Transcendental Idealism in Wittgenstein

8/13/2019 Transcendental Idealism in Wittgenstein

1/22

ThePhilosophical QuarterlyVol. 35 No. 139ISSN 0031-8094 $2.00

TRANSCENDENTAL IDEALISM INWITTGENSTEIN, AND THEORIESOF MEANINGB Y A . W . M O O R E

IThe new theory of meaning, like the old, points in the direction oftranscendental idealism, and shares also the problem of our beingdriven to state it in forms which are required to be understood, if tall, in the wrong way.

Th is is how Bernard Williams represen ts what he takes to be a fundam ental continuity between the Tractatusand W ittgenstein's later work.1 H eargues that there is, in Wittgenstein's progression of thought, a commonelement of transcendental idealism,2which arises from each of the differentconceptions of meaning . It is a species of idealism which has to be stated interm s of a limit of the world, which is no t itselfinthe world. In theTractatus,this limit is the m etaphysical subject, or transcenden talself- what Wittgenstein means by T. 3 In the later work, Williams urges, it is a plural descendant of that: the essential shift is from T to 'we' (and from 'my language'to 'our language'). But there is a predicament inherent in such idealism.Since the world is to be understood as containing everything that canmeaningfully be spoken of, the re is no legitimate way of talking about a limitof it. If we do attempt to talk about such a limit, we shall produce eitherarrant nonsense or (at best?) irrelevant claims about items within the world,which, in purporting to concern something other than those items, willinevitably include down-right falsehoods. We are told explicitly in theTractatusthat such attemp ts are attem pts to say that which canno t be said but

1 Wittgenstein and Idealism , reprinted inMoral Luck (Cambridge, 1981), p. 163.2 Th e term transcenden tal idealism was coined by Kant, and defined by him as a particulardoctrine about space and time. (See his ritique of Pure Reason,trans. Norman Kem p Smith(London, 1933), A369.) The current trend, which I shall follow in this essay, is to use the termin an extended sense that is supposed to arise naturally out of Kant's. I shall not attempt toredefine it. The essay itself ought to cast some light on how it is nowadays understood.5 TractatusLogico-Philosophicus, trans. D. F. Pears and B. F. McGuinness (London, 1961),5.632 and 5.641.

byguestonJanuary9,2014

http://pq.oxfordjournals.org/

Downloadedfrom
http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/


2/22

T R A N S C E N D E N T A L I DE A L IS M IN W I T T G E N S T E I N 3 5which makes itself manifest in (or is shown by) what can be said. OnWilliams' conception, this predicament is implicit in the later work too.

He might have added that it can be found throughout the history ofphilosophy. Th is is one of philosophy's perenn ial problem s: how to pu t asupposed philosophical truth w hich, if it is uttered , m ust be taken to mean anempirical falsehood, or w orse .5 As a predicament issuing from transcenden tal idealism, it can already be found in Kan t. Its impact for him can bemade graphic through the question: do objects in space and time depend fortheir existence on our perceptions?6It seems to be a crucial feature of Kan t'sspecies of idealism that they do not. For, as he continually reminds us, histranscendental idealism is meant to coincide with empirical realism, a connection for which he argues in several places;7and by empirical realism ismeant precisely the view that, through perception, we are aware of (spatio-temporal) reality outside us which is independent of that perception. Yetthere does no t appea r to be any way of stating Ka nt's transcendental idealismwhich does not stand in direct conflict with this. Kant himself whenexpounding his idealism, claims that objects in space necessarily presupposeperception and that they are nothing outside us.8He resolves this paradox byappeal to a deep ambiguity in the use of expressions such as outside us .9They may be taken either in an empirical sense or in a transcendental sense,that is either in such a way that we are ourselves construed as objects in the(phenomenal) world or in such a way that we are not. 10 (In its empiricalsense, outside us implies in space .) T h e claim that objects in space andtime are n othing outside us indep end ent of our perceptions is then emp irically false bu t transcendentally true . But this is why Kant faces the samepredicament as Wittgenstein. The transcendental interpretation of these

4 5.61-5.62.5 This phrase is quoted from Williams, op.tit.,p. 163.6 Cf. P. F. Strawson, The Bounds of Sense: An Essay on Kant's Critique ofPure Reason(London, 1966), p. 259.7 E.g. in the ritiqueo fPureReason,A369 ff. It is interesting to compare this with 5.64 of theTractatus.8 ibid. A372-376.9 ibid. A373.10 We must obviously construe empirical and transcend ental , so used, as terms of art.Later on in this essay, I shall take up the usage again, with a slightly different application. Giventhat the term s are being us ed in a special way, we must expect some strange cons equen ces. O ne

such, per hap s, is tha t the se nten ce Even if we all believed otherwise , seven plus five would stillequal twelve mu st be regarded as expressing an empirical truth.11 In theProlegomena toA ny FutureM etaphysics, trans. Lewis White Beck (Indianapolis andNew York, 1950), Kant's first published work after the ritiqueofPureReason,he shows himselfto be very sensitive about being interpreted as the wrong kind of idealist, a fate which he claimshas already befallen him. (See the end of the First Part.) But if it is true that two diametricallyopposed doctrines have to be stated in the same way, then it is only to be expected that this willgenerate confusion. Nor does Kant's suggested remedy, that his own view should henceforth be



Downloadedfrom


3/22

136 A. W. MOOREterms is unavailable, a second crucial feature of Kant's critical philosophybeing diat there can be no intelligible application of concepts beyond theworld of possible experience, or, to bring ou t die connection ofthisprinciplewith its Wittgensteinian echo, to anything not in die world.12 Thus Kant 'stranscendental idealism shares the fate of the kind of transcendental idealism to be found in W ittgenstein: that it cannot be stated save in terms wh ich,by its own lights, are unintelligible unless construed as embodying straightforward empirical falsehoods.

A good case can be made out for saying mat this fate is also shared bycertain anti-realist views in the philosophy of language (views which gain muchof their impetus from Wittgenstein's later work). The motivating thoughtbehind these views is diat linguistic meaning depends ultimately upon what ismanifest in our linguistic and non-linguistic practices, and that the meaningof a declarative sentence must (therefore) not be identified with conditionswhich may obtain independently of our being able to recognize diat theydo.13 But trudi conditions, as classically conceived, are conditions of diiskind. So die traditional identification of the meaning of a declarative sentence with its truth conditions should give way to an identification of itsmeaning widi, say, the conditions which we recognize as justifying anassertion of the sen tenc e. Different kinds of anti-realism are concern ed withworking ou t the implications which this has for different areas of discou rse.The problem is that, given what it would beto identify die meanings ofdeclarative sentences with dieir trudi conditions in the traditional way (forexample, uncritically to apply the law of the excluded middle to them), it isan integral part of our linguistic practice that we have a propensity to do justdiat. And diis is more dian superficially embarrassing for any anti-realist. Ofcourse, it is not yet an insuperable difficulty. No anti-realist is, as such,committed to the doctrine that every aspect of our linguistic practice issacrosanct. H er e, no ne the less, is some indication of why (some) anti-realistviews might fall victim to the fate we have been talking about. It may be thatany philosophical statement of them is bound to conflict widi certain basicclaims which are integral to our linguistic practice and which they might beexpected to ratify. There is evidence for this in Michael Dummett's treatment of anti-realism concerning the past.14 He characterizes the dispute

known not as transcenden tal idealism but as critical idealism , seem anything other thancosmetic. The problem is far too deep to be merely terminological. What is significant is that inhis later writings, including the second edition of the ritique of Pure Reason,Kan t states hisidealism in these bold b ut am biguous term s m uch less frequendy. It is as if he begins to sensethe impossibility of shrugging off their empirical interpretation.12 See, e.g., the ritique of Pure Reason,A696/B724.13 Cf. Michael Du mm ett, T he Philosophical Basis of Intuitionism , reprinted in Truth andOther Enigmas (London, 1978), pp. 216-17.14 Th e Reality of the Past , reprinted inibid.



Downloadedfrom


4/22

TRANSCENDEN TAL IDEALISM IN WITTG ENSTE IN 37between the realist and the anti-realist in terms which, he explicitly acknowledges, ought to be rejected by both parties but especially (and moreimportantly, from our point of view) by the anti-realist. Fo r, according to hischaracterization, the a nti-realist maintains that the past exists only in thetraces it has left on the present .15 But this is something which can beunderstood only in realist terms, as betrayed in our standard ways ofspeaking, and, so understood, it is false. Nevertheless, the characterizationdoes,D um m ett con tinues, succeed in conveying something of the psychological effect of the two opinions . Should it turn out that, ultimately,onlyacharacterization of this unsatisfactory kind is possible, then the point wouldbe established. Anti-realist views would be among those whose fate is thatthey have to be stated in ways that are not right.

In this essay, I hope to draw certain conclusions from Williams' contention that the predicament is also latent in the later work of Wittgenstein. Ishall offer no substantial arguments for the contention, since I am principallyinterested in its consequences. It does however seem to me to be correct,subject to Williams's own qualification that the element of idealism in thelater work of W ittgenstein which gives rise to the predicam ent is concealed,qualified, overlaid with other things .16Th e p redicam ent is certainly nomorethan latent. It arises with no ne of the dramatic urgency that accom panies it inKant, for example. Given Wittgenstein's later conception of the correct wayto do philosophy, this is hardly surprising. Philosophy, he tells us, consists ofassembled reminders for particular purposes, not of theses.17 So we shouldnot expect to find him directly propounding some philosophical theory, stillless one which merits the tide 'transcendental idealism' and of such a kindthat the question of how precisely it ought to be stated should be such adelicate and bedevilling issue. What we do find are hints and suggestions,backed up by platitudes or rather mundane empirical observations, whichare intended to remove the allure of what would otherwise be enticinglygrand and ambitious philosophical theses, whose perversion of the termswithin the m is checked by W ittgenstein's rem inders as to how the terms a re,and ought to be, used. This is indicative of the therapeutic role whichphilosophy is supposed to play. (It is instructive to compare this with 6.53 ofthe Tractatus) The insights which the later work affords are then bound toresist any direct statement. To try to state them direcdy would be tosuccumb to the very temptation which they ought to be counteracting. Butuiis in turn is why the predicament explicitly faced in the Tractatus is alsolurking in the later work; we are be ing induced into a certain way of thinking

,s ibid. p. 370.16 op . at., p. 147.17 PhilosophicalInvestigations, trans. G. E. M. Anscombe (Oxford, 1958), 127-128.



Downloadedfrom


5/22

138 A . W . M O O R Eto which, by the very lights of that work, we cannot give voice. This isWilliams' contention.

I think the contention w ould be impossible to justify, however, were it notfor die fact diat Wittgenstein himself sometimes goes against his ownprinciples precisely by attempting to give voice to that way of thinking and byattempting to express certain general philosophical truths. The predicamentis not entirely dormant. And when, radier than just uttering banalities,Wittgenstein attempts to put their implications direcdy - to say, as it were,wha t mey show - the n we can view him as providing elucidations of die kindwhich, according to 6.54 of the Tractatus, that work consists of: pieces ofnonsense which have to be transcended before the world can be seen aright.Perhaps die best examples are to be found in his work on madiematics. Hesuggests diat, in some sense, mathematical truth is determined by what webelieve.'8Bu t what we wou ld characteristically say,when engaging in madiematics and not when doing philosophy, is diat mathematical trudi is in nosense de term ined by what we believe.19 Again, his claim that '3 + 3 = 6 ' is arule as to the way in which we are going to talk,20 if it means anydiing at all,means something which it is in die spirit of mathematics to deny, surely notWittgenstein's intention.21Here we do indeed see his theory of meaning, asapplied to madiem atics, being stated in forms which are requ ired to beunderstood, if at all, in the wrong way .

IIMy principal aim is to establish that there is one extremely importantconsequ ence of all of tiiis: that W ittgenstein's laterwork if right, puts paid toan alluring pipe -dream which philosophers of language have long had . Th is

is the dream of being able to state what die words and phrases in ourlanguage mean, so as to reveal, in some philosophically substantial sense,what it is to understand them. When described like this, the dream soundssuitably heady. What exacdy it amounts to ought to become clearer in duecourse. But already it is possible for me to sketch what I have in mind. Inorder to remain faithful to the spirit of Wittgenstein's later philosophy, onecan legitimately theorize about language only by commenting descriptivelyon various actual practices in which we engage, practices which are alreadypermeated with meaning and which can be described only by one who

18 E.g. ibid. pp. 226-7.19 This point is borrowed from Williams, op.cit.,p . 163, and slightly adap ted.20 This claim is quoted by G. E. Moore in hisPhilosophical Papers (London, 1959), p. 279.21 Cf. the Philosophical Investigations, 124. There is something to the idea that, whenWittgenstein makes these rather bold claims, he can regard them as meaningful and appositeprecisely insofar as they further the aims of the rest of his work. But the idea contains anelement of disingenuousness which Wittgenstein himself would not have found acceptable.



Downloadedfrom


6/22

TRANSCEN DENTAL IDEALISM IN WITTG ENSTE IN 139already knows how to participate in them. That is, one can legitimatelymeorize about language only from within. As soon as one attempts totranscend that, by conveying the significance of these practices and by statingthe meanings of expressions from without, then the predicament which Ihave been talking about comes to the fore. In terms borrowed from theTractatus: the meanings of the words and phrases in our language are a limitto our world, revealed in our practices. The se meanings (unlike die prac ticesthemselves) are not apart of the world, and any attempt to state or analyzethem in the way suggested fails precisely by treating them as such. M eaningscan only be shown, not stated.22T h e ma tter can also be expressed in terms borrowed from Kan t. Jus t asKant distinguished between an empirical and a transcendental interpretationof expressions such as 'outside us', so too we can distinguish between anempirical and a transcendental interpretation of expressions such as 'ourlanguage', again according to whether or not we are ourselves being construed as objects in the world.23 In its empirical sense, our language isEnglish (or German, or whatever); in its transcendental sense, our languageis roughly ou r world-view, pe rhap s even our world.24This ambiguity arises inconnection with a host of expressions, for example 'how we carry one', butsupremely in connection with the pronoun 'we'. Obviously this pronoun canbe used to pick out a particular g roup of human beings in the world, perhapsa group which shares some one (empirical) language. But it can also be usedas the plural desce nda nt of the T of the Tractatus.15 To be sure, Wittgenstein himself would never have recognized these distinctions. And even ifuiey have a viable application when it comes to interpreting his work, wemust be extremely wary of supposing that at any point in his later work it ispossible to say, without betraying gross exegetical insensitivity, that he isusing one of these expressions definitely in one sense rather than the other.The attribution to him of transcendental idealism is far too circumspect forthat to be the case. Besides, it is arguable that he often makes capitalprecisely out of disregarding such ambiguities.26 It is nevertheless useful tohave the relevant distinctions ready to hand. We can use them to provideanother rather glib formulation of what it is which Wittgenstein's laterphilosophy precludes, namely any account of how our empirical languageengages with aspects of our transcendenta l language. The se aspects manifestthemselves; they cannot be spoken of. To revert to the earlier dictum:meanings can only be shown, not stated.

22 For links with the Tractatus conception of meaning, cf. 4.022 and 4.12 ff.23 Cf. footnote 10.24 Cf. thePhilosophicalInvestigations, 241. Cf. also Williams, op .cit.,p. 154.25 For links with Kant, cf. Strawson, op.cit.,p. 235.26 Cf. Williams, op .cit.,pp . 147 and 160.



Downloadedfrom


7/22

140 A . W . M O O R EThis is still too glib, however. There is, after all, a perfecdy harmless andcommonplace sense in which nobody would deny diat meanings can be

stated. One says, for example, that 'procrastinate' means the same as 'putthings off, diereby stating die meaning of 'procrastinate'. A G erm an /English dictionary states the meanings of German words and phrases foranybody who knows English. To state die meaning of an expression in thissense is to produce another expression which has die same m eaning. Q ualmsabout synonymy aside, there can be nothing philosophically objectionableabout the idea of stating meanings in tiiis way. Somediing else must be atstake in this contrast between showing and stating. Clearly it has somethingto do with the idea of conveying meanings directly, which in turn, perhaps,calls to mind the idea of laying bare what exactiy it is which one knows whenone unders tands a given expression. But we are owed a more precise accountdian mat.

T h e allusion to knowledge helps. Th e d iought that thereisanydiing whichone knows when one understands a given expression is, it seems to me, fundamentally ou t of keeping witii W ittgenstein's later conception of meaning andin particular with his conception of understanding as a practical capacity.(Knowledge is here being taken as propositional knowledge.) What is not soclear, however, is the role which diis diought need play within what I havedubbed the pipe-dream of philosophers of language, as we shall see shortly.

It has been customary, in recent work in die philosophy of language, todraw a distinction between two senses of the phrase 'theory of meaning'. Inits grander sense (tiiat used by Williams in the quotation widi which Iopened tiiis essay), a theory of m eaning is a general philosophical account ofwhat it is for the expressions in a language to have m eaning, an answer to diequestion: what is meaning? In its narrow er sense (principally a ssociated widithe work of Donald Davidson), a dieory of meaning is a theory of meaningfor a particular language, a body of trutiis which, in a sense which needs tobe made precise, systematically gives the meanings of all the expressions inthe language and thereby acts as an answer to the question: what do dieseexpressions mean? If what I am proposing is correct, men we have reason todoubt whedier Wittgenstein's later work could accommodate any dieory ofmeaning in die wider se nse. T h e quotation from Williams is thus no t entirelyfelicitous, though the infelicity is as much Wittgenstein's as his. However, inorder to get clearer on diese issues, we should temporarily focus on theoriesof meaning in the narrow sense.

The main reason why this will help is diat they pose an immediate threatto my thesis, as is evidenced by the nature of much of die interest shown indiem.27 For it appears that diey can provide us widi a way of realizing die27 S ee, e.g., Davidson, Tru th and M eaning , reprinted inInquiriesInto Truth and Interpretation (Oxford, 1984); Du mm ett, Wh at is a The ory of M eaning? , in Mind and Language, ed.



Downloadedfrom


8/22

T R A N S C E N D E N T A L I D E A LI S M I N W I T T G E N S T E I N 1 4 1dream which I have been talking about without containing anything whichhas the (on Wittgensteinian grounds) unacceptable consequences which Ihave suggested would accrue. A theory of meaning for a particular languagewas loosely charac terized above as giving the meanings of llthe expressions in the language. Le ss loosely, such a theory is supposed to be a body oftruths such that, if anyone knew them all, that person would ipso factounderstand the language.28 They (theories of meaning) are certainly intended to give me anings in a m ore substantial sense than a Ge rm an/En glishdictionary. (Possessing all the information in a German/English dictionarywould not confer understanding of German on a monolingual speaker ofFrench.) They are, to that extent, intended to do the rightkindof thing. Nowwe have to broach the question of whether, by admitting the possibility ofsuch theories, we are committed to anything mat is repudiated in the laterwork of Wittgenstein. Given what has been said so far, it is not at all clearthat we are. The claim that there could be a particular theory statingknowledge, possession of which suffices for understanding a given language,does not obviously entail that possession of such knowledge is a necessarycondition of understanding the language, or even that anybody who understands the language thereby knows anything at all. So we do not run upagainst that particula r W ittgensteinian scrup le. Again, it has frequendy beenargued that such theories need only draw on very Spartan resources, resources which appear incapable of offending against any philosophicalqualms a bout stating meanings o r expressing transcendental tru ths. Typically, they are conceived as formal (finitely) axiomatizable theories, whoseaxioms and theorems treat only of items in the language and undisputedtracts of reality, relating these to one another by means of various semanticconcepts founded on the idea of truth (or some kindred idea such as mat ofwarranted assertibility), together with a certain amount of set-theoreticalapparatus. No axiom or theorem concerning a given expression need evenpurport to be a statement of its meaning. It is not in that sense that suchtheories are intended to give meanings. Indeed, all the resources of such atheory are supposed to be purely extensional and hence capable of beingunderstood and mastered without any prior grasp of the concept of meaningat all (otherwise these theories would lose much of their intended explanatory power). And the theorems of the theory need no more be thought of as

Samuel Gutten plan (Oxford, 1975); and various essays inTruth andMeaning:Essays in Semanticseds. G. Evans and McDowell, 1976.28 Cf. Davidson, Reply to Fos ter , reprinted inop. cit. p . 171; the introduction to Evans andMcDowell,op . cit.,p. ix; and McD owell, On the Sense and Reference of a Proper Nam e , inMind 86, p. 159. A related but importantly different characterization is offered by Dummett inWhat is a Theory of Meaning? , p. 99. It is the characterization offered in the main text that Ishall be presupposing.



Downloadedfrom


9/22

142 A . W . M O O R Eexpressing transcendental truths than is any of the sentences in the object-language. In short, they need contain nothing which violates any Wittgen-steinian taboos against trying to express grandiose philosophical truths.Moreover, current orthodoxy has it that they are particularly aptdevices foranybody who is convinced that meanings can only be shown, not stated. Foronce the smoke has cleared, we are told, the sense in which such a theorywill give meanings isdiat its dieorem s will serve to show the meanings of diewords and phrases in its object-language.2''

W e should bew are of trying to read too mu ch into die orthodoxy, however.It involves a use of the word 'show' which is not identical to that in theTractatus. According to the latter, a sentence shows its own meaning,30 andthis leaves no obvious room for die suggestion mat a particular body of truthsabout the expressions in a language will be specially suited to showing dieirmeanings. Similarly, the kind of diingwewant to say on any interpreta tion of'show' which can be appropriated to fit in with the later work, provided thatit rests on a genuine analogy with die Tractatus, is diat die meanings of ourwords are shown in how we use diem or diat the meaning of mathematicalsentence is shown in how it is proved.31T hi s does no t issue very naturally indie idea of theory of me aning, particularbody of truths designed to showthe meanings of our words and sentences. The significant characterization ofwhat dieories of meaning should do remains the earlier one: they shouldstate truths, knowledge of which suffices for understanding dieir object-languages. Wh en it is claimed diat they should show meanings, what is beingclaimed is usually just diis. But then in order to extract die kind of philosophical interest from them which would be relevant to what I have beencalling the pipe-dream of philosophers of language (part ofwhich, remember, was to reveal what it is to understand die expressions in our language),one would surely need to say something about what knowledge of diesetruths would consist in and about what itisfor things to be the way that onewould then know diem to be.32 It remains an open question whether onecould do thiswidiout falling foul of the scrup les which em erge from W ittgenstein's later work. However innocuous the content of the theoremsthemselves might appear to be, the claims made on behalf of them wouldcome sufficiently close to statements of meaning, in die relevandy substantialsense, for a Wittgensteinian sense of disquiet to be stimulated once again.

29 Cf. Dummett,Frege:PhilosophyofLanguage(London, 1973), p. 227; Evans, TheVarietieso fReference(Oxford, 1982), p. 26; and Colin M cG inn , Th e Structu re of Co nten t , inThought andObject ed. Andrew Woodfield (Oxford, 1982), p. 223.30 4.022.31 Cf. thePhilosophical Investigations, p. 220.32 T hi s is a point which Du m m ett has often insisted upon , e.g. in W hat is a The ory ofMeaning ? , p. 1 21, and W hat is a Theory of Meaning? (II) , in Evans and M cDowell,op .cit.,pp . 70-71.



Downloadedfrom


10/22

TRANSCEN DENTAL IDEALISM IN WITTG ENSTE IN 143This is not to suggest that the later W ittgenstein could not have taken suchtheories seriously (though my own view is that he could not). The point is

rather that he could not have taken seriously a certain kind of interest inthem. A more radical conclusion would have to be argued on independentgrounds. For suppose we deliberately refrain from making such claims aboutthese theories as will leave us open to the charge of trying to expresstranscendental truths. The possibility then remains open that they cancontinue to be of interest to us in what they reveal about various relationsbetween language and reality; or about the formal properties of truth; orabout logical form; or about structural validity; or about die composition ofcomplex expressions and about how their semantic values are determined bythe semantic values of their parts; or whatever. Somewhat more cynically,one might argue that ifwe hold back from making the relevant extravagantclaims about them, then we are left widi somediing that cuts so littlephilosophical ice that it could not possibly come into conflict with anyplausible philosophical reflections, still less those in the later work of Wittgenstein. Th e coun ter-arg um ent would have to be, more or less, that there isstill something of substance in the very composition of these theories,according to which language is made up of morphemes which stand, atemp-orally, in certain determinate semantic relations with aspects of the world, onthe strength of which the semantic properties of combinations of thesemorphemes can be determined by fixed rules; and that this in itself wouldhave been anathema to the later Wittgenstein, who sought to draw ourattention to the flexibility, indeterminacy and o pen -texture dness of meaning,to the creativity of language-use, to the disparities between language and arule-governed calculus, and to the inappropriateness of systematization andgenerality in theorizing about language.33 Be that as it may, any interest inthese theories founded on the belief that, through them, we can realize thegiven philosophical dream, would ultimately have to be interest in certainpowerful claims being made on behalf of them, and diis would make thesupposed innocuousness of dieir own resources an irrelevance. We are leftwith the thought that, given what we find in the later work of Wittgenstein,the dream cannot be realized.Nevertheless, diis discussion of theories of meaning in the narrow sensemay at least have given us a be tter idea of wha t die dream is. It is the dreamof being able to co nstruc t a theory of m eaning in the wide sense, which willenable us to specify, for any meaningful expression in our language, somefeature of it (such as the idea it stands for in the m inds of those who use it, orthe intentions with which it is habitually or conventionally used, or itscontribution to the truth conditions of declarative sentences in which it

33 Cf. Crispin Wright,Wittgenstein ontheFoundationso fMathematics (London, 1979), ChapterXV. See also, e.g., thePhilosophicalInvestigations., 81.



Downloadedfrom


11/22

144 A. W. M OOREoccurs, or its contribution to die assertibility conditions of declarative sentences in which it occurs), together w ith som e accoun t of what it would be toknow that it possessed diis feature, which will jointly serve to give itsmeaning and to provide, if not a literal account, at least a model of how it isunderstood. (Specifying die feature may require diat other expressions betaken into account. T he dream is not meant to embrace atomism.) I shall callsuch a theory an m-theory, and the re levant feature of an expression itsm-feature.It is my principal contention, then, diat Wittgenstein's later work,appropriately construed, is incompatible with any belief in die possibility ofan m-dieory.

IllAny m-theory compatible with Wittgenstein's later work would have toincorporate the broad intuition diat die meaning of an expression is afunction of how it is actually used, or, to put the intuition even more looselybu t in a form that we shall come back to, that m eaning is a m atter of how wecarry on. I am going to present for consideration an argument to the effectthat the m -feature of an expression would therefore have to be a feature such

that no-one actually knew the expression to possess it; no-one's understanding of the expression would in fact consist in knowing that it possessed thatfeature. Admittedly, the argument will trade on an uncompromising conceptof knowledge, something involving incorrigibility. This makes it easier, and,correlatively less interesting, to argue that a given individual does not havediis or that item of knowledge. But assessing the argument will provide auseful route to understanding (some of) the tension between Wittgenstein'slater work and a belief in die possibility of m-theories. (This is indeed all Ishall claim on behalf of the argument. I do not myself wish to endorse it.)The argument is this. On a later Wittgensteinian view, any expression inour language means what it does, in the last analysis, because of how wecarry on over time, because of our continuing linguistic and non-linguisticpractices. The whole temporal spread is relevant. At any point in time, wemay carry on in divergent ways, more than one of which could count aspreserving the meaning of the expression. That the expression possesses itsm-feature must therefore entail that it is subject to continuing practices ofcertain kind. T hus , anybody who knew that it possessed its m-feature wouldipso facto be capable of having knowledge of the future, which (given asuitably tight construal of knowledge) nobody can have. Take the word'gre en '. Its meaning what it does entails that it stands in a certain relation togreen things, which we may express, roughly, by saying that it applies tosomething at a given time if and only if that thing is then green, or, evenmore roughly, that it applies to green things. (What exacdy this amounts to



Downloadedfrom


12/22

TRANSCEN DENTAL IDEALISM IN WITTG ENSTEIN 145need not detain us now. Suffice it to remark that the notion of applicationwhich appears here will simply be some refinement of our intuitive notion ofapplication, subject to the proviso that it is to be understood extensionally.Thus 'green' also applies to things whose colour can be obtained by mixingblue and yellow paint.) But it is purely contingent that this is so. Had wecarried on differendy, 'green' could have applied to blue things, or tonodiing at all, or to gr ue things, wh ere a thing is gru e at a given time ifand only if either the thing is then green and die time is prior to timetor thetiling is the n blue and th e time is not prior to time /. (Here tis unde rstood tobe some time in the future.)34 Bu t if it did apply to grue things, then nothingwhich has happened in the world so far need appear any different. So it iscompatible with everything that anybody knows (incorrigibly) mat 'green'does indeed apply to grue things. This is true even of people who have beenspeaking English successfully for many years, who will continue to do so formany years to come, perhaps beyond t, and who believe, quite righdy, that'green' applies to green things. Not even these people knowhow the word'green' will be used in the future. But a future use of the word 'green' invirtue of which it applied to grue things would be equally meaning-preserving, albeit by preserving a meaning other than that which 'green'actually has. So nothing which anybody now knows can serve to determinethe meaning of the word 'green'. Its m-feature must lie beyond anybody'sken.

As it stands, this argument is obviously far too quick. But by assessing it,in the light of various possible reactions and objections to it, we shall findourselves bein g drawn to the h ear t of the m atter - especially when it comesto adjudging the argument from a later Wittgensteinian point of view. Onenatural reaction to the argument rests, I think, on a muddle. Consider (G),understood in the way oudined in the last paragraph:

(G) 'Green' applies to green things.If we distinguish be tween the senten ce (G) and the claim that it is being u sedto make, then we find tha t, however non -trivial the claim may be, thesentenceenjoys an indefeasibility which makes the charge of triviality an appropriateone. (It is almost as if (G) is an analytic contingency, although this is aneedlessly paradoxical way of putting it.) Whatever 'green' applied to, (G)would still be true, provided that all other linguistic conventions were heldconstant. And by the same token

(G*) 'Green' applies to grue things34 Th is is an adaptation of the definition first proposed by Nelson G oodm an, due toS. F. Barker and Peter A chinstein, in O n the New Riddle of Induction , in PhilosophicalReview, 69 (I960), p. 511.



Downloadedfrom


13/22

146 A . W . M O O R Ewould be false.35I f gree n' applied to other than green things, then (G) wouldexpress a different truth, and (G*) a different falsehood, but they would betrue and false respectively none die less.36 The reaction to the argumentwhich I have in mind is this. Since I clearly have no more insight into thefuture dian anybody else, it follows that, if the comments made about'green' towards the end of the argument are correct, then, for all / know,'green' applies to grue things. Yet surely I cannot acknowledge this possibility without temporarily misusing English (for I can be confident that (G*) isnot a true sentence of English, given the points just noted, just as I can beconfident that (G) is a true sentence of English, which confidence has beenbetrayed in the very writing of uiis essay). But to admit that I cannotacknowledge a possibility without misusing the language in which I attemptto do so is tantamount to admitting that I cannot acknowledge it at all. Itfollows that either th e backg round view of meaning is wrong or the argum entitself is unsound.

I think the muddle here can be teased out. Reconsider the possibility that'green' applies to grue things. Suppose that this is the case. Then I for onehave a false belief about 'green', and use it incorrecdy. For I believe uiat itapplies to green things, and diis in part informs my usage, for exam ple whenI assert 'Grass is green' or even 'I believe that 'green' applies to greenthings'. If there is any fact of the matter as to what claims I take myself tobemaking when I assert these sentences, then certainly, on the given hypothesis,I am not asserting die right sentences in order to make them. Similarly,I am not using the right sentence even to express the hypodiesis. I ought tobe using (G) radier than (G*). (This reinforces the idea that (G) enjoys acertain kind of stability.) B ut none of diis rules ou t the possibility that, for allI know, 'g reen ' does apply to grue tilings. It is just that, */it does, then I havenot be en (indeed am no t) using English correctly. It by no mea ns follows thatjust by saying 'For all I know, 'green' applies to grue things' I must temporarily cease to use English correctly. When I do say that, I am a lluding to thepossibility tha t I am misusing the language, bu t only because the possibility isalready there to allude to. I am notrepresenting myself as a possible misuser

3 51 am assuming diat (G*) is indeed false, and dierefore presupposing that it is not the casethat something is green if and only if it is grue, which would be the case if, say, nothing wereeither green or blue at or beyond t.It is also worth po inting out that one of the conventions thathas to be held constant, for (G*) to be gua ranteed its falsehood, is that 'gr een ' and 'blu e' shouldnot apply to any one monochromatic thing at the same time.36 Cf. D um me tt, Wh at is a Theo ry of Mean ing? , pp. 10 6-1 07 . It is a closely related factthat (G) could never be used to inform somebody that the given relation actually holds betweenthe word 'green' and green things. For somebody who was not already aware of this would notknow which fact (G) was being used to state, precisely because the word 'green' occurs withinit. And by the same token, anybody who did know which fact (G) was being used to state wouldalready appreciate that it was true, which is indeed, righdy or wrongly, often taken to be ahallmark of triviality.



Downloadedfrom


14/22

TRANSCENDENTAL IDEALISM IN WITTGENSTEIN 147of the language - as dioug h I were somehow offering die sentence 'Fo r all Iknow 'green' applies to grue dungs' as one which, at least in my moudi,could be taken to express a trudi, if not in English then in some idiolect ofmy own which I mistakenly take to be English. For a variety of reasons,many of diem implicit in Wittgenstein's later work on language, this idea isincoherent. Unless I succeed in saying somediing in English by uttering thesentence, which is after all wha t I am trying to do, then I do not succeed insaying anydiing at all. But conversely, diere is no reason to suppose (on thestrengdi of anything said above) diat I cannot meet widi such success,thereby making a true claim, in English, about die limitations of myknowledge.Let us address anodier possible reaction to the argument, which takes dieform of an objection.37 T o say diat 'gre en ' applies to green tilings is to take asemantic claim about it which is true in virtue of die meaning which it hasnow,even diough some of die green things to which it applies may be greenonly when it has ceased to have diat meaning. O ne way of putting the ma tter(not die only way) would be as follows. The relation of application can bediought of as holding atemporally between die members of certain orderedquadruplets of exp ressions, languages, times and tilings, diose, namely, suchmat die thing satisfies the expression relative to its meaning in the languageat die time; and if the expression is 'green', die language English, and dietime now, the n d ie thingwillitself be an ordered pair consisting of a physicalobject and a time, die physical object being one which is green at the time.(No doubt this should be m ade more precise, but it will serve our purposes.)Certainly, if 'green' now applied to grue things, this would be a differencewhich concerned future times, but the times would be those after t whichoccurred in the specified ordered pairs, and not die time occupying thirdposition in the quadruplet, which remains by stipulation as now. It is unclear,men, why the different meaning which 'green' would have to have needdepend in any way on what is going to happen at those future times. It is alsounclear why the actual and current meaning of the word 'green' cannotreside in some feature of die word to which we nowhave access. Th e factdiat nobody knows how this word will be used in the future is irrelevant.That may go to show only diat nobody knows whether it will continue topossess diis feature and thus retain its current meaning.

This objection, like the original argument, is too quick as it stands butpoints in the direction of something important. It would certainly be decisiveagainst the part of die argument which specifically concerns the word'green', had that part stood alone. But it did not stand alone. It was onlyintended to illustrate die point which had already been made:grantedthat37 I am grateful to Simon Blackburn for drawing my attention to this objection and its

importance.



Downloadedfrom


15/22

148 A. W. M OOR Emeaning itselfisa matter of how we carry on, over time, me n nothing whichanybody now knows can serve to determine die meaning of nyexpression inour language (mat is, no feature which any expression is known to possesscan be its m-feature). Of course, one may not want to grant mis. But that isbeside the point, so long as one is prepared to grant mat it is implicit in dielater work of Wittgenstein. For the argument itself was designed to showmerely what that work will tolerate. (One prob lem with the objection is thusdiat it is spelt out in terms which are not at all Wittgensteinian.)

The objection nevertheless forces us to ask whemer the comments specifically concerning the word 'green' do succeed in illustrating the core ofthe argument preceding diem; and this question also turns out to be centralto any appraisal of die argument from a later Wittgensteinian point of view,which means that we are getting to the heart of the matter. From diat pointof view, die core of the argument is essentially acceptable. It does, after all,tell against die view that our understanding is a species of prepositionalknowledge. But the subsequent comments about 'green' do not have aWittgensteinian flavour. On die contrary, they elicit the following objection,which does (in the nature of its motivation, if not in style).It would be impossible for 'gree n' to apply to grue tilings if everything elseremained the way itis.T he re would have to be further differences, includingdifferences in the language. For example, either 'blue' would have to ceaseapplying to blue tilings, or the convention that 'green' and 'blue' should notapply to the same monochromatic thing at the same time would have to bedropped. But, the objection runs, various apparently extralinguistic featuresof reality would have to be radically different too. If, after t, we found thatreality did not conform at all well with those express prior expectations ofours which involved die word 'green', such as the expectation embodied in

the sentence 'Grass is green', then, as certain familiar philosophical arguments testify, we would no longer be able to use the word for purposes ofsuccessful communication; the word would cease to have any meaning, andit would certainly not be true to say, at that stage, that it applied to gruethings.38 But for reality to conform with these expectations, while 'green'applied to grue things, would involve marked differences from what isactually the case. For example, grass would need to be blue after t.Bu t nowthe objection can be pushed a little further. Sup pose that realitydidconformwith enough of our expectations for successful communication to be possible, so that 'green' still applied to grass after t but diat up to die present38 This leaves open the possibility that t should in any case be some time after the word'green' has ceased to be used with any meaning, say millions of years after the demise of allnatural languages. But, for reasons which will eventually become clear, Wittgenstein's laterconception of meaning certainly precludes the thought that, for all anyone knows, 'gr e e n 'applies tothose grue things.



Downloadedfrom


16/22

TRANSCENDENTAL IDEALISM IN WITTGENSTEIN 149everything rem ained the way it actually is. W e have no firm grip on what itcould mean to say that 'green' applied in such circumstances to other thangreen things. In these hypothesized circumstances, our continuing practices(how we carry on) would in all relevant and non-question-begging respectsbe the same as our actual continuing practices. But what it isfor somethingto be green, and how we carry on, particularly with our use of the word'green', are far too delicately interwoven for us to be able to say, withoutfurther ado, that these are circumstances in which 'green' applies to thingswhich are blue beyond a certain point in time. On the contrary, to imagineour practices starting out just as they actually have, and then continuing toissue in the same degre e of successful communication but in such a way thatthe meaning of the word 'green' is not thought of as changing, just is toimagine, inter alia, that the word 'green' should continue to apply to greenthings. Adm ittedly, in saying that there is nothing I know which rules out thepossibility that 'g re en ' applies to grue things, I may be attempting to expressthe perfectly correc t thought that noth ing I know rules out the possibility thatI have a mistaken belief abou t the application of the word 'g ree n' which mayeventually come to light. But then, why not justsaythat? To say that, for all Iknow, 'g re en ' applies to grue things is to say som ething altoge ther different.It is to say something absurd.

Thus the objection. If it is correct, it has one immediate consequence:given Wittgenstein's later conception of meaning, the mere fact that 'green'applies to green things cannot constitute its m-feature (for no feature of anexpression which it is known to have can constitute its m-feature, whereasenough known to guarantee that 'green' applies to green things); whateverexactly is involved in knowing that 'green' applies to green things, suchknowledge could not possibly model, or act as any kind of surrogate for,understanding the word. It is revealing that the objection has this consequence. It highlights how it (the objection) stands with respect to what waspretty m uch an unargu ed assum ption of the original argum ent, namely that itis a purely contingent fact that 'green' applies to green things. A distinctivekind of doubt has been cast on how this assumption was exploited. But wehave to be very careful wh en spelling out what is involved he re. N othing hasbeen said to suggest that the contingency claim is straightforwardly false. Tobe sure, when one looks at a sentence like (G) in a suitably Wittgensteinianframe of mind, one is tempted to see the sentence's own triviality (whateverthat might consist in) as necessitating some kind of triviality for any use tomight it might be put. Why, after all, would one want to assert a thing likethat? Not to impart substantial information.39 But it is a long way fromnoticing this to denying that the sentence expresses a contingent fact or eventhat it has perfectly legitimate uses. One might, for example, assert it simply

39 Cf. footnote 36.



Downloadedfrom


17/22

150 A. W. M OOR Eas a way of emphasizing diat the application of the word 'green'canonly besatisfactorily specified by using 'green' itself in this essentially unilluminat-ing way (that we do not have the linguistic resources to offer a weightieraccount of how 'green' hooks up with the world, that 'green' means what itdoes and not another thing). The contingency claim has not itself beendenied outright, men. Indeed, denying it outright (insisting diat 'green'necessarily applies to green diings) would be tantamount to denying eitherthe evident arbitrariness of how we use the word 'green' or the still moreevident fact that being green has nothing to do widi die English language.One could, however, deny mat it is appropriate to bring the contingent/necessary distinction to bear on such a proposition in die first place; andcertainly, less dramatically, one could challenge what seems to be a corollaryof the view diat die proposition is not necessary, namely diat any English-speaker can, in principle, form a conception of what it is for som ething to begreen completely independently of considerations about how die word'green' is used. It is somediing like diis diat has gone on above. What hasbeen denied is diat die claim mat 'green' applies to green diings has any realbite.It is this which lends the objection its W ittgenstein ian flavour.

Before we assess more carefully what diis amounts to, we can profitablylook at its repercussions for any attempt to exploit, as die basis for anm-dieory, formal dieories of die kind discussed above, diose conceived asdieories of meaning in the narrow sense. I am tiiinking more specifically ofdieories which are expected to include, as axioms, formal counterparts ofsentences such as (G). Relatedly, and more fundamentally, diey are dieorieswhose dieorems will include suitably formalized embellishments (widi dueaccount taken of relativization to times, speakers, language and so forth) ofsentences such as die following:

(T ) 'G rass is green ' is true grass is green.40Although this biconditional appears banal to the untrained eye, it is, again,more or less a philosophical commonplace nowadays diat it can be used tostate a pure ly con tingent fact. It is a fact wh ich takes som e discovering, whichis capable of being forgotten, and which die majority of people, having noEnglish, do not even believe.41For the sen tence 'Gra ss is green' to be true is,at least in part, for English-speakers to indulge in a certain kind of linguisticbehaviour. Fo r grass to be green, by contrast, is somed iing wholly extraling-uistic. The fact diat die sentence and the stuff stand in diis particularrelation to one another is as non-trivial as the fact that die se ntence 'Snow isblack' stands in the same relation to blood. Or such is die orthodoxy. Buteven if m e contingency claim itself is unassailable, we can now give voice to a

40'' is a truth functor standing for material equivalence.41 Cf. the introduction to Evans and McDowell,op. at.,p . xi.



Downloadedfrom


18/22

TRANSCEN DENTAL IDEALISM IN WITTG ENSTE IN 151strong Wittgensteinian reservation about diis way of justifying it: it is not atall clear that for grass to be green iswholly extralinguistic. Another characteristically Wittgensteinian misgiving about the orthodoxy is that thepredica te ' . . . is tru e' is most naturally taken as a purely disquotationaldevice (so that calling a declarative sentence true is just an indirect surrogatefor coming straight out with it). If that is so, then the left-hand side of anybiconditional such as (T) will be too close in content to the right-hand sidefor it to pack any real pun ch . Bu t the reservation expressed above (as it were,the reverse of that) cuts much deeper, and (unlike that) cannot be quelled bysome such simple device as construing the predicate ' . . . is tru e' as atechnical term, requiring relativization to a language. Of course, as soon asone attem pted to unpack and defend the claim that knowing a theory of thiskind for English would suffice for understanding it, one would ipso factobeinvolved in the construction of an m-theory. To that extent, we have yet tosee where such an attempt would run aground on a later Wittgensteinianconception of language. But it is already clear diat something would beamiss. There would be a real problem in showing that knowledge of such atheory could so much as approximate to understanding the language. Itstheorems, to recall what was said above, would lack any real bite.

IVThe motivating thought here, implicit in much of Wittgenstein's laterwork, is that there is no clear demarcation of the extralinguistic.42In particular, it is no easy matter to disentangle what is involved in something's beinggreen from wha t is involved in ou r using language in certain ways. The re is astrong inclination to balk at this, to insist diat what it is for something to be

green is purely a ma tter of the world'sbeing certain way irrespective of whatwe get up to, pe rha ps even that it can be known (only) by focussing attentionon a green object and thinking: to be green is to be likethat.But for the laterWittgenstein, this would not (and could not) give one insight into what it isfor something to be green. Anybody who wants to gain such insight mustobserve us com municating widi one anothe r and exercising various discriminatory capacities which we possess. Not only that. Such a person mustunderstand ou r acts of com munication, see the point of our classifications -in short, become one of us.How coherent is this? How coherent is the claim diat this is required inorder to know what it is for somediing to be green? It would be perfecdycoherent (indeed boring) if, although we seemed to be using the word'green' straightforwardly, we were in fact talkingaboutit and saying what was

42 Cf. thePhilosophicalInvestigations, the final paragraph of 136 and 381.



Downloadedfrom


19/22

152 A. W. MOORErequired in order to unde rstand it. (Perhaps some such illusion is involved inthe sentence 'By 1857, she had become George Eliot' or, to take a moreclosely related exam ple, 'H e believes that a carp is a kind of bird '.) If (G) tooinvolved such a n illusion, we m ight even be able to acqu iesce in th e view thatit expressed a necessary and utterly trivial truth. But there is no reason tosuppose that we have been subject to any such illusion. More importandy, ifthe idea of an m-theory were to be defended within the framework ofWittgenstein's later work, then similar claims would be made in contextswhere such an illusion would have to be ruled out. For at some stage inspecifying the m-feature of 'green' and in saying what it would be to knowthat it possessed its m-feature, we would want to disclose how it stood inrelation tobeing green.Whatever conclusions we had come to about the statusof (G), and whether or not we chose even to make use of it, we wouldultimately want to reveal somediing about the relation of 'green' to greenthings which required that they should be thought ofasgreen things and notjust as satisfying the English word 'green'. M-theories thus bring to the foreprecisely such q uestions as: wha t is it for so mething to be green? A W ittgen-steinian answer is that it is (at least in part) for u s to carry on in a certain way(or that it is for our language to be a certain way); and this answer raisesessentially intractable problems.But what are these problems? It is not that whenever we want to say how'green' stands semantically in relation to extralinguistic features of reality, wesomehow find that we cannot relate it to anything other than itself or at anyrate to anything not linguistic. We do not find that, and nothing has beensaid to suggest that we do. After all, we can always use (G). Nor is there aproblem to the effect that the only claims we can make about how 'green'stands semantically in relation to extralinguistic features of reality are suchthat knowing diem could not, for the later Wittgenstein, am ount to anything.For all that has be en said in this essay, knowing that 'gr ee n' applies to greenuhings may, for him, amount to quite a lot. The problems lie much deeper.They concern the very coherence of some of the claims which have beenmade about what it is for something to be green. We need have no qualmsabout saying that 'green' applies to green things; we may even have verygood reason to say this - indeed, as I suggested earlier, as a way ofemphasizing what else cannotbe said. But as soon as we attem pt to state themean ing of'g re en ', in the sense that has come to preoccupy us, or as soon aswe attempt to say what it is for something to be green, we encounter(literally) transcendental problems.

'Transcendental': that was one of thedefinienda in the distinction drawnseveral pages back bu t subsequently little heed ed . Th ose expressions involving the first-person plural which were described as ambiguous ('our Ian-



Downloadedfrom
http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/http://pq.oxfordjournals.org/


20/22

TRANSCENDENTAL IDEALISM IN WITTGENSTEIN 153guage', 'how we carry on', 'we') have featured prominendy in dais recentexegesis of Wittgenstein, and the time has come to broach die question oftheir interpretation. If the exegesis is not to be regarded as totally insensate,then diis question will not admit of any straightforward answer. One thing,certainly, which has been hinted at several times is that anything as parochialas a straightforward empirical interpretation must in many cases be ruledout. If diere is any substance at all to the claim that for something to begreen is (even in part) for our language to be a certain way, then ourlanguage had better not be thought of sjust English - not if the claim itselfdoes not conce rn any particular linguistic item. It had better be d iought of, tosome extent, transcende ntally. T o discuss what being green actually consistsin, the kind of thing one has to discuss when constructing an m-dieory, andto discuss it within the framework of W ittgenstein's later philosophy, ne cessarily involves talking about aspects of our transcendental language.

But we cannot talk about diese aspects. They manifest themselves in ourempirical practices. When we try to talk about them, we make claims whichare either incoherent or empirically false. That is the predicament inherentin transcendental idealism, and it has now finally been brought to thesurface. It has taken the following form. The question of what being greenconsists in has had to be understood as the question of what being greenconsists in, not as some anaemic surrogate which might be rather sloppilyexpressed in me same way. It has not, therefore, been possible to construe itas a question in semantics about the meaning of the E nglish word 'gre en ', tobe answered by citing a dictionary definition or stating its application; nor asa question in anthropology about the classifications we effect in describingour environment, to be answered in terms of human needs and interests (orperhaps the needs and interests of particular group); nor yet as a questionin natural science about the physical phenom ena in virtue of which a thing isgreen, to be answered in terms of pigment, wavelength and retinas; nor evenas a question in the philosophy of language about the conditions which haveto be satisfied for th e word 'gree n' to apply to som ediing, to be answered byproviding a schematic analysis of application, perhaps in terms oftruth. Thequestion of what being green consists in has had to be construed as aquestion about the way the world is for us, a question about our world-view.It is not a qu estion abou t any thing we come across; it is a question about ourway of coming across tilings. Bu t dien die answer born of the transcend entalidealism in Wittgenstein's later work is that something's being green consists(partly) in our carrying on in a certain way, or in our using language in someparticular way. It follows that ifwe did not use language in that way, grass(say) could n ot be green. B ut we have no way of understanding th is, except asan empirical falsehood. What does the colour of grass (real grass, that stuff



Downloadedfrom


21/22

154 A. W. MOOREout there) have to do withlanguage}By striving to captu re, in W ittgensteinianterms, the essence of a reality which is empirically independent of us, welapse into unintelligibility or simple error.It is not that Wittgenstein's later work makes the philosophy of languagecompletely pointless. There remains a wide range of permissible activitieswhich could merit that label: making certain distinctions which are notusually recognized, and imposing a certain systematization on our ordinaryunderstanding of words, in order to meet various new linguistic needs;describing particular linguistic practices, in order to gain a clear view ofhowcertain expressions are used, in the hope that this will lead us away fromthose alluring misuses of them which issue in philosophical perplexity;feeling our way around inside our world-view, in order to gain a sense forwhere the limits of comprehensibility lie; reflecting on why certain thingscount as conceptually impossible, for example that 3 + 3=7, and therebycoming to a better understanding of the relevant underlying linguistic practices, those involved in counting and doing arithmetic, perhaps.43 But thelater work will not tolerate the idea of an m-theo ry, th at is to say a theory ofmeaning which enables us to state how the words and phrases in ourlanguage engage with our world-view.44T hi s is not to deny that Wittgensteinhimself in his later work, produces comments which have the flavour ofbelonging to such a theory. For, as has already been intimated, he sometimesseems prepared, in the spirit of the Tractatus, to say things which , by his ownprinciples, ought not to be said, but which will serve as elucidations - therungs of ladde r to be throw n away after it has been climbed.45 Still, if whathe produces is elucidatory nonsense, it is none the less nonsense.

To conclude: it is of the essence of the later work of Wittgenstein todisplay the limits of our language as the limits of our world.46 Any theorywhich allowed for a philosophically substantial statement of what the expressions in a language mean would have to make provision for acknowledging this, if it were to be compatible with tha t work. But n o theorycouldmakeprovision for acknowledging this, for it is possible to acknowledge only whatit is possible to say, and it is possible to say only how things areinthe w orld.Similarly, nothing in the world (no fact) encapsulates what is shown to

43 Cf. thePhilosophicalInvestigations, 122 and 132.44 Apart from the them es dwelt on in this essay, we mus t not forget that an m-theory w ould bedesigned to help us u nders tand understa nding; it would be intended to reveal die kind of thingsone could know about die words and phrases in a language as a way of coming to understanddiem, or as a way of modelling such understanding. However, for die later Wittgenstein,understanding is first and foremost acapacity, manifested in a certain kind of practice, and itought not to be diought of as a species of interpretation. (See diePhilosophicalInvestigations, 199 and 201.) So there may already be enough here to fuel die conclusion diat his later workprecludes die idea of an m-theory.45 T h e Tractatus, 6.54.46 Cf. the Tractatus, 5.62 and Williams, op .cit.,pp. 150-151.



Downloadedfrom


22/22

TRANSCEN DENTAL IDEALISM IN WITTG ENSTEIN 155somebody who un derstan ds the expressions in a language. Th e meanings ofthose expressions canno t be stated, though the expressions can, of course, beused with their meanings. Let us turn for the last word to Wittgenstein:47The limit of language is shown by its being impossible to describethe fact which corresponds to . . . a sentence, without simply repea ting the sentence. (This has to do with the Kantian solution of theproblem of philosophy.)48University College, Oxford

47 Culture and Value, trans. Peter Winch (Oxford, 1980), p. 10.48 Many people made helpful comments on earlier drafts of this essay. I am especiallyindebted to Simon B lackburn, Joh n M cDow ell, Leslie Stevenson, Gar rett Thom son and PhilipTuretzky.



Downloadedfrom

Documents

Transcendental Idealism in Wittgenstein