications.) But for many type-sentences, such as those just mentioned, the question
whether they, the sentences, are true or false is one that has no natural application: it is not the invariant type-sentences themselves that are naturally said to be true or false, but rather the systematically varying things that people say, the propositions they express, when they utter those sentences on different particular occasions. But if the notion of truthvalues is in general inappropriate to typesentences, how can the notion of truthconditions be appropriate? For presumably the truth-conditions of something are the conditions under which it is true. The difficulty, however, is quite easily resolved. All that needs to be said is that the statement of truth-conditions for many typesentences-perhaps most that are actually uttered in ordinary conversation-has to be, and can be, relativized in a systematic way to contextual conditions of utterance. A general statement of truth-conditions for such a sentence will then be, not a statement of conditions under which that sentence is a truth, but a general statement of a type of conditions under which different particular utterances of it will issue in different particular truths. And there are other more or less equivalent, though rather less natural, ways of resolving the difficulty. So now, at last, to the central issue. For the theorists of formal semantics, as I have called them, the whole weight, or most of the weight, both of a general theory of meaning and of particular semantic theories, falls on the notion of truth-conditions and hence on the notion of truth. We agree to let it rest there. But we still cannot be satisfied that we have an adequate general understanding of the notion of meaning unless we are satisfied that we have an adequate general understanding of the notion of truth. There is one maneuver here that would completely block all hope of achieving adequate understanding; and, if I am not mistaken, it is a maneuver which has a certain appeal for some theorists of formal semantics. This is to react to a demand for a general explication of the notion of truth by referring us back to a Tarski-like conception of truth-ina-given-language, L, a conception which is
elucidated precisely by a recursive statement of the rules which determine the truth-conditions for sentences of L. This amounts to a refusal to face the general philosophical question altogether. Having agreed to the general point that the meanings of the sentences of a language are determined, or largely determined, by rules which determine truth-conditions, we then raise the general question what sort of thing truth-conditions are, or what truth-conditions are conditions of; and we are told that the concept of truth for a given language is defined by the rules which determine the truth-conditions for sentences of that language. Evidently we cannot be satisfied with this. So we return to our general question about truth. And immediately we feel some embarrassment. For we have come to think there is very little to say about truth in general. But let us see what we can do with this very little. Here is one way of saying something uncontroversial and fairly general about truth. One who makes a statement or assertion makes a true statement if and only if things are as, in making that statement, he states them to be. Or again: one who expresses a supposition expresses a true supposition if and only if things are as, in expressing that supposition, he expressly supposes them to be. Now let us interweave with such innocuous remarks as these the agreed thoughts about meaning and truth-conditions. Then we have, first: the meaning of a sentence is determined by those rules which determine how things are stated to be by one who, in uttering the sentence, makes a statement; or, how things are expressly supposed to be by one who, in uttering the sentence, expresses a supposition. And then, remembering that the rules are relativized to contextual conditions, we can paraphrase as follows: the meaning of a sentence is determined by the rules which determine what statement is made by one who, in uttering the sentence in given conditions, makes a statement; or, which determine what supposition is expressed by one who, in uttering the sentence in given conditions, expresses a supposition; and so on. Thus we are led, by way of the notion of truth, back to the notion of the content of such
speech acts as stating, expressly supposing, and so on. And here the theorist of communication-intention sees his chance. There is no hope, he says, of elucidating the notion of the content of such speech acts without paying some attention to the notions of those speech acts themselves. Now of all the speech acts in which something true or false may, in one mode or another, be put forward, it is reasonable to regard that of statement or assertion as having an especially central position. (Hot for certainties, we value speculation primarily because we value information.) And we cannot, the theorist maintains, elucidate the notion of stating or asserting except in terms of audience-directed intention. For the fundamental case of stating or asserting, in terms of which all variants must be understood, is that of uttering a sentence with a certain intention-an intention wholly, overt in the sense required by the analysis of utterer's meaning-which can be incompletely described as that of letting an audience know, or getting it to think, that the speaker has a certain belief; as a result of which there may, or may not, be activated or produced in the audience that same belief. The rules determining the conventional.meaning of the sentence join with the contextual conditions of its utterance to determine what the belief in question is in such a primary and fundamental case. And in determining what the belief in question is in such a case, the rules determine what statement is made in such a case. To determine the former is to determine the latter. But this is precisely what we wanted. For when we set out from the agreed point that the rules which determine truth-conditions thereby determine meaning, the conclusion to which we were led was precisely that those rules determined what statement was made by one who, in uttering the sentence, made a statement. So the agreed point, so far from being an alternative to a communication theory of meaning, leads us straight in to such a theory of meaning. The conclusion may seem a little too swift. So let us see if there is any way of avoiding it. The general condition of avoiding it is clear. It is that we should be able to give an account of
the notion of truth-conditions which involves no essential reference to communicative speech acts. The alternative of refusing to give any account at all-of just resting on the notion of truth-conditions-is, as I have already indicated, simply not open to us if we are concerned with the philosophical elucidation of the notion of meaning: it would simply leave us with the concepts of meaning and truth each pointing blankly and unhelpfully at the other. Neither would it be helpful, though it might at this point be tempting, to retreat from the notion of truth-conditions to the less specific notion of correlation in general; to say, simply, that the rules which determine the meanings of sentences do so by correlating the sentences, envisaged as uttered in certain contextual conditions, with certain possible states of affairs. One reason why this will not do is that the notion of correlation in general is simply too unspecific. There are many kinds of behavior (including verbal behavior)-and many more kinds could be imagined-which are correlated by rule with possible states of affairs without its being the case that such correlation confers upon them the kind of relation to those possible states of affairs that we are concerned with. Another reason why it will not do is the following. Consider the sentence "I am tired." The rules which determine its meaning are indeed such as to correlate the sentence, envisaged as uttered by a particular speaker at a particular time, with the possible state of affairs of the speaker's being tired at that time. But this feature is not peculiar to that sentence or to the members of the class of sentences which have the same meaning as it. For consider the sentence "I am not tired." The rules which determine its meaning are also such as to correlate the sentence, envisaged as uttered by a certain speaker at a certain time, with the possible state of affairs of that speaker's being tired at that time. Of course the kinds of correlation are different. They are respectively such that one who uttered the first sentence would normally be understood as affirming, and one who uttered the second sentence would normally be understood as denying, that the state of affairs in
question obtained; or again they are such that one who utters the first sentence when the state of affairs in question obtains has made a true statement and one who utters the second sentence in these circumstances has made a false statement. But to invoke these differences would be precisely to give up the idea of employing only the unspecific notion of correlation in general. It is not worth labouring the point further. But it will readily be seen not only that sentences different, and even opposed, in meaning are correlated, in one way or another, with the same possible state of affairs, but also that one and the same unambiguous sentence is correlated, in one way or another, with very many different and in some cases mutually incompatible states of affairs. The sentence "I am tired" is correlated with the possible state of affairs of the speaker's being at the point of total exhaustion and also with the state of affairs of his being as fresh as a daisy. The sentence "I am over 40" is correlated with any possible state of affairs whatever regarding the speaker's age; the sentence "Swans are white" with any state of affairs whatever regarding the color of swans. The quite unspecific notion of correlation, then, is useless for the purpose in hand. It is necessary to find some way of specifying a particular correlation in each case, viz. the correlation of the sentence with the possible state of affairs the obtaining of which would be necessary and sufficient for something true to have been said in the uttering of the sentence under whatever contextual conditions are envisaged. So we are back once more with the notion of truth-conditions and with the question, whether we can give an account of this notion which involves no essential reference to communicative speech acts, i.e. to communication-intention. I can at this point see only one resource open, or apparently open, to the theorist of meaning who still holds that the notion of communication-intention has no essential place in the analysis of the concept of meaning. If he is not to swallow his opponent's hook, he must take some leaves out of his book. He sees now that he cannot stop with
the idea of truth. That idea leads straight to the idea of what is said, the content of what is said, when utterances are made; and that in turn to the question of what is being done when utterances are made. But may not the theorist go some way along this path without going as far along it as his opponent? Might it not be possible to delete the reference to communication-intention while preserving a reference to, say, belief-expression? And will not this, incidentally, be more realistic in so far as we often voice our thoughts to ourselves, with no communicative intention? The maneuver proposed merits a fuller description. It goes as follows. First, follow the communication-theorist in responding to the challenge for an elucidation of the notion of truth-conditions by invoking the notion of, e.g. and centrally, statement or assertion (accepting the uncontroversial point that one makes a true statement or assertion when things are as, in making that assertion, one asserts them to be). Second, follow the communication-theorist again in responding to the challenge for an elucidation of the notion of asserting by making a connection with the notion of belief (conceding that to make an assertion is, in the primary case, to give expression to a belief; to make a true assertion is to give expression to a correct belief; and a belief is correct when things are as one who holds that belief, in so far as he holds that belief, believes them to be). But third, part company with the communicationtheorist over the nature of this connection between assertion and belief; deny, that is, that the analysis of the notion of asserting involves essential reference to an intention, for example, to get an audience to think that the maker of the assertion holds the belief; deny that the analysis of the notion of asserting involves any kind of reference to audience-directed intention; maintain, on the contrary, that it is perfectly satisfactory to accept as fundamental here the notion of simply voicing or expressing a belief. Then conclude that the meaning-determining rules for a sentence of the language are the rules which determine what belief is conventionally articulated by one who, in given contextual
conditions, utters the sentence. As before, determining what this belief is, is the same thing as determining what assertion is made. So all the merits of the opponent's theory are preserved while the reference to communication is extruded. Of course, more must be said by this theorist, as by his opponent. For sentences which can be used to express beliefs need not always be so used. But the point is one to be made on both sides. So we may neglect it for the present. Now will this do? I do not think it will. But in order to see that it will not, we may have to struggle hard against a certain illusion. For the notion of expressing a belief may seem to us perfectly straightforward; and hence the notion of expressing a belief in accordance with certain conventions may seem equally straightforward. Yet, in so far as the notion of expressing a belief is the notion we need, it may borrow all its force and apparent straightforwardness from precisely the communication situation which it was supposed to free the analysis of meaning from depending on. We may be tempted to argue as follows. Often we express beliefs with an audience-directed intention; we intend that our audience should take us to have the belief we express and perhaps that that belief should be activated or produced in the audience as well. But then what could be plainer than this: that what we can do with an audience-directed intention we can also do without any such intention? That is to say, the audience-directed intention, when it is present, is something added on to the activity of expressing a belief and in no way essential to it-or to the concept of it. Now what a mixture of truth and falsity, of platitude and illusion, wq have here! Suppose we reconsider for a moment that analysis of utterer's meaning which was roughly sketched at the beginning. The utterer produces something-his utterance x-with a complex audience-directed intention, involving, say, getting the audience to think that he has a certain belief. We cannot detach or extract from the analysis an element which corresponds to his expressing a belief with no such intentionthough we could indeed produce the following
description and imagine a case for it: he acts as if he had such an intention though as a matter of fact he has not. But here the description depends on the description of the case in which he has such an intention. What I am suggesting is that we may be tempted, here as elsewhere, by a kind of bogus arithmetic of concepts. Given the concept of Audience Directed Belief Expression (ADBE), we can indeed think of Belief Expression (BE) without Audience Direction (AD), and find cases of this. But it does not follow that the concept of ADBE is a kind of logical compound of the two simpler concepts of AD and BE and hence that BE is conceptually independent of ADBE. Of course these remarks do not show that there is no such thing as an independent concept of belief-expression which will meet the needs of the anti-communication theorist. They are only remarks directed against a too simple argument to the effect that there is such a concept. This much is clear. If there is such an essentially independent concept of beliefexpression which is to meet the needs of the analysis of the notion of meaning, we cannot just stop with the phrase 'expressing a belief'. We must be able to give some account of this concept, to tell ourselves some intelligible story about it. We can sometimes reasonably talk of a man's actions or his behavior as expressing a belief when, for example, we see those actions as directed towards an end or goal which it is plausible to ascribe to him in so far as it js also plausible to ascribe to him that belief. But this reflection by itself does not get us very far. For one thing, on the present progrdm, we are debarred from making reference to the end or goal of communication an essential part of our story. For another, the sort of behavior we are to be concerned with must be, or be capable of being, formalized or conventionalized in such a way that it can be regarded as subjected to, or performed in observance of, rules; and of rules, moreover, which regulate the behavior precisely in its aspect as expression of belief. It will not do to say simply: we might suppose a man to find some satisfaction (unspecified) or some point
(unspecified) in performing certain formalized (perhaps vocal) actions on some occasions, these actions being systematically related to his having certain beliefs. For suppose a man had a practice of vocalizing in a certain way whenever he saw the sun rise and in another, partly similar, partly different, way whenever he saw it set. Then this practice would be regularly related to certain beliefs, i.e. that the sun was rising or that it was setting. But this description gives us no reason at all for saying that when the man indulged in this practice he was expressing the belief that the sun was rising or setting, in accordance with a rule for doing so. We really have not enough of a description to know what to say. As far as we could tell, we might say, he just seems to have this ritual of saluting the rising' or the setting sun in this way. What need of his it satisfies we don't know. Let us suppose, however-for the sake of the argument-that we can elaborate some relevant conception of expressing a belief which presupposes nothing which, on the present program, we are debarred from presupposing; and that we draw on this concept of expressing a belief in order to give an account, o r analysis, on the lines indicated, of the notion of linguistic meaning. Then an interesting consequence ensues. That is, it will appear as a quite contingent truth about language that the rules or conventions which determine the meanings of the sentences of a language are public or social rules or conventions. This will be, as it were, a natural fact, a fact of nature, in no way essential to the concept of a language, and calling for a natural explanation which must not be allowed to touch o r modify that concept. There must be nothing in the concept to rule out the idea that every individual might have his own language which only,he understands. But then one might ask: Why should each individual observe his own rules? or any rules? Why shouldn't he express any belief he likes in any way he happens to fancy when he happens to have the urge to express it? There is one answer at least which the theorist is debarred from giving to this question, if only in the interests
of his own programme. He cannot say: Well, a man might wish to record his beliefs so that he could refer to the records later, and then he would find it convenient to have rules to interpret his own records. The theorist is debarred from giving this answer because it introduces, though in an attenuated form, the concept of communication-intention: the earlier man communicates with his later self. There might be one way of stilling the doubts which arise so rapidly along this path. That would be to offer possible natural explanations of the supposed natural fact that language is public, that linguistic rules are more or less socially common rules; explanations which successfully avoided any suggestion that the connection of public rules with communication was anything but incidental and contingent. How might such an explanation go? We might say that it was an agreed point that the possession of a language enlarges the mind, that there are beliefs one could not express without a language to express them in, thoughts one could not entertain without a rule-governed system of expressions for articulating them. And it is a fact about human beings that they simply would not acquire mastery of such a system unless they were exposed, as children, to conditioning or training by adult members of a community. Without concerning ourselves about the remote origins of language, then, we may suppose the adult members of a community to wish their successors to have this mind-enlarging instrument at their disposal-and evidently the whole procedure of training will be simplified if they all teach the same, the common language. We may reasonably suppose that the learners, to begin with, do not quite appreciate what they will ultimately be doing with language; that it is for them, to begin with, a matter of learning to do the right thing rather than learning to say the true thing, i.e. a matter of responding vocally to situations in a way which will earn them reward or avoid punishment rather than a matter of expressing their beliefs. But later they come to realize that they have mastered a system which enables them to perform this (still unexplained) activity whenever they wish to; and then they are speaking a language.
Of course it must be admitted that in the process they are liable also to acquire the secondary skill of communicating their beliefs. But this is simply something added on, an extra and conceptually uncovenanted benefit, quite incidental to the description of what it is to have mastered the meaning-rules of the language. If, indeed, you pointedly direct utterances, of which the essential function is belief-expression, to another member of the community, he will be apt to take it that you hold whatever beliefs are in question and indeed that you intend him to take this to be so; and this fact may give rise, indeed, it must be admitted, does give rise, to a whole cluster of social consequences; and opens up all sorts of possibilities of kinds of linguistic communication other than that which is based on beliefexpression. This is why, as already acknowledged, we may have ultimately to allow some essential reference to communication-intention into outlying portions of our semantic theory. But this risk is incurred only when we go beyond the central core of meaning, determined by the rules which determine truthconditions. As far as the central core is concerned, the function of communication remains secondary, derivative, conceptually inessential. I hope it is clear that any such story is going to be too perverse and arbitrary to satisfy the requirements of an acceptable theory. If this is the way the game has to be played, then the communication theorist must be allowed to have won it. But must the game, finally, be played in this way? I think, finally, it must. It is indeed a generally harmless and salutary thing to say that to know the meaning of a sentence is to know under what conditions one who utters it says something true. But if we wish for a philosophical elucidation of the concept of meaning, then the dictum represents, not the end, but the beginning, of our task. It simply narrows, and relocates, our problem, forcing us to inquire what is contained in the little phrase ". . . says something true." Of course there are many ways in which one can say something which is in fact true, give expression, if you like, to a true proposition, without
thereby expressing belief in it, without asserting that proposition: for example when the words in question form certain sorts of subordinate or coordinate clauses, and when one is quoting or playacting and so on. But when we come to try to explain in general what it is to say something true, to express a true proposition, reference to belief or to assertion (and thereby to belief) is inescapable. Thus we may harmlessly venture: Someone says something true if things are as he says they are. But this "says" already has the force of "asserts." Or, to eschew the "says" which equals "asserts," we may harmlessly venture: Someone propounds, in some mode or other, a true proposition if things are as anyone who believed what he propounds would thereby believe them to be. And here the reference to belief is explicit. Reference, direct or indirect, to beliefexpression is inseparable from the analysis of saying something true (or false). And, as I have tried to show, it is unrealistic to the point of unintelligibility-or, at least, of extreme perversity-to try to free the notion of the linguistic expression of belief from all essential connection with the concept of communication-intention. Earlier I hinted that the habit of some philosophers of speaking as if "true" were a predicate of type-sentences was only a minor aberration, which could readily enough be accommodated to the facts. And so it can. But it is not a simple matter of pedantry to insist on correcting the aberration. For if we are not careful, it is liable to lead us totally wrong. It is liable, when we inquire into the nature of meaning, to make us forget what sentences are for. We connect meaning with truth and truth, too simply, with sentences; and sentences belong to language. But, as theorists, we know nothing of human language unless we understand human speech.
NOTES
1. Not the only way; for to say that a concept cannot be adequately elucidated without reference to a concept I,is not the same thing as to l say that it is possible to give a classical analysis
115
of 4 in terms of I. But the simplest way; for / the classical method of analysis is that in terms of which, in our tradition, we most naturally think. 2. "Truth and Meaning," Synthese, 1967, p. 310. My italics. [Reprinted in this volume.] 3. This acknowledgement is probably implicit, though not very clearly so, in Austin's concept of locutionary meaning (see How to do things with
Words, Oxford: 1962); it is certainly implicit in Grice's distinction between what speakers actually say, in a favored sense of 'say', and what they imply (see "Utterer's Meaning, SentenceMeaning and Word-Meaning," in Foundations of Language, 1968); and again in Searle's distinction between the proposition put forward and the illocutionary mode in which it is put forward (see Speech Acts, Cambridge: 1969).
Grice, H. P., Meaning Revisited, in Mutual Knowledge, ed. N. V. Smith (New York: Academic Press, 1982), pp. 223-243.
Grice, H. P.. Utterer's Meaning and Intentions, Philosophical Review 78 (1969), 147-177. Grice, H. P., Utterer's Meaning, Sentence-Meaning and Word-Meaning, Foundations of Language 4 (1968), 1-18. Kempson, Ruth, Semantic Theory (Cambridge: Cambridge University Press, 1977). Linsky, Leonard, Semantics and the Philosophy of Language (Urbana, Ill.: University of Illinois Press, 1952). McDowell, John, Meaning, Communication, and Knowledge, in Philosophical Subjects, ed. Zak Van Straaten (Oxford: Clarendon Press, 1980), pp. 117-139. Platts, Mark, Ways of Meaning (London: Routledge & Kegan Paul, 1979). Quine, W. V., Cognitive Meaning, Monist 62 (1979), 129-142. Searle, John, The Background of Meaning, in Speech Act Theory and Pragmatics, ed. John R. Searle and Ferenc Kiefer (Dordrecht, Holland: D. Reidel, 1980), pp. 221-232. Shiffer, Stephen, Meaning (Oxford: Clarendon Press, 1972). Schiffer, Stephen, Remnants of Meaning (Cambridge: MIT Press, 1987). Strawson, P. F., Truth, Analysis 9 (1949), 83-97. Tarski, Alfred, The Concept of Truth in Formalized Languages, in Logic, Semantics and Metamathematics (Oxford: Clarendon Press, 1956), pp. 152-278. Ziff, Paul, On H. P. Grice's Account of Meaning, Analysis 28 (1967), 1-8.