Kripke's Wittgenstein and the Impossibility of Private Language: The Same Old Story?

I. Some Background on the Debate over Kripke's Wittgenstein

Long before the existence of Kripke's W on R & PL, Wittgenstein commentators wondered whether and how Wittgenstein's PLA establishes the impossibility of a private language. Their efforts focused on PI §243 and the next 80 or so sections of PI. Notoriously, Kripke suggests that previous Wittgenstein scholars have misunderstood Wittgenstein's PLA. For one thing, they were unaware that the PLA is really a response to a new form of scepticism. According to Kripke, Wittgenstein's rule following considerations (roughly, PI §§143-242) contain a sceptical paradox which purports to show "all language, all concept formation, to be impossible, indeed unintelligible." (K, p. 62). Wittgenstein, a la Hume, is supposed to give a sceptical solution to his sceptical paradox, and this solution "contains the argument against 'private language'; for allegedly, the solution will not admit such a language" (K, p. 60). More specifically, "the sceptical solution does not allow us to speak of a single individual, considered by himself and in isolation, as ever meaning anything." (K, pp. 68-9).

Since it is KW's PLA that is the focus of my paper I will not here rehearse the arguments and counter-arguments connected with KW's sceptical paradox, save en passant. Of course, the question of what is and what is not to be found in sections 143-242 of PI is neither trivial nor unimportant. I have elsewhere provided my two cents worth to this debate and suffice to say that I read these sections of PI very differently from Kripke. Because of this I have chosen to ignore the paradox and focus on KW's PLA to see to what extent it can be brought in line with standard accounts of Wittgenstein's PLA. Because KW claims that the sceptical solution contains his PLA I need only look to the solution in order to do this. Nicely enough, KW himself can be seen to ignore the paradox. After all, he claims that it cannot really be answered (i.e., there is no straight solution to it) but can at best be sidestepped via a sceptical solution.²

Critics have not, of course, spared KW's sceptical solution, insisting that it distorts Wittgenstein's PLA. Most importantly, and vehemently, they charge that KW is wrong to insist "on community-aid for solitary rule-followers". Eschewing community, KW's critics stress the importance of "regularities of action of sufficient complexity" to yield the requisite rough ground for the normativity required for talk of rule following and meaning generally. In a word, KW is accused of supposing that obeying a rule is a social practice whereas his critics claim that all that Wittgenstein requires of a practice is that it be public, but not necessarily social.³

This debate, as suggested above, is sometimes framed in terms of a contingently/necessarily private distinction, in the following way: Call rules or languages which only one person can possibly understand "necessarily private" and any other type of private rules or languages, "contingently private". KW's critics claim that he denies the impossibility of contingently private rules and languages whereas "the real" Wittgenstein rules out necessarily private rules and languages. This same debate is also carried on in a jargon-less way by focusing on the question: Does Wittgenstein rule out the possibility of a Crusoe figure, isolated from birth, following rules and a fortiori, speaking a meaningful language?⁴ KW's critics have spilled much ink in trying to show both that he is committed to denying that such a person can be said to follow rules or speak a language and that such a denial is absurd, and so no part of the real Wittgenstein's argument.

Since I am sympathetic to the idea that contingently private languages are possible, I admit to finding it difficult to believe that "the real Wittgenstein" denies that we can talk of a lifelong Crusoe following rules or speaking a language. If KW claims that we are barred from saying that Crusoe speaks a language then I do not see how KW can speak for the real Wittgenstein. For not only is such a claim implausible in its own right, but it seems especially so in the case of Wittgenstein. Indeed, the claim has been shown to be incompatible with passages from Wittgenstein's manuscripts.⁵ However, the worm in the apple of this attack on KW is that he does allow us to talk about a solitary individual, a Crusoe, e.g., following rules. In a passage that seems to be either totally ignored or only dimly appreciated by his critics, KW says of the sceptical solution:

Does this mean that Robinson Crusoe, isolated on an island, cannot be said to follow any rules, no matter what he does? I do not see that this follows. What does follow is that if we think of Crusoe as following rules, we are taking him into our community and applying our criteria for rule following to him. The falsity of the private model need not mean that a physically isolated individual cannot be said to follow rules; rather that an individual, considered in isolation (whether or not he is physically isolated), cannot be said to do so. (K, p. 110).⁶

This passage makes clear that KW does not deny Crusoe his contingently private language. But an important question remains, viz., what sort of language does an individual "considered in isolation" speak, contingently private or necessarily private? The answer to this question is important since the novelty of KW's PLA is entirely bound up with it. If an individual considered in isolation speaks a necessarily private language then KW, despite appearances, has reasserted a bit of very old news. Most importantly, criticizing KW for being committed to the impossibility of contingently private languages will no longer be possible.

While it is true that many critics of KW seem unaware of this distinction between a "physically isolated individual" (hereafter, 'PII') and an "individual considered in isolation" (hereafter, 'ICI'), let alone the question of what sort of language an ICI speaks, they ought not take all of the blame for this. It can be said of this distinction what is often said of the passage in Camus' The Stranger where Meursault announces his concern for all humanity, viz., nothing in the book seems compatible with it, so much so that readers of both books no doubt feel betrayed. It is quite easy for readers of W on R & PL to suppose that KW is giving an argument to establish the impossibility of someone on a desert island speaking a language. How else to take Kripke's claim on p. 69 that "the sceptical solution does not allow us to speak of a single individual, considered by himself and in isolation, as ever meaning anything"?

II. ICIs and the impossibility of rule-following: First Pass

Before we can answer questions about the nature of the language of an ICI we need to get clearer on ICIs themselves. Doing this will, of course, require a bit of digging and backtracking in Kripke's text. We will see however that despite appearances to the contrary, KW's ICI cannot be said to follow rules or speak a language precisely because s/he is confined to necessarily private rules and a necessarily private language.

To begin with, we know that an ICI cannot be said to follow rules whereas a PII can be said to do so, but what exactly is the source of this difference? KW's answer, contained in the following passage from p. 110, seems to be that we cannot say that an ICI follow rules because we cannot apply our criteria for rule following to him, and there is no similar bar in the case of a PII:

Remember that Wittgenstein's theory is one of assertability conditions. Our community can assert of any individual that he follows a rule if he passes the tests for rule following applied to any member of the community.

This is not, however, terribly informative for we still need to know why we cannot apply our criteria for rule following to an ICI or why an ICI cannot pass our tests. We can see that an ICI cannot pass our tests for rule following on pain of undermining the key result of the sceptical solution but what we want to know is why an ICI cannot pass our tests.

An answer to this requires getting clear on KW's use of 'he passes the test for rule following'. KW's critics seem to have understood this phrase in two very different ways. The different readings are best distinguished by appeal to the function quus. Does KW use, 'he passes the test for rule following', so that a quus-follower can be said to pass our tests for rule following, or does he not, i.e., does KW force us to admit as rule followers all and only those who use our actual rules? Some critics of KW suppose that he is committed to the latter reading, while others do not. For me, the suggestion that KW forces all rule followers to follow our rules is quite tendentious.

The key difficulty with committing KW to the claim that all who pass our rule following tests must be using our actual rules is that it makes the link between, (i)being an ICI, and (ii) the impossibility of an ICI following rules, quite mysterious. For if all and only those who can pass our tests for rule following are those who are using our actual rules then an ICI turns out to be someone who is not following our actual rules. But it is unclear how or why the mere fact of being an ICI must result in being, e.g., a quus-follower or a non-plus-follower. In short, the cart has got before the horse here. Instead of showing how the fact of being an ICI forces such a person to be a follower of bizarre rules we get the latter fact being used to identify an ICI!

KW himself nowhere suggests that being an ICI is either necessary or sufficient for being a follower of bizarre, nonstandard rules. And rightfully so, I think, for if he had tried to make either connection it would have rendered his claim that an ICI cannot be said to follow rules rather implausible. It is quite a leap to go from the claim that someone is following rules that strike us as bizarre or nonstandard to the claim that such a person is following rules that we cannot possibly understand and so cannot be said to be following rules at all. Admittedly, it may take awhile before we can recognize the rule that a nonplus-follower follows but there doesn't seem to be any logical bar to our doing so. In fact, such possibilities seem necessary if there are to be new rules or improvements in and evolution of, our standard rules. For new rules and alterations in old rules necessarily begin life as nonstandard. Since we cannot deny that we have over time extended and improved our rules the proper conclusion is that reading KW as forcing us to admit as rule followers all and only those who use, e.g., plus but not quus, is to misread him.

We ought then to read KW as allowing us to ascribe rule following to anyone we could, with luck and persistence if need be, recognize to be following rules. But this reading leads straight to the conclusion that an ICI follows necessarily private rules. For an ICI is not someone who follows bizarre but intelligible rules (e.g., quus) but is rather someone who follows rules that no one else in our community can possibly recognize. Such being the case however, an ICI is someone who follows "private rules" in the standard sense, viz., necessarily private rules. For similar reasons an ICI will also speak a private language, in the standard sense. Despite appearances then, KW's ICI seems to be covered by standard accounts of the PLA: An ICI cannot be said to follow rules precisely because s/he is following necessarily private rules , i.e., rules that no one else can possibly understand.

III. ICIs and the impossibility of rule-following: Second Pass

Those who claim that KW's PLA gives us the impossibility of contingently private languages (and not just the impossibility of necessarily private languages) might suggest that my conclusion is a bit hasty since it results from focusing on the tail end of Kripke's argument. Such a focus, it could be said, leads to identifying an ICI simply with someone who cannot be "taken in by the community" or who "cannot pass our tests for rule following". However, this identification is misleading and overly simplistic, it could be said, because there is more to being an ICI. In particular, whereas standard commentaries on Wittgenstein's PLA stress that a private language is, by definition, one that cannot in principle be understood by another person, an ICI is a private rule follower per accidens, as it were. If this is so then there is a significant difference between KW and more standard Wittgensteins. But is it so?

The place to look for the answer is KW's account of the assertability conditions for meaning talk. (K, pp. 87-92). The following passage seems to contain KW's most important claims concerning the nature of an ICI:

[T]he 'assertability conditions' that license an individual to say that, on a given occasion, he ought to follow his rule this way rather than that, are, ultimately, that he does what he is inclined to do. The important thing about this case is that, if we confine ourselves to looking at one person alone, his psychological states and his external behavior, this is as far as we can go. . . . There are no circumstances under which we can say that, even if he inclines to say '125', he should have said '5', or vice versa. By definition, he is licensed to give, without further justification, the answer that strikes him as natural and inevitable. Under what circumstances can he be wrong, say, following the wrong rule?
No one else by looking at his mind and behavior alone can say something like, "He is wrong if he does not accord with his own past intentions"; the whole point of the sceptical argument was that there can be no facts about him in virtue of which he accords with his intentions or not. All we can say, if we consider a single person in isolation, is that our ordinary practice licenses him to apply the rule in the way it strikes him." (K, p. 88).

The remainder of this section will be devoted to showing that KW's argument for why an ICI cannot be said to follow rules, contained in the passage above, is either unsound or else presupposes the standard notion of privacy.

Formally, KW's argument can be cashed out as follows:

(1) Our assertability conditions license an individual to follow a rule as s/he is inclined to follow it (i.e., "thinking one was obeying the rule is the same thing as obeying the rule").

(2) If we "confine ourselves" to a solitary individual's psychological states and external behavior then we can go no further than the assertability condition mentioned in (1);

therefore

(3) If we consider a single person in isolation, all we can say is that our ordinary practice licenses him to apply rules in the way it strikes him, and such being the case, there is no room to legitimate talk of this person following rules.

One problem with this argument is that premise 1 is false.⁷ Oddly, KW himself recognizes that it is false. As he says on p. 88: "It is by no means the case that, just because someone thinks he is following a rule, there is no room for a judgement that he is not really doing so." I have long been befuddled at this obvious conflict between KW's claim about the nature of the assertability conditions for an individual and his apparent recognition that an individual's inclinations to follow a rule are by no means the last and only word on the matter. But now I see he may have a way out.

The fact is that the falsity of (1) poses no real problem for KW's argument. This is because (1) can be dropped without affecting the argument. As strange as it may seem, (2) can be true even though (1) is false, and (2) may be adequate to insure the conclusion (3). To see why, note that the key difference between these premises is that in (2), but not in (1), we are "confined" to the psychological states and external behavior of one person alone. It is precisely this confinement that permits (2) to be true despite the falsity of (1).

Nicely enough, it is precisely this confinement that is also the key aspect of an ICI. We will see that an ICI comes into existence when we confine ourselves to a single individual's psychological states and external behavior. And this confinement is also what permits KW to conclude that ICIs cannot speak a meaningful language.

Now how, exactly, does the notion of confinement serve to bring ICIs into existence and to render them unable to speak a meaningful language? Following KW, we know that a PII can be said to follow rules whereas an ICI cannot. However, our ascribing rule following to a PII seems to involve nothing more than appealing to the external behavior of the individual in question.⁸ For example, in trying to determine whether Crusoe is following rules, what more do we appeal to than his external behavior? If the answer is, "Nothing", then it seems that KW's confining us to, e.g., Crusoe's psychological states and external behavior, is not going to prevent us from ascribing rule following to him. But I suggested above that it is precisely by way of this confinement that a physically isolated Crusoe is supposed to be turned into a Crusoe who is considered in isolation. But if this confinement does not prevent us from ascribing rule-following to a Crusoe who is considered in isolation there is an obvious dilemma: Confined to Crusoe's external behavior we seem capable of saying that he follows rules (because we need appeal to nothing more than his behavior in ascribing rule following to him) and cannot say that he follows rules (since being confined to Crusoe's external behavior makes him an ICI)!

The only way out of this dilemma, it seems, is to take the notion of confinement very seriously indeed. In particular, when Crusoe is "considered in isolation" our confinement to his psychological states and external behavior could mean that we are no longer allowed to make use of our psychological states, our external behavior, nor, most importantly, our "socially imbibed" criteria for rule following in trying to making sense of Crusoe's behavior, linguistic or otherwise. In short, whereas we are allowed to spectate on a physically isolated Crusoe and describe his behavior in our terms, (our concepts and our language) we are not permitted to do so once we have considered Crusoe in isolation. Thus, the reason we can ascribe rules to a PII but not to an ICI is because in talking about the latter we are not allowed to use our language but are rather confined to seeing and saying all and only that which the ICI sees and says. Since the ICI cannot, according to Kripke, justifiably ascribe rules to himself, it follows that an ICI cannot be said to follow rules.

Now I admit that this understanding of KW's distinction between a PII and an ICI is not explicitly emphasized by KW himself. In fact, I'll admit that this way of understanding the distinction is not at all obvious and even a bit bizarre. I would also suggest that it takes much of the sting out of KW's case against private language. However, it seems to me to be the only way to understand KW's claim that although an ICI cannot be said to follow rules there is no such impossibility in the case of a PII. I see no other way of distinguishing between ICIs and PIIs that will preserve this key claim of KW.

The most interesting consequence of this reading of KW's "confinement" is that it effectively renders an ICI (qua ICI) the sole arbiter of his actions. If anyone other than the ICI says anything about the ICI then that person has not, ipso facto, confined himself to the psychological states and external behavior of the ICI and so is not talking about an ICI at all. A useful analogy here is Thrasymachus' ill-fated distinction between rulers and rulers qua rulers (or more generally, craftsmen qua craftsmen) in Book I of Plato's Republic. According to Thrasymachus, rulers qua rulers cannot make mistakes about ruling, nor can mathematicians qua mathematicians make mathematical errors, etc. His point is not that, e.g., Pythagoras never made a mathematical error, but simply that insofar as he did so, he could not at the time of the error properly count as a mathematician. In the same way, we can say things about ICIs but when we do so we are no longer officially talking about an ICI. An ICI qua ICI cannot be talked about by anyone other than himself. For in considering an individual in isolation we are forced to speak that individual's language and only that individual's language. In essence, we can say all and only that which the ICI can say, thus making the ICI the sole legitimate arbiter of his actions. At this point, the question whether an ICI can be said to follow rules comes down to the question whether an ICI is unable to distinguish following a rule from merely thinking s/he is following a rule. (This is simply another way of asking whether premise 2 from p. 9 above is true when the solitary individual is an ICI). If there is no way for an ICI to distinguish between following a rule and merely thinking s/he is following a rule then an ICI cannot be said to follow rules. Of course, KW claims that an ICI is unable to make this distinction. Clearly, if KW is correct then an ICI cannot be said to follow rules or speak a meaningful language.

For my purposes, the interesting question is not whether an ICI can make the distinction between following and merely thinking s/he is following the rule but rather the nature of an ICI's language. I think it is clear that an ICI speaks a private language in the standard sense, i.e., s/he speaks a language which no one else can in principle understand, and furthermore, an ICI speaks such a language by definition. For recall that being an ICI is a matter of being considered in a particular way. It is only by adhering to our complete confinement to the ICI's Weltanschauung that an ICI is brought into existence. In other words, an ICI is, by definition, the one and only legitimate speaker of and about his language and activities respectively. As soon as another speaker says anything different from the ICI or uses a "language" different from the ICI's then the ICI has thereby ceased to be an ICI. Given this, an ICI is by definition speaking a language which no one else can possibly understand, i.e., a necessarily private language, a private language in the standard sense.

IV. Concluding Remarks

So long as it is supposed that KW's ICI does not speak a necessarily private language, the claim that it is impossible for an ICI to follow rules or speak a meaningful language is quite startling. It is so because it threatens standard accounts of the PLA by threatening the standard account's notion of 'private language'. I have deflated KW's startling claim by showing that an ICI does in fact speak a necessarily private language. Thus, far from challenging the standard account of the PLA, Kripke's work seems simply to echo it: KW's ICI cannot follow rules or speak a meaningful language precisely because an ICI follows necessarily private rules, i.e., rules which no one else can in principle understand.

There is, of course, still the matter of the impossibility of an ICI succeeding in the task of legitimate self-ascription of rule following. Above, I simply ignored the question whether such self-ascription by an ICI is impossible. But is it? By my lights, it is not. For I see nothing to prevent an ICI from distinguishing between merely thinking he is following a rule and following a rule. Or at least, whatever problems an ICI may have in making such a distinction apply, mutatis mutandis, at the community level as well. Consequently, I do not accept KW's claim that it is impossible for an ICI to follow rules or possess a language.

Given this I am inclined to allow that there are important differences between Wittgenstein's PLA and that of KW. I find the key difference to be that whereas Wittgenstein's private linguist is confined to a "language" in which the meanings are mental objects of a peculiar sort, KW's ICI is someone who can be described only in his own terms. That is, all that can be said about an ICI is exactly what the ICI can say about himself and in the very same terms that the ICI uses. This difference explains why Wittgenstein's PLA is untouched by the possibility of ICI's following rules and speaking a genuine language whereas KW's PLA is undermined by such a possibility.

One final point. If I am correct then Kripke's work has the interesting and significant (albeit unintended) consequence of establishing that Wittgenstein's own PLA is nontrivial or not analytic. For on my reading, KW has established that the impossibility of a language does not necessarily follow from the fact that the language is necessarily private.⁹

1 Saul Kripke. Wittgenstein on Rules and Private Language. Cambridge: Harvard University Press, 1982. (Hereafter W on R & PL or K). The main critics of Kripke's account are G. P. Baker and P.M.S. Hacker in their Scepticism, Rules and Language (Oxford: Basil Blackwell, 1984; hereafter, SRL) and Wittgenstein: Rules, Grammar and Necessity (Oxford: Basil Blackwell, 1985; hereafter, RGN); and Colin McGinn in his Wittgenstein on Meaning (Oxford: Basil Blackwell, 1984). The main supporter of Kripke's account is Norman Malcolm. See his Nothing is Hidden (Oxford: Basil Blackwell, 1986), and "Wittgenstein on Language and Rules" in Philosophy, 64, 1989 (hereafter, WLR).

A note on pronouns: Most often, my use of pronouns follows Kripke's.

² Recall that Kripke claims that the sceptical paradox does not admit of a straight solution but only a sceptical solution. What this means is that the sceptic's assertions cannot be answered but can at best be defused. That is, the sceptical solution shows how or why we can get on with meaning talk in the absence of a straight answer to the sceptic. Clearly then it is not the sceptical paradox that threatens the propriety of standard accounts of the PLA. If standard accounts of the PLA go wrong then they do so because they are at odds with the sceptical solution to the paradox. Also, the paradox itself seems to contain or reveal little or nothing that almost all pre-Kripke commentators of the PI were not fully aware of, viz., that many of the pre-PLA sections contain attacks on a Platonistic (or, more generally, an overly psychologistic) conception of meaning. For these reasons it seems safe to say that the failure of previous commentators to notice the sceptical paradox seems quite irrelevant to the matter of their failure to do justice to the PLA.

³ Norman Malcolm, op. cit. and Colin McGinn, op. cit.

⁴ See Malcolm, op. cit. See also A.C. Grayling's "Wittgenstein's Influence: Meaning, Mind and Method" in Wittgenstein: Centenary Essays, Supplement to Philosophy 1990, #28, p. 78.

⁵ See Baker and Hacker's SRL, p. 41 and their RGN, pp. 172-180, where they cite and discuss the various passages from Wittgenstein's Nachlass. Baker and Hacker's reading is challenged by Malcolm, especially in his WLR.

⁶ See McGinn, op. cit., p. 197, fn. 84, for evidence that the import of this passage has been neglected.

⁷ Kripke seems to suppose that premise 1 follows from the sceptical paradox. It does not. Neither before nor after the sceptical paradox is premise (1) true. Also, prima facie, the argument looks patently circular as (1) and (3) seem to say the same thing. The difference, brought out below, lies in the use of 'isolated individual' in the conclusion but not in premise (1).

⁸ It is odd that no commentator on Kripke's book seems to have appreciated that KW's sceptical challenge, in offering us access to any information we like about the psychological states and external behavior of others, has probably offered us much less than it seems. Being a Quinean both by temperament and training, I regard Kripke's offer of access to another's psychological states in trying to determine what s/he meant as, in one sense, on a par with being offered all of the land north of the North Pole. (Cf. PI §141).

⁹ I would like to thank Ray Elugardo for helpful comments on the original version of this paper, as well as the sceptical students who tried, along with me, to make sense of Kripke's arguments during my course on Wittgenstein's rule-following considerations in Winter Quarter 1993.

Critical Discussions| Top of Page | WRPL Unofficial HomePage