Some Oddities in Kripke's Wittgenstein on Rules and Private Language

Oddity One:  Kripke claims that Wittgenstein has invented "a new form of scepticism", one which inclines Kripke "to regard it as the most radical and original sceptical problem that philosophy has seen to date, one that only a highly unusual cast of mind could have produced" (K, p. 60).  However, Kripke also claims that there are analogies (and sometimes the analogies look very much like identities) between Wittgenstein's sceptical argument and the work of at least three and maybe four other philosophers, viz., Quine, Goodman, Hume and Berkeley.  Very strange stuff, methinks. 

The originality of Wittgenstein's work is especially difficult to see after Kripke claims that Wittgenstein presents a problem concerning the nexus between past . . . 'meanings' and present practice" (K, p. 62), and says that Hume is said to have questioned "the causal nexus whereby a past event necessitates a future one, and the inductive inferential nexus from the past to the future." (K, p. 62).  Wither the originality?  And the connection with Goodman's work with 'grue' is even closer than that between Kripke's Wittgenstein and Hume.  Given that Kripke had read Goodman before "discovering" the rule-following paradox in Wittgenstein, one ought to be sceptical of Kripke's claim that Wittgenstein has invented a new form of scepticism.  It would be much more accurate to say that Kripke has strapped Wittgenstein with a hybrid scepticism drawn from Hume and Goodman.

Oddity Two:  Kripke claims that Wittgenstein's private language argument is not focused on showing how or that private language is impossible. As Kripke sees things, Wittgenstein's main problem is NOT, "How can we show private language -- or some other special form of language -- to be impossible?". (K, p. 62). Rather, says Kripke, Wittgenstein's private language argument falls out of the attempt to answer the question, "How can we show any language at all (public, private, or what-have-you) to be possible?" (K, p. 62). According to Kripke, "Wittgenstein's main problem is that it appears that he has shown all language, all concept formation, to be impossible, indeed unintelligible." (K, p. 62). Kripke lays out the details of his Wittgenstein's modus operandi as follows:

Of course I am suggesting that Wittgenstein's argument against private language has a structure similar to Hume's argument against private causation. Wittgenstein also states a sceptical paradox. Like Hume, he accepts his own sceptical argument and offers a 'sceptical solution' to overcome the appearance of paradox. His solution involves a sceptical interpretation of what is involved in such ordinary assertions as "Jones means addition by '+'." The impossibility of private language emerges as a corollary of his sceptical solution of his own paradox, as does the impossibility of 'private causation' in Hume. It turns out that the sceptical solution does not allow us to speak of a single individual, considered in isolation, as ever meaning anything. (K, p. 68).

Ignoring for now any worries about the accuracy of Kripke's understanding of Hume's sceptical anlace, (of which I have many), this passage finds Kripke insisting that Wittgenstein will offer a "sceptical interpretation" of our meaning assertions (presumably, this means an interpretation that respects key elements of the sceptical attack on meaning but nonetheless somehow permits us to avoid the intolerable conclusion that such assertions make no sense), which interpretation will restore substance to our ordinary meaning attributions but which will be seen to be inapplicable in the case of solitary individuals, individuals considered in isolation.

However, an examination of the appropriate sections of Kripke's book (roughly, pp. 87 ff.) reveals something quite different from what Kripke advertised on p. 68.  On p. 88, Kripke gives an argument the conclusion of which is: "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."  Although there are well-known problems, exegetical and otherwise, with KW's notion of a person "considered in isolation" (ICI), it is at least clear that the ICI is the subject of KW's case against the possibility of private language.  Thus the passage from p. 88 shows that for KW, the alleged private linguist is licensed to apply rules as it strikes him to apply them.

Kripke then claims, rightly I think, that this result effectively makes the private linguist someone for whom the notion of rule following can be given "no substantive content".  (K, p. 89).  As Kripke notes, unless we are able to judge the legitimacy of someone's claim to be following a rule, or someone's claim to have provided an answer in accord with the rule s/he previously followed, our rule following talk is without substantive content.  Since KW claims that we can do neither in the case of the ICI, it is clear to him that substantive rule following is impossible in the private case, i.e., the case of the ICI. In short, it is not possible to obey a rule "privately".  What's as significant as it is obvious here is that KW has argued for the impossibility of private rule following and so private language prior to arguing for the possibility of some language or other.

For immediately after arguing that the ICI is someone who cannot be a substantive rule follower because "all we can say is that he is licensed to follow . . . rule[s] as it strikes him", Kripke claims:

The situation is very different if we widen our gaze from consideration of the rule follower alone and allow ourselves to consider him as interacting with a wider community.  (K, p. 89).

Clearly then KW is not first showing us how some language or other is possible and then as a corollary discovering that this possible language has features which for some reason cannot be applied in the case of the private linguist. In fact, KW's reasoning runs in the opposite direction.  That is, KW first argues for the impossibility of rule following and language in the private case ("all we can say, if we consider one person in isolation . . . "), which argument is then alleged to lose its force in the "non-private" case. In particular, KW claims that once other people are allowed in the picture, it is no longer the case that "all we can say of someone is that s/he is licensed to follow rules as it strikes him/her".

Contrary to Kripke's advertisements then, it is clear that his Wittgenstein's case against private language is no mere corollary to his sceptical solution.  Indeed, KW's case against private language is the foil for his argument to show that some language or other (in this case a public one) is indeed possible, despite the best efforts of his sceptic.  And because of this, I think we end up with much more than an oddity.  We end up with a substantive difficulty that threatens the force and legitimacy of KW's entire discussion.

The difficulty is made clear by cashing out KW's sceptical solution’s modus operandi.   KW begins his solution by insisting on a shift from a truth-conditional account of meaning to an account of meaning based on assertion conditions (or justification conditions).  The rationale for this maneuver is that it allegedly renders the sceptic's case against meaning facts irrelevant.  As Kripke says:

Now if we suppose that facts, or truth conditions, are of the essence of meaningful assertion, it will follow from the sceptical conclusion that assertions that anyone ever means anything are meaningless.  On the other hand, if we apply to these assertions the tests suggested in Philosophical Investigations [i.e., the tests associated with an assertion conditions conception of meaning], no such conclusion follows.  All that is needed to legitimize assertions that someone means something is that there be roughly specifiable circumstances under which they are legitimately assertable, and that the game of asserting them under such conditions has a role in our lives. K, pp. 77-8).

I won't here pause to critique the picture contained in this passage.  I cite it as evidence that KW believes that the shift in meaning conceptions is the key to solving, albeit sceptically, the sceptical problem of the meaningless of our meaning talk.  This passage makes clear why KW believes he can grant the sceptic's negative assertions (i.e., no fact shows Jones ought to say '125' rather than '5'; no fact shows that Jones meant plus rather than quus) and yet avoid the meaningless of our meaning talk, viz., facts are irrelevant to the legitimatization of our meaning attributions.

After advocating a shift in meaning conceptions, KW then appears to claim that even given the shift in meaning conceptions, it is impossible to give substance to talk of an ICI following rules, meaning anything by his/her words, or speaking a language.  (Of course, it goes without saying that on a fact-based conception of meaning, private language is impossible for anyone who accepts the sceptic's negative assertions, as KW does.  After all, on a fact-based conception of meaning, all language is impossible, if one accepts the sceptic's negative assertions.  For these reasons, we must read KW's case against private rule following as presupposing an assertion conditions conception of meaning).  KW then, of course, suggests that "widening our gaze" to a community permits us to give substance to our rule following talk.  However, while this conclusion establishes that substantive talk of rule following is possible (thereby answering the sceptic's claim that all language is impossible), it should be clear that KW has not provided us with a viable case against private language.  He has at best shown that, on an assertions conditions conception of meaning, talk of private rule following cannot be rendered substantive, cannot be said to be possible.  But clearly this result does not warrant saying that private rule following is impossible under any and all possible conceptions of meaning talk.


Oddity Three:  The third oddity arises in connection with KW's contention that "widening our gaze" from the ICI to a community enables us to give substance to our rule following talk, or why KW believes that he has offered us a solution to this problem that does not apply in the case of the ICI.  As noted above, KW's so-called sceptical solution is not a solution of the problem of the impossibility of language but rather offers a "solution" to the alleged problem of the impossibility of private rule following.  The solution's key move is the claim that in a community setting, there will be a host of "justification conditions for attributing correct or incorrect rule following to . . . subject[s] . . . and these will not be simply that the subject's own authority is unconditionally to be accepted."  (K, p. 89).  [Q: What is it that we do not have to allow the subject to be unconditionally authoritative about?  That the subject is following the rule for plus?  That the subject's answer to a computation problem is correct?  That the subject means plus by '+'?].  But how or why does the mere existence of competing justification conditions yield substantive rule following or solve the problems that allegedly rendered the ICI incapable of substantive rule following?  KW's answer is obvious but hardly enlightening: it prevents "the subject's own authority" from being "unconditionally accepted".

But how does this allow us to determine whether someone means plus and not quus prior to his/her giving an answer to '68 + 57'? The answer, I believe, is that it does no such thing.  It only allows us to say of an individual's answer that it is at odds with our own.  And while this is, in one sense at least, more than we're permitted to say of an individual considered in isolation, (as I note elsewhere, we seem unable to say ANYTHING at all about an ICI without thereby undermining his/her status as an ICI; however, if we are allowed to say anything at all about people considered people in isolation, we can say that such a person is, or is not following a particular rule, and we will make such a claim based on whether s/he has given answers in accord with the rule in question) it obviously does not enable us to claim that the individual ought to say 125 rather than 5, or vice versa.  KW, in short, seems to confuse being able to say something more of a putative rule follower than "s/he is licensed to apply the rule in the way it strikes him/her" (K, P. 88), with having given substance to rule following talk.  At best, KW has offered us a flimsy notion of substantive rule following or an individual considered in isolation.  At worst, he has failed to appreciate that we can say of an ICI exactly what we can say of individual community members, viz., that s/he is (or is not) following particular rules.

Oddity Four:  Actually, this is an oddity and a half, but who's counting?  On p. 90, Kripke offers us "rough assertability conditions for such a sentence as "Jones means addition by 'plus'." Never mind that Kripke first floated the notion of Wittgenstein's alternative construal of such sentences on p. 66!).  However, he starts by laying out the assertability conditions for Jones making such an attribution. Again, never mind that the conditions offered here are of no use.  (First and foremost, is there anything in these acs which would permit them to be distinguished from say, the acs for Jones to say of himself that he means quus, or that he means minus, or times, etc., by 'plus'. Why anyone would regard these as viable and legitimate acs is a complete mystery to me.).  According to KW, all that allegedly prevents Jones from being allowed to rely on nothing more than his feelings of confidence about the correctness of his understanding and his "new responses", is that Jones's claim that he means addition by 'plus' is "subject to the correction of others", as are his claims about the correctness of his "new responses". Absurdity one here is that Jones appeals only to feelings of confidence, or his inclinations in declaring his meaning or in calling his new responses correct.  Surely the answers themselves are what Jones looks to, as well as their working in whatever circumstances Jones may find himself adding.  His inclination and confidence are at best byproducts of correctness, not the source and certainly not their justification or legitimization.

The second absurdity is talk of "correction by others" here. It sounds nice but it is illegitimate.  We can see this by asking how one go would about correcting Jones's claim that he means addition by 'plus'. Does one claim that Jones is lying or mistaken about his "feeling of confidence"?  If so, how is the charge of lying or mistake to be legitimated? That's quite mysterious indeed.  Obviously, Kripke is supposing that we check the legitimacy of Jones's meaning attribution (to himself) and the legitimacy of his claims of correct answers by comparing Jones's answers to our own.  But what can checking mean here?  Do we check of our respective inclinations with those of others?

Ultimately, however, these problems with talk of "correcting" Jones's claims are overridden by the lack of any account of what justifies someone other than Jones to say of Jones that, e.g., he is adding.  Kripke's discussion of assertion conditions offers us an account of how the community is able to "correct" Jones's "mistaken" claims about his meaning and his answers but it never offers us an account of the assertion conditions for ascribing a particular meaning to Jones.  In particular, Kripke tells us the assertability conditions for the community to say of Jones that s/he doesn't mean plus by '+' (roughly, when Jones gives answers different from those his community gives for plus) but he never explains how the community is justified to say of Jones that he does mean plus).  

Oddity four and a half then is that Kripke offers us neither ACs for others to ascribe meanings to Jones nor does he appreciate that he offers us nothing that explains how we are to rule out alternative meanings for Jones's use of '+'.  In particular, as far as we can tell, the conditions under which a community would be legitimated to say that Jones means plus by '+' would also permit that community to say that Jones means quus by '+' as well.  As such, the celebrated AC conception of meaning is unable to solve the problem of showing how someone is justified to say 125 rather than 5.

Also, why do we need an ac conception of meaning in order to avoid having "a subject's own authority from being unconditionally accepted"?  The correct answer to this is that we do not.  For a fact based conception of meaning is also able to avoid having a subject's own authority about what s/he is doing from being unconditionally accepted.  If this is all the ac conception can offer us, it is not offering us anything that a fact based conception of meaning cannot offer.  The bottom line here is that KW avoids having the ac conception solve the original sceptical problem by having it solve the problem of showing how to avoid having a subject's own authority from being unconditionally accepted.  Allegedly, solving this problem allows us to have substantive rule following.

However, it is obvious that solving the private language/rule problem is something a fact based conception can do.  (Indeed, by my lights, the alleged ac conception of KW is knee deep in facts; the truth is that the nonexistence of facts sought by the sceptic are not a sufficient basis for rejecting a fact based conception of meaning and so it's no surprise that facts make their way into KW's alleged ac conception of meaning).  And so a fact based conception also permits us to have substantive rule following. The only thing it couldn't do is show how we can justify saying '125' rather than '5', to '68 + 57', assuming that the problem '68 + 57' is "new" to us.  And the reason it could not do this is because there is no fact that shows we mean plus rather than quus.  And the reason there's no fact is because the existence of such a fact is tied to showing how to justify saying '125' rather than '5'. (Sounds circular, doesn't it?).  However, KW's ac conception also cannot show how to justify saying '125' rather than '5' to 68 = 57.

The argument in KW's favor may go something like this: If I say '125' and this answer is agreed to by everyone else in my community then it's justified and answering '5' is not justified because it would be rejected by my community.  The problem with this answer is that it fails to appreciate that answering '5' is justified if I mean quus. And my community's inclination to mean plus by '+' does not, cannot, prevent me from meaning quus by '+'.  Consider: If it does, then what sense is there in the community's declaring that I don't mean plus by '+'?  Obviously, I do not have to mean by '+' what the community means by '+'.  In order for my community to play a role in the justification of my claim that 68 + 57 = 125, the community must somehow make it possible to determine whether I mean plus or quus prior to my giving my answer.  If it cannot aid in making this determination, and it cannot, for the determination is impossible, then I am not justified to say 125 rather than 5.  

To repeat: a proper justification for giving one answer rather than another to '68 + 57' requires being able to justify a particular sort of meaning claim (or so KW claims when he is attacking fact based conceptions of meaning. For example, in order for us to allow that Jones is justified to say 68 + 57 is 125 rather than 5, we must also somehow be convinced that Jones meant plus rather than quus in the past (or so KW claims in sceptical argument).  But nothing, according to KW, not just no fact but nothing at all, can show that Jones meant plus rather than quus in the past.  As such, it's a complete mystery as to how KW can claim to have SOLVED the sceptical problem.


Oddity Five:  Perhaps the most serious oddity of them all. KW's sceptical solution begins with the recommendation that we switch from a truth conditional conception (tc conception) of meaning to a justification or assertion conditions conception of meaning (ac conception).  The following passages are representative:

If . . . we allow ourselves to adopt an oversimplified terminology . . . we can say that Wittgenstein proposes a picture of language based, not on truth conditions, but on assertability or justification conditions: under what circumstances are we allowed to make a given assertion? (K, p. 74);

Now if we suppose that facts, or truth conditions, are of the essence of meaningful assertion, it will follow from the sceptical conclusion that assertions that anyone ever means anything are meaningless.  On the other hand, if we apply to these assertions the tests suggested in Philosophical Investigations, no such conclusion follows.  All that is need to legitimize assertions that someone means something is that there be roughly specifiable circumstances under which they are legitimately assertable, and that the game of asserting them under such conditions has a role in our lives. K, pp. 77-8).

Both passages clearly suggest that assertability conditions or justification conditions are going to somehow legitimize or justify our meaning claims, in ways which tcs or facts cannot. However, all that KW offers us is an appeal to what people actually say.  That is, KW offers us an appeal to what people actually say or do as the way to solve the sceptical problem. But how can an account of what people actually say constitute a justification of what they actually say?  The short answer is that it cannot. Several Kripke passages are instructive here.  First, one from pp. 86-7:

Following Wittgenstein's exhortation not to think but to look, we will not reason a priori about the role [meaning attributions] ought to play; rather we will find out what circumstances actually license such assertions and what role this license actually plays.

Now, this "answer" has some merit (who, after all, wants to "reason a priori" about the role meaning attributions ought to play; I don't) but it is also misleading.  First, the use of 'rather' is odd here. For there is no contrast at all between what Kripke says we ought not do (viz, reason a priori about the role meaning attributions ought to play) and what he is going to do (viz., determine the role which the licensing of meaning attributions "actually" plays).  If he had contrasted reasoning a priori about the role meaning attributions ought to play with determining the role they actually play, (or even contrasted the circumstances that ought to license meaning attributions with those that pass for licensing circumstances among actual language users) the 'rather' would be legitimate.  But as things stand, it is not.  Another oddity is that it seems trivial to say what role the licensing of meaning attributions actually plays. Indeed, it seems I can determine this from my easy chair (and so a priori?!): THE ROLE THAT LICENSING OF MEANING ATTRIBUTIONS ACTUALLY PLAYS IS THAT OF LICENSING THE ASSERTION OF MEANING ATTRIBUTIONS.

Besides these oddities, the statement above is ambiguous. On the one hand, the expression, "what circumstances actually license such assertions" could be taken to refer to the circumstances that genuinely legitimate meaning assertions, where the genuine legitimating circumstances need not be circumstances that any actual speaker actually uses. On the other hand, the expression could be taken to refer to the circumstances that actual speakers in fact make use of in legitimating meaning attributions, irrespective of whether the circumstances in question ought to be so taken. In this latter case, it is quite possible to find that people in fact accept certain circumstances as legitimating meaning attributions that they ought not.  In particular, it's clear that we take someone using '+' in accord with plus as legitimating the claim that s/he means plus by '+'. 

But KW's sceptic contends, of course, that we are wrong to do so, on the grounds that someone using '+' in accord with plus is also, at the same time, someone who can be taken to be using '+' in accord with quus. Now, if one of the key steps of KW's sceptical solution is to turn his back on questions about what circumstances REALLY justify meaning attributions in favor of simply ascertaining which circumstances in fact pass as legitimating circumstances among (benighted?) laypersons, then it becomes mysterious why a similar maneuver is unable to save fact-based conceptions of meaning from the sceptic's attack.  After all, if we ask which facts are actually regarded as licensing the meaning attribution, "Jones means plus by '+'", the correct answer is we regard the fact that Jones has answered in accord with plus as licensing this attribution.  The fact that the sceptic claims that we ought to find such a fact to be an inadequate justification or license of "Jones means plus by '+'" could then be seen for what they are, viz., the irrelevant and unsatisfiable demands of a lunatic.

At the very least, it should be clear that we need to know how Kripke is using the expression, "what circumstances actually license such assertions".  If he uses it in the latter sense noted above, it's clear that there is no room for a challenge to our practice, whether that challenge be sceptical or not.  That is, once we find the circumstances which people in fact accept as legitimating a particular meaning attribution, there is no room for a claim like, "Yes, people accept circumstances x, y and z as legitimating meaning attribution M, but they ought not, for x, y and z don't really justify M."  So long as one is interested only in the question of what passes for legitimization of meaning attributions in the marketplace, the sceptic's complaint is out of bounds.  Most importantly, however, there is no reason we cannot or ought not rule the sceptic's demands out of bounds from the very start. If we do so, there is no reason to reject a fact-based conception of meaning.

I am not, however, recommending that we do this, i.e., that we rest content with an account that tells us what people in fact accept as licensing meaning attributions.  Simply looking at our meaning attribution practice (whatever that would be) and seeing what people actually demand or accept as licensing particular meaning attributions is not good philosophy. For the existence of a practice is not sufficient to justify it.  We should not allow KW to take what people actually demand or accept as licensing meaning attributions to be an adequate answer to the question:  What licenses our meaning attributions? For what people accept as justified or properly licensed may not be. (Isn't this the insight which led Socrates to pursue the role of gadfly?!). However, this is precisely what KW offers us.

The following passage, believe it or not, is a succinct account of KW's "answer" to the question, “What licenses our meaning attributions?”:

We say of someone else that he follows a certain rule when his responses agree with our own and deny it when they do not . . . .  (K, p., 92).

I do not dispute that this is, in part, what we do.  But it also fails to note another crucial and important element in our determination of who is or is not following a rule, viz., the nature of the rule itself.  We know, e.g., that plus is a function that demands that 68 + 57 = 125 and that quus is a function that demands that 68 + 57 = 5.  As such, it's more accurate to say that we say of Jones that he's plussing when he says 68 + 57 = 125 and we deny that he's plussing if he says 68 + 57 = 5.  And while it is certainly the case that 125 is "our response" to "68 + 57", this is because we take '+' to mean plus. So it's too simple to say that it's mere agreement in responses that leads us to declare that someone is following a certain rule.  For I can declare Jones a quusser, if he says "68 + 57" is 5, despite the fact that his answer to "68 + 57" does not agree with my own.  Of course, were I quussing, i.e., using '+' to mean quus, I would also say that "68 + 57 = 5". So there is a sense in which my response agrees with Jones's response.  What Jones and I disagree about is whether '+' means plus or quus. And that seems a rather silly disagreement because it is either readily solved (we agree to use '+' to refer to one of the functions and then invent a new symbol to represent the other function) or else it cannot be solved.   (For more on the agreement issue, see Chomsky's Knowledge of Language).


Top of Page | WRPL Ultimate HomePage

Last modified September 9, 2011
JAH, Professor
Dept. of Philosophy