This problem seems worse than that faced by Buridan's famous ass, for we are faced with many more than two alternatives. Also, most of the usual responses to that dilemma have to do with the fact that it's not a "real" problem, either because it did not happen ("we'll cross that bridge when we come to it") or because it can never happen, either because of Divine perfection or the complexity of the material world. Our problem is amenable to neither of these solutions since the choice really confronts us as a formal problem where we are unhampered by material constraints.

Withstanding the (apparent) force of these arguments, which is (therefore) an easier feat than it might appear, it seems that there is no right choice of the set of equivalence classes. In fact there is no choice at all. We find ourselves in much the same situation as when we had our discussion with Quine's McX: we do not wish to merely deny that there is a right choice, but that there is any choice at all, i.e. to assert that all sets of equivalence classes are, for all practical purposes, the same one.

This surprising conclusion is actually the result of the combined force of the two original arguments. The first focuses on the differences between the various classes which are, of course, undeniable. The second, on the other hand, focuses on the similarities of the various classes, viz. that all serve equally well as choices for a universe for interpretation.

Although they reach opposite conclusions -- the first that all classes are different, the second that they are all the same -- there is no factual disagreement between the two either about the nature of the classes or on the nature of the distinction between the various classes. The facts of the matter are that all classes serve equally well as potential hypotheses for any translation while they are, at the same time, radically different. It would seem that the only reason that one would consider these classes is for translation, so that one might truly say that for all practical purposes they are the same in light of the fact that all serve equally well for translation.

One might reverse the tables and consider the opposite situation: Confronted by an attempt at translation which we must evaluate, what reason do we have for thinking that it is wrong?

This situation is actually the more practical one; translations -- radical and otherwise -- have been done before the problem of indeterminacy was posed. Thus, confronted by some enthusiastic linguist, our challenges to his conjectured translation are strangely sterile. We can certainly not say for sure that his proposal is wrong (as long as it is adequate given the class he has chosen) although we might be unwilling to certify it as correct. This is because of the observation which underlies the second argument, that all classes serve equally well.

As to the first argument, which makes much of the fact that the classes differ so radically, we may ask what it is that can serve to distinguish the "right" translation from all others. It can certainly not be something internal since, as we have been observing all along, all reasonable candidates are internally indistinguishable. But then we must look for some external standard against which to measure our candidate, a search we know to be doomed to failure because of the universal relativity and ultimate inscrutability of analytic hypotheses.

In fact, for all we know, our hypothetical linguist may actually have the "right" one, whichever that is. In fantasy, since he is after all hypothetical, we may suppose that he has in fact come up with the "one and only" right translation. We would have no way of knowing this and could (and certainly would if we believe in them) make all the same arguments against his proposal.

The search in general for some distinguished translation among all the adequate ones is a truly absurd task. It is a labor that we have no reason to believe that we must undertake, towards a goal that we would not know we had achieved (if we do make it -- if we are not already there).

It is this last feature when makes it so awesome when we are charged by Quine with doing the feat. But it is this aspect which might also make us suspect. If we would actually search for round squares we may also look throughout eternity while philosophical reflection will allow us to complete that task much sooner by concluding that none exist anywhere. In our case too, we might reflect that universal relativity makes the acquisition of the "right" translation worthless, since it too would be relative, and furthermore, in view of the locus of all this indeterminacy, it becomes meaningless as well.

*

It is an easy consequence of Quine's observations on indeterminacy beginning at home that there is no right translation. If there were such a thing, it would presumably be a translation constructed in such a way as to enable a native speaker of either language to determine the partitioning used in the other language through the use of the correspondence rules which make up the translation. In order for this to be possible it is necessary that both partitionings be fixed.

This seems like such an obvious and reasonable requirement that one feels almost rude in pointing out that it is a requirement that tends toward the nonsensical.<1> There is absolutely no way that we can talk about the partitioning used by "a" native speaker when, for all we know, each one has a different set of equivalence classes, and even that some speakers of different languages have identical partitionings.

One might say that the indeterminacy arguments can be said to show that it would be desirable if we did not have to concern ourselves with partitionings of the universe. One hopes that they have been seen to demonstrate that we cannot meaningfully deal with the partitionings in any absolute sense.

We will try to give some reasons why we need not discuss them. One caveat: Otto Neurath's characterization of the task of a philosophy of language is doubly appropriate. For one because we are using language to construct a theory of itself, and also because language can represent only serially, which we are trying to make it do of itself while it is being used and perceived holistically.

If we turn our attention to that component of language that we have previously called the "universe," we find something that may appear somewhat surprising. It is not the public space composed of middle sized dry goods and the like, but rather a more private area, differing from speaker to speaker, made up of his particular (mental) representations of that outside world.

It is the idea of an object which I carry in my head, that I use to interpret the symbol representing that object -- not the object itself. That these are not identical should need no argument. Although no claim is made about the gap separating the two we may assume that they are as close as each person can make them except that the object is concrete while my representation of it is not. This is merely an observation of the way things work; it has no epistemological consequences.

The fact is that even these mental entities play no role in the actual exchange of information, despite the fact that the goals of that process are framed wholly in terms involving those entities. One who wishes to communicate intends that the mental representation of a particular part of the world that he has before him also be in front of the person with whom he is communicating.

However, these mental representations are not themselves capable of transmission (in general) so that they must be further converted into some form which allows of exchange, like the symbols, oral or graphic, which make up language. These linguistic symbols are transmittable via aural and visual patterns. In a complementary situation, the receiver of these patterns is not interested in them intrinsically, but only because he is able to convert them into the matter that he uses for his mental representation of the world.

The fact that the universe plays no direct role in the exchange of information is the cause of all the relativity. The universe -- the part that is outside the language -- is wholly subjective, hence relative to the people involved. Since the nature of the universe is not transferable, there is no way to tell someone, child or adult, which one to use. Since an indefinite number of distinct universes is adequate to the processing of any given set of linguistic strings, there is no may to deduce which partitioning has been picked by anyone.

It has been my experience that many people find it disconcerting to be told that their mental representations are not part of the transfer of information, so I would like to repeat a few earlier remarks. Even if there would be some distinguished way of partitioning the world (say a Divine way, or something like that) nobody would be any the worse, in practical terms, for not using that particular set of equivalence classes.

Continues . . .

Previous page . . .