7/30/2019 4106594

1/33

7/30/2019 4106594

2/33

SYMPOSIUM: " CAN THERE BE A PRIVATELANGUAGE? "PROFESSORA. J. AYER. MR. R. RHEES.

I.-By A. J. AYER.IN a quite ordinary sense, it is obvious that there can beprivate languages. There can be, because there are. Alanguage may be said to be private when it is devised toenable a limited number of persons to communicate withone another in a way that is not intelligible to anyoneoutside the group. By this criterion, thieves' slang andfamily jargons are private languages. Such languages arenot strictly private, in the sense that only one person usesand understandsthem, but there may very well be languagesthat are. Men have been known to keep diaries in codeswhich no one else is meant to understand. A privatecode is not, indeed, a private language, but rather a privatemethod of transcribing some given language. It is,however, possible that a very secretive diarist may not besatisfied with putting familiar words into an unfamiliarnotation, but may prefer to invent new words: the twoprocessesare in any case not sharply distinct. If he carrieshis invention far enough he can properly be said to beemploying a private language. For all I know, this hasactually been done.From this point of view, what makes a language privateis simply the fact that it satisfies the purpose of beingintelligible only to a single person, or to a restricted set ofpeople. It is necessary here to bring in a reference topurpose, since a lafiguage may come to be intelligibleonly to a few people, or even only to a single person, merelyby falling into general disuse: but such "dead" languagesare not considered to be private, if the limitation of theiruse was not originally intended. One may characterize aprivate language by saying that it is not in this sense meantto be alive. There is, however, no reason, in principle,why it should not come alive. The fact that only one

7/30/2019 4106594

3/33

64 A. J. AYER.person, or only a few people, are able to understand it ispurely contingent. Just as it is possible, in theory, thatany code should be broken, so can a private languagecome to be more widely understood. Such private languagesare in general derived from public languages, and even ifthere are any which are not so derived, they will still betranslatable into public languages. Their ceasing to beprivate is then just a matter of enough people becomingable to translate them or, what is more difficult but stilltheoretically possible, not to translate but even so to under-stand them.If I am right, then, there is a use for the expression"private language" which clearly allows it to have appli-cation. But this is not the use which philosophers havecommonly given it. What philosophersusuallyseemto havein mind when they speak of a private language is one thatis, in their view, necessarily private, in as much as it isused by some particular person to refer only to his ownprivate experiences: for it is often held that for a languageto be public it must refer to what is publicly observable:if a person could limit himself to describinghis own sensa-tions or feelings, then, strictly speaking, only he wouldunderstand what he was saying; his utterance mightindirectly convey some information to others, but it couldnot mean to them exactly what it meant to him. Thus,Carnap who gives the name of "protocol language" toany set of sentences which are used to give "a direct record"of one's own experience argues,in his bookleton The Unityof Science,' that if an utterance like "Thirstnow", belongingto the protocol language of a subject S, is construed asexpressing " only what is immediately given " to S, itcannot be understood by anyone else. Another subject S2may claim to be able to recognize and so to refer to Si'sthirst, but " strictly speaking" all that he ever recognizesis some physical state of S,'s body. " If by 'the thirstof S 1' we understand not the physical state of his body,but his sensations of thirst, i.e., something non-material,then S's thirst is fundamentally beyond the reach of1 76 ff.

7/30/2019 4106594

4/33

CAN THERE BE A PRIVATE LANGUAGE ? 65S2's recognition."2 S2 cannot possibly verify any state-ment which refers to S,'s thirst, in this sense, and conse-quently cannot understand it. " In general," Carnapcontinues, " every statement in any person's protocollanguage would have sense for that person alone...Even when the same words and sentences occur in variousprotocol languages, their sense would be different, theycould not even be compared. Every protocol languagecould therefore be applied only solipsistically: there wouldbe no intersubjective protocol language. This is theconsequence obtained by consistent adherence to the usualview and terminology (rejected by the author)."3Since Carnap wishes to maintain that people canunderstand one another's protocol statements, if only onthe ground that this is a necessarycondition for statementsmade in what he calls the physical language to be inter-subjectivelyverifiable,he drawsthe inferencethat " protocollanguage is a part of physical language". That is, heconcludes that sentences which on the face of it refer toprivate experiencesmust be logically equivalent to sentenceswhich describe some physical state of the subject. Otherphilosophers have followed him in giving a physicalistinterpretation to the statements that one makes about theexperiences of others, but have stopped short of extendingit to all the statements that one may make about one'sown. They prefer to hold that certain sentences do serveonly to describe the speaker's private experiences, andthat, this being so, they have a different meaning for himfrom any that they can possiblyhave for anybody else.In his Philosophical InvestigationsWittgenstein appearsto go much further than this. He seems to take the viewthat someone who attempted to use language in this privateway would not merely be unable to communicate his mean-ing to others, but would have no meaning to communicateeven to himself; he would not succeed in saying anythingat all. " Let us", says Wittgenstein,4 magine the following

2 p. 79.3 p. 80.4 Philosophical nvestigations, .258.

F

7/30/2019 4106594

5/33

66 A. J. AYER.case: " I want to keep a diary about the recurrence of acertain sensation. To this end I associate it with thesign 'E' and write this sign in a calendar for every dayon which I have the sensation.-I will remarkfirst of allthat a definition of the sign cannot be formulated.-Butstill I can give myself a kind of ostensivedefinition.-How ?Can I point to the sensation? Not in the ordinary sense.But I speak,-write the sign down, and at the same time Iconcentrate my attention on the sensation-and so, as itwere, point to it inwardly.-But what is this ceremonyfor? for that is all it seems to be! A definition surelyserves to establish the meaning of a sign.-Well, that is doneprecisely by the concentration of my attention; for inthis way I impresson myself the connexion betweenthe signand the sensation. But 'I impress it on myself' can onlymean: this process brings it about that I remember theconnexion right in the future. But in the present case Ihave no criterion of correctness. One would like to say:whatever is going to seem right to me is right. And thatonly means that here one can't talk about 'right'."Again, " What reason have we for calling ' E' the signfor a sensation For ' sensation' is a word of our commonlanguage, not of one intelligible to me alone. So the useof the word standsin need of ajustificationwhich everybodyunderstands."5

This point is then developed further: " Let us imaginea table (somethinglike a dictionary) that exists only in ourimagination. A dictionary can be used to justify thetranslation of a word X into a word Y. But are we alsoto call it ajustificationif such a table is to be looked uponlyin the imagination?-' Well, yes: then it is a subjectivejustification'.-But justification consists in appealing tosomething independent.-' But surely I can appeal fromone memory to another. For example, I don't know if Ihave remembered the time of departure of a train rightand to check it I call to mind how a page of the time-table looked. Isn't it the same here? ' No; forthis processhas got to produce a memory which is actually correct.If the mental image of the time-table could not itself be

6 Op. cit., . 261.

7/30/2019 4106594

6/33

CAN THERE BE A PRIVATE LANGUAGE ? 67tested or correctness,how could it confirm the correctnessof the first memory? (As if someone were to buy severalcopies of the morning paper to assure himself that what itsaid was true.)Looking up a table in the imagination is no morelookingup a table than the image of the result of an imaginedexperiment is the result of an experiment."'The case is quite different, Wittgenstein thinks, whenthe sensation can be coupled with someoutward manifesta-tion. Thus he maintains that the language which weordinarily use to describe our " inner experiences " isnot private because the words which one uses to refer toone'ssensationsare " tied up with [one's] naturalexpressionsof sensation"' with the result that other people are in aposition to understand them. Similarly he grants that thepersonwho tries to describehis private sensationby writingdown the sign " E " in his diary might find a use for thissign if he discovered that whenever he had the sensationin question it could be shown by means of some measuringinstrument that his blood pressure rose. For this wouldgive him a way of telling that his blood pressurewas risingwithout bothering to consult the instrument. But then,argues Wittgenstein, it will make no difference whetherhis recognition of the sensation is right or not. Providedthat wheneverhe thinks he recognizes t, thereis independentevidence that his blood pressurerises, it will not matter ifhe is invariably mistaken, if the sensationwhich he takes tobe the same on each occasion is really not the same at all." And that alone shows that the hypothesis that [he]makes a mistake is mere show."'Let us examine this argument. A point to whichWittgenstein constantly recurs is that the ascription ofmeaning to a sign is something that needs to be justified:the justification consists in there being some independenttest for determining that the sign is being used correctly;independent, that is, of the subject'srecognition,or supposedrecognition, of the object which he intends the sign to

6 Op. cit., . 265.? Op. cit., 2.256.8 Op. cit., . 270. v2

7/30/2019 4106594

7/33

68 A. J. AYER.signify. His claim to recognize the object, his belief thatit really is the same, is not to be accepted unless it can bebacked by furtherevidence. Apparently, too, this evidencemust be public: it must, at least in theory, be accessible toeveryone. Merely to check one private sensation byanother would not be enough. For if one cannot betrusted to recognize one of them, neither can one be trustedto recognize the other.But unless there is something that one is allowed torecognize, no test can ever be completed: there will be nojustification for the use of any sign at all. I check mymemory of the time at which the train is due to leaveby visualizing a page of the time-table; and I am requiredto check this in its turn by looking up the page. Butunless I can trust my eyesight at this point, unless I canrecognize the figures that I see written down, I am still nobetter off. It is true that if I distrust my eyesight I havethe resource of consulting other people; but then I haveto understand their testimony, I have correctly to identifythe signs that they make. Let the object to which I amattempting to refer be as public as you please, let the wordwhich I use for this purpose belong to some commonlanguage, my assurancethat I am using the word correctly,that I am using it to refer to the " right " object, must inthe end rest on the testimony of my senses. It is throughhearing what other people say, or through seeing what theywrite, or observing their movements, that I am enabled toConclude that their use of the word agrees with mine.But if without further ado I can recognize such noises orshapes or movementswhy can I not also recognizea privatesensation? It is all very well for Wittgenstein to say thatwriting down the sign "E", at the same time as I attendto the sensation,is an idle ceremony. How is it any moreidle than writing down a sign, whether it be the conven-tionally correct sign or not, at the same time as I observesome " public " object ? There is, indeed, a problem aboutwhat is involved in endowing any sign with meaning,but it is no less of a problem in the case where the objectfor which the sign is supposed to stand is public than in the

7/30/2019 4106594

8/33

CANTHEREBE A PRIVATE ANGUAGE 69case where it is private. Whateverit is about my behaviourthat transformsthe making of a sound, or the inscriptionof a shape, into the employment of a sign can equally welloccur in either case.

But, it may be said, in the one case I can point to theobject I am tryingto name, I can give an ostensivedefinitionof it; in the other I cannot. For merely attending to anobject is not pointing to it. But what difference does thismake? I can indeed extend my finger in the directionof a physical object, while I pronounce what I intend tobe the object's name; and I cannot extend my finger inthe directionof a private sensation. But how is this extend-ing of my fingeritself anything more than an idle ceremony?If it is to play its part in the giving of an ostensivedefinitionthis gesture has to be endowed with meaning. But if Ican endow such a gesture with meaning, I can endow aword with meaning, without the gesture.

I suppose that the reason why the gesture is thought tobe important is that it enables me to make my meaningclear to others. Of course they have to interpret mecorrectly. If they are not intelligent, or I am not careful,they may think that I am pointing to one thing when Ireally intend to point to another. But successfulcommuni-cation by this method is at least possible. The object towhich I mean to point is one that they can observe. Onthe other hand, no amount of gesturing on my part candirect their attention to a private sensation of mine, whichex hypothesihey cannot observe, assuming further that thissensationhas no " naturalexpression". So I cannot give anostensive definition of the word which I wish to stand forthe sensation. Nor can I define it in terms of other words,for how are they to be defined? Consequently I cannotsucceed in giving it any meaning.This argument is based on two assumptions, both ofwhich I believe to be false. One is that in a case of thissort it is impossible, logically impossible, to understand asign unlessone can either observe the objectwhich it signifies,or at least observe something with which this object isnaturally associated. And the other is that for a person

7/30/2019 4106594

9/33

70 A. J. AYER.to be able to attach meaning to a sign it is necessarythatother people should be capable of understanding it too.It will be convenient to begin by examining the second ofthese assumptionswhich leads on to the first.Imagine a Robinson Crusoe left alone on his islandwhile still an infant, having not yet learned to speak.Let him, like Romulus and Remus,be nurturedby a wolf, orsome other animal, until he can fend for himself; and solet him grow to manhood. He will certainly be able torecognize many things upon the island, in the sense thathe adapts his behaviour to them. Is it inconceivablethat he should also name them? There may be psycho-logical grounds for doubting whether such a solitary beingwould in fact invent a language. The development oflanguage, it may be argued, is a social phenomenon. Butsurely it is not self-contradictoryto suppose that someone,uninstructedin the use of any existing language, makes upa language for himself. After all, some human being musthave been the first to use a symbol. And even if he did soas a member of a group, in order to communicate with theother members, even if his choice of symbols was sociallyconditioned, it is at least conceivable that it should originallyhave been a purely private enterprise. The hypothesis ofG. K. Chesterton's dancing professor about the origin oflanguage, that it came " from the formulatedsecretlanguageof some individual creature" is very probably false, butit is certainly not unintelligible.But if we allow that our Robinson Crusoe could inventwords to describe the flora and fauna of his island, whynot allow that he could also invent words to describe hissensations? In neither case will he be able to justify hisuse of words by drawing on the evidence provided by afellow creature: but while this is a useful check, it is notindispensable. It would be difficult to argue that thepower of communication, the ability even to keep a privatediary, could come to him only with the arrival of ManFriday. His justification for describing his environment inthe way that he does will be that he perceives it to havejust those featureswhich his words are intended to describe,

7/30/2019 4106594

10/33

CAN THERE BE A PRIVATE LANGUAGE ? 71His knowing how to use these words will be a matter of hisremembering what objects they are meant to stand for,and so of his being able to recognize these objects. Butwhy should he not succeed in recognizing them? Andwhy then should he not equally succeed in recognizing hissensations? Undoubtedly, he may make mistakes. Hemay think that a bird which he sees flying past is a bird ofthe same type as one which he had previouslynamed, whenin fact it is of a different type, sufficientlydifferent for himto have given it a different name if he had observed itmore closely. Similarly, he may think that a sensationis the same as others which he has identified, when infact, in the relevant aspects, it is not the same. In neithercase may the mistake make any practical differenceto him,but to say that nothing turns upon a mistake is not to saythat it is not a mistakeat all. In the case of the bird, thereis a slightly greater chance of his detecting his mistake,since the identical bird may re-appear: but even so he hasto rely upon his memory for the assurance that it is theidentical bird. In the case of the sensation, he has only hismemory as a means of deciding whether his identificationis corrector not. In this respect he is indeed like Wittgen-stein's man who buys several copies of the morningpaper to assure himself that what it says is true. But thereason why this seems to us so aburd is that we take it forgranted that one copy of a morning paper will duplicateanother; there is no absurdity in buying a second news-paper, of a different type, and using it to check the first.And in a place wheretherewas only one morningnewspaper,but it was so produced that misprints might occur in onecopy without occuring in all, it would be perfectly sensibleto buy several copies and check them against each other.Of course there remains the important difference that thefacts which the newspaper reports are independentlyverifiable, in theory if not always in practice. But verifica-tion must stop somewhere. As I have already argued,unless something is recognized, without being referred toa further test, nothing can be tested. In the case ofCrusoe's sensation, we are supposing that beyond his

7/30/2019 4106594

11/33

72 A. J. AYER.memory there is no further test. It does not follow thathe has no means of identifying it, or that it does not makesense to say that he identifies it right or wrong.So long as Crusoe remains alone on the island, so long,that is, as he communicatesonly with himself, the principaldistinction which he is likely to draw between " external "objects and his " inner " experiencesis that his experiencesare transient in a way that external objects are not. Hewill not be bound to draw even this distinction; his criteriafor identity may be differentfrom our own; but it is reason-able to suppose that they will be the same. Assuming,then, that his language admits the distinction, he will findon the arrivalof Man Friday that it acquiresa new import-ance. For whereas he will be able to teach Man Fridaythe use of the words which he has devised to stand forexternal objects by showing him the objects for which theystand, he will not, in this way, be able to teach him the useof the words which he has devised to stand for his sensations.And in the cases where these sensationsare entirely private,in the sense that they have no " natural expressions" whichMan Friday can identify, it may well be that Crusoefails tofind any way of teaching him the use of the words which heemploys to stand for them. But from the fact that hecannot teach this part of his language to Man Friday it byno means follows that he has no use for it himself. In acontext of this sort, one can teach only what one alreadyunderstands. The ability to teach, or rather the abilityof someone else to learn, cannot therefore be a prerequisitefor understanding.Neither does it necessarilyfollow, in these circumstances,that Man Friday will be incapable of learning the meaningof the words which Crusoe uses to describe his privatesensations. It is surely a contingent fact that we dependupon ostensive definitions, to the extent that we do, forlearning what words mean. As it is, a child is not taughthow to describehis feelings n the way he is taught to describethe objects in his nursery. His mother cannot point to hispain in the way that she can point to his cup and spoon.But she knows that he has a pain because he cries and

7/30/2019 4106594

12/33

CAN THERE BE A PRIVATE LANGUAGE ? 73because she sees that something has happened to him whichis likely to cause him pain; and knowing that he is inpain she is able to teach him what to call it. If therewere no external signs of his sensation she would have nomeans of detecting when he had them, and thereforecouldnot teach him how to describe them. This is indeed thecase, but it might easily be otherwise. We can imaginetwo personsbeing so attuned to one another that whenevereither has a private sensation of a certain sort, the otherhas it too. In that case when one of them described whathe was feeling the other might very well follow the descrip-tion, even though he had no " external " evidence to guidehim. But how could either of them ever know that hehad identified the other's feeling correctly? Well, howcan two people ever know that they mean the same by aword which they use to refer to some " public" object.Only because each finds the other's reactions appropriate.Similarly one may suppose that Man Friday sympathizeswhen Crusoe's private sensation is painful, and congratu-lates him when it is pleasant, that he is able to say whenit begins and when it stops, that he correctly describesit as being rather like such and such another sensation,and very different from a third, thereby affording proofthat he also understands the words that stand for thesesensations. Admittedly, such tests are not conclusive.But the tests which we ordinarily take as showing that wemean the same by the words which we apply to publicobjects are not conclusive either: they leave it at leasttheoretically open that we do not after all mean quite thesame. But from the fact that the tests are not conclusiveit does not, in either case, follow that they have no forceat all. It is true also that such tests as the expressedagreementabout the duration of the experiencerequirethatthe two men already share a common language, whichthey have no doubt built up on the basis of common obser-vations. It would indeed be difficult, though still, I think,not necessarily impossible, for them to establish communi-cation if all their experiencewere private, in Wittgenstein'ssense. But even if their understandingeach other's use of

7/30/2019 4106594

13/33

74 A. J. AYER.words could come about only if some of the objects whichthese words described were public, it would not followthat they all must be so.It is not even necessary to make the assumption thatMan Friday comes to know what Crusoe'ssensations are,and so to understandthe wordswhich signifythem, throughhaving similar sensations of his own. It is conceivablethat he should satisfyall the tests which go to show that hehas this knowledge, and indeed that he should actuallyhave it, even though the experience which he rightlyascribesto Crusoeis unlike any that he has, or ever has had,himself. It would indeed be very strange if someonehad this power of seeing, as it were, directly into another'ssoul. But it is strangeonly in the sense that it is somethingwhich, on causal grounds,we should not expect to happen.The idea of its happening breaksno logical rule. An ana-logous case would be that of someone's imagining, orseeming to remember, an experience which was unlikeany that he had ever actually had. To allow that suchthings are possible is, indeed, to admit innate ideas, in theLockean sense, but that is not a serious objection. Theadmission is not even inconsistent with the prevalentvarieties of empiricism. It can still be made a rule thatin order to understand a word which signifies a sensationone must know what it would be like to have the sensationin question: that is, one must be able to identify thesensation when one has it, and so to verify the statementwhich describes it. The peculiarity of the cases which weare envisaging is just that people are credited with theability to identifyexperienceswhich they have not previouslyhad. There may indeed be causal objections to thehypothesis that this can ever happen. The point whichconcernsus now is that these objections are no more thancausal. The ways in which languages are actually learneddo not logically circumscribethe possibilitiesof their beingunderstood.If the sort of insight which we have been attributing toMan Friday were commonly possessed, we might well beled to revise our concepts of publicity and privacy. The

7/30/2019 4106594

14/33

CAN THERE BE A PRIVATE LANGUAGE ? 75mistake which is made by philosophers like Carnap isthat of supposing that being public or being private, inthe senseswhich are relevantto this discussion,arepropertieswhich are somehow attached to different sorts of objects,independently of our linguistic usage. But the reason whyone object is publicly and another only privately accessibleis that in the one case it makes sense to say that the objectis observed by more than one person and in the other itdoes not. Tables are public; it makes sense to say thatseveral people are perceivingthe same table. Headachesare private: it does not make sense to say that severalpeople are feeling the same headache. But just as we canassimilate tables to headaches by introducing a notationinwhich two different persons' perceiving the same tablebecomes a matter of their each sensing their own private" tabular" sense-data, so we could assimilate headachesto tables by introducing a notation in which it was correctto speak of a common headache, which certain peopleonly were in a condition to perceive. As things are, thisnotation would not be convenient. But if people were soconstituted that they werecommunally exposedto headachesin the way that they arecommunallyexposedto the weather,we might cease to think of headaches as being necessarilyprivate. A London particular might come to be a localheadache as well as, or instead of, a local fog. Certainpersons might escape it, just as certain persons, for onereason or another, may fail to perceive the fog. But thefog exists for all that, and so, given this new way of speaking,would the public headache. The conditions which wouldmake this way of speaking useful do not, indeed, obtain;but that they do not is, once again, a purely contingent fact.The facts being what they are, we do not have a usefor such expressionsas " S2's feeling Sx's thirst" or " S2'sobserving the sensation of thirst which S, feels." On theother hand, we do attach a meaning to saying that the samephysical object, or process, or event, for instance a stateof Sx's body, is observed by S2 as well as by S,. Does itfollow, as Carnap thinks, that for this reason S2 cannotunderstanda statementwhich refers to Sl's feeling of thirst,

7/30/2019 4106594

15/33

76 A. J. AYER.whereas he can understand a statement which refers to thecondition of S,'s body? Suppose that we modified ourrules for identity, in a way that many philosophers haveproposed, and allowed ourselves to say that what wasordinarily described as S, and S2's observing the samephysicalevent was " really " a case of each of them sensinghis own sense-datawhich, while they might be qualitativelysimilar,could not be literally the same. Should wetherebybe committed to denying that either could understandwhatthe other said about this physicalevent? Surelynot. Andequally the fact that S2 cannot feel, or inspect, Sl's feelingsin no way entails that he cannot understand what S, saysabout them. The criteria for deciding whether two peopleunderstand each other are logically independent of thefact that we do, or do not, have a use for saying that literallythe same objects are perceived by both.I conclude, first, that for a person to use descriptivelanguage meaningfully it is not necessary that any otherpersonshouldunderstandhim, and, secondly,that foranyoneto understand a descriptive statement it is not necessarythat he should himself be able to observe what it describes.It is not even necessary that he should be able to observesomething which is naturally associated with what itdescribes,in the way that feelings are associatedwith their" naturalexpressions". If we insiston makingit a necessarycondition for our understandinga descriptivestatementthatwe are able to observe what it describes,we shall find our-selves disclaiming the possibility of understanding notmerely statements about other people's private sensations,but also statements about the past; either that, or re-inter-preting them in such a way that they change their reference,as when philosophers substitute bodily states for feelings,and the future for the past. Both courses, I now think,are mistaken. No doubt it is a necessary condition formy understanding a descriptive statement that it shouldbe, in some way, verifiable. But it need not be directlyverifiable, and even if it is directly verifiable, it need notbe directly verifiableby me.

7/30/2019 4106594

16/33

II.-By R. RHEES.THE problem about private languages is the problem ofhow words mean. This is much the same as the questionof what a rule of language is.When we talk about something, our language does notpoint to it, nor mirror it. Pointing or mirroring couldrefer to things only within a convention, anyway: onlywhen there is a way in which pointing is understoodand away in which mirroring is understood. I point for thesake of someone who understands it. Apart from thatit were an idle ceremony; as idle as making sounds infront of things.Our words refer to things by the way they enter indiscourse; by their connexions with what people are sayingand doing, for instance, and by the way they affect what issaid and done. What we say makes a difference. Whatexpressionswe use makes a difference. And the notion ofa rule goes with that. If it made no differencewhat soundyou made or when, you could not be understood and youwould have said nothing. If you have said something,your utterance will be taken in one way and notin another.In many cases you will have committed yourself to sayingother things, to answeringin certain ways if you are asked,or to doing certain things. That belongs to the regularuse of your words, and that is why it would not have beenjust the same if you had used others instead. That is alsowhy it is possible to learn the language.When we speak of " use " we may think of generalpractice and we may think of rules. Sometimes these canbe left together, but sometimes there are differences weought to notice. When I learn the use of an expression,or learn what it means-that is how other people speak.Yet I do not say I have learned what other people do;I have learned what it means. I may learn what it meansbyobservingwhat other people do, and of course if I knowwhat it means I know that others who speak the languagewill use it in that way. But I have not learned whatgenerally happens. I have learned a rule.

7/30/2019 4106594

17/33

78 R. RHEES.That is in some ways like learning the rules of a game,although in some ways it is very different. It is differentfrom learning the rules of a calculus, too. In fact in someways it is misleading to talk of rules at all here. But itdoes make some things clearer-that it is possible to usean expressionwrongly, for instance.A rule is something that is kept. That is why we canknow what we are talking about. When you have learnedhow the expression s used, then you can not merely behave

as other people do, you can also say something. That isnot a matter of behaving in a particular way. " This isred " does not mean " Everyone calls this red". If thatwere all there were to it nothing would mean anything.And yet, that there should be rules at all does dependon what people do, and on an agreement in what they do.If you teach someone the meaning of a colour word byshowing him samples of the colour, then he will probablyunderstand; and if he understands he will go on to usethe word in new situations just as you would. If heremembered your instruction all right but differed wildlyfrom you in what he called " the same as" the samplesyou had shown him, and if this went on no matter howoften you repeated your explanation, then he could neverlearn what that colour word means. And this holdsgenerally, not just with colours. It is a point to whichWittgenstein is referring in Investigations 42. Of coursethat situation practically never arises. And if it were atall general we could not speak.I am not saying, " People see that their reactions tally,and this makes communication possible". That wouldassume considerable understanding and language already.The agreementof which I am speakingis somethingwithoutwhich it would not be possible for people to " see " thattheir reactions tallied or that anything else tallied. Wesee that we understand one another, without noticingwhether our reactions tally or not. Becausewe agree inour reactions, it is possible for me to tell you something,and it is possiblefor you to teach me something.The consensusof reactions is in this sensepriorto language,

7/30/2019 4106594

18/33

CAN THERE BE A PRIVATE LANGUAGE ? 79but the reactionsthemselvesare not languages,nor are theylanguage. Neither does the agreement in reactions comefirst or anticipate language. It appears as the languagedoes, it is a common way of taking the expressionsof thelanguage. They are common reactions within the courseof language-not to anything there might have been beforelanguage or apart from it.Because there is this agreement we can understandoneanother. And since we understand one another we haverules. We might perhaps speak of being " trusted" togo on in the way that is for us the only natural one. Butif you have learned the language you take it for granted.If any one did not, we could never understandhim.Because there is this agreement it is possible to saysomething. When I tell you that the patch on the patient'sskinis red, I am not sayingthat it is called red, but that it isred. But I could mean nothing definite by that, and youcould not understand me, unless people who have learnedthe words as we have would agree in calling this red. Ifpeople could not be brought to use the word in any regularway, if one man who had been taught as we have shouldgo on to give the name to what we should call the comple-mentarycolour, if another used it as we do on Monday butin a different way on Tuesday, and if others did not showeven these degrees of regularity-then it would not meananythingto say that someonehad used the word mistakenly.There would be no distinction between mistakenly andcorrectly. And there would be no distinction betweensaying that it is red and saying anything else.It is not a statement about what I do or about whatpeople generally do. But unless the words had a regularuse I should not know it was red, and I should not knowwhat colour it was, becausethere would be nothing to know.I know what colour it is because I know red when I see it;I know what red is. A bull may charge at a red flag, andrats may be trained to react in one way to red lights andin another way to blue lights, but neither the bull northe rat knowswhat red is, and neither knows that this is red.We might put this by saying that neither of them has the

7/30/2019 4106594

19/33

80 R. RHEES.

concept "red" and neither of them has the concept" colour". No one can get the concept of colour just bylooking at colours, or of red just by looking at red things.If I have the concept, I know how the word " red " is used.There must be a use, though; there must be what I havebeen calling common reactions. The phrase " the samecolour " must mean somethingand be generallyunderstood,and also "a different colour". I must know when itmakes sense to talk about different shades of the samecolour; and so on. Unless I did know what it makessense to say, unless I were used to talking about coloursand to understandingpeople when they did, then I shouldnot know what red is and I should not know red whenI see it.Of course the colour red is not the word " red". And Isuppose if a man cannot see he will never know what it is.But the colour red is not this, either. This is red. But ifI say " This is the colourred", that is a definition-I amgiving you a definition by showing you a sample. Andthe point of that depends upon the definition'sbeing takenin a particular way; and also on its connexion with otheruses of language. If I had just shown you that samplewithout saying anything, and without your asking-whatwould you have learned from this ? Not what the colourred is, anyway.

Someone might say, "I know what I mean by 'red'.It is what I experience when I look at this. Whether Ihave this experience under the same circumstances as leadyou to use the word-that is a furtherquestion, which maybe important in deciding the descriptionof physical objects.But I know what colour I see in these circumstances". (Itwould be hard to keep from asking, " Well, what colour doyou see ? ") I suppose the point would be that I knowthis independentlyof having learned the (public) language.If I know what I mean, in this way-if I know what colourI am referringto-then apparently I have done somethinglike giving myself a definition. But I must also haveconfused giving a definition and following a definition. Itis this which allows me to evade the difficultyof what I am

7/30/2019 4106594

20/33

CAN THERE BE A PRIVATE LANGUAGE ? 81going to call " following the definition". Which is a realdifficulty: what could it mean to say that I had followedthe definition-" my " definition-incorrectly ? But if thathas no sense, then what on earth is the point of the defini-tion ? And what does the definition establish

Suppose someone asked " What colour is red ? " andthought it was like asking" What colour is blood ? " This,he might think, is something which I can learn only by myown experience, my immediate experience. And althoughI can tell you what colour blood is, I cannot tell you whatcolourredis. I can only suggestthings that may enable youto find out for yourself. Well, but in this case what is thesense of "what colour red is" ? If it is somethingnobodycansay, then nobody can askit either. Suppose I ask it only ofmyself-but whatever is it I am asking ? Something Ishould like to know ? But if that has no sense, then thereis nothing I tell myself either. Perhaps I say " What acolour ! ", but that is all.I cannot learn the colour unless I can see it; but Icannot learn it without language either. I know it becauseI know the language. And it is similar with sensations.I know a headache when I feel it, and I know I felt giddyyesterday afternoon, because I know what giddiness is.I can remember the sensationI had, just as I can rememberthe colour I saw. I feel the same sensation,and that is thesame colour. But the identity-the sameness--comes fromthe language.A rule is somethingthat is kept. The meaningof a wordis something that is kept. It is for this reason that I cansay this is the same colour I saw a moment ago. I can seethe same colour just because I know red when I see it.And even with shades for which we have no special names,the same thing holds: I know the same colour when Isee it.It is similar, I have said, with sensations. I can saywhat I felt and I can say what I feel, and I can say it is thesame sensation this time-because I know what sensationsI am speaking of. It might be said that I can know it isthe same only if itfeels the same; and that is something no

0

7/30/2019 4106594

21/33

82 R. RHEES.language can tell me. Nor can I know whether you arefeeling the same as you have felt before. Only you can tellme that, because you are the only one who knows what itfeels like. Well I agree that no language can tell mewhether this feels the same. No language can tell mewhether those two are the same colour, either. And myfamiliarity with methods of measurement will not tell mewhether those two plots have the same area before I havemeasured them. But without language I could not havetold whether this feels the same, either; if only becauseI could not have asked.Of course recognizing a sensation is a different sort ofthing from recognizing a colour. This holds whether I amspeaking of my own or another's. It is different fromrecognizing what anything looks like or what is going on.When I say the dog is in pain I am not describingwhat thedog is doing, any more than I describe what I am doingwhen I give expressionto pain. It is morelike an expressionof pity. At any rate, feeling pity, trying to ease him andso on-or perhaps turningaway from the sight-is all partof believing that he is in pain. And to say that I wasobviously justified in that-or maybe that I was mistaken-is a different sort of thing fromsaying that I wasjustifiedormistaken in believing that he had a fracture. " Mistake"means something different here, although it is just asdefinite. If I made a mistake in thinking the boy was inpain, well he was shamming and my pity was misplaced.The mistake was not that I supposed something was goingon in him when nothing was. I may have supposed thattoo, perhaps that he had a cramp, but that is a differentmistake. The dog's pain is not something going on. Itis just his being n pain. I know for certain that he is inpain, and I know this because I know what pain is andwhat suffering s. There is an importantdifferencebetweenseeing that he is in pain and being in pain myself, becauseI do not see that I am in pain, and while it is conceivablethat I am mistaken about him, that makes no sense in myown case. But this does not mean that I know somethingabout myself which I cannot know about him.

7/30/2019 4106594

22/33

CAN THERE BE A PRIVATE LANGUAGE ? 83We do not speak of sensations in the same way as we

speak of processes or of colours. The name of a sensationis a different sort of name from the name of a colour. Butif it means anything to say I am in pain again or that he isin pain again, this is because the word " pain " has a regularuse and because we know this when we know what painis. If it were something I knew only in myself, then Imight say " This is something different now " or " Thisis the same again " or I might say neither, and in any caseit would not make any difference. This is not a questionof whether I can trust my memory. It is a question ofwhen it makes sense to speak of remembering; either ofa good memory or a faulty one. If I thought I could nottrust my memory, then of course I might look for confirma-tion. But there cannot be any question of confirmationhere, nor any question of doubting either. There is justno rule for what is the same and what is not the same;there is no distinction between correct and incorrect; andit is for that reason that it does not make any differencewhat I say. Which means, of course, that I say nothing.I cannot say anything unless I know the language. ButI cannot know the language-any language-privately.I may have a secret code, but that is not the point here.It is a question of whether I can have a private under-standing; whether I can understand something whichcould not be said in a language anyone else could under-stand. (" He may understand the language I speak,but he will not understand what I understand.") I sayI cannot know a language privately, for what would therebe to know? In language it makes a difference what yousay. But how can it make any difference what you sayprivately? (I do not mean talking to yourself.) It seemsthat in a private language everything would have to be atonce a statement and a definition. I suppose I maydefine a mark in any way I wish. And if every use of themark is also a definition-if there is no way of discoveringthat I am wrong, in fact no sense in suggesting that Imight be wrong-then it does not matter what mark Iuse or when I use it.

G2

7/30/2019 4106594

23/33

84 R. RHEES.One might ask, " Why can I not give myself a definitionand decide for myself what following the definition isgoing to be?" But when? Each time? If I decideonce and for all, that only renews the problem: what is" according to my decision " ? But what would the deci-sion be anyway? In ordinary language I may decideto use an expressionin a particular way, and I know howto keep to this. I do this in connexion with establishedusages and rules. That is why " in a particular way "

means something. That is also why I can decide to usethe expressionsof a secret language or the signs of a codein a particular way. For I am dealing with expressionsthat can be understood, and I know how the matter couldbe said in ordinarylanguage. I know whether I am sayingthe same as I said before, and I know what I am deciding.But not when it is something which could not be said inordinary language. Here there would be no point insaying, for instance, " I am going to use S to mean that",because I do not know what " meaning that " could be.The reason is not that others must see what my wordsreferto. It isjust that if my words are to referto anythingthey must be understood. They cannot refer at all exceptin connexion with a use, a use which you learn when youlearn what the word means. They cannot refer to any-thing unless there is a way in which the language is spoken.That is why there cannot be a private understanding.If it makes no difference what is said, nothing is understood.There is of course no reason why I should not give anaccount of something which only I can see. Or of some-thing which only I can feel: as when I tell a doctor what Ifeel in my abdomen. He does not feel my sensations(if that means anything), but he knows what I am talkingabout; he knows what sensationsthey are.

Ayer asks why Crusoe should not invent names for hissensations. (He actually says " names to describehissensations", but I do not understand this.) I can inventnames for my sensations. But that is because I speak alanguage in which there are names for sensations. I knowwhat the name of a sensation is. Inventing a name or

7/30/2019 4106594

24/33

CAN THERE BE A PRIVATE LANGUAGE ? 85giving it a name is something that belongs to the languageas we speak it.It is possible, certainly, to invent new expressions,andeven in one sense new languages. But it is a differentquestion whether anyone could have invented language.If language were a device or a method which people mightadopt, then perhaps he could. But it is not that. Andyou could as easily speak of someone'sinventing commerce;more easily, in fact. For he would have to invent what wecall use and meaning. And I do not say so much that thiswould be beyond anyone's powers as rather that it isunintelligible.The expressions of a language get their significanceand their force from their application, from their extensiveuses. Many of them enter in almost everything we do.And this gives them the force and obviousness they havein new contexts. So even if someonedreamedof a languagebefore there was any, how could he put that forward as"a practical proposition"? Or what would he putforward? Marks and sounds would be so much gibberish.To invent a vocabulary he would have at least to inventways of using these sounds in various circumstances-incircumstances of a social life which has in fact grown upwith language and could no more be invented than languagecould. And people would have to understand them.They would have to see not just that this sign occurs hereand that there; they would have to see the difference itmakes if you use the one or the other. And once again thedifficulty is that there would be nothing to understand;because there would be no established use, and nothingwe should call " the difference it makes".

Wittgenstein did not say that the ascriptionof meaningto a sign is something that needsjustification. That wouldgenerally be as meaningless as it would if you said thatlanguage needs justification. What Wittgenstein did holdwas that if a sign has meaning it can be used wrongly.On the other hand, if anyone had tried to invent languageand teach it to others, then you might say the languageand the use of expressionsdid stand in need of justification.

7/30/2019 4106594

25/33

86 R. RHEES.

But why could not a dominant individual have broughtpeople to behave as the people in one of Wittgenstein'sprimitive language games do? Why could he not forcethem to that as we train animals ? Why could he not trainthem to respond in regular ways when ordered, and perhapsto answer ?

Well, no animals have been trained to do even theprimitive things that are done in those language games.Those people are not just going through a complicatedtrick; what they say depends upon what they need andwhat they find. They are not just carrying out orders.They use the expressions they do because they have some-thing to say, and because that use is understood by allparties. Whereas although you may train animals to makethe " correct " responses to different words or signs, theanimals themselves do not use different words. A dog mayrespond in one way to " Slippers ! " and in another wayto " Basket! ", but he does not himself have one soundfor the one and a different sound for the other; neither doeshe do anything like always giving two barks when he wantsfood and one bark when he wants a drink. No traininghas brought an animal to speak, even in a primitive way.This is not a question of the capacities of animals. If anyanimals do learn to speak, they will not learn it just as theylearn tricks. A dog " knows what you want him to do "when you utter the word, but he does not know what itmeans.

If people merely carried out orders and made certainutterances when they were ordered-if this were " makingthe signs they were supposed to make when they weresupposed to make them "-they would not be speaking.I suppose people might be trained to do that with Greeksentences without knowing Greek. And the people in ourexample would not understand what they were saying.They could not do that unless they used the expressionsthemselves, and using them is not just doing what you aretold with them. What we call following a rule in languageis not following orders. That is why we talk about " takingpart in " a language-the language is not any one man's

7/30/2019 4106594

26/33

CAN THERE BE A PRIVATE LANGUAGE ? 87doing more than another's, and the rules, if they are rulesof language, are not one man's rules. This is essential forunderstanding.It might be that when people had been trained as weimagined they would eventually begin to speak. But thatwould not be what was invented, and it would not havecome about by invention. It would have grown up throughthe initiative and spontaneous reactions of varous people,none of whom was inventing language.

We might ask for whomwould anyone invent language ?Or for what ? For animals, for instance ? Or for peoplewho have a social life as we have ? If it is the latter, heneed not trouble, for we have it. But unless it is for thosewho have the kind of social life people with languages dohave-then what is the point and what is he inventing ?What would a " language " for a flock of parrots be, forinstance ? Can you get anywhere except by absurdlyimagining them to live as human beings do, as in children'sstories ?The point is that no one could invent just language.Language goes with a way of living. An invented languagewould be a wallpaper pattern; nothing more.A man might invent marks to go with various objects.That is not language. And when Ayer's Crusoe inventsnameso describelora and fauna, he is taking over more thanhe has invented. He is supposed to keep a diary, too.Ayer thinks that if he could do that when Friday waspresenthe could surelyhave done it when he was still alone.But what would that be-keeping a diary ? Not justmaking markson paper, I suppose (or on a stone or what itmight be). You might ask, " Well what is it when I doit ? And why should it not be the same for him, only abit more primitive ? " But it cannot be that. My marksare either marksI use in communicationwith other people,or they stand for expressions I use with other people." What difference does that make ? He can use them justas I do." No, becauseI use them in theirvariousmeanings.He cannot do that.What is it he cannot do ? What is it that I can do and

7/30/2019 4106594

27/33

88 R. RHEES.he cannot ? There seems to be nothing logically absurdin supposingthat he behavesjust as I do. To a large extentI agree. But it is absurdto supposethat the marks he usesmean anything; even if we might want to say that he goesthrough all the motions of meaning something by them.I should agree that if " meaning something " weresomething psychological, he might conceivably do that.If it were a question of what is put into my mind by myassociationwith other people, then there is nothing logicallyabsurd in supposingthis to come into someone'smind with-out that association." What is it that I can do . . . ? " To say that mean-ing something must be something I do is rather like sayingit is something that happens at the moment. The point isthat I speak a language that is spoken. What I say hassignificancein that language, not otherwise. Or in otherwords, if I say anything I must say it in some language.If there were no more than my behaviour, the marks Imake and so on, then I should not mean anything either.If I say there is " more " than that--it is that I use theexpressionsin the meanings they have. If Crusoeused thesame expressionshe would not do that. Nor can he usedifferent expressionsbut in these meanings. He does notuse expressions n any meaningsat all.Using them in their meanings is what we call followinga rule. For language there must be " the way the expres-sions are used", and this goes with the way people live.I need not live that way myself when I use them. Defoe'sCrusoe could have kept a diary, but Ayer's could not.Defoe's Crusoe'sdiary need never be read by anyone, andthe meaning of what he writes does not depend on that.What he writes down may never play a part in the lives ofother people. But the language in which he has written itdoes. And for that reason he can understand what hewrites, he knows what he is saying. He knows the use orapplication of the expressionshe uses, and it is from thatthey get the significance they have for him. He knowswhat he is talking about. Ayer's Crusoe does not andcannot.

7/30/2019 4106594

28/33

CAN THERE BE A PRIVATE LANGUAGE ? 89Ayer's Crusoe may use marks for particular purposes-

to show where he has hidden something, perhaps-andwith as great regularity as we care to think. This is notwhat we mean by the regularuse of an expressionin a lan-guage. If he should suddenly do something which weshould call using these marks entirely differently, it wouldhave no sense to say that he had done anything wrong oranything inconsistent with what he had done before.We could not speak of his using them in the same meaningor in a different meaning. If he always uses them for thesame purpose-as he might always gather wood for thesame purpose-this is not what we mean by using an expres-sion in the same way. Using an expressionn the samewaydoes not meanusing it for the samepurpose. (What I saidabout identity is connected with this.) And if there is anysort of discrepancy between what I said at one time andwhat I say at another-this does not mean that what I dowith a mark or sound at one time is different from what Idid with it before. If I have always done this with themark,thereis nothing of a rule of language in that." But if he uses them just as they would be used bysomeone who spoke the language, so that they could beunderstood, what is the trouble with saying that he usesthem in their meanings? " The first trouble is that hedoes not understand them. And this really means thathe does notuse themjust as someonewho spokethe languagewould. For he cannot be guided by his signs in just theway in which you and I may be guided by words.This is not a question of something beyond his powers.If we ask whether a machine could follow words,or whethera machine might speak, we are not askingwhat a machinemight be designed to do. It is not a question of capacityor performanceat all.If you say something to me I understand you. If atape recorderplays back what you have said, I understandwhat I hear but I do not understand the tape recorder.Which is a grammatical statement: I do not fail to under-stand either. If I say that you have said something butthe tape recorderhas not, I am not saying that something

7/30/2019 4106594

29/33

90 R. RHEES.

has happened in your case which did not happen in theother. But I do have an entirely different attitude towardsyou and towards what I hear from you, and I behavetowards you in a host of ways as I should never behavetowards a machine-for instance I may answer you, andI should never answer the recorder. I should not try toanswer you either, nor should I suppose you had saidanything, unless I assumed you knew the language; orunless I thought you said something in a language I didnot understand. And I take it for grantedyou arespeakingthe language as it is spoken.I am hardly ever in doubt whether you said something,if I have heard you. But I should begin to doubt if Ifound that you did not follow my answer, and that youdid not seem to know anything about the matters to whichyour words referred. What I should be doubtful about,in that case, would not be whether something went on inyou. I should be doubtful whether you knew what youwere saying. But for all I know you may have " done "all that you would have done if you had. The trouble isthat your utterancewas not a move you were making in theconversation or in the language at all.If I doubt whetheryou know the language, or if I doubtwhether you ever know what you are saying, then in manyways I must regard you more as I should regard the taperecorder. This is not because you do not do anythingthat other people do. It is because you do not take partin what they do. You do not speak the language theyspeak. And speaking he language they speak is not justuttering the words; any more than understanding thelanguage is just " recognizing" the words. It is carryingon a conversation,for instance; or it may be writing reports,or listening to a play in a theatre. It is being someoneto whom the rest of us can speak and get an answer; towhom we can tell something and with whom we canmake a joke and whom we can deceive. All this, and ofcourseimmeasurablymore,belongsto speakingthe language.And it belongs to being able to follow words. You canfollow words because you know how to speak. And for

7/30/2019 4106594

30/33

CAN THERE BE A PRIVATE LANGUAGE ? 91

the same reason a machine cannot follow words. Thishas nothing to do with any question of what physics andengineering may achieve. It is just that it makes nosense to say that a machine might follow words.One can say that absolutely of a machine, but not ofCrusoe, because Crusoe might learn a language. But solong as he never has learned a language, in the sense oftaking part in a language, it is as meaningless to say of himthat he follows words as it would be to say this of an electroniccomputor.I cannot ask whether a machine has made a mistake orwhether it meant what it said. A machine may be out oforder, and then you cannot rely on it. But it is not makinga mistake. (And when I make a mistake myself there isnothing out of order.) A machine may " correct mistakes "in connexion with the operation of negative feed-back.But there is nothing there like a mistake in understanding;nor like a mistake in calculation either. This is one reasonwhy a machine cannot follow words-why that makes nosense. I can follow words only where a mistake or amisunderstanding is at least conceivable. (" Yes, of coursethat's what it means.") Otherwise there would be nothinglike what we call understanding them.I may react to words, rightly or wrongly, when I do notunderstand them. They may be words in a language I donot know, but I may have been taught to obey the ordersof someone who shouts them. Maybe no one else woulduse them in these orders as he does, and that is of no conse-quence to me. It would have been exactly the same ifhe had used sounds of his own instead of words. I mayreact wrongly, as an animal might. But if I call this makinga mistake, it is not like mistaking the meaning of the wordshe uses; any more than I have shown I understand thewords if I make no mistake. I know what he wants, thatis all. (I know enough to get out of the way of a barkingdog, too.) If I had understood the words I should probablyknow what they would mean in other situations; and atany rate I should know what they would mean if somebodyelse used them too. The latter is the important point.

7/30/2019 4106594

31/33

7/30/2019 4106594

32/33

CAN THERE BE A PRIVATE LANGUAGE ? 93been the words of a language. I call their meanings" independent", partly because they have to be learned.That is characteristicof language.Unless the meanings of words were independent-unless they had to be learned-they could not be mis-understood. You do not misunderstand just a sound.You mistake the cry of an animal for the cry of a bird.You may mistake the call of an enemy for the call of a friend.That is not misunderstanding, not in the present sense.If one spoke of learning the meaning of a sound, that wouldnot be like learning the meaning of a word. Perhaps itwould not be nonsense to say that he " knew instinctively"that it was the cry of an animal. But it is nonsense to saythat he knew instinctivelythe meaning of a word.You can misunderstandwhat you can learn. And youare misunderstandinga rule-not a matter of fact. Mis-taking the cry of a bird for an animal cry is not misunder-standing a rule.If one spoke of the independent existence of a tree,this might mean partly that I could think there was notree there and be wrong. But the meanings of wordsare not quite comparable with that, and by theirindependence I do not mean quite the same. If I am wrongabout the tree, I may run into it. If I am wrong about themeaning of a word, it is not like that. It is just that I usethe word incorrectly, or understand it incorrectly. Andthat seems almost like saying that if I am wrong I amwrong. Which in a sense is just what I do mean. Thatis why it is better in this case to say that " the meanings areindependent" meansjust that they have to be learned; asa rule has to be learned. And that is why it is natural tospeak of misunderstandingere; as it is not, so much, whenyou are speakingof a mistakein fact.If anyone did not undersand what kind of mistake it is,he would not understand the difference between correctand incorrect; and vice versa. But then he would notunderstand what words are.Now since you have learned the meaningsof the expres-sions you use, it may happen that you do not mean what

7/30/2019 4106594

33/33

94 R. RHEES.you say. At least it makes sense to ask of anyone who hasspokenwhether he meant it. If he does not mean what hesays, this is familiar and definite enough, but you cannotdescribe it by describing what he is doing. You candescribeit only by taking into account his relation to otherpeople. In this case it is not simply that various peopleuse the same words, although that is a large part of it.What is important is the special r61le r part played by theperson in saying them. That is what his " not meaningthem" is. And it is as characteristic and essential forlanguage as independent meanings are. I have said it isessential that different people may use the same words.But if those people were all doing the same thing, it wouldnot be language. There must be something more like anorganisation, in which different people are, as we may putit, playing different r6les. (The simile limps, but it hassomething important too. It must serve here.) Thatbelongs to the use of language. Without it there wouldnot be words and there would not be meaning.Language is something that is spoken.