Quine on Naturalized Semantics and Ontological Relativity

My thinking in philosophy has for a fairly long time now been animated by thinking about WVO Quine’s writings.  His views have, of course, been tremendously influential over the past 50+ years within philosophical circles.  My readers with more background in mathematics and less in philosophy and philosophical logic, however, may not be familiar with Quine’s central theses and arguments.  Since I’ll be working through responses to my comps questions in future posts, and will sometimes need to mention Quinean ideas without fully explicating them, it will be useful to have up some basic exegetical posts.  This is one of them, a response to a final exam question in Greg Landini’s Spring 08 Quine course.  The intro is a bit cheezy, but it gets the point across how Quine’s challenge to the “myth of the museum” attacks presuppositions of both platonism and the mentalism shared alike by enlightenment rationalists and empiricists.  In place of the myth Quine proposes a naturalized semantics motivated by Dewey, Wittgenstein, and behaviorist psychology.  Ok, here we go.

The location of the museum had been in dispute for centuries. Some held that meanings inhabited a mysterious third realm. Some thought it was in our minds. Some thought that it was in the world we inhabit. There was little doubt that there was a museum, however, wherein meanings were kept for consultation on the occasion of any philosophical dispute. So goes the myth Quine says was busted by Dewey and Wittgenstein. In place of mythology Quine promises to develop the ideas of Dewey and Wittgenstein regarding alternatives to the myth of the museum along thoroughly modern and scientific lines. To understand a word will not be to be in some relation with its meaning, whatever that may be, but rather to have a kind of know-how.

Quine indicates two aspects to knowing a word: (a) the phonetic part: knowing its sound, and (b) the semantic part: knowing how to use it. For a simple example, one may know how to use the word to refer to an object in one’s environment. To refer is to draw another’s attention to the object, so learning how to do this includes learning how to bring about desired behavior in one’s fellows. What one does, in general, in learning how to use language is learn about the behavior of one’s fellows. Quine demands that in naturalized semantics one must recognize that “even in the complex and obscure parts of language learning, the learner has no data to work with but the overt behavior of other speakers.” Quine argues that one consequence of this view of meaning is ontological relativity: the doctrine that there is no language transcendent fact of the matter regarding “what there is”.

Relativism, or opposition to it, was one of the reasons for the introduction of the myth in the first place. Meno presented Socrates with the following dilemma: If we know the meaning of “virtue” then we don’t need to search for a definition, but if we don’t know the meaning then we won’t recognize a good definition if we find it. Either way, it’s pointless to go looking for a definition. The puzzle presented to Socrates is now known as the “paradox of analysis”, which states that philosophical analysis is pointless because it is either trivial or false. Plato presents Socrates as solving the problem through the theory of the forms and anamnesis (i.e., recollection). We recognize an analysis of a term such as “virtue” as adequate by recalling our former acquaintance with a certain museum item: viz., the form of virtue. The analysis is non-trivial in light of the difficulty of recollection. The account has left many a bit uneasy, including Socrates himself. After giving his answer to the dilemma, Socrates confesses to Meno that the idea of a third realm of forms that we inhabited in a past life seems a bit dubious. The alternative, Socrates fears, is that talk of virtue, justice, etc. is just a bunch of squawk. The pernicious Callicles and Thrasymachus would jump on such a result.

Quine is less concerned about pernicious relativism and more concerned about giving a scientifically plausible and naturalistic semantics. Platonism is out. Mentalism is also out. The mentalist places the museum in our minds rather than a mysterious third realm. She gains some ground on Socrates epistemologically, it would seem. Some have learned from Descartes to insist that nothing could be plainer than one’s own mind. The terminology has varied regarding the exhibits in the mental museum, but Hume’s “ideas” and “impressions” are typical. Impressions are sense experiences. An idea is thought of as a “dim copy” of an impression. An expression like “blue” is understood by reference to an exemplar idea bearing the label “blue” in one’s mental museum. The problems with mentalism are numerous. Wittgenstein’s Philosophical Investigations offers trenchant criticisms of the view that concept words refer to private mental exemplars, and there are well-known metaphysical difficulties with mind-body dualism (though it is perhaps debatable that the mentalist is committed to dualism). Quine has been criticized for dogmatism in his embrace of naturalistic alternatives and for dismissing mentalism too quickly without much argument. I think that a fairer assessment of Quine’s work is to view him as more interested in the positive project of developing the implications of alternatives to mentalism, such as behaviorism, than in beating what he plausibly took to be a dead horse.

Quine’s case for ontological relativity builds from his theses of the inscrutability of reference and the indeterminacy of translation. Indeed, the inscrutability of reference would show that the physical world, no more than the platonic or mental realms, cannot serve as a museum of fixed meanings. One way to present the case is consider the maneuvers available to an ethno-linguist clinging to a seemingly strange hypothesis: that the native’s ‘gavagai’ uttered regularly in the presence of rabbits actually means ‘undetached rabbit parts’. As Quine notes, it’s a rather unnatural hypothesis and one that actual practicing ethno-linguists would never and should never make; that is beside the point though. How could we ever disprove the hypothesis? Quine insists that based on the behavior of the natives alone, we could not. We might be able to show that the hypothesis is inconsistent with an established translation manual specifying, for instance, the individuative apparatus of the language, but our ethno-linguist may simply cling to his hypothesis by abandoning the translation manual, which functions evidentially as an auxiliary hypothesis, and retranslating the relevant native utterances in a way that is consistent with the clung-to hypothesis. The opportunities for revision of auxiliary hypotheses seem endless and the suggestion that behavioral data alone can uniquely determine one correct translation seems hopeless.

If one wished, one could deal with the stubborn ethno-linguist as follows: simply come up with a translation manual for him. One could translate his ‘undetached rabbit parts’ as ‘rabbit’, supposing that he speaks a bizarre dialect of English.  One will have to undertake some considerable gerrymandering the terms of his apparatus of individuation for the hypothesized dialect: viz., the use of pronouns, plurals, identity, etc. It seems, however theoretically, possible to translate our ethno-linguist so that the composition of his translation of the natives (call this t1) with our translation of him (call this t2) yields the more natural translation of the natives call this (t3): i.e., t2(t1(‘gavagai’)) = t3(‘gavagai’) = ‘rabbit’. He no longer has a bizarre hypothesis, just bizarre linguistic dispositions. This reflection brings radical translation home. We will have heterophonically translated our neighbor. As Quine notes, our ordinary practice is to homophonically translate one another, but there are exceptions. Frequently, for instance, when I hear someone say that they are “anxious” about something they seem to anticipate without the least bit of anxiety, I translate them as meaning they are “eager”.

Nevertheless, heterophonic translation is the exception and homophonic translation is obviously a rule that we cannot get by without. Suppose we want to test t2, our proposed translation manual of the ethno-linguist. How will we test it? We can ask the ethno-linguist whether by ‘undettached rabbit parts’ he means rabbit or undetached rabbit parts. Of course, he’ll answer “I mean undettached rabbit parts, obviously”. This neither confirms nor disconfirms t2. In fact, an answer to the question whether the ethno-linguist has a bizarre hypothesis or speaks a bizarre language cannot be, Quine argues, empirically determined at all. So, if we are verificationists, we should reject the question as senseless. The point is not restricted to the term ‘rabbit’. Which of the numerous series satisfying the Peano axioms are the numbers? We could always interpret some person’s specification of which series they are talking about by heterophonically translating their utterances to have them come out referring to our preferred series; again, the question whether they have a different view in philosophy of mathematics or a different language comes out senseless if we have the strict verificationist/empiricist criteria of meaning.

We have considered the ethno-linguists’ radical translation of the native and our own radical translation of the ethno-linguist. This all must stop somewhere. Could one radically translate oneself? Does the question we have not been able to answer empirically, the question whether the hypothesis or the language is bizarre, have an answer for us “from the first-person”? Ask yourself: “Self, by ‘rabbit’ do you mean rabbit or undetached rabbit parts?” Answer yourself: “I mean rabbit, obviously.” The answer is no more informative when you ask yourself than it was when you asked the ethno-linguist. If one could put the question thusly, the matter might be resolved: “Self, by ‘rabbit’ do you mean this (mentally pointing to an idea) or that (mentally pointing to another idea)?” You could then confidently hold the former idea before your mind as what you mean. It would be the same if one could consult forms. However, if we are to reject platonism and mentalism for naturalism we do not have such recourse, and I am with Quine in finding neither the platonist nor the mentalist picture to be particularly accurate, illuminating, or useful.

As long as one is taking seriously the possibility of heterophonically translating oneself, it won’t help to make reference to the apparatus of individuation to convince oneself that one means rabbit by ‘rabbit’. Just as with the natives and the ethno-linguist, that will be just as up for grabs. These are just as immanent to our natural language as all the rest. The tempting conclusion is that one cannot tell even of oneself whether one is speaking of rabbits or undetached rabbit parts. Quine’s way of resisting this temptation is to point out that the whole business of translation presumes a language to be translated into. To pose the question—“Self, by ‘rabbit’ do you mean rabbit or undetached rabbit parts?”—one has to use, not merely mention, the term ‘rabbit’. In even posing this question to oneself, one acquiesces in the language one was brought up in, and that is where radical translation stops. In particular, recall that it was crucial to the argument for indeterminacy of translation that the individuative apparatus of the language be subject to translation. Acquiescence in this apparatus, which in its regimented is just first-order logic, provides a fixed frame of reference in which to draw distinctions.

So, Quine maintains that just as it is only relative to an inertial frame that a thing can be said to be in motion, it is only relative to a linguistic frame that we can say what a term refers to. Quine extends the analogy further, claiming that just as according to the relational theory of space it makes sense only to speak of relations between inertial frames and relations between things and not of absolute position or absolute motion, it also only makes sense to speak of relations of interpretation and reinterpretation between languages and not of absolute interpretation. When “interpretation” above is taken, as Quine means it, in the sense of Tarski the wide implications of Quine’s arguments for the theory of truth are laid bare. The correspondence theory can be understood as a theory of absolute interpretation. Truth in the sense of correspondence is truth in the sense of Tarski plus the specification of an intended interpretation. The facts correspond to the sentence when and only when the sentence is Tarski-true on the intended model. The upshot of Quine’s reflections on ontological relativity for the correspondence theory of truth is that there may be no way of specifying an intended interpretation for the language in which one finally acquiesces.