Literature on the Private Language Argument


(Pears(1) p.142) "...and an argument, which is based ultimately on certain empirical facts about the meaning of the word 'language', shows that (the language in which we express and describe our sensations) cannot be private in this sense (of being unteachable)".

Comment: We agree that this is the ultimate source of the PLA. But nothing substantial follows from facts about language. If we use 'language(0)', then it is indeed an empirical linguistic fact that a private language(0) is contradictory, since the usual meaning of 'private' includes 'uncheckable', while that of 'language(0)' includes being checkable. This is not substantially interesting. I respond "So what?".
If instead we use 'symbolic system', then it is an empirical linguistic fact that a private symbolic system is not contradictory, since the chosen meaning of 'symbolic system' does not include being checkable. This isn't substantially interesting either.

(p.146) "...public criteria are needed..."; p.147 "meaning is linked...with public criteria". This is the checkability.
On (p.159) is Pears' summary of the argument: "(i) There would be, for any given statement that you might make, only two possibilities: either you would be under the impression that it was true, or you would be under the impression that it was false. (ii) Neither of these two possibilities would subdivide into two further cases, the case in which your impression was correct, and the case in which your impression was incorrect. (iii) For since your statements would have been cut off from their teaching links, there would be no possible check on the correctness of your impressions. (iv) But it is an essential feature of any language that there should be effective rules which a person using the language can follow, and know that he is following. (v) Yet in the circumstances described, there would be no difference between your being under the correct impression that you were following a rule, and your being under the incorrect impression that you were following a rule, or, at least, there would be no detectable difference even for you. (vi) So there would be no effective rules in this so-called 'language'. (vii) Anything that you said would do. (viii) Therefore, it would not really be a language..." {My numbering}

Comment: (i), (iii), (v), (vi), and (vii), are correct; they express uncheckability. (ii) depends on (iv), which is either a statement of linguistic essentialism, or an empirical linguistic fact (the tendentious word 'essential' should be removed). Pears has already said he thinks that the latter is the best interpretation - in which case (viii) follows, preferably rewritten as (viii*): "Therefore it is not a 'language', in the ordinary meaning of the word".
This conclusion is trivial. The argument "is based ultimately on certain empirical facts about the meaning of the word 'language'; it reduces to: "Since 'language', as normally used, implies the existence of checkable, effective, rules, 'private language' is contradictory". To which I respond: "True. I propose the existence of a private symbolic system, with no checkable associations. This is not contradictory".
Similarly, Pears writes that for a person supposing a language just of sensations alone (p.160) "it would be difficult, if not impossible, ... to rebut the contention that this so-called 'language' would not be a language at all. For nothing that the speaker said could ever be checked." This is again only the trivial fact about normal linguistic use of 'language'.

Pears makes clear, in his chapter 'Sensations', that a sophisticated empiricist view ('C-subtle', he calls it) involves accepting that the overall meaning of 'red' or 'pain' - its usual use - involves *two* aspects:
(i) the set of teaching links: public, observable, patterns of behaviour
(ii) the inner reference: the private sensation.
Only a primitive empiricist would think it involves just the sensation.
Neither aspect is automatically dominant, in establishing the truth of the consequently ambiguous claim "I am in pain". For example it is mistaken to think that this view allows (p.155) "that a person might always have had from birth the wrong kind of sensation correlated with the teaching links of pain". This is because (p.156) "it would not be the wrong kind of sensation, and we would not say that he was not in pain. On the contrary, we would say that he was in pain, whatever the precise further character of his sensations might be. That is the function of the variable." On this view - as I would express the point - there is no right or wrong kind of sensation; what is in each person's box does not matter: (p.154) "in the specification of the type of sensation, the variable stands with promiscuously open, but never really embracing, arms. What fills them, in each person's case, is a matter of indifference to the meaning of the phrase".
Pears also makes clear the possibility of accepting,as a consequence that a language only about (ii) would be necessarily unteachable (p.158): "Someone might accept .. that our language for sensations would be necessarily unteachable, but...(this) only shows that, contrary to appearances, our language of sensations is not really teachable, and that we do not ever really communicate about such matters".
He recognises the role of verifiability in the argument. The critic of the argument will make (p.161) "the charge that Wittgenstein has simply assumed that, if it is impossible to verify directly, whether the speaker's impression that he is following a rule for the use of a sensation word is correct or incorrect, then these are not two distinct possibilities...The general question of the validity of the verification principle is raised. Is the meaning of a statment the method of its verification, and is an unverifiable statement therefore meaningless?" Good question. Few people accept the logical positivist's verification principle.

I conclude that David Pears' presentation and analysis of the PLA seems to be consistent with mine.

Any comments?