Dummett1986_A Nice Derangement of Epitaphs-Some Comments on Davidson and Hacking

5/13/2018 Dummett1986_A Nice Derangement of Epitaphs-Some Comments on Davidson and Hacking

1/10

45 8 L im its o f th e L iter al, ". , . . ,

1 ". ', . / , ")1 , J .', "

L a ng ua ge g am es 26A Nice Derangement of Epitaphs:Some Comments on Davidson and

HackingA disunified picture of language could lead on to the first real use, in many ayear, of Wit tgenstein's idea of a mult itude of loosely inte rlocking languagegames. But a moment 's ref lect ion on Wittgenstein's ideas would also lead us toquery another aspect of Davidson' s s tance. The issue of a private language hasmattered lit tle to Davidson, who is concerned with communication. (But I did,above, quote the phrase 'pr ivate vocabulary' which is perhaps not so innocent,even in context). It takes two for there to be interpretation, interpreter andspeaker. For Davidson, only two. If we think there have to be criteria ofcorrectness, will two people suffice?I said above that malaprops and the ketch/yawl error were two different kinds

of error. It i s a larger background of Engl ish that makes them so. Davidson'spassing theories suddenly seem too permissive. He seems to admit ?nlydiffe rences in pract ice between two people, as opposed to error posed 11 1 alarger linguistic community.

By a community I do not mean some communion of all of the speakers of'English', nor an elitist group of standard speakers. My divide-and-communicate strategy rejects any such fantasies. It assumes only the truism thatmy neighbour and I have some communities in common, and other s that do notoverlap much, although we can join up if so moved. The notion of correctnessarguably arises within those communities._ Such communities must, I think, have numerous members. Lovers, ofcourse, evolve their own ways of talking to each other . They are no counterex-ample . Love is prec isely the relat ionship in which error and correc tness fadeaway. Notor ious ly each lover usually knows what the other is thinking before i tis said.

I think that Davidson's present ph ilosophy has something in common withsolipsism. The solips is t thinks that a private language is not only possible, butthe only one. Davidson is infini te ly far away from that limit point , but he haspassed only from one to two. We might call him a duetist. The possibility ofe rror impl ies that there have to he more than duc ts.We can readily accommodate this possibility within Davidson's observations,but only ifwe become less attached to holism than he has been. Let us concludenot that 'there i sno such thing as a language ' that we bring to interact ion wi thothers. Say rather tha t there is no such thing as the one total l anguage that webring. We bring numerous only loosely connected languages from the looselyconnected communities that we inhabit.

MICHAEL DUMMETT

, i 1 !~:. j, '. !",~-'

Fir st , two point s concerning which Hacking has, I beli eve, misunderstoodDavidson. I lacking cast doubt on Davidson' s idea that 'l inguis tic abili ty shouldbe modelled ... by a recursive theory'; but I suspect that this apparentdisagreement turns on no more than divergent interpretations of the word'recursive'. It is an interesting question whether a syntact ic property such asthat of being a well- formed or grammatical sentence of some natural language,or a semantic relat ion such as that ofbeing synonymous in that language, can beconsidered, at l east to a second approximat ion, to be recursive in the sense of'effectively decidable'. I do not believe, however, that Davidson has everintended to rai se , le t alone to insist on a parti cula r answer to, this quest ion. Ithink, rather, that he uses the term ' recursive ' to mean 'specif ied inductively ' ' .More relevant to our present purpose is Hacking's understanding 01

Davidson's terminology of 'prior theory' and 'pass ing theory'. Davidson treatsof communicat ion in te rms of a conversation between two parti cipants, thespeaker and the hearer, whom, to avoid pronoun trouble, I shall designate Sand II. On Davidson's account, both Sand H have a prior theory and a passingtheory; let us, for the moment, confine our attention to 1 - 1 . I-I 's theories arcabout how to understand S when S is addressing H. The use of the adjective'pr ior' is, however , mis leading, and has misled lIacking about the dis tin,ct ionbetween the two theories, prior and passing, that II has, and the two tha t S has.I l acking takes the prior theo ry of each partic ipant tobe that which he brings tothe conversation at the outset, and his passing theory to be that into which itevolves in the course ofthe conversation. On this account, I I begins with a priortheorv, which comprises his initial propensities to understand in particular wayswhatever S may say to him. In the course of the conversation, he revises thistheory, the theory that thus evolves being his passing theory. This is notDavidson's picture, however; for he speaks of II's prior theory itself asundergoing modif icat ion. Rather, I I has , at every s tage, both a prior theory anda passing theory, both being subject to continual revis ion.What, then, dist inguishes II' s prior theory from his passing theory? If I have

understood correctly, his prior theory is a theory about how, in general, tounderstand S when S addresses II; and this theory may obviously change in

. r . .~h~( ,.. ,. I

~.:'.I J'I I , 'l. . " '- .' ; .'

f " _ . . _I 0 ('


2/10

4 6 0 L im its o f t he L ite ra lresponse to wha t S says in the course of the particula r conversation. His passingtheory, on the other hand, is a theory about how to understand specificutterances of S made during tha t conversa tion. H may gue ss, for example, tha tin a particular sentence S is deliberately using some word in a non-standardway, without any intention of so using it in future, and hence with no wish thatH be prepared for such a future use of it by S. II's ascription to S of anon-standard usc of that sentence then becomes part of his passing theory,whic h conc erns how to understand that pa rticular utterance; but, construing S'sintentions as he docs, Iwill refrain from incorporating this non-standa rd useinto h is evolving 'prior ' theory of how, in gene ra l, to unders tand S when he talksto H.

The terms 'prior' and 'passing' arc in large part responsible for I lacking'smisunderstanding, and arc hetter replaced by 'long-range' and 'short-range'. Ifmy understanding of the distinction between II's prior theory and his passingtheory is correct, there is, however, some confusion in the way Davidson drawsthe distinction between S's two theories, which has contributed to I l acking'smisinterpretation. S's prior theory, Davidson says, is wha t he be lieves II's priortheory to be. In my terminology, then, S's initial long-range theory compriseshis expectations concerning how, in general, I I is disposed to understand Swhen S addresses H. Davidson remarks, however, that S may deliberatelyspeak in such a way as to prompt I I to modify his prior theory; in view of this, heexplains S's passing theory as that which S intends II to usc. Basing thedistinction between S's prior and passing theories on this contrast between hisexpectations and his intentions has the effect of drawing that distinction in adifferent way from that between II's prior and passing theories. The sameprinciple of distinction ought to be used in both cases; and that applied to I Iseems the more illuminating. At any stage before the conversation ends, S willbe making certain suppositions about what, at that moment, arc II's long-termdispositions to understand in particular ways what S says to him. He will alsohave expectations about what these dispositions on H's part will become as soonas IIhas heard what S intends immediately to say to him. It seems a matter ofindifference whether we fdcntity the long-rangc theory held by S at thatmoment with thc former or the latter; but svmmetrv should inhibit us fromdescribing the latter as S's passing theory, since it is concerned with long-termdispositions, that is, with what S intends and expects II to incorporate into hislong-rangc theory. We should take S's passing or short-range theory to relate tohow he intends, and expects, IIto understand particular utterances he makesduring the conversation, when he docs not intend, or expect, H to incorporatehis interpretations of those utterances into his long-range theory.

Davidson evidently regards the distinction between prior and passingthcories as important: if it is important, it is worth while to draw it withprecision,Itwould be natural to think that Davidson is concerned with the well-known

distinction between what a speaker means and what his words mcan; but thiswould be an oversimplification. In an early passage, Davidson refers to thenotion of incorrect usage, saying that it is not mysterious or opcn tophilosophical suspicions, but that it is philosophically uninteresting. Thecontrast here is between what a word means in the common language and what

Comments 01 1 Davidson and H acking 461the speaker wrongly takes it to mean in the common language; but Davidsonattaches little philosophical importance to the notion of the common language.In place of the shallow conception, yielded by this notion, of what words mean,'we want,' he says, 'a deeper notion of what words, when spoken in a context,mean.' This deeper notion will yield a different, and deeper, distinction'between what a speaker, on occasion, means, and what his words mean'. If Iunderstand him aright, this deeper distinction, which has nothing to do withthat between correct and incorrect uses of words, relates to thc divergence,when thcrc is one, between the speaker's long-range theory and his short-rangethcorv: between how hc wants the hearer usually to understand certain wordsthat he has uttered, and how he wants him to understand that particularutterance of thcm. We have not only to draw this distinction, Davidson says, butto safeguard it, since it is thrcatened by thc generalized phenomenon ofmalapropism; to do so, 'we must revise our ideas of what it is to know alanguagc.'

Ilc raises thc topic again by alluding to Donnellan's distinction betweenreferential and attributive uses of descriptions, which hc endorses, saying thatwe need 'a firm sense of the difference between what words mcan or refer toand what speakers mean or rcfcr to'. I think, however, that we need tounderstand the occurrences of the word 'mcan' in this particular sentencedilfcrcutlv from all other occurrences of it in Davidson's papcr, namely as nomore than a synonym of 'refer to'. What hc is hcrc endorsing is Donnellan'sdistinction between two uscs of descriptions; but, hc obse rves, that distinc tionhas 'nothing to do with words changing their meaning or reference'. llc arguesthat if Jones employs the description 'Smith's murderer' in its referential usc,thus referring to someone who did not murder Smith, 'the rcfercncc isnonetheless achieved by way of the normal meanings of the words', which'therefore, must have their usual reference.' He thus allows a distinctionbetween what the phrase refers to, in virtue of its meaning and of thecircumstances, namely the person who in fact murdered Smith, and what Joneswas referring to, in virtue of his heliefs and his having made a referential usc ofthc phrase, namely thc person hc falsely supposed to have murdered Smith. ButI take him to be denying that Jones means anything unusual by thc words'Smith's murderer', in thc morc normal sense of attaching any unusual mcaningto them: in this morc normal scnsc of 'mean', there will therefore bc no roomfor anv distinction between what the words mean and what jones means bythem .. Donnellan's distinction thus proves to bc a distraction: it docs notexemplify either the shallow or the deep distinction of speaker's mcaning fromword-meaning treate d in the ea rlier pa ssage.

Davidson is concerned with a phenomenon distinct from that of rcfcrcntialuse, namely a hearer's understanding of a word or phrase in a sense unfamiliarto him. A speaker cannot attach a meaning to an expression, Davidson says,unless hc intends his hearer to understand him as doing so, and he cannotintend him to understand him thus unless he has a reasonable expectation thathe will. The phenomenon occurs in three kinds of case: that in which, unknownto thc speaker, the hearer is unfamiliar with the expression; that in which thespeaker inadvertently makes a deviant use of a word (the malapropism); andthat in which his deviant use is deliberate. In the third case, hc will drop clues


3/10

46 2 L im its o f th e L it er alprompting the hearer either to modify his long-range theory, or to adopt thecorrect short-range theory, as the case requires, and may therefore wellsucceed in conveying his meaning. In the first two cases, he has a falselong-range theory concerning the hearer, that is, a false idea of the hearer'slong-range theory, but the hearer may sti ll succeed in understanding him as heintended.Deviant uses of language illustrate what Davidson has dismissed as the

shal low distinction between what a speaker means and what his words mean. Ientirely sympathize with Ilacking's recoil from Davidson's dismissive treatmentof the notion of the common language, which underl ies his characterization ofthis way of drawing the distinction as shallow. As Hacking emphasizes,Davidson's paper drives towards the conclusion that ' there is no such thing as alanguage;' but, in the earlier stages of the paper, he deploys the notion of alanguage. The notion of a language that he deploys is not, however, thataccording to which English is one language and Farsi another, hut oneaccording to which 'no two speakers need speak the same language.' It is,rather, that which ident if ies a language with an idiolect, indeed the languagespoken by a particular individual a t a particular time; Davidson speaks of 'mylanguage next year'. He is thus in the tradition of Frcge, for whom, if twopeople attach different senses to a name, they do not speak the same language.This approach does not enta il repudiat ing that notion of a language accordingto which Engl ish and Dutch are languages; but i t involves taking the not ion ofan idiolect as fundamental to an account of what language is , and explaining acommon language as a range of largely overlapping idiolects.Hacking applauds Davidson for adopting, in this paper, a two-sided accountconcerned with 'interaction between speaker and hearer', as opposed to the

one-sided account, exclusively from the viewpoint of the interpreter, adopted inother papers. Davidson is indeed here crucially concerned with the use oflanguage for communication; and he pays as much attent ion to the theories ofthe speaker as to those of the hearer. It is, however, an exaggeration to descrihehim as concerned with interaction. In his picture, there is no interaction, noexchange of the roles or speaker and hearer: the hearer remains mutethroughout the conversation, or, rather, monologue. The hearer can thereforeseek from the speaker no elucidation of what he has said: and i t is this arti fic ia lrestriction that deprives the notion of incorrect use of its interest for Davidson.Figures of speech and other del iberate ly non-standard uses apart, a speakerholds himself responsible to the accepted meanings ofwords and expressions inthe language or dialect he purports to he speaking; his willingness to withdrawor correct what he has said when made aware of a mistake about the meaning ofa word in the common language therefore dist inguishes erroneous uses fromintent ional ly deviant ones. He cannot become aware of any mistake i f the onlyother party to the dialogue is constrained to silence, and for this reasonDavidson takes no note of this feature of his l inguist ic behaviour. For the samereason, Davidson can take no account of the phenomenon called by Putnam'the division of linguistic labour,' or of the closely related and pervasiveexploitation by speakers of the possession by some word of an acceptedmeaning in the common language, even though they themselves have only aPartial knowledge of it.

Comments 01 1 Davidson and Hacking 463A valuable but modest project would be to investigate the means by which ahearer comes to understand non-standard uses of expressions and uses ofexpressions unfamiliar to him, taking for granted as a background an extensiveknowledge of the language on his part. From his denial that he is 'concernedwith the infinitely difficult problem of how a first language is learned ' it might

appear that th is was Davidson's project in 'A Nice Derangement of Epitaphs '.'By comparison', he says, 'my problem is simple. I want to know how peoplewho already have a language ... manage to apply their skill or knowledge toactual cases of in terpreta tion. All the th ings I assume an interpreter knows orcan do depend on his having a mature set of concepts, and being at home withthe business of communication.'If Davidson were engaged on this modest project, I lacking would be quiteright to dismiss as 'pedantic' his insistence that a t ransla tion only tel ls you thattwo expressions of different languages mean the same, not tol iat either means.lie would be wrong, however, to criticize Davidson's continual use, in thispaper, of the term 'interpretat ion' ; for interpretat ion isjust what a hearer has toengage in when the speaker uses some expression unfamil iar to him or in a wayunfamiliar to him. As Hacking complains, Davidson in fact gives 'interpreta-tion' a far more extended application, to any case inwhich a hearer understandswhat a speaker is saying, even when he is making a straightforward use of alanguage the hearer knows. This is one of many things that show Davidson 'sproject to be a more ambi tious one. Another is the final conclusion: the modestproject could not possibly ar rive at the result that there is no such thing as alanguage. A third is what Davidson immediately goes on to say after the passagequoted above about the business of communication: 'the problem is to describewhat is involved in the idea of "having a language" or being at home with thebusiness of communication.' I Ic is not taking for granted the hearer's priorunderstanding of standard uses of expressions of the language, and seeking toexplain in terms of i t how he comes to understand non-standard uses. Rather,he is aiming, by extending the results of a study of the understanding ofnon-standard uses, to arrive at an account of the understanding of language ingeneral.Such an account must explain what communicative abilities are possessed bysomeone who has acquired his first language; all that Davidson can have meant

to forswear, as an infinitely difficult problem, was a claim to explain the processof acquiring it. Even this is rendered dubious by his characterization of his task,in an early passage of the paper, as being to 'describe what an interpreter knowsthat al lows him to understand a speaker, and show how he could come to knowthis on the basis of evidence plausibly available before interpretation hasbegun. ' If the project were the modest one, 'before interpreta tion has begun'could mean 'before the particular ut terance has been made.' Since the project isthe ambi tious one, what the interpreter knows must include what he learnedwhen he acquired his first language, which cannot rank, in this project, asassumed background knowledge; and so 'before interpretation has begun' maybe taken to mean 'before he has learned to talk.' Admittedly, in this earlypassage, Davidson is describing the intention of earlier papers, and is recantingsome of the solut ions proposed in them; but the recantation surely is not meantto cover the sta tement of the problem.


4/10

46 4 Limits o f t he L i te ra lSince the project is ambitious, Hacking cannot reasonably object to David-

son's contrast between a translation manual and a theory of meaning. Aphilosopher oflanguage isnot interested in a theory of meaning as enabling himto master another language, a purpose better served bya translation manual: heis interested in the idea of a theory of meaning as throwing light on whatmeaning is . An actual theory of meaning would specify' the meaning of everyexpression of the language directly, rather than in terms of the meanings ofother expressions, taken as given; the form to be taken by such a theory willtherefore show what it is for an expression to have a meaning. Since atranslation manual docs something quite different, enquir ing what form itshould take cannot lead to an answer to the question what meaning is.Hacking is justi fied, on the other hand, in cavil ling at Davidson's extendedusc of the term 'interpretation'. Where I have spoken of the speaker and thehearer, Davidson speaks throughout of the speaker and the interpreter; where Ispoke of the hearer as understanding the speaker' s words in a particular way,Davidson speaks of his adopt ing a particular interpretat ion of them. A crucia lobservation made by Wit tgenstein in his discussion of following rules is that'there is a way of grasping a rule which is no t an interpretation' tlmxstigations,20I). Similarly , there is a way of understanding a sentence or an ut terance thatdoes not consist in put ting an interpreta tion on it . Whatever the full content ofWit tgenstein's distinction between an AI~O{lSSlIlIl( (way of grasping) and aDeutung (inteprctation) may be, it at least means that one who grasps a rule orunderstands a sentence need not be able to s~ y how he understands it . I Ie docsnot have to be able to say it even to himself: we must not make the mistakeWittgenstein attributes to St Augustine, of describing a child's learning hismother-tongue 'as ifhe already had a language, only not this one' (Inrcst igati tnts ,32) .It is just this that Hacking objects to in Davidson's usc of the word' in terpreta tion '. lie al lows, grudgingly, that the word may be used in anycase inwhich the hearer needs to guess at or search for the speaker 's meaning. In suchcases the hearer may say to himself, 'A ketch must be a two-masted boat inwhich the mizzen is stt!pped forward of the rudder', or, 'lie must mean'''autodidact''', or, 'lie means that it is showy and over-ornate.' But when thehearer does not have to search for the speaker' s meaning, but takes for granted

that he is using words in just the way with which he is familiar, there is, asHacking says, no process of interpretat ion going on; to assimilate this case tothe other by describing it as an instance of 'homophonic' interpretationdisguises the radical difference between them.Homophonic interpretat ion is not impossib le . Listening to Mrs Malaprop, Imight say to myself with surprise, 'By "incorrigible" she really docs mean"incorrigible"' , or, 'She means by "incorrigible" what is normally meant by it .'In normal cases, I should have no such thought; even in this case, though myhaving it explains my attaching that meaning rather than some other to MrsMalaprop's ut terance of the word, i t makes no contribut ion at a ll to explainingin what my attaching that meaning to it consists. The reason is obvious: myhaving that thought wil l not result in my attaching the standard meaning to herutterance unless I know what that standard meaning is; and the thought, asformulated, docs not embody the content of that knowledge.

Comment s all Davidson and H acking 465Just because interpretation depends upon having a language in which theinterpretation is stated, the difference between a translation and a proposition

of a theory of meaning isnot significant when the term 'interpretation' istruly inplace. Itmakes no difference whether the hearer says to himself, ' ''Ketch' ' mustmean "a two-masted boat in which ... "', or, 'A ketch must be a two-mastedboat in which ... " or, 'Itmust be that "is a ketch" is true of a thing if and only ifthat thing is a two-masted boat in which ... '. It would indeed be pedant ic toinsist that he might have the first of these three thoughts without understandingthe definition that he had formulated: what is common to all three is that hisadoption of the in terpreta tion explains his understanding of what the speakersays only because we assume that he grasps the sense ofthe phrase in which theinterpretation is stated. Davidson claims that ordinary speakers may plausiblybe said to 'know certa in theorems of a theory of t ruth' for thei r language, ' suchas "An utterance of 'The cat is on the mat' is true if and onlv if the cat is on themat".' Here Frcgc's pedantic usc of distinct terms for the sentence stating atheorem and the thought expressed by that sentence would stand us in goodstead. A speaker 's knowledge of the t ruth of the sentence sta ting the theorem inno way guarantees his understanding of the sentence 'The cat is on the mat.'1li s knowing that an utterance of the la tter sentence is true just in case the cat ison the mat may possibly do so; but, from this thesis, unsupplemented by anaccount of what his having that knowledge consists in, we gain little insight intothe character of his understanding.The occurrence of the phenomena that interest Davidson isincontrovertible:hut how can an investigation ofthem lead to the conclusion that there is no suchthing as a language? Oppressive governments, such as those of Franco andMussolini, attempt to suppress minority languages; under such regimesteachers punish chi ldren for speaking those languages in the playground. InIndia, crowds demonstrate against the proposal to make Hindi the sale offic iallanguage. Bretons, Catalans, Basques and Kurds each declare that theirlanguage is the soul of their culture. The option does not seem to be open to usto declare that such governments and such peoples are under an illusion thatthere is anything they arc suppressing or cherishing. I know that there arcpeople' who know Swahili , Burmese and Slovene, and that I know none of theselanguages; moreover, I know just what I have to do to acquire such knowledge.In what docs the fact that I know no Yoruba consist? Not , surely, in the fact thatI have no tendency to form the same short-range (passing) theory of theutterances of a Yoruba speaker on a particular occasion as he has. Malav is alanguage whose speakers cultivate allusiveness, and so, with 'Malav' substitutedfor 'Yoruba ', that might be qui te a good description of someone who knew onlvwhat is cal led bazaar Malay: but my ignorance of Yoruba is more fundament~1than that. It consists, in Davidson's terms, in my lacking any long-range theorywhatever for Yoruba speakers; or, rather, s ince a great many Yoruba speakersalso know English, in my lacking any such theory for them when they arespeaking Yoruba. It thus appears that there is no way to characterize th is pieceof ignorance Oil my part without appeal to the concept of a language. It is truethat Davidson tags to his conclusion that 'there is no such thing as a language'the proviso 'if a language is anything like what philosophers ... have supposed':but he offers no al ternative account of what a language is. Whatever force his

t~-


5/10

46 6 L im it s o f t he L it er alarguments may have , they cannot sustain the bald conc lusion, but cry ou t forsome account of an indispensable concept .Davidson' s argument is this . The apparatus ofpr ior and passing theor ies, or,

as I have called them, long-range and short-range theories, is required toexplain the phenomena, of malapropisms and of deviant and unfamiliar uses,that interes t him. It is also sufficient to explain linguis tic communication ingeneral . But ne ither the short -range nor the long-range theo ry has the rightform to be described as a theory of a language . Hence the notion of a languageis no t needed for the phi losophy of language , but only tha t o f language.As applied to a short-range (passing) theory, this argument is cogent. Ashort-range theory is a theory of how specific utterances are to be understood.Davidson appears to conceive of i t as massively reduplicating the long-range

theory ( 'Mrs Malaprop's theory, prior and passing, is that . .. ' ), but there seemsno good reason for this: i fwe think of the two theories as used in conj unc tion,the short-range theory may be taken as bearing only on those utterances forwhich the long-range one docs not yield the correct interpretation. Theshort-range theory, so viewed, will not be a structured theory, but only acollect ion of disconnected propositions. A theory of meaning for a language, bycontrast, relates to words ami expressions as types, and must be st ructured inthe manner on which Davidson insist s when he demands tha t i tbe recursive. Along- range theo ry, on the other hand, wil l resemble a theory of mean ing for alanguage in both these respects .At th is stage we may well wonder just mhidi notion of a language it is thatDavidson is repudiating. I f i t isour ordinary notion of a language, according towhich one per son may be sa id to speak six languages ami another to know nonebut his own, then the breast-beating is out of place ('I must revise "!)I ideas','the trouble I am in'). For I Javidson's resume of the enterprise of his earlierpapers sets aside this ordinary notion as shallow, and treats as fundamental thequite different one of an idiolect; and, though the long-range theories of 'ANice Derangement of Epitaphs' cannot be equated to idiolects, they do notdif fer f rom them so sharp ly as to warrant cries of 'mea culpa'.

In this ta lk of theories, \~eneed to take care what we consider each theory tobe a theory of. According to Davidson, the speaker's long-range theory is atheory of what the hearer's long- range theory is; we may ca ll any theory abou twhat another theory is a second-order theory. There is certainly a place forsecond-order theor ies, s ince eavesdroppers, as well as speakers and hearers ,need to engage in interp retation. In order to form a t rue pic ture of what is go ingon, the eavesd ropper needs to dec ide what the speaker intends to convey to thehearer; he also needs to dec ide what the heare r takes the speaker to mean . Hemust therefore, in Davidson's terms, frame a theory of what the speaker's(long-range and short -range) theo ries a re, and he must a lso frame a theory ofwhat the hearer's theories arc; the two theories he frames arc both second-order theor ies. A literary cri tic is an eavesdropper on a monologue bythe authoraddressed to those of his contemporaries, or those members of hi s poste rity,whom he conceived as his readers; his criticism consists in presenting asecond-order theory and giving reasons for i t. The literary histor ian presentssecond-order theories about the interpretation of literary works by theircontemporary readers . We must beware, however , of representing all theor ies

Comments 01 1 Da vi ds o n a n d Ha ck in g 46 7as of second order. We cannot say both that the speaker's theory is a theoryabout wha t the heare r's theory i s, and tha t the heare r's theory is a theory aboutwhat the speaker 's theory is, without fal ling into an inf inite regress: there mustbe some first-order theory. A first-order theory iss imply a theory of meaning: i ti s i rrelevant for present purposes whether, as Davidson bel ieves, it takes theform of a truth theory, or some other.What does mat te r is it s scope. We must distinguish between three things: alanguage ; a theory of meaning for that language; and a second-order theo ry. :\

language isan exist ing pattern of communicative speech: i t is not a theory, but aphenomenon. A theory of meaning for that language, as conceived by Davidson,is a theory of the content of expressions belonging to it - what that content isand how it i s determined by the composit ion of the expressions. It docs notitself employ the notion of meaning, but can be recognized as an adequater epresentation of the meanings of words and expressions of the language. Itserves to explain how the language functions, that is, to explain the phe-nomenon of speech in that language; but it does so only indirectly. This is sobecause the theory it se lf conta ins no refe rence to speakers or to thei r beli efs,intentions or behaviour, linguistic or non-linguistic. Instead it uses theoreticalt erms such as 'tr ue' applying to certa in expressions of the language. In order,the refore , to assess a theory of mean ing for a language as corr ect or incorr ect ,we III l1st have some principles, implici t or explici t, that make the connectionbe tween the theoret ical notions and what the speakers of the language say anddo. These principles arc not regarded by Davidson as part of the theory ofmeaning itself: they arc presumably constant over all theor ies of meaning andall languages . We may call them the linking principles. Save in their presence, i twil l make no sense to speak of beli eving or disbe li eving a theory of meaning,s ince they alone provide a cri terion for i ts being correct .

A second-o rder theory is of a qu ite dif ferent character. If i t is a long-rangetheory, i t consi st s of a se t of be li efs about wha t expressions of some languagemean, or about what certain individuals intend or take them to mean; if it is ashort-range theory, then about what cer tain specif ic utterances were intendedor taken to mean.Two unresolved questions leave i t uncertain whether we ought to restri ctsecond-order theor ies to those constitut ing interpretat ions in the s tr ict sense.

One arises from our having left the linking principles tacit: we do not knowwhether o r not they connect the theory of meaning to the l inguisti c prac ti ce byimputing theories to the speakers, or, if so, whether these are first- orsecond-order theories. We get no help here from Davidson, who docs notdistinguish between second-order theories and theories of meaning, norbe tween interp retations in the stri ct sense and ways of understanding some-th ing in general (between a Del/tung and an Aullassung). He says, for instance,that 'Mrs Malaprop 's theory . .. i s that "a nice derangement of epi taphs" meansa nice arrangement of epithets. ' It is plainly Davidson's (correct) theory aboutMrs Malaprop that by 'a nice derangement of epitaphs' she means 'a nicearrangement of epithets'. This is a second-order theory, or part of one, andconstitutes an interpretation in the strict sense: there is therefore nothingproblematic in attributing this theory to Davidson, or in describing him asbelieving it. It is less clear that Mrs Malaprop herself may be said to have a


6/10

468 L im its o f th e L it er alsecond-order theory. It is indeed highly natural to say that Mrs Malapropbe lieves tha t 'a nice derangemen t of ep itaphs' means 'a n ice a rrangement ofepithets'; but there is then a problem about how she represents this belief tohersel f, since she certa inly does not represent i t in those words. We are heretalking, not o f an inte rpre tat ion in the str ict sense, but of how Mrs Malapropunderstands the words , or ofwhat she means by them. I t isunclear whether thisis a matter of her holding some belief, properly so called, at all, or, if so,whether the belief is of second or of first order: whether, that is, it is part ofsome second-order theory which she holds, or should be schematicallyrepresented by attributing to her some theory of meaning. .Only the linking principles, when made explici t, wil l reveal whether a theoryof meaning for a language isgenuinely an object ofknowledge, in some mode of

knowledge, on the par t ofeveryone who knows that language. Independently ofwhether it is or not, however, someone's knowledge of the language may beschematically represented by attributing such a theory to him. In knowing thelanguage, he thereby attaches to express ions belonging to it the meanings theyhave in that language. The theory of meaning purports to provide the onlypossible analysis ofwhat i t is lor those expressions to have those meanings. Thespeaker himself may be ignorant of that analysis: but , by attributing to him thecorresponding theory of meaning, we simultaneously represent him as attachingthose meanings to the express ions and supply an analysis of their having thosemeanings. In the same way, a second-order theory i s not l ite ral ly to the effec tthat such-and-such a theory of meaning is correct, or is the theory held by acertain ind ividual . Its li ter al content i s tha t certain expressions have certainmeanings , or that some individual intends them to have or takes them as havingthose meanings. Since a theory of meaning makes explici t what i t isor would befor them to have those meanings, however, the second-order theory can herepresented as a theory about that theory of meaning.The second unresolved question is in what manner we should, in the first

instance, identify the languages to which the apparatus of theor ies of meaningand second-order theories i s to apply. We must make such a choice, becausethe application of the entire-apparatus waits upon a selection of the languages towhich the theories of meaning relate. Whatever our choice, we shall have toleave a place for broader or narrower ways of using the te rm 'languages'. Wemust, however , take one way of using it as fundamental for our account. It willbe theor ies of meaning for languages in this fundamental sense that we shal lneed to consider; we sha ll not need to take account o f theories of meaning forlanguages in a broader or narrower sense , which we sha ll be able to exp lain interms of languages in our fundamental sense and thei r associated theories ofmeaning.One natural choice for the fundamental notion of a language is that of acommon language as spoken at a given time - either a language properly socall ed, such as Engl ish o r Russian, or a dia lect of such a language. If we makethis choice , we sha ll have to acknowledge the part ia l, and part ly e rroneous,grasp of the language that every individual speaker has. The theory of meaningfor the language wi ll give the meanings tha t i ts words and expressions in fac thave. Davidson, as we saw, docs not consider this notion open to philosophicalsuspicion, j ust uninteresting. Actua lly , it is ne ither so uninte resting nor so

C om m en ts o n D av id so n a nd H ac kin g 469unproblematic as he supposes; but let us take it, for present purposes, as given.In any account of actual l inguis tic practice, we must recognize that no speakerknows every word of the language or uses correctly every word he does know: inother words, the linking principles for a theory ofmeaning for a language in thissense of ' language' must employ the notion of an idiolect. The obvious way fori t to do so isby treating an idiolect asa second-order theory: a par tial , and par tlyincorrect, theory about what the meanings of the expressions are in thecommon language, that may be represented as a partial theory of what thecorrect theory ofmeaning for the language is. Such a second-order theory is 1I0tan interpretat ion in the s tr ict sense; given this choice ofthe fundamental notionof a language, then, we cannot restrict second-order theories to thoseconsti tut ing such inte rpretat ions. To be quite accu rate , the second-o rdertheory in question will not give a perfect representat ion of a speaker 's idiolect,since, strict ly speaking, his idiolect comprises only his own speech habits, andall speakers of, say, English both understand certain expressions that theywould never use and recognize them as correct English. But this is anunimpor tan t detai l: on the p resent view, a common language is r ela ted to anid iolect essent ia lly as the ru les of a game arc rela ted to a player's he li efs aboutwhat they arc.

Davidson's o riginal choice for the fundamenta l not ion of a language wasqui te d iffe rent , namely to take the language as const ituted by the l inguisti chabits of an individual at a given time. Here the idiolect becomes fundamental ,and hence no longer explicable as a second-order theory: it is, rather, that towhich a (first -ordcr) theory of meaning relates. 'A Nice Derangement ofEpitaphs' insists that this choice of fundamental notion is defective. Theground offe red for this is that we a ll adapt our mode of speech to our audience .The observation is just: once we have rejected the common language as thefundamental notion, we cannot stop at the idiolect, but must make a finerdiscriminat ion. We must take a language in the fundamental sense to consist ,no t of the genera l speech hab its of an individual at a parti cular time but rather

" ,his habi ts of speech when addressing a part icula r hearer at that time .None of this jus tifies the claim that we must jet tison the notion of a languagealtoge ther . We cannot dispense wi th it, fo r, if we do, ou r theories of meaninghave no subj ec t-mat ter. Davidson has not shown how to do wi thou t it : he hasmerely made a proposal for a new way of picking out a language, in thefundamental sense, those languages, namely, of which we are to take thetheor ies of meaning as treat ing. Given this new principle for circumscribing thelanguage, a ll goes ahead as before . Granted, Davidson has introduced somenew features into the apparatus. The short-range theories, which relate tonon-standard interpretat ions of par ticular utterances - interpretat ions deviantby the standard of the long-range theories - are one such novel ingredient. Ifthe distinction between first- and second-order theories be accepted, thesecond-order theor ies are another . But the f irst-order long-range theor ies aresimply theories of mean ing for languages ident ifi ed by means of a fine r gridthan before.

It might be obj ec ted tha t I have missed the point. I exp lained a language as apattern ofcommunicative speech. Davidson isnot concerned to deny either thatcommunica tive speech occurs, or tha t the re a re pat te rns discernible in it: he is


7/10

47 0 L im its o f th e L ite ra ltherefore not denying that there are languages, if this is what is meant by'languages'. Rather, he is equating a language with a theory of meaning; andwhat he is saying is that neither a long-range nor a short-range theory isanything like a theory of meaning, as he formerly conceived of it. That ashort-range theory does not resemble a theory of meaning isclear: but what elsecan a long-range theory be, if it is of first order? Since it seems that there can be no retort to this question, this defence ofDavidson must claim that all long-range theories are of second order. Thisclaim makes no sense, however. Since a theory of higher order is a theory aboutwhat some other theory is, to say that no theory is of first order is to say thatevery theory isof infini te order; but there can be no theories of infin ite order. I tis not inco;lceivable that the theories we need to pay most attention to are oforder much higher than the second: for instance, a speaker 's theory of what hishearer 's theory of what the speaker' s theory of what the hearer' s theory ofwhatthe speaker 's theory is. It is in principle impossible that there should be anytheories a sta tement of the subject -mat ter of which would require a phrase likethe foregoing, but one that could never be completed; let alone that all theoriesshould be of this k ind . Ifsomeone holds a theory of fini te order, then the theorylast referred to in the phrase sta ting i ts subject -mat ter must be a theory of firstorder: and this will be a theory of meaning. Davidson refers to the Gr iceanci rcle: he cannot give a coherent account if he converts it in to an exit less loop.Davidson's tendency, in 'A Nice Derangement of Epi taphs ', to suppose thatevery theory must be of higher order is due to his adoption of a quasi-Griccanaccount of what it i s to mean something by an expression. There arc two naturalpictures of meaning. One depicts words as carrying meanings independently ofspeakers. It was to this conception that Alice was appealing when she objectedto Humpty Dumpty, 'But "glory" doesn't mean a "nice knock-down argu-ment". ' According to Al ice, Humpty Dumpty could not mean that by the word,because the word itsel f d id not have i t in i t to bear that meaning. The oppositepicture is that which Humpty Dumpty was using. On this conception, it i s thespeaker who attaches the meaning to the word by some inner mental operation;so anyone can mean by~glory' whatever he chooses. Each picture is crude; eachis easily ridiculed by a phi losopher or a linguist. But each theorist of languagetends to offer a more sophisticated version of one picture or the other.Davidson's is a version of the second picture. His reason for denying that, by'glory', Humpty Dumpty could, in speaking to Alice, mean 'a nice knock-downargument ' i s qui te different from Alice's: it is that Humpty Dumpty knew thatAlice would not understand him as meaning that. If he had expected her so tounderstand him, and had intended her to do so, then his intention, or, perhaps,his expecta tion , would have conferred that meaning on the word as ut tered byhim. Given this picture ofwhat meaning is, it appears to follow that the relevanttheories are all of higher order, being theories about the theories held by otherparticipants in the dialogue.Language is both an instrument of communication and a vehicle of thought;it is an important question of orienta tion in the philosophy of language whichrole we take as primary. I welcome Davidson's attention to the communicativefunction of language, since I am disposed to take that as its primary role:language is a vehicle of thought because it is an instrument of communication,

COII/II/el/ts 01 1 D avi ds on a nd H acki ng 47 1and not convcrselv. And vet it is an error to concentrate too exclusively oncommunication. Wittgenstein iswell known to have taken language primarily asa social activity; and yet his challenge to say one thing and mean something elsethereby has to do with language in its role as a vehicle of thought. The difficultyof saying, ' It' s cold here' , and meaning, ' It' s warm here' i Im-est igations, 510), orof saying, 'The sky is clouding over ', and meaning, 'There is no odd perfectnumber', does not lie in the presence of an audience, but is just as great if youarc saying it to yourself, as in the experiment Wittgenstein surely intends.Davidson's mode of explanation isimpotent here: the difficulty is not that I haveno reason to expect myself to know what I mean. Alice was more nearly right:the difficulty lies in the fact that 'The sky is clouding oyer' does not mean'There is no odd perfect number. 'The choice of idiolects , or, rather, of linguistic practices yet more narrowlycircumscribed, as languages in the fundamental sense, makes the matter easierto understand in the first instance, but in the end, I think, renders itincomprehensible. It makes it easier at the outset because such a choice of thefundamental notion of a language leaves us free to treat all higher-order theoriesas interpretations in the strict sense, that is, as pr inciples for construing thespeech of others that the subject can state to himself: such a principle, forexample, as that when one whose mother-tongue is German says, ' I mean that... " he is likclv to mean, 'I think that ... " or that when he says, 'If I would do. .. " he probably means, 'If I were to do ... '. Now a speaker 'who suspects hishearers' knowledge of the language to be imperfect, or his usc of it in someother way to diverge from his own, wi ll be careful to avoid expressions l iable tobe misunderstood, but will very rarely misuse any expression in order to accordwith his hearer's faulty understanding of it, or even employ an expression whichhe (the speaker) finds unnatural. Hence the speaker's knowledge of thelinguistic abi li ties and habits of the hearer wil l, for the most part, have only aninhibiting ef fect, prompting him to avoid certain forms of expression. I listheory about the hearer' s l inguistic propensit ies - his second-order theory -will thus not be that in virtue of which what he does say has the meaning that itdocs; that, rather, i f i t is tobe called a theory at a ll , wil lbe his first -order theory- his theory of meaning for the language he isusing to address that hearer. Thehearer , on the other hand, I llay well appeal to principles of interpretation inorder to hit on the speaker 's meaning. In describ ing the speaker as employing along- range theory of what the hearer's theory is, Davidson has therefore putthe matter the wrong way round. It is the speaker's first-order theory thatdetermines the meanings of his remarks; the hearer may advert to his ownthcorv of what the speaker' s first -order theory is in order to in terpret him.Al l this, however, is a deta il . The essential point is that in terpreta tion, in thest rict sense, i s of necessi ty an exceptional occurrence; and, by the same token,the appeal to a second-order theory is of necessi ty the exception. In the normalcase, the speaker simply says what he means. By this I do not mean that he firsthas the thought and then puts i t into words, but that , knowing the language, hesimply speaks. In the normal case, likewise, the hearer simply understands.That is , knowing the language, he hears and thereby understands; given that heknows the language, there is nothing that his understanding the words consistsin save his hearing them. There are, of course , many except ional cases. They


8/10

47 2 Limits o r the Literal Comments on Davidson and Hacking 47 3speaker need have no reasonable expectation of being understood, as anyonewho has observed the behaviour of tourists can tes tify. Second-order theor iesmay be brought into play when the speaker wants to communicate and isconscious of obstacles to understanding: in the normal case, he takes forgranted that hi s heare r speaks his language, just as a chess-player takes forgranted that his opponent knows the moves of the pieces.

A language is a practice in which people engage. There can be solitarypractices, in the sense of those in which one engages on one's own; but apractice is essentially social, in the different sense that i t is learned from othersand is constituted by rules which it is part of social custom to follow. Theconception of a practice in which only one person ever engages is thus veryquestionably coherent. We might reinterpret Davidson as taking, as hisfundamenta l notion of a language, the language spoken between any pair ofindividuals who converse with one another , so that each language constitutes apractice in which just two people engage. This is plainly contrary to Davidson'sintentions, according to which the roles of speaker and hearer arc notinterchanged; and i t docs not tal ly wi th the idea that the speaker is opera tingonly with a second-order theory, or, under our proposed emendation, that thisis true of the hearer. The natural choice for the fundamental notion of alanguage, from the viewpoint that sees a language as a practice, isa language inthe ordinary sense in which English isa language, or, perhaps , a dialect of sucha language.The view that I am urging against Davidson is an adaptation of Alice'spicture, according to which words have meanings in themselves, independentlyof speakers. Of course, they do not have them intrinsically, and henceindependently of anything human beings do. They have them in virtue ofbelonging to the language, and hence in virtue of the existence of a socialpractice. But they have them independently of any particular speakers. Nospeaker needs to form any express intent ion, or to hold any particular theoryabout his audience, or, indeed, about the language, in order to mean by a wordwhat it means in the language : he has only toknow the language and tou tter theword i ll an appropriate context , such as that of a sentence. In particular cases,he may have some interpretat ion in mind; hut he could not have an interpreta-tion in mind in every case. This is to say that we cannot grasp senses withoutany vehic le for them, and associa te those senses to the words as semaphoreassociates f lag posit ions to let ters: to inves t a word with a sense is jus t to graspthe pat tern of its usc.All this rai ses interest ing quest ions about how a language changes, which

must here be se t aside. These questions have much to do with the phenomenathat interest Davidson. They are not, however, to be answered in such a wav asto make expression of meaning depend wholly upon the intentions of thespeaker , and thus to liberate speakers from all responsibili ty to the language as asocial institution. The crude version of such an answer is the picture oflanguage favoured by Ilumpty Dumpty. Davidson 's is a ref inement, which stil lcommits the same essential error . Davidson' s ref inement consists in insist ingthat the speaker must have regard to the hearer' s discernment of his intention.Without the background of the common prac tice of speaking the language , i tremains mysterious how the hearer can discern it; with that background, it

occur, for the speaker, whenever he uses a word with which he suspects thehearer may be unfamiliar, whenever he very deliberately uses a subtle orhumorous figure of speech, when he is consc ious of knowing the languageimperfectly or believes his hearer to do so. They occur , for the hearer, when thespeaker misuses an ex-pression, or uses an unfamiliar one, or employs a difficultf igure ofspeech, or , again, when one or other isspeaking in a language he isnotfully at home in. These arc just the cases that aroused Davidson's interest.Th~y are, however, in the nature of things, atypical cases: if taken as prototypesfor l inguis tic communication, they prompt the furmulation of an incoherenttheory.H~w can there be cases of the kind I have labell ed 'normal '? In such a case,speaker and hearer treat the words as having the meanings that they do in thelanguage. Their so treating them docs not consist in thei r having any beliefsabout the other person, but, rathe r, in the ir engaging in the way they do in theconversat ion, reac ting as they do to what the other says, and, perhaps, act ingaccordingly after i t is over. They may he compared to players ofa game. A gamewi th two or more players i s a soc ial act ivity; but the players' grasp of the rulesdoes not consist in any theories they have about the knowledge of the rules onthe part of the other players. It is manifested by their playing the game inaccordance with the rules (or cheating bysurreptitiously breaking them), that is,in their act ing in a manner that makes sense only in the light of the rules . Thereisa greater problem in the linguis tic case. The players of the game can probablystate the rules , perhaps only haltingly, ifchallenged to do so, even though theyse ldom actua lly advert to those rules when playing. Speakers of a languageknow the meanings of the words belonging to i t, but arc frequently unable tostate them; it is i l l principle imposs ible for their knowledge ofthe language, if i tbe their mother-tongue, to consist in i ts entirety in knowledge that they couldstate. We may represent thei r knowledge as a grasp of a theory of meaning forthe language; but we face a grea t difficulty in explaining the characte r of thisrepresentation. This, however, is a problem which, though central to thephilosophy of language, is not to be solved by replacing the theory of meaningby a second-order theory about the theories he ld by other speakers.What, in the normal case, both speaker and hearer are going 011 arc theirbeliefs ( if thev can he called beliefs) about what the words mean, not about whatthe other takes or intends them to mean; these second-order beliefs arcoperative only in the exceptional cases. To say that they go on these beliefs is tosay that they are engaging in a practice that they have learned, just as thegames-players do: they speak and act in accordance wi th the meanings of thewords in the language, as the players act in accordance with the rules of thegame. But is i t poss ible to mean something by an express ion without intendingone's hearer to understand it? If the hearer is a mere eavesdropper , i t cer tainlyis, for instance when soli loquiz ing aloud or speaking in an aside ; i f Gali leereally said, 'Eppur, si I11UOVc', that would be a case. It is even possible when thehearer is the person addressed. In To p flat, the character played by Eric Bloreaddresses a string of insults in English to an I talian policeman, and is dismayedto find that he has been understood. Naturally, one cannot fail to intendSomeone to understand what one is attempting to convey to him. That,however, isa trivial tautology, not a deep truth about meaning; and, even so, the


9/10

474 Limits ofthe Literal Comments Oil Daoidso lind Hacking 475cannot be defined geographically: it cannot be defined without using theconcept of a language. But with what possible right can Davidson makeessential appeal to this concept in the course ofan argument purport ing to showit to lack application?Taking a language in the fundamenta l sense to be such a thing as Japanese orBengali involves allowing that a second-order theory need not be an intcprcta-

t ion. We shall need to dis tinguish between the theory of meaning for a languageand all individual's imperfect grasp of that theory. The lat ter isof second orderin so far as that individual purports to be speaking the given language, andhence is di sposed to modify what he says i f he becomes aware of a linguisti ce rror. Since it is not an interpreta tion, however, it opera tes as we have beenconsidering a first-order theory to do: it is not the content of a piece oftheore ti cal knowledge which the speaker has and to which he can advert. Thelinking principles f in a theory ofmeaning for a language of this kind willbe verycomplex, since they have to describe an immensely complex social practice: theywill treat , among other things, of the divis ion of l inguis tic labour, of the usuallvi ll -def ined sources of l inguis tic authori ty, of the different modes of speech andthe relat ions between the parent language and var ious dialects and s langs. Butthis is no obj ect ion to such a choice of the fundamental not ion of a language; noother choice can find any place for phenomena of this sort, wh ich arc genuinefea tures of the human usc of language .The attribution to a speaker of a theory of meaning, or of an imperfect andpartial grasp of one, is only schematic: he in no sense literally knows such a

theory . What he has done i s to master a practi ce ; and the notion of a p ract ice isone requiring rather careful philosophical character izat ion. To the questionwhether mastery of a practice is theoretical or practical knowledge we can onlyreply that the categorization is too crude: it falls between. A character inWodchousc is asked whether she can speak Spanish, and replies , ' I don 't know:I've never tri ed.' Suppose she had been asked whether she could swim, and hadgiven the same answer. The answer would sti ll have been absurd, but it wouldhave been absurd for an empirical reason only, namely that we know fromexperience that human beings arc unable to swim unless they have been taught,something that is not true of dogs. This fact - that , if you are to swim, you mustlearn to do so - appears tobe the only reason for our speaking of knonnng how toswim; and thi s i s so for al l pure cases of prac tical knowledge. It is not the samefor knowing Spanish. Even i fyou have not been taught to swim, you can t1) ' toswim, though you will surely fail. You can try because you know what swimmingis; you can tell, for sure, whether someone else is swimming or not. If you donot know Spanish, on the other hand, you cannot so much as tly to speakSpanish, and you cannot tell, for sure, whether someone else is speakingSpanish or not. You cannot do so because you do not know what it is to speakSpanish. This is what characterizes mastery of a practice: you cannot knowwhat the pract ice is - though you may be aware tha t i t resembles certain othersin sta table respec ts - unti l you have mastered it . But , a lthough this makes i t agenuine case of knowledge, unlike practical knowledge, which is knowledgeonly by courtesy, i t docs not make it theoret ical knowledge.Why, then, is it so compell ing to represent maste ry of a language as if i t were

theoretical knowledge? Although there is no one who knows what i t is to speak

becomes easy to explain how he can do so in the exceptional cases in whichdiscernment is necessary. That I am not car icatur ing Davidson is shown b)' hisconcluding remarks, directed against the role of convention in language.Conven tions, whether they be expressly taught or picked up piecemea l, arcwhat constitute a social practice; to repudiate the role of convention is to denythat a language is in this sense a practice. In the exceptional cases - those inwhich it is in place to speak of interpreting what someone says - there areindeed no rules to fol low: that i s what makes such cases except iona l. In suchcases, the hearer has to apprehend the speaker's intention much as he has toapprehend the intent ions behind non-l inguistic ac tions (the par t played byconvent ion in which i t is easy to overlook). This feature is a lways present indiscourse, since, bes ides grasping what someone means, in the sense ofwhat heis saying, we must make out why he says i t - what re levance he takes i t to have,wha t point he i s d riving at, and so on. Davidson would l ike to be lieve tha t ourwhole understanding ofanother' s speech iseffected without our having to knowanything: 'there is no such thing' as a language, he says, 'to be learned ormastered', and the implication is that there is nothing to be learned ormastered. He claims to 'have erased the boundary between knowing a languageand knowing our way around in the world general ly ': we understand somconc'sspeech by divining his intentions, as we divine those informing his actions . Butthe distinction is that ordinary actions make a difference to the worldindependently of any conventional signif icance they may have, whereas utter-ances seldom make any noticeable difference independently of their meanings:without a knowledge of the language, we therefore do not have the sa lli e cluesto the intentions behind utterances, because, without understanding, we do notknow what the speaker has done by utter ing those sounds. Since the meaning isa ll, or a lmost a ll, that mat te rs, i t would make no d iffer ence what sounds weremade , as long as the intent ion behind them remained constant, if things weretruly as Davidson supposes, and there were no language to which theybelonged. But, to quote Euclid , this is absurd.The need fo r the not ioq of a language i s apparent if we ask what long-rangetheory someone brings to a f irst l inguist ic encounter with another . Davidsonanswers th is quest ion by saying, 'We may know so lit tl e abou t our intendedinterpreter that we can do no better than to assume that he will interpret ourspeech along what we take to be s tandard lines .' This assumption is dreadfullyfamiliar: i t is that which used to made by many British tourists, often expressedin the words , 'They will understand you perfectly well as long as you speak loudenough. ' This, i t may natural ly be said, is unfair: Davidson was not thinking ofan encounter with a randomly selected human being, but of one with afe llow-count ryman. The word 'fe llow-count ryman' represents an evasion,however. It would be quite inappropriate in Belgium, Switzer land, the SovietUnion, India or most African countries: even the United States is not asl11onoglot as it pretends. What i s rea lly meant is an encounter wi th someonepresumed to speak the same language. I twould, after all , be r idiculous to speakof the standard manner of in terp reting the speech of human beings in genera l;there is no such thing. All this, indeed, is express ly acknowledged by Davidsonin this very passage, for he inser ts the proviso 'assuming we know he belongs toour language community '. What, then, is a language community? It obviously


10/10

4 7 6 L im it s o f t he L ite ra lSpanish other than those who speak Spanish, i t i s not str ic tly true that there isno other way to come to know what speaking Spanish is. Theoreticalknowledge, the content of which was a ful ly explici t descrip tion of a pract ice,would also amount to knowing what the pract ice was. In the case of a language,such a description would be a theory of meaning for that language, completewith all the linking principles. To give such a description is exceedinglydi fficult . To attain one would be the real izat ion of the phi losopher's hopes ofanalysing the complex activity of using a language, which is the chief vehicle ofour mental li fe and the chief medium of our interactions with one another. Weshall go astray, however, if we make a literal equation of the mastery of apract ice with the possession of theoretical knowledge of what that practice is:even so far astray as to conclude that there is no such thing as a language.

27Indeterminacy of French

Interpretation: Derrida andDavidson

I// / ( ) and Da r id s o n?

SA!\lUEL C. WHEELER I I I

My conclusions in this paper will be scarcely news to those who are alreadyfamiliar with both Davidson's and Derrida's work. Since there are relatively fewof those people, however, there is some point to stating the obvious. I hope topersuade a Icw of you, who are already presumably familiar with Davidson'swork, that Dcrrida supplies important, if dangerous, supplementary argumentsand considerat ions on one of the topics on which Davidson has done exci tingwork.

I show how a line of thought IJ f the French philosopher Jacques Derrida is aversion of the thesis of indeterminacy of radical interpreta tion as purified byDavidson. I claim that the basic conception of language and its relation tothought and to what passes fin reality is much the same for Derrida and forDavidson. More importantly, J maintain that some of the basic ideas and(perhaps) insights that move their respective arguments are the same for the twothinkers.This similari ty in where the)' have arrived is partly obscured by the fact thatthese thinkers come from rather different traditions. By a 'tradition' I mean a

family of theories and a complex of shared metaphors, shared formulat ions ofproblems, shared texts, shared shorthands, and shared styles which more or lessshape the way phi losophy is done. Just as our mention of 'canonical notation'alludes to a familiar but complex part of our canon, so a continental's usc of'primordial ' animates a reader' s prior absorpt ion of texts from that t radit ion.

II F reg e a nd l lusserl an d the presence o f meaningI begin with some simple-minded reminders and questions about Frege'sremarks on sense. For Frege, a given term on a given occasion of use designates

Thi s pape r ha s p ro fi ted fr om the comments of John Vi cke rs on t he paper as a whole and fr om,I

Documents

Dummett1986_A Nice Derangement of Epitaphs-Some Comments on Davidson and Hacking