9
NOMINALISM AND THE PRIVATE LANGUAGE ARGUMENT’ Michael Hodges Vanderbilt University The best known version of the Private Language Argument (PLA) makes it depend essentially on the problem of memory, and there is substantial textual evidence to support the claim that this is what Wittgenstein had in mind. On this version (or these versions) the speaker of a PL can only appeal to memory and further “his memory” in order to check his use of terms. But in such a situation there would be no way to distinguish being right from simply thinking that one is right. So whatever seems right will be right. Even this argument can be given two forms. The first is a weak or skeptical argument. It simply rests its case on the claim that we could never know if we were using terms correctly. Clearly Wittgenstein wants to push the argument further than this. A second or strong version combines the skeptical argument with some form of the verification principle and thus argues that where there can be no evidence for or against an hypothesis the hypothesis makes no sense. Since there can be no evidence for distinguishing between cases of “seeming to be right” and “being right” there can be no distinction, but such a distinction is essential to the concept of “being right.” Thus the concept could have no place in a PL. But if that concept has no place in a PL then a PL is not possible for it is essential to the idea of a language that the concepts of “correct” and “incorrect” have a place in it. The final implication of this strong argument goes as follows. The privacy which we have been examining is a function solely of the objects which the language is supposed to be about. It is not a special feature of the language itself which makes it private but rather features of the supposed subject matter. Thus by showing that a PL is impossible we have, in effect, shown that the notion of a private object is impossible. The thrust of the PLA is not to show that there are private objects but that we cannot say anything about them (which is an absurd view in any case).* Rather it is to show that there can be no such objects in the special sense that the concept is logically incoherent. Now the above argument may be valid but even if it is, it is not convincing simply because it depends essentially on the questionable verification principle. There is another way to understand the PLA which frees it from its supposed dependence on the problem of memory Michael Hodges received his Ph.D. from the University of Virginia and is currently Associate Professor of Philosophy at Vanderbilt University. His papers have appeared in various professional journals. 283

NOMINALISM AND THE PRIVATE LANGUAGE ARGUMENT

Embed Size (px)

Citation preview

NOMINALISM AND THE PRIVATE LANGUAGE ARGUMENT’ Michael Hodges Vanderbilt University

The best known version of the Private Language Argument (PLA) makes it depend essentially on the problem of memory, and there is substantial textual evidence to support the claim that this is what Wittgenstein had in mind. On this version (or these versions) the speaker of a PL can only appeal to memory and further “his memory” in order to check his use of terms. But in such a situation there would be no way to distinguish being right from simply thinking that one is right. So whatever seems right will be right.

Even this argument can be given two forms. The first is a weak or skeptical argument. It simply rests its case on the claim that we could never know if we were using terms correctly. Clearly Wittgenstein wants to push the argument further than this. A second or strong version combines the skeptical argument with some form of the verification principle and thus argues that where there can be no evidence for or against an hypothesis the hypothesis makes no sense. Since there can be no evidence for distinguishing between cases of “seeming to be right” and “being right” there can be no distinction, but such a distinction is essential to the concept of “being right.” Thus the concept could have no place in a PL. But if that concept has no place in a PL then a PL is not possible for it is essential to the idea of a language that the concepts of “correct” and “incorrect” have a place in it.

The final implication of this strong argument goes as follows. The privacy which we have been examining is a function solely of the objects which the language is supposed to be about. It is not a special feature of the language itself which makes it private but rather features of the supposed subject matter. Thus by showing that a PL is impossible we have, in effect, shown that the notion of a private object is impossible. The thrust of the PLA is not to show that there are private objects but that we cannot say anything about them (which is an absurd view in any case).* Rather it is to show that there can be no such objects in the special sense that the concept is logically incoherent.

Now the above argument may be valid but even if it is, it is not convincing simply because it depends essentially on the questionable verification principle. There is another way to understand the PLA which frees it from its supposed dependence on the problem of memory

Michael Hodges received his Ph.D. from the University of Virginia and is currently Associate Professor of Philosophy at Vanderbilt University. His papers have appeared in various professional journals.

283

and the verification principle. However this alternative version of the argument is not free of presuppositions and I will try to show that it too depends on a highly questionable thesis.

In order to understand the version I offer it is necessary to distinguish between two ways in which “rules” can be related to actions, including linguistic acts. If I simply set out to take a walk without any real objective in mind, then none of my particular steps can be assessed as correct or incorrect. This is true even though there will be some “rule” (actually there will be indefinitely many “rules”) which correctly describes my actual steps and of course there will be a number of “rules” with which my steps do not accord. That is, I may walk eighty steps north and then thirty-five steps southeast, etc. The important point here is that even if I had walked eighty-one steps north it would have been absurd to say that I had made a mistake. We might put this in a somewhat misleading way by saying that whatever I do is correct. This is misleading precisely because it does not make sense to describe my steps as either correct or incorrect. If it makes sense to speak of a “rule” that my steps accord with in this sort of case then such a rule will be purely descriptive. It will not prescribe the course I should, but may fail to, take. Rather it will describe the course I do take. In this sense the“ru1e” waits on my actions and does not determine or guide them. Notice that there is no question here of my remembering anything or of anything’s seeming right to me in the relevant sense. With regard to this particular walk there is simply no rule which I can fail or succeed in following, and this is true because although my actions may be in accord with some rule, I am not following any rule.

Compare the above case with the following, in which someone hands me a set of instructions (a treasure map, for example) and I set out to follow them. In this circumstance we can talk about taking the right course or making a mistake. The new element that has been introduced is something independent of the actual steps that I take, in terms of which they can be assessed. Here the rule or instruction does not simply describe what I do, it prescribes what I should do. The rule does not wait on the actions; rather, the actions conform or fail to conform to the rule. What is needed in order to transform aimless walking into rule governed and therefore potentially correct or incorrect actions is something independent of the particular steps themselves by means of which they can be assessed. In the case we are considering the treasure map is something independent.

Now the thrust of the PLA is simply that a PL user has nothing independent of his actual applications of a given term by means of which to assess those applications. Therefore his applications cannot be correct or incorrect. Although they may be in accord with some rule they are not governed by any rule. But where behavior is not governed by a rule it is not rule governed and where behavior is not rule governed there is no possibility of error, not because a PL user is uniquely placed so as

284

not to ever be wrong or because he could never tell if he was wrong, but simply because, like our aimless walker, he is not following a rule at all.

If there is some already existing system of classification in terms of which actual applications of a term can be understood, the situation is radically changed. Now there is something independent of my particular uses in terms of which those uses can be evaluated. The only way to make room for the notion of the correct use of a term is by appealing to something independent of the uses to be evaluated. Thus Wittgenstein argues that public criteria and public custom provide the necessary background for the possibility of meaningful language. At 198 Wittgenstein says, “I have further indicated that a person goes by a sign- post [follows a rule] only in so far as there exists a regular use of sign- posts, a custom.” What the custom provides is precisely the “something independent” which makes possible behavior which is not merely in accord with a rule but which is rule governed.

Wittgenstein’s point can be put as follows. From the point of view of a user of a private language every use of a term is like a christening. If I am to name a ship and if I am properly authorized to do so then there can be no question of my getting the name right. Whatever I call it isjust what it is. Of course in the case of any ordinary naming, once it has been accomplished, further uses of the name can be correct or incorrect. This is precisely what can never happen in the case of a private language.

To see this we must consider the case of common names, as opposed to proper names. A common name is applicable to more than one individual. Of course there may be, as a matter of accident, only one individual of a kind, but that is not the point. The class of individuals to which the term applies is open in the sense that it is always logically possible that there are other individuals which are members of the class in question. Now let us suppose that I am in the same position with regard to a common name as I was with regard to the christening of a ship. First I christen a given individual “S.” This first use cannot be correct or incorrect, for I have simply decided to call this “S” and I am free to call it what I will. The question is whether this fact, the fact that I call this “S” in any way determines my future acts. What of some second individual? Is it an s! Well, if I call it an S then it is an S.

We may be tempted at this point to appeal to the notion of sameness and say that if the second individual is the same as the first then it is an S. However this is an empty wrinkle because there is no independent standard of sameness. If I call a second individual Sthen it is the same as the first. It is an Stoo. For Wittgenstein the appeal to sameness is useless because “The use of the word ‘rule’ and the use of the word ‘same’ are interwoven.” (225) Sameness comes into play with rules but so far we have no rule. Nothing determines what I am to do, for all that constitutes something being an S is that I call it an S.

It is just this feature of a PL that Rush Rhees is trying to explicate when he says, “It seems that in a private language everything would have to be at once a statement and a definiti~n.”~ If, as the PL theorist

285

contends, meaningful use of language is to be possible, then some uses of “S’ must be statements. Instead each new use is simply a further interpretation of the rule for using “S,” a further element in the definition of “S.” There is nothing except those very uses against which they can be evaluated. As Wittgenstein said, it is “as if someone were to buy several copies of the morning paper to assure himself that what it said was true” (265). My practice establishes itself as a yardstick against which it measures itself.

While discussing this whole question of how rules can determine action Wittgenstein says,

This was our paradox: no course of action could be determined by a rule, because every course of action can be made out to accord with the rule. The answer was: if everything can be made out to accord with the rule, then it can also be made out to conflict with it. And so there would be neither accord nor conflict here. (201)

So far the notion of action which is “determined by a rule” has not been sufficiently grounded. So there is no accord or conflict. There is no correct or incorrect, no sense to “getting it right.” But after all we can “get it right” which “shows that there is a way of grasping a rule which is not an interpretation, but which is exhibited in what we call ‘obeying the rule’ and ‘going against it’ in actual cases.” (201) It is interesting to note that these passages are followed immediately by what seems to me to be the PLA in short form:

And hence also ‘obeyinga rule’is a practice and to think one is obeying a rule is not to obey a rule. Hence it is not possible to obey a rule ‘privately’: otherwise thinking one was obeying a rule would be the same thing as obeying it. (202)

The groundwork for the PLA is set by the extensive and sustained discussion of how rules determine actions which begins around 186 and goes right up to 242. Throughout these sections Wittgenstein deals with what it is “for the future steps to be determined by an order.” (1 89) “How it is possible for a rule to show me what I am to do at this point.” (198) “HOW actions can be determined by rules” (20 1-2), “the connection between sameness and rules.” (225) All of this leads directly into the discussion of the PLA at 243 and following. Interpreters have focused too much attention on the sections following 243 and not enough on the essential groundwork that precedes. No doubt it is possible to read the text say from 243 to 300 in terms of the verificationist argument as I mentioned at the beginning, but this reading simply ignores the fundamental and groundbreaking discussion of the nature of rule governed behavior that sets the real context of the PLA. In fact the verificationist’s version connects up Wittgenstein’s argument with a position or positions which Wittgenstein nowhere explicitly discusses. The great advantage of my interpretation is that it connects the PLA and the central themes which are discussed and rediscussed throughout The Philosophical Investigations.

286

If we return now to the situation in which a PL user attempts to christen an individual “S,” we can see that having called the first individual “S” he is free to call every new individual S or to call no new individual S. The use of “S” may be in accord with some rule but it is not governed by any rule precisely because there is nothing independent of each of the actual applications of “S“ by which they can be assessed. But if this is true the term “S“ is not meaningful at all. It is not a question of remembering anything as some philosophers have thought. So far there is nothing to remember. This version of the PLA does not depend on the concept of memory at all. In the crucial sections of the Investigations where Wittgenstein discusses the issues, at 258 and 265, for example, he does not stress the term “memory” but rather the terms “right” and “correct .”

At 258 Wittgenstein says that the process of christening an individual and, as he puts it, impressing “on myself the connection between the sign and the sensation” has got to bring it “about that I remember the connection right in the future. But in the present case I have no criterion of correctness. One would like to say: what is going to seem right to me is right. And that only means that we can’t talk about ‘right’.” The emphasis on ‘right’ is Wittgenstein’s. If we interpret this passage in the light of our previous analysis Wittgenstein’s point is not that one has no way of telling whether he is correct but rather that so far there simply is nothing to be correct about.

Then did the man who made the entry in the calendar make a note of nothing whatever? Don’t consider it a matter of course that a person is making a note when he makes a mark-say in a calendar. For a note has a function, and this “S‘ so far has none. (260)

Again at 262 Wittgenstein says,

It might be said: if you haven’t given yourself a private definition of a word, then you must inwardly undertake to use the word in such-and-such a way. And how do you undertake that? Is it to be assumed that you invent the technique of using the word; or that you found it ready-made?

If we suppose that we “invent the technique” as we go along, then as we have seen there will be no function for the term, no rule against which I can measure my applications. If, on the other hand, we do find the technique “ready-made,” then we have tied into a practice, a common language and our activities will be understood in that context. Throughout this discussion Wittgenstein draws on the material already discussed in the earlier sections of the Investigations. What is central here is the question of action being determined by rules.

Even at 265 where he does talk explicitly of justification he makes the point that a justification can only make sense if it means appealing to something that is “actually correct.” In short what is presupposed is that there is something to be correct about and, of course, where the question is one of train departures this presents no special problem. It makes

28 7

sense to talk of remembering the departure time, of “being right” simply because there is something to be right about. However, where the question is one of the application of “S,” there is nothing to be correct about since “S’ has no use thus far.

The crux of the PLA, as I have presented it, is the thesis that, for a PL user, there can be no independent criterion of sameness. But what is meant by “criterion” here? We must be careful for on one plausible reading the version I have offered transforms itself into the more classical verificationist version mentioned at the beginning. If we understand by “criterion” a test, then the force of the argument is that a PL user can appeal to nothing beyond his own claim to recognize a similarity between two instances and therefore whatever he thinks is right is right. But there is an important ambiguity in this position. It is certainly true that from the point of view of verifying the claim that A and B are similar there is nothing beyond the supposed recognition af that fact to which an appeal can be made. If one chooses to call into question such basic acts of recognition, then perhaps skepticism cannot be avoided. However, we are not concerned with skeptical con- siderations here but with the intelligibility of a PL user’s making a mistake or correctly identifying a particular individual. On the view that we are now considering, it is not the case that thinking that A and B share a common property constitutes their sharing a common property, and therefore it is not true that whatever I think is right is right. On this view one might perfectly well think that A and B share some particular feature when in fact they don’t. In short, so long as one is willing to accept the view that there are extra-linguistic similarities or shared properties one can drive a conceptual wedge between thinking that he is right and being right, although it may be impossible to determine when in fact one is right if the claim to recognize a common feature is called into question.

Now it may be suggested that a difference that it is impossible to detect is no difference. That is, if there is no way in principle to determine when I just think I am right as opposed to being right, then I am not entitled to talk of such a difference. This line of defense for the PLA transforms it into the well-worn and well-known version mentioned at the beginning, for the argument will now essentially depend on the verification principle.

These considerations bring us back to the question of the meaning of “criterion” in the claim that there can be no independent criterion of sameness for a PL user. Perhaps the question to ask is why it is that the fact that A and B are similar or share some common feature is not an independent criterion of sameness. If there is some such extra-linguistic fact it would seem to provide the necessary background against which the PL user’s claims to classification can make sense. Consider the following point made by Rush Rhees: I cannot learn the colour unless I can see it; but Icannot learnit without language either. I know it because I know the language. And it is similar with sensations .... 1 feel the same

288

sensation and that is the same colour. But the identity-the sameness-comes from the language. (p. 94, Philosophy and Ordinary Language)

This gives us a hint as to the fundamental issues involved. Apparently, if we are to follow Rhees, Wittgenstein’s position is that any classification, that is any sameness among individuals, presupposes a linguistic or conventional element for “the sameness comes from the language.” No doubt there is also an objective or extra-linguistic component for “I cannot learn the colour unless I can see it” but such an element is not sufficient. Now it is precisely this conventional element that is and must be missing in the case of the PL user and it is also this conventional element-the rule-which would constitute the “something indepen- dent” which is necessary if it makes sense to talk of correct or incorrect use. Perhaps we can get clearer here if we consider an array of differing color shades laid out in a spectrum, each shading into the next. Now while considering any two of these individual shades suppose we ask if they are shades of the “same color.” The answer to such a question will make reference not only to the color samples themselves but also and perhaps more importantly to the system of color classification employed. This is true simply because there are innumerable ways of dividing up such a spectrum of shades and no particular way can by itself lay claim to being the right way. Until we know which division or classification is being employed the question concerning whether two individuals are the same color cannot be answered simply because it has no answer outside such a system.

Now if the “sameness comes from the language” in the way Rhees suggests it should be clear why a PL user cannot use the supposed fact that A and B are similar or share some common feature as an independent criterion. That A and B are the same in some way necessarily involves a linguistic or conventional component but in the case of a PL user that element is simply his own actual uses. Since his uses constitute the practice and since his uses are to be evaluated by that practice it is as though one were using a yardstick to measure itself. “Imagine someone saying: ‘But I know how tall I am.’ and laying his hand on top of his head to prove it” (279).

Let us look more closely at the thesis that we have uncovered here. It is certainly true that our actual classifications are conditioned by the language that we employ. Such a claim is by now a commonplace of both .psychology and philosophy. Our language provides us with a flexible set of categories which, no doubt, greatly informs our classifications and views of the world. Certainly our divisions do not simply reflect natural lines to be found in an extra-linguistic world, but in large part reflect our interests and purposes as enshrined in the concepts of our shared language. This much classificatory conven- tionalism is, I think, perfectly unobjectionable. However such a position does not provide the support necessary for the PLA. So far no claims have been made with regard to the necessity of a conventional element.

289

All that has been claimed is that in the normal cases a good deal of conventionalism is combined with an extra-linguistic element. In order to support the PLA, as I have presented it, one of two stronger theses would have to be defended. First, one could maintain that in the end there simply is no objective or extra-linguistic ground for a system of classification. I will call this view “nominalism” although I do not mean to suggest that all philosophers who have called themselves nominalists would be committed to such a conclusion. A second and more plausible line of support would be to maintain that although there is some extra- linguistic component, it must be combined with a linguistic or conventional component in order to provide a basis for a system of classification. On this second position there can be no purely extra- linguistic justification for classification. I think we have now uncovered the real nerve of the PLA. The PL theorist cannot simply recognize a similarity and use that as a basis of his linguistic usage. To “recognize a similarity” presupposes a linguistic element. There can be no recognition without language.

It is not my purpose in this paper to provide a full-scale defense or refutation of the conventionalist thesis which underlies the PLA as I have developed it. However it will be interesting to see how this position sits with Wittgenstein’s own treatment of classification in terms of family resemblances. Although there is a prima facie compatibility, I want to argue that the family resemblance account of classification provides us with nothing new as support for the PLA. In fact either it provides the PL user with the necessary extra-linguistic background for classification, or, if it does provide support for the PLA, it must be in the form of the radically nominalistic position I mentioned.

At first glance Wittgenstein’s family resemblance analysis seems to provide us with a general model for combining both the extra-linguistic and conventional elements in classification. On this view a class is constituted not by the presence of some one property in all of its members but rather by an overlapping web of similarities and differences. However, once one has abandoned the single essential property view it seems clear that something more than a n extra- linguistic element is necessary for classification. The web of overlapping similarities and differences has no natural breaks. It can be divided at innumerable points and no point is The Right Point. Thus if we are to have a class of individuals “S,” not only will we have to appeal to the objective, extra-linguistic overlapping similarities and differences but also to the linguistic conventions governing “S.” The situation envisioned could be represented as follows:

s1 s2 s3 s4

Pz P3 P4 P3 P4 Ps P4 Ps p6 PI P2 P3

290

Obviously there is no natural break in this progression and just as obviously if “S‘ is to specify a class the progression must be broken at some point. What “breaks into” the extra-linguistic continuity is precisely the conventions of the language.

Now supposing we grant the plausibility of this account for the term “S,” what are we to say of “Pz” and its relation to SI and S Z ? ~ Certainly SI and S2 form a class even if it is a sub-class of S. Will the same sort of family resemblance account be applied to “P?? If so we will be left with a new level of overlapping similarities and differences to deal with, and any attempt to deal with them will commit us to still further overlapping similarities and differences. However, if we stop short at any level and admit some non-family properties then the plausibility of the PLA is destroyed, for with regard to that non-family property, call it “F,” it seems to make sense to suppose a prelinguistic recognition of it in two or more instances, and we have seen that this much provides the PL user with the “something independent” that is demanded.

On the other hand if we push the family resemblance analysis through completely we will be left finally with individuals and no properties at all. In short the family resemblance account of classification pushed to its logical conclusion leads to the denial of the view that there is an extra- linguistic justification at all for classification. This position is precisely what I called nominalism earlier and while such a view may support the PLA, it strikes me as totally implausible. What seems to flow from all this is that if Wittgenstein’s doctrine of family resemblances is to provide us with the sort of support we had hoped for it must be interpreted in a radically nominalistic way.

That the doctrine of family resemblances allows for a nominalistic interpretation and that such an interpretation would support the PLA is some indication that there are philosophical theses hidden in The Investigations and a further reminder that it does no good to leave one’s presuppositions unexamined, for the unexamined presupposition is not worth holding.

NOTES

’ A version of this paper was presented at the APA, Eastern Division, December, 1973. G. Pitcher seems to take the PLA in this quite paradoxical fashion. See The

Philosophy of Wittgensrein, p. 300, where he says “The point is not that private sensations ... do not exist ... the point is rather that nothing can be said about them.” ’ R. Rhees, “Can There Be a Private Language?” Philosophy and Ordinary Language, p. 96i

For a full discussion of this point see my “Wittgenstein on Universals,” Philosophical Studies, Vol. 24, No. I , January, 1973.

29 1