What is the connection between thought and language? The dependence of speaking on thinking is evident, for to speak is to express thoughts. This dependence is manifest in endless further ways. Someone who utters the sentence "The candle is out" as a sentence of English must intend to utter words that are true if and only if an indicated candle is out at the time of utterance, and he must believe that by making the sounds he does he is uttering words that arc true only under those circumstances. These intentions and beliefs are not apt to be dwelt on by the fluent speaker. But though they may not normally command attention, their absence would be enough to show he was not speaking English, and the absence of any analogous thoughts would show he was not speaking at all.
The issue lies on the other side: Can there be thought without speech? A first and natural reaction is that there can be. There is the familiar, irksome experience of not being able to find the words to express one's ideas. On occasion one may decide that the editorial writer has put a point better than one could oneself. And there is Norman Malcolm's dog who having chased a squirrel into the woods, barks up the wrong tree.' It is not hard to credit the dog with the belief that the squirrel is in that tree.
A definite, if feebler, intuition tilts the other way. It is possible to wonder whether the speaker who can't find the right words has a clear idea. Attributions of intentions and beliefs to dogs smack of anthropomorphism. A primitive behaviorism, baffled by the privacy of unspoken thoughts, may take comfort in the view that thinking is really "talking to oneself" – silent speech.
Beneath the surface of these opposed tendencies run strong, if turgid, currents, which may help to explain why philosophers have, for the most part, preferred taking a stand on the issue to producing an argument. Whatever the reason, the question of the relationship between thought and speech seems seldom to have been asked for its own sake. The usual. assumption is that one or the other, speech or thought, is by comparison easy to understand, and therefore the more obscure one (whichever that is) may be illuminated by analyzing or explaining it in terms of the other.
The assumption is, I think, false: neither language nor thinking can be fully explained in terms of the other, and neither has conceptual priority. The two are, indeed, linked, in the sense that each requires the other in order to be understood; but the linkage is not so complete that either suffices, even when reasonably reinforced, to explicate the other. To make good this claim, what is chiefly needed is to show how thought depends on speech, and this is the thesis I want to refine, and then to argue for.
Thought and Talk 233
We attribute a thought to a creature whenever we assertively employ a positive sentence the main verb of which is psychological - in English, "believes," "knows," "hopes," "desires," "thinks," "fears" are examples - followed by a sentence and preceded by the name or desription of the creature. (A "that" may optionally or necessarily follow the verb.) Some such sentences attribute states, others report events or processes: "believes," "thinks," and nts" report states, while "came to believe," "forgot," "concluded," "noticed," "is prov-' report events or processes. Sentences that can be used to attribute a thought exhibit lat is often called, or analyzed as, semantic intensionality, which means that the attribution My be changed from true to false, or false to true, by substitutions in the contained tntences that would not alter the truth value of the sentence in isolation. ' I do not take for granted that if a creature has a thought, then we can, with resources of 1C kind just sketched, correctly attribute that thought to him. But thoughts so attributable Fit least constitute a good sample of the totality.
It is doubtful whether the various sorts of thought can be reduced to one, or even to a few: tare, knowledge, belief, fear, interest, to name some important cases, are probably logically iWependent to the extent that none can be defined using the others, even along with such Kfarther notions as truth and cause. Nevertheless, belief is central to all kinds of thought. If BWneone is glad that, or notices that, or remembers that, or knows that, the gun is loaded, i*4en he must believe that the gun is loaded. Even to wonder whether the gun is loaded, or Bspeculate on the possibility that the gun is loaded, requires the belief, for example, that a on is a weapon, that it is a more or less enduring physical object, and so on. There are good sons for not insisting on any particular list of beliefs that are needed if a creature is to ider whether a gun is loaded. Nevertheless, it is necessary that there be endless inter-d beliefs. The system of such beliefs identifies a thought by locating it in a logical and mic space.
jfcHaving a thought requires that there be a background of beliefs, but having a particular KWght does not depend on the state of belief with respect to that very thought. If I consider j to a certain concert, I know I will be put to a degree of trouble and expense, and I have : complicated beliefs about the enjoyment I will experience. I will enjoy hearing loven's "Grosse Fugc," say, but only provided the performance achieves a reasonable lard, and I am able to remain attentive. I have the thought of going to the concert, but 11 decide whether to go, I have no fixed belief that I will go; until that time, I merely irtain the thought.
p We may say, summarizing the last two paragraphs, that a thought is defined by a system
|fbeliefs, but is itself autonomous with respect to belief.
p'We usually think that having a language consists largely in being able to speak, but in
Mut follows speaking will play only an indirect part. What is essential to my argument is the a of an interpreter, someone who understands the utterances of another. The considera-is to be put forward imply, I think, that a speaker must himself be an interpreter of others, 11 shall not try to demonstrate that an interpreter must be a speaker, though there may be od reason to hold this. Perhaps it is worth pointing out that the notion of a language, or of > people speaking the same language, does not seem to be needed here. Two speakers lid interpret each other's utterances without there being, in any ordinary sense, a common iiage. (I do not want to deny that in other contexts the notion of a shared language may
IP The chief thesis of this paper is that a creature cannot have thoughts unless it is an Bterpreter of the speech of another. This thesis does not imply the possibility of reduction, uvioristic or otherwise, of thought to speech; indeed the thesis imputes no priority to
language, epistemological or conceptual. The claim also falls short of similar claims in I it allows that there may be thoughts for which the speaker cannot find words, or for \ there are no words.
Someone who can interpret an utterance of the English sentence "The gun is •»•
must have many beliefs, and these beliefs must be much like the beliefs someone b
have if he entertains the thought that the gun is loaded. The interpreter must, we I
suppose, believe that a gun is a weapon, and that it is a more or less enduring physical <
There is probably no definite list of things that must be believed by someone who it
stands the sentence "The gun is loaded," but it is necessary that there be endless interll
An interpreter knows the conditions under which utterances of sentences are true,! often knows that if certain sentences are true, other must be. For example, an interpr English knows that if "The gun is loaded and the door is locked" is true, then "The (i locked" is true. The sentences of a language have a location in the logical space created bjj pattern of such relationships. Obviously the pattern of relations between sentences BJ[ much like the pattern of relations between thoughts. This fact has encouraged the \ it is redundant to take both patterns as basic. If thoughts are primary, a language Decani serve no purpose but to express or convey thoughts; while if we take speech as primaryja tempting to analyze thoughts as speech dispositions: as Sellars puts it, "thinking ttj distinctly human level... is essentially verbal activity."2 But clearly the parallel betweei structure of thoughts and the structure of sentences provides no argument for the prill of either, and only a presumption in favor of their interdependence.
We have been talking freely of thoughts, beliefs, meanings, and interpretations; or la freely using sentences that contain these words. But of course it is not clear what entitifl sorts of entity, there must be to make systematic sense of such sentences. However,! apparently of thoughts and sayings does belong to a familiar mode of explanation of hu behavior and must be considered an organized department of common sense which I well be called a theory. One way of examining the relation between thought and lang by inspecting the theory implicit in this sort of explanation.
Part of the theory deals with the teleological explanation of action. We wonder whyai raises his arm; an explanation might be that he wanted to attract the attention of a fri This explanation would fail if the arm-raiser didn't believe that by raising his arm he ww attract the attention of his friend, so the complete explanation of his raising his arm, orati rate a more complete explanation, is that he wanted to attract the attention of his friend * believed that by raising his arm he would attract his friend's attention. Explanation oft familiar kind has some features worth emphasizing. It explains what is relatively apparer an arm-raising — by appeal to factors that are far more problematical: desires and beliefs. ] if we were to ask for evidence that the explanation is correct, this evidence would in thet consist of more data concerning the sort of event being explained, namely further behavn which is explained by the postulated beliefs and desires. Adverting to beliefs and desire* I explain action is therefore a way of fitting an action into a pattern of behavior made cohere by the theory. This does not mean, of course, that beliefs are nothing but patterns i behavior, or that the relevant patterns can be defined without using the concepts of belief • desire. Nevertheless, there is a clear sense in which attributions of belief and desire, i hence teleological explanations of belief and desire, are supervenient on behavior broadly described.
A characteristic of teleological explanation not shared by explanation generally is thei in which it appeals to the concept of reason. The belief and desire that explain an action i
»vil that anyone who had that belief and desire would have a reason to act in that way. pit's more, the descriptions we provide of desire and belief must, in ideological explana-gexhibit the rationality of the action in the light of the content of the belief and the object t desire.
c cogency of a teleological explanation rests, as remarked, on its ability to discover a aviit pattern in the behavior of an agent. Coherence here includes the idea of rationality I in the sense that the action to be explained must be reasonable in the light of the 1 desires and beliefs, and also in the sense that the assigned desires and beliefs must ui one another. The methodological presumption of rationality does not make it Bible to attribute irrational thoughts and actions to an agent, but it does impose a .01 on such attributions. We weaken the intelligibility of attributions of thoughts of any I to the extent that we fail to uncover a consistent pattern of beliefs and, finally, of Ms, for it is only against a background of such a pattern that we can identify thoughts, tsee a man pulling on both ends of a piece of string, we may decide he is fighting against elf, that he wants to move the string in incompatible directions. Such an explanation I require elaborate backing. No problem arises if the explanation is that he wants to It the string.
om the point of view of someone giving teleological explanations of the actions of ner, it clearly makes no sense to assign priority either to desires or to beliefs. Both are initial to the explanation of behavior, and neither is more directly open to observation than I Other. This creates a problem, for it means that behavior, which is the main evidential I for attributions of belief and desire, is reckoned the result of two forces less open to ic observation. Thus where one constellation of beliefs and desires will rationalize an n, it is always possible to find a quite different constellation that will do as well. Even a this sample of actions threatens to leave open an unacceptably large number of alterna-I explanations.
Fortunately a more refined theory is available, one still firmly based on common sense: |theory of preference, or decision making under uncertainty. The theory was first ride precise by Frank Ramsey, though he viewed it as a matter of prov iding a foundation IT the concept of probability rather than as a piece of philosophical psychology.' msey's theory works by quantifying strength of preference and degree of belief in such a jras to make sense of the natural idea that in choosing a course of action we consider not y how desirable various outcomes are, but also how apt available courses of action are to induce those outcomes. The theory does not assume that we can judge degrees of belief (make numerical comparisons of value directly. Rather it postulates a reasonable pattern ol ferences between courses of action, and shows how to construct a system of quantified iefs and desires to explain the choices. Given the idealized conditions postulated by : theory, Ramsey's method makes it possible to identify the relevant beliefs and :s uniquely. Instead of talking of postulation, we might put the matter this way: to the t that we can see the actions of an agent as falling into a consistent (rational) pattern of ain sort, we can explain those actions in terms of a system of quantified beliefs and wres.
I We shall come back to decision theory presently; now it is time to turn to the question of 1 speech is interpreted. The immediate aim of a theory of interpretation is to give the ming of an arbitrary utterance by a member of a language community. Central to rpretation, I have argued, is a theory of truth that satisfies Tarski's Convention T Mudified in certain ways to apply to a natural language). Such a theory may be taken as Wng an interpretation of each sentence a speaker might utter. To belong to a speech
Thought and Talk
community - to be an interpreter of the speech of others - one needs, in effect, to know s a theory, and to know that it is a theory of the right kind.4
A theory of interpretation, like a theory of action, allows us to redescribe certain evei in a revealing way. Just as a theory of action can answer the question of what an agent is doin when he has raised his arm by redescribing the act as one of trying to catch his friendV| attention, so a method of interpretation can lead to redescribing the utterance of < sounds as an act of saying that snow is white. At this point, however, the analogy bream down. For decision theory can also explain actions, while it is not at all clear how a theoryc interpretation can explain a speaker's uttering the words "Snow is white." But this is, afl all, to be expected, for uttering words is an action, and so must draw for its teleolo explanation on beliefs and desires. Interpretation is not irrelevant to the teleological exp nation of speech, since to explain why someone said something we need to know, amo other things, his own interpretation of what he said, that is, what he believes his words meaa| in the circumstances under which he speaks. Naturally this will involve some of his belid about how others will interpret his words.
The interlocking of the theory of action with interpretation will emerge in another waytfjj we ask how a method of interpretation is tested. In the end, the answer must be that it hef bring order into our understanding of behavior. But at an intermediate stage, we can see t the attitude of holding true or accepting as true, as directed toward sentences, must play I central role in giving form to a theory. On the one hand, most uses of language tell I directly, or shed light on the question, whether a speaker holds a sentence to be true. If» speaker's purpose is to give information, or to make an honest assertion, then normally I speaker believes he is uttering a sentence true under the circumstances. If he utters I command, we may usually take this as showing that he holds a certain sentence (clo related to the sentence uttered) to be false; similarly for many cases of deceit. When,| question is asked, it generally indicates that the questioner does not know whether a < sentence is true; and so on. In order to infer from such evidence that a speaker holdil sentence true we need to know much about his desires and beliefs, but we do not have I know what his words mean.
On the other hand, knowledge of the circumstances under which someone holds ! tences true is central to interpretation. We saw in the case of thoughts that although I thoughts are not beliefs, it is the pattern of belief that allows us to identify any thou analogously, in the case of language, although most utterances are not concerned with I it is the pattern of sentences held true that gives sentences their meaning.
The attitude of holding a sentence to be true (under specified conditions) relates I and interpretation in a fundamental way. We can know that a speaker holds a sentence to I true without knowing what he means by it or what belief it expresses for him. But if we 1 he holds the sentence true and we know how to interpret it, then we can make a i attribution of belief. Symmetrically, if we know what belief a sentence held true expn we know how to interpret it. The methodological problem of interpretation is to i how, given the sentences a man accepts as true under given circumstances, to work out \ his beliefs are and what his words mean. The situation is again similar to the situation fl| decision theory where, given a man's preferences between alternative courses of action, i can discern both his beliefs and his desires. Of course it should not be thought that a th of interpretation will stand alone, for as we noticed, there is no chance of telling when I sentence is held true without being able to attribute desires and being able to describe| actions as having complex intentions. This observation does not deprive the theory <
Tpretation of interest, but assigns it a place within a more comprehensive theory of action I thought.5
It is still unclear whether interpretation is required for a theory of action, which is the ition we set ourselves to answer. What is certain is that all the standard ways of testing ones of decision or preference under uncertainty rely on the use of language. It is Ltively simple to eliminate the necessity for verbal responses on the part of the subject: he i be taken to have expressed a preference by taking action, by moving directly to achieve t end, rather than by saying what he wants. But this cannot settle the question of what he i chosen. A man who takes an apple rather than a pear when offered both may be pressing a preference for what is on his left rather than his right, what is red rather than ow, what is seen first, or judged more expensive. Repeated tests may make some readings This actions more plausible than others, but the problem will remain how to determine i he judges two objects of choice to be identical. Tests that involve uncertain events -oices between gambles — are even harder to present without using words. The psycholo-it, skeptical of his ability to be certain how a subject is interpreting his instructions, must I a theory of verbal interpretation to the theory to be tested. If we think of all choices as aling a preference that one sentence rather than another be true, the resulting total ory should provide an interpretation of sentences, and at the same time assign beliefs and sires, both of the latter conceived as relating the agent to sentences or utterances. This nposite theory would explain all behavior, verbal and otherwise.
All this strongly suggests that the attribution of desires and beliefs (and other thoughts) ist go hand in hand with the interpretation that of speech, that neither the theory of cision nor that of interpretation can be successfully developed without the other. Ut it remains to say, in more convincing detail, why the attribution of thought depends on interpretation of speech. The general, and not very informative, reason is that out speech we cannot make the fine distinctions between thoughts that are essential to explanations we can sometimes confidently supply. Our manner of attributing ttitudes ensures that all the expressive power of language can be used to make such itinctions. One can believe that Scott is not the author of Waverley while not doubting at Scott is Scott; one can want to be the discoverer of a creature with a heart without nting to be the discoverer of a creature with a kidney. One can intend to bite into the apple I the hand without intending to bite into the only apple with a worm in it; and so forth. : intensionality we make so much of in the attribution of thoughts is very hard to make of when speech is not present. The dog, we say, knows that its master is home, tdoes it know that Mr Smith (who is his master), or that the president of the bank (who I that same master), is home? We have no real idea how to settle, or make sense of, these stions. It is much harder to say, when speech is not present, how to distinguish universal ughts from conjunctions of thoughts, or how to attribute conditional thoughts, or lights with, so to speak, mixed quantification ("He hopes that everyone is loved by
? These considerations will probably be less persuasive to dog lovers than to others, but in fly case they do not constitute an argument. At best what we have shown, or claimed, is that i there is behavior that can be interpreted as speech, the evidence will not be adequate ) justify the fine distinctions we are used to making in the attribution of thoughts. If we sist in attributing desires, beliefs, or other attitudes under these conditions, our attribu-s and consequent explanations of actions will be seriously underdetermined in that many trnative systems of attribution, many alternative explanations, will be equally justified by
the available date. Perhaps this is all we can say against the attribution of thoughts to du creatures; but I do not think so.
Before going on I want to consider a possible objection to the general line I have bccq
pursuing. Suppose we grant, the objector says, that very complex behavior not observed a
infants and elephants is necessary if we are to find application for the full apparatus availab
for the attribution of thoughts. Still, it may be said, the sketch of how interpretation worlij
does not show that this complexity must be viewed as connected with language. The reason!
is that the sketch makes too much depend on the special attitude of being thought true.'
most direct evidence for the existence of this attitude is honest assertion. But then it wo
seem that we could treat as speech the behavior of creatures that never did anything w
language except make honest assertions. Some philosophers do dream of such dreary triba;
but would we be right to say they had a language? What has been lost to view is what may
be called the autonomy of meaning. Once a sentence is understood, an utterance of it may I
used to serve almost any extralinguistic purpose. An instrument that could be put to only on
use would lack autonomy of meaning; this amounts to saying it should not be counted as I
language. So the complexity of behavior needed to give full scope to attributions of thoug
need not, after all, have exactly the same complexity that allows, or requires, interpretati
as a language. .;
I agree with the hypothetical objector that autonomy of meaning is essential to language; indeed it is largely this that explains why linguistic meaning cannot be defined or analyzed on the basis of extra linguistic intentions and beliefs. But the objector fails to distinguish between a language that could be used for only one purpose and one that is used for only one purpose. An instrument that could be used for only one purpose would not be language, But honest assertion alone might yield a theory of interpretation, and so a language that, though capable of more, might never be put to further uses. (As a practical matter, the event is unthinkable. Someone who knows under what conditions his sentences are socially true cannot fail to grasp, and avail himself of, the possibilities for dishonest assertion or for joking, story-telling, goading, exaggerating, insulting, and all the rest of the jolly" crew.)
A method of interpretation tells us that for speakers of English an utterance of "It is raining" by a speaker x at time / is true if and only if it is raining (near x) at /. To be armed with this information, and to know that others know it, is to know what an utterance means independently of knowing the purposes that prompted it. The autonomy of meaning also helps to explain how it is possible, by the use of language, to attribute thoughts. Suppose someone utters assertively the sentence "Snow is white." Knowing the conditions under which such an utterance is true I can add, if I please, "1 believe that too," thus attributing a belief to myself. In this case we may both have asserted that snow is white, but sameness of force is not necessary to the self-attribution. The other may say with a sneer, expressing disbelief, "Snow is white" - and I may again attribute a belief to myself by saying, "But/ believe that." It can work as well in another way: If I can take advantage of an utterance of someone else's to attribute a belief to myself, I can use an utterance of my own to attribute a belief to someone else. First I utter a sentence, perhaps "Snow is white," and then I add "He believes that." The first utterance may or may not be an assertion; in any case, it does not attribute a belief to anyone (though if it is an assertion, then I do represent myself as believing that snow is white). But if my remark "He believes that" is an assertion, I have attributed a belief to someone else. Finally, there is no bar to my attributing a belief to myself by saying first, "Snow is white" and then adding, "I believe that."
In all these examples, I take the word "that" to refer demonstratively to an utterance,
ler it is an utterance by the speaker of the "that" or by another speaker. The "that" lot refer to a sentence, both because, as Church has pointed out in similar cases, the :nce would then have to be relativized to a language, since a sentence may have different mings in different languages;6 but also, and more obviously, because the same sentence ly have different truth values in the same language.
What demonstrative reference to utterances does in the sort of case just considered it can as well when the surface structure is altered to something like "I believe that snow is te" or "He believes that snow is white." In these instances also I think w;e should view the it" as a demonstrative, now referring ahead to an utterance on the verge of production, lus the logical form of standard attributions of attitude is that of two utterances paratac-lly joined. There is no connective, though the first utterance contains a reference to the )nd. (Similar remarks go, of course, for inscriptions of sentences.) I have discussed this analysis of verbal attributions of attitude elsewhere, and there is no to repeat the arguments and explanations here.7 It is an analysis with its own difficul-s, especially when it comes to analyzing quantification into the contained sentence, but I these difficulties can be overcome while preserving the appealing features of the idea. I want to stress a point that connects the paratactic analysis of attributions of attitude nth our present theme. The proposed analysis directly relates the autonomous feature of meaning with our ability to describe and attribute thoughts, since it is only because (he interpretation of a sentence is independent ot its use that the utterance of a sentence can lerve in the description of the attitudes of others. If my analysis is right, we can dispense with the unlikely (but common) view that a sentence bracketed into a "that"-clause needs an entirely different interpretation trom the one that works for it in other contexts. Since sentences are not names or descriptions in ordinary contexts, we can in particular reject the assumption that the attitudes have objects such as propositions which "that"-clauses might be held to name or describe. There should be no temptation to call the utterance to which reference is made according to the paratactic analysis the object of the attributed attitude.
Here a facile solution to our problem about the relation between thoughts and speech suggests itself. One way to view the paratactic analysis, a way proposed by Quine in Word and Object, is this: when a speaker attributes an attitude to a person, what he does is ape or mimic an actual or possible speech act of that person.8 Indirect discourse is the best example, and assertion is another good one. Suppose I say, "Herodotus asserted that the Nile rises in the Mountains of the Moon." My second utterance - my just past utterance of "The Nile rises in the Mountains of the Moon" - must, if my attribution to Herodotus is correct, bear a certain relationship to an utterance of Herodotus': it must, in some appropriate sense, be a translation of it. Since, assuming still that the attribution is correct, Herodotus and I are satnesayers, my utterance mimicked his. Not with respect to force, of course, since I didnt assert anything about the Nile. The sameness is with respect to the content of our utterances. If we turn to other attitudes, the situation is more complicated, for there is typically no utterance to ape. If I affirm "Jones believes that snow is white," my utterance of "Snow is white" may have no actual utterance of Jones's to imitate. Still, we could take the line that what I affirm is that Jones would be honestly speaking his mind were he to utter a sentence translating mine. Given some delicate assumptions about the conditions under which such a subjunctive conditional is true, we could conclude that only someone with a language could have a thought, since to have a thought would be to have a disposition to utter certain sentences with appropriate force under given circumstances.
We could take this line, but unfortunately there seems no clear reason why we have to.
We set out to find an argument to show that only creatures with speech have thoughts. What has just been outlined is not an argument, hut a proposal, and a proposal we need not accept The paratactic analysis of the logical form of attributions of attitude can get along without the mimic-theory of utterance. When I say, "Jones believes that snow is white" I describe Jones's state of mind directly: it is indeed the state of mind someone is in who could honestly assert "Snow is white" if he spoke English, but that may be a state a languageless creature could also be in.
In order to make my final main point, I must return to an aspect of interpretation so far
neglected. I remarked that the attitude of holding true, directed to sentences under specified'
circumstances, is the basis for interpretation, but I did not say how it can serve this function. /
The difficulty, it will be remembered, is that a sentence is held true because of two factorei
what the holder takes the sentence to mean, and what he believes. In order to sort things out, I
what is needed is a method for holding one factor steady while the other is studied. *
Membership in a language community depends on the ability to interpret the utteranc of members of the group, and a method is at hand if one has, and knows one has, a theory! which provides truth conditions, more or less in I arski's style, for all sentences (relativized,! as always, to time and speaker). The theory is correct as long as it entails, by finitely statr1* means, theorems of the familiar form: " 'It is raining' is true for a speaker x at time t iff only if it is raining (near ,v) at /." The evidential basis for such a theory concerns scnten held true, facts like the following: "'It is raining' was held true by Smith at 8 a.m. on Aug 26 and it did rain near Smith at that time." It would be possible to generate a correct th~ simply by considering sentences to be true when held true, provided (1) there was a th which satisfied the formal constraints and was consistent in this way with the evidence,; (2) all speakers held a sentence to be true just when that sentence was true - provided, t is, all beliefs, at least as far as they could be expressed, were correct.
But of course it cannot be assumed that speakers never have false beliefs. Error is'
gives belief its point. We can, however, take it as given that most beliefs are correct.'
reason for this is that a belief is identified by its location in a pattern of beliefs; it is I
pattern that determines the subject matter of the belief, what the belief is about. Before SC
object in, or aspect of, the world can become part of the subject matter of a belief (true(
false) there must be endless true beliefs about the subject matter. False beliefs tend.)
undermine the identification of the subject matter; to undermine, therefore, the validity.!
a description of the belief as being about that subject. And so, in turn, false beliefs undermio
the claim that a connected belief is false. To take an example, how clear are we that M
ancients - some ancients - believed that the earth was flat? This earth? Well, this earth of(T^
is part of the solar system, a system partly identified by the fact that it is a gaggle of I"
cool, solid bodies circling around a very large, hot star. If someone believes none of this a
the earth, is it certain that it is the earth that he is thinking about? An answer is not called!
The point is made if this kind of consideration of related beliefs can shake one's confij~
that the ancients believed the earth was flat. It isn't that any one false belief nece
destroys our ability to identify further beliefs, but that the intelligibility of such identifi
tions must depend on a background of largely unmentioned and unquestioned true bdi
To put it another way: the more things a believer is right about, the sharper his error* I
Too much mistake simply blurs the focus. ?!
What makes interpretation possible, then, is the fact that we can dismiss a priori | chance of massive error. A theory of interpretation cannot be correct that makes a man ?—* to very many false sentences: it must generally be the case that a sentence is true w speaker holds it to be. So far as it goes, it is in favor of a method of interpretation I
Thought and Talk241
f counts a sentence true just when speakers hold it to be true. But of course, the speaker maybe wrong; and so may the interpreter. So in the end what must be counted in favor of a method of interpretation is that it puts the interpreter in general agreement with the speaker: according to the method, the speaker holds a sentence true under specified conditions, and these conditions obtain, in the opinion of the interpreter, just when the speaker holds the lenience to be true.
No simple theory can put a speaker and interpreter in perfect agreement, and so a I Workable theory must from time to time assume error on the part of one or the other. The | basic methodological precept is, therefore, that a good theory of interpretation maximizes Iagreement. Or, given that sentences are infinite in number, and given further considerations I to come, a better word might be optimize.
I Some disagreements are more destructive of understanding than others, and a sophisti-icated theory must naturally take this into account. Disagreement about theoretical matters pfflay (in some cases) be more tolerable than disagreement about what is more evident; •(disagreement about how things look or appear is less tolerable than disagreement about how W>ey are; disagreement about the truth of attributions of certain attitudes to a speaker by that 1C speaker may not be tolerable at all, or barely. It is impossible to simplify the considera-s that are relevant, for everything we know or believe about the way evidence supports slief can be put to work in deciding where the theory can best allow error, and what errors t least destructive of understanding. The methodology of interpretation is, in this respect, (thing but epistemology seen in the mirror of meaning. || The interpreter who assumes his method can be made to work for a language community BU strive for a theory that optimizes agreement throughout the community. Since easy nmunication has survival value, he may expect usage within a community to favor simple
non theories of interpretation. |:,If this account of radical interpretation is right, at least in broad outline, then we should (knowledge that the concepts of objective truth, and of error, necessarily emerge in the ntext of interpretation. The distinction between a sentence being held true and being in Rtrue is essential to the existence of an interpersonal system of communication, and when I individual cases there is a difference, it must be counted as error. Since the attitude of "Ming true is the same, whether the sentence is true or not, it corresponds directly to belief, e concept of belief thus stands ready to take up the slack between objective truth and the d true, and we come to understand it just in this connection. jpWe have the idea of belief only from the role of belief in the interpretation of language, Iras a private attitude it is not intelligible except as an adjustment to the public norm •Mvided by language. It follows that a creature must be a member of a speech community El is to have the concept of belief. And given the dependence of other attitudes on belief, (Scan say more generally that only a creature that can interpret speech can have the concept |t thought.
i a creature have a belief if it does not have the concept of belief? It seems to me it )t, and for this reason. Someone cannot have a belief unless he understands the •ability of being mistaken, and this requires grasping the contrast between truth and error Due belief and false belief. But this contrast, I have argued, can emerge only in the context terpretation, which alone forces us to the idea of an objective, public truth, t is often wrongly thought that the semantical concept of truth is redundant, that there J difference between asserting that a sentence s is true, and using s to make an assertion. It may be right is a redundancy theory of belief, that to believe that fi is not to be languished from the belief that p is true. This notion of truth is not the semantical notion:
language is not directly in the picture. But it is only just out of the picture; it is part of ti frame. For the notion of a true belief depends on the notion of a true utterance, and I turn there cannot be without shared language. As Shakespeare's Ulysses puts it:
no man is the lord of anything, Though in and of him there be much consisting, Till he communicate his parts to others; Nor doth he of himself know them for aught Till he behold them formed in th'applause Where they're extended.
(Troilus and Cresstda, III. iii. 115-20)
N. Malcolm, "Thoughtless Brutes," Proceedings and Addresses of the American Philosophical/
W. Sellars, "Conceptual Change," in G. Pearce and P. Maynard, eds, Conceptual Chi
(Dordrecht: Reidel, 1973), p.82.
F. P. Ramsey, "Truth and Probability," reprinted in Foundations of Mathematics (New Yo
Humanities Press, 1950).
See my "Radical Interpretation," Dialectica 21 (1973), pp. 313-28; and "Belief and the Basis (
Meaning," Synthese 27 (1974), pp. 309-23.
The interlocking of decision theory and radical interpretation is explored also in "Belief and the I
Basis of Meaning," in Essay 12 of Kssays in Actions and Events (Oxford: Clarendon Press, 1980), and!
in "Toward a Unified Theory of Meaning and Action," Grazer Philosophische Studien 2 (1980), f
A. Church, "On Carnap's Analysis of Statements of Assertion and Belief," Analysis 10 (1950), f
See "On Saying That," Synthese 19 (1968-9), pp. 130-46.
W. V. Quine, Word and Object (Cambridge, MA: MIT Press, 1960), p. 219.