1 Wittgenstein, Ludwig Philosophical investigations, tr. Anscombe, G.E.M. (New York: MacMillan Co. 1958);Google Scholar all references to this work will be indicated in the text of the paper using paragraph numbers.

2 Ayer, A.J. in The Foundations of Emprirical Knowledge (New York: St. Martin's Press 1969)Google Scholar expressed this very well when he argued for the need to recognize two kinds of linguistic rule:

… it is necessary that, besides the rules which correlate symbols with other symbols, our language should also contain rules of meaning, which correlate symbols with observable facts. (112)

Ayer's motivation for this is primarily epistemological; the threat of scepticism can only be removed if it can be shown that there is a direct and immediate link somewhere between language (or some part of language) and the world.

3 Malcolm, Norman ‘Review of Wittgenstein's Philosophical Investigations,’ in Pitcher, George eel., Wittgenstein (Garden City, NY: Doubleday & Co. 1966)Google Scholar

4 Ibid., 68

5 Malcolm's interpretation of the private language argument has been repeatedly attacked; one of the most influential of these argues that the private language argument is simply a new application of the Principle of Verification, and, as such, subject to the same difficulties any application of the Principle is. This argument is developed most forcefully by Thomson, Judith ‘Private Languages,' American Philosophical Quarterly, 1 (1964),Google Scholar and put to considerable work by Fodor, J.A. The Language of Thought (New York: Thomas Y. Crowell 1975) 68–71,Google Scholar in his defense of an innate system of representations.

Thomson offers an essentially correct analysis of Malcolm's argument but she misdiagnoses what goes wrong with it. She argues that there are three stages in the development of Malcolm's argument:

(1) Classificatory terms must be rule-governed.

(2) (1) requires that it be posible to distinguish between following the rule and seeming to follow the rule.

(3) (2) requires that if one seems to follow the rule, it must be possible to determine whether the rule is actually followed.

The third stage is crucial, for a private ‘rule,’ it is alleged, is not susceptible to such a determination. Thomson treats this third stage as an application of the Principle of Verification - to determine whether a rule is correctly followed is to determine what objects truly fall unde.r the classificatory term. As she puts it, ‘A sign “K” is not a kind-name in a man's language unless it is possible to find out whether or not a thing is of the kind associated with “K“’ (137). So, she concludes, the private language argument is Just a new version of verification ism.

I have two things to say about this analysis of Malcolm's argument: (1) Even if Thomson is right that Malcolm does appeal to the Principle of Verification, it is not necessary for the argument, and so, to construe the private language argument as a reapplication of this Principle misses the point. The third stage of the argument requires only that there be some way of drawing a distinction between following a rule and seeming to follow a rule - it does not require that the distinction be sound. For Malcolm is arguing that it cannot be known whether a private rule is applied consistently, and consistent application does not require true application. That this is the fundamental issue can be seen from Malcolm's opening question in the passage quoted above: ‘Now how is it to be decided whether I have used the word consistently?'. (2) To insist that a distinction must make a difference may be a form of verificationism, but it is hardly positivist verificationism. The peculiar flavor of the private language argument comes in showing Just how the adherent of private language fails to draw a distinction with a difference.

6 For a detailed and excellent discussion of the various convolutions this line of argument can take, see Saunders, John Turk and Henze, Donald F. The Private Language Problem (New York: Random House 1967).Google Scholar

7 This point is made especially clearly by Fogelin, Robert Wittgenstein (London: Routledge & Kegan Paul 1976),Google Scholar Ch. XIII, ‘The Private Language Argument.’ He argues that Wittgenstein is trafficking in what he calls ‘general sceptical arguments’ at this point. Such arguments, he maintains - and correctly so - cannot be safely contained; in as far as they are successful against certain solutions to a problem, they are successful against all solutions to the problem. In 265, Wittgenstein seems to think that a public time-table will do the Job of determining whether one is correct about the time of departure in a way that memory cannot. To this, Fogelin suggests that we ‘examine Wittgenstein's own method for checking memory reports. Supposedly, in the time-table example, I can check my recollection by looking at a genuine time-table. To pick one sceptical doubt of any number available, what is my criterion for saying they match?: is it that they seem to match? That doesn't help, for things may seem to match without matching, so we appear to need yet another standpoint for deciding whether my recollection really matches or only appears to match the real timetable' (163).

8 This point needs some qualification, or perhaps clarification. The issue of uncheckable checks does lead to a blind alley; the issue of whether there are privileged objects of representation which are (to use Sellars’ phrase) ‘selfauthenticating' is the real target though it is mistakenly confused with the uncheckable checks issue. I discuss the real target in section 3.

9 Thomson thinks that Wittgenstein holds that there is something especially wrong with private ostension (130·1). This is a mistake. Wittgenstein deliberately reminds the reader of his earlier discussion of ostension in the paragraph (257) immediately preceding his characterization of a private language (258).

10 See footnote 7.

11 Kenny, Anthony Wittgenstein (Cambridge, MA: Harvard University Press 1973), especially pp. 192-193,Google Scholar suggests an alternative reading of this passage: the appeal to a ‘subjective Justification’ for the use of a private sign ‘S’ is viciously circular, for what is used as a Justification (memory of what'S’ means) is the very thing that is to be Justified. In other words, Kenny claims, the independence Wittgenstein insists on is not sceptically motivated but is trivially acceptable: one's ground cannot be the very thing that is being Justified. This is a simple logical point, not a sceptical remark about the adequacy of memory as a check. This is a promising suggestion; however, I do not think that it is correct. It is not the case, as Kenny claims, that ‘the memory of the meaning of “S” is being used to confirm itself (193). This appears to be the case, for if I have remembered the time of departure correctly it will match my memory of the time-table. But this would equally be true if one were matching the memory of the time of departure with a public time-table. The weight of Wittgenstein's argument seems to rest on the claim that the memory of the time-table cannot be checked for correctness. Of course, any attempt to check this memory (without appealing to public criteria) would be circular, as Kenny claims, but this is the point at which the empiricist would stop the chain of Justification arguing that this memory is in some way self-justificatory. So Kenny's interpretation embroils him in the same debate that I argued leads up a blind alley.

12 That Wittgenstein speaks of the standard meter and color-charts in the same connection does not mean that he sees no differences between the highly conventional use of the standard meter and the way in which we identify colors. His discussion of the peculiarity of basic color Judgments (as both Judgments and standards for correctness) in On Certainty (New York: J. & J. Harper Editions 1969) 522-548, 624-628, and Remarks on Colour (Berkeley: University of California Press 1977), displays the way in which color-judgments are fully human institutions without being simply given to us in our sensory experience nor being matters of decision, as in the case of the standard meter.

13 Pears, David Ludwig Wittgenstein (New York: Viking Press 1969) 161.Google Scholar Pears clearly thinks this is the road to the preservation of private languages. He argues inCh. viii ‘Sensations; that Wittgenstein is only successful in showing that meaning cannot be private, but that he cannot show that reference cannot be private. Wittgenstein tackles this problem explicitly and succeeds in showing that private reference is as vacuous as private meaning. Also see Alan Donagan, Wittgenstein on Sensation; in Pitcher, for a similar view. This position, it seems to me, is Just another version of the ‘two meaning’ theory defended by A.J. Ayer in Language, Truth and Logic, 2nd ed. (New York: Dover Publications 1946). Instead of talking of the first-person meaning and the third-person meaning of a word like ‘pain,' one talks of the public meaning and the private reference.

14 Descartes, R. ‘Meditations on First Philosophy,’ in The Philosophical Works of Descartes, Vol. 1., tr. Elizabeth Haldane, S. and Ross, G.R.T. (Cambridge: Cambridge University Press 1972) 148Google Scholar

15 Ibid., 149

16 If one thinks that the private sensation does have a use, namely, to guide the subject in his application of the word ‘pain; then we have come full circle, for this is the full-fledged version of a private language.