From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

1. The Semantic Argument

What most people think of as "the Private Language Argument" receives its clearest expression at #258 of Philosophical Investigations. There Wittgenstein writes:

"Let us imagine the following case. I want to keep a diary about the recurrence of a certain sensation. To this end I associate it with the sign "S" and write this sign in a calendar for every day on which I have the sensation. -I will remark first of all that a definition of the sign cannot be formulated. -But still I can give myself a kind of ostensive definition. -How? Can I point to the sensation? Not in the ordinary sense. But I speak, or write the sign down, and at the same time I concentrate my attention on the sensation -and so, as it were, point to it inwardly. -But what is this ceremony for? For that is all it seems to be! A definition surely serves to establish the meaning of a sign. -Well, that is done precisely by the concentrating of my attention; for in this way I impress on myself the connexion between the sign and the sensation. -But "I impress it on myself" can only mean: this process brings it about that I remember the connexion right in the future. But in the present case I have no criterion of correctness. One would like to say: whatever is going to seem right to me is right. And that only means that here we can't talk about 'right'."

A reconstruction of (what I shall call) this Semantic Argument against private language might run as follows:

(1)

In any (possible) language, there must be an is right/seems right distinction;

(2)

In a private language, no such distinction can be drawn; so

(3)

There cannot be a private language.

The idea behind (1) is presumably the unobjectionable one that meaning is normative. To say that a word has meaning is thereby to say that there are (or can be) uses of the word which are correct (right) and uses which are incorrect (wrong). What makes, e.g., 'table' a word (part of a language) rather than a meaningless squiggle is that there are uses of it which are clearly correct, and uses of it which are clearly incorrect.

It is just this distinction which Wittgenstein thinks cannot be made out by the private linguist: in trying to give meaning to a sign by 'inner' ostensive definition, " … I have no criterion of correctness. … whatever is going to seem right to me is right. And that only means that here we can't talk about 'right'." (#258) Hence, the rationale for (2).

Of course, strictly speaking, what Wittgenstein has provided a rationale for is not (2), but:

(2*)

No private linguist can make out to himself an is right/seems right distinction.

Some have objected that the step from (2*) to (2) is verificationist: why should the fact that the private linguist cannot make out to himself such a distinction imply that there is no such distinction? This step may be just worth calling verificationist, but it should be noted how weak it is. Verificationism about meaning is much less controversial than verificationism about truth. It does seem bizarre to suppose that there could be distinctions of meaning in a language necessarily inaccessible to speakers of that language, whereas it is far from bizarre to suppose that there can be forever inaccessible truths about the external world. Further, the possibility of conceding (2*) yet denying (2) is hardly a congenial one for a defender of private language.

Moreover, the case for (2*) has not been made out. Wittgenstein considers and rejects two ways in which a private linguist might make out an is right/seems right distinction: at #265 he considers an appeal to memory, and at #270 he imagines that one might use the connection between the occurrence of a sensation and some external effect (e.g., a manometer registering a rise in blood-pressure). However, Wittgenstein's objections to these suggestions, and his presentation at #258, suffer from the same defect: an over-simplified view of the resources a private linguist might draw on.

As Crispin Wright has emphasised, the private linguist might turn theorist, naming not just one sensation but many, and noting regularities in their occurrence.¹ The linguist as theorist can then make perfectly good sense of the possibility that he might be mistaken in one of his judgements. (For example, he might notice that S1, S2 and S3 always occur in a certain sequence. His identification of a sensation as type S3, may make him realise he was wrong to classify his previous sensation as S1.) In this way, the linguist's theory can provide the materials for the construction of an operational is right/seems right distinction.

Nor is (1) unobjectionable. A. J. Ayer wondered whether even English admitted of an is right/seems right distinction, in the sense that Wittgenstein requires.² Certainly, one can be corrected by one's fellows, but ultimately agreement about usage can only be based in individuals' memories and appearances, and how can we then sustain an 'objective' is right/seems right distinction? It may be replied that correct usage simply consists in conformity with one's fellows. But this view is counterintuitive: surely an is right/seems right distinction for a public language like English ought to allow for the possibility that we all might go 'off the rails' in our use of a word. Moreover, simply to assert this definition of 'correct usage' is to beg the question against a defender of private language.

Further, one might ask why ordinary self-ascriptions (e.g., 'I have a headache') are not counterexamples to (1). Such ascriptions are perfectly meaningful, yet seem not to admit of any is right/seems right distinction. It may be replied that a word like 'headache' features in first- and third-person attributions, where grasp of both kinds of attribution is essential to understanding the word, and the third-person or criteria-based use does allow for an is right/seems right distinction (e.g., 'I thought he had a headache, but now I see he was only acting'). But how does that help with the first-person case, which still seems resistant to our target distinction?

There is also a problem with contingently private languages in the case of the #258 argument. It is generally agreed that contingently private languages are possible. Does the #258 argument not exclude such languages? That is, is premise (2) not indifferent to whether the private language is contingently private or logically private? After all, a socially isolated Crusoe, who invented his own language from scratch, can only rely on his memory and senses, just like the private sensation linguist. How then can Crusoe make out an is right/seems right distinction? It may be replied that Crusoe does indeed speak a language when he names the trees and mountains around him, and that his language admits of an is right/seems right distinction in virtue of the fact that we could, in principle, learn his language and so make sense of what it would be for some of his uses to be mistaken.

However, a defender of the possibility of a logically private language might reasonably reply that this response begs the question: it simply assumes that an is right/seems right distinction is available only for languages which others can in principle learn. Why should this be conceded? Second, if Crusoe speaks a language, he surely does not do so in virtue of the fact that we might understand and correct him; he does so in virtue of intrinsic facts about him and his world.

The problem raised by contingently private languages reinforces our conclusion that the #258 argument cannot be cogent.

2. The Metaphysical Argument

The discussion in the previous section has been somewhat negative. Let me now try to extract from Wittgenstein a better argument against private language, one which gets to the heart of the issue. The textual support for this reading is not extensive, but it fits with many of Wittgenstein's ideas and intentions. Here I draw heavily on John Cook's excellent article 'Wittgenstein on Privacy'.³

What we are trying to show to be impossible is a logically private language, that is, a language which necessarily only one person can understand. What could be as much as a (putative) example of such a language? Only a private sensation language, since any language about 'outer' objects could, in principle, be learnt by others. But this is only the right answer on a certain conception of sensations, viz., sensations as private objects, in terms of which sensation words are to be defined. Without that conception of the terrain, there is no subject matter for a logically private language to be about. And the claim will be that Wittgenstein does make suggestions as to how we might resist the view of sensations as private objects, and thus undermines the only rationale there could be for believing in the possibility of a logically private language.

At #253 Wittgenstein writes:

Another person can't have my pains." Which are my pains? What counts as a criterion of identity here? Consider what makes it possible in the case of physical objects to speak of "two exactly the same", for example, to say "This chair is not the one you saw here yesterday, but is exactly the same as it".

    In so far as it makes sense to say that my pain is the same as his, it is also possible for us both to have the same pain. (And it would also be imaginable for two people to feel pain in the same -not just the corresponding - place. That might be the case with Siamese twins, for instance.)

    I have seen a person in a discussion on this subject strike himself on the breast and, say: "But surely another person can't have THIS pain!" The answer to this is that one does not define a criterion of identity by emphatic stressing of the word "this". Rather, what the emphasis does is to suggest the case in which we are conversant with such a criterion of identity, but have to be reminded of it."

The root of the idea of sensations as private is that, e.g., however similar our headaches might be, you and I cannot (literally) have the same headache. On this conception of headaches, and only on this conception, if I define the word 'headache' ostensively, I have invested that word with a logically private meaning. If we can undermine this conception of headaches (and other sensations), we can finally lay to rest the idea of a logically private language.

How might we do this? The first two paragraphs of #253 are suggestive, and Cook is the philosopher who has most fully worked out this suggestion. The standard account of sensations assumes that two people cannot literally have the same sensation, however similar (in duration, intensity, etc.) those sensations might be. According to Wittgenstein, this account misconstrues our language-game with sensations: " … it is … possible for us both to have the same pain." And he gives, as illustration, the case of Siamese twins who feel a pain in a common part. However, Wittgenstein surely intended his comments to apply quite generally, and not just to the case of Siamese twins. Cook suggests how such a generalisation might be motivated.

As noted, on the standard view, the sentence 'I have the same headache as my father' (where we both have a headache of exactly similar duration, intensity, etc), understood literally, expresses an impossibility. What we should say is: my father and I have similar headaches. It is this piece of reasoning which, in the name of Wittgenstein, Cook wants to ridicule. He suggests that this reasoning is as wrong-headed as the following: "the sentence 'I have the same build as my father' (where we have an exactly similar build), understood literally, expresses an impossibility. Builds are, literally, private. What we should say is: my father and I have similar builds."

According to Cook, the just mooted line of reasoning is a travesty, and contrary to the 'grammar' of the word 'build'. In the case of builds, unlike that of tables and chairs, there is no space for a distinction between 'same F' and 'exactly similar but distinct Fs'. There is no (strict or literal) sense in which I fail to have my father's build. In the only sense there is, I have the same build as my father (which I can also express by saying that we have similar builds). It's not as if I have something (my build) which my father cannot have, however physically similar to me he may be. Builds are simply not private in that way.

Cook's suggestion is that the conception of sensations as private or unshareable, without which there could be no logically private language, gained currency because the language-game with sensations was wrongly assimilated to that for ordinary physical objects, with its firm distinction between numerical and qualitative identity. A better model of the language-game for sensations is that for builds (or character, gaits, senses of humour, etc.). Once we see that, we see there is simply nothing to the idea (vividly expressed in the final paragraph of #253) that what I have in having a pain is (literally) something that you cannot have. And now, at last, we have a cogent argument against private language: there is simply nothing that could be the subject-matter of a logically private language; so there could not be such a language.