Title Final Paper PDF Quine Fodor IT.pdf ✅

Stephen Priest December 2023 1 Quine, Fodor, and the Indeterminacy of Translation Quine’s philosophy of language brilliantly works out the implications of three main viewpoints, taken together: physicalism, empiricism, and behaviorism. Two towering theses emerge from his considerations of language: the indeterminacy of translation (IT) and from it, ontological relativity (OR). In this essay, after detailing the relevant portions of Quine’s theories, I explore Jerry Fodor’s views about language, meaning, and translation in order to determine how they might restrict Quine’s conclusions, and I give a few reasons why his views might be better motivated than Quine’s. Fodor’s theories extend Chomskian claims about innateness of language to semantics, claiming that all humans have an innate, unlearned metalanguage that allows us to learn other natural languages. I argue that Fodor’s view successfully limits Quine’s conclusions to the rather uncontroversial underdetermination of translation thesis (UT) and that it makes better sense of language learning in general. 1. Quine’s Views and Arguments Quine’s main initial assumption is physicalism: nothing besides the physical exists, so, besides sets, there are no abstract entities, including mental ones. Another of Quine’s driving views is empiricism, although it is an empiricism stripped of certain assumptions Quine argues against, such as the analytic-synthetic distinction.1 As I explain later, physicalism and empiricism together support behaviorism about meaning and language, Quine’s final driving view in his philosophy of language. Behaviorism rejects the existence of meaning as anything other than information about how language is used. Quine, following Dewey, rejects mentalist semantics, saying that 1 See Quine, “Two Dogmas of Empiricism.”
Stephen Priest December 2023 2 “‘Meaning ... is not a psychic existence; it is primarily a property of behavior’ ... there cannot be, in any useful sense, a private language.” Meaning “is not determinate in [a man’s] mind beyond what might be implicit in his dispositions to overt behavior.”2 Behaviorism thus blends with pragmatism: knowing the meaning of a word is nothing more than knowing how to use it, or—to say even less—knowing how it is used. Quine denies the existence of any other kind of meaning, including that words and sentences can have a meaning at all, if “meaning” is meant to refer to some kind of private definition or to imply relations of synonymy with other meanings. Mentalist semantics, as opposed to behaviorism, assumes that at the very least, each person has some determinate meaning stored in their minds for terms and sentences, which meaning could stand in relations of synonymy to other terms and sentences. Importantly, however, Quine thinks that behaviorism applies even to the meaning each of us thinks our terms have in our own minds, not just to a person’s attempt to understand someone else’s meaning. It is worthwhile to spell out some of the reasoning behind linguistic behaviorism, since these points are directly relevant for understanding the IT thesis and arguments against it. Straightforward support comes from Quine’s physicalism and empiricism. Because there are no abstract minds or mental entities, and because, therefore, we can have no access to knowledge of the world other than by causal interactions between it and our sensory systems, learning anything is an empirical enterprise based on observation alone.3 Language can be no exception: learning language can come by no other process than seeing its use; thus, knowing the meaning of utterances is nothing more than being familiar with and able to reproduce others’ use of those 2 Quine, “Ontological Relativity,” 185-6 3 If one objects that some things can be discovered by a priori reasoning, Quine responds in “Two Dogmas of Empiricism” that this is simply based on the false distinction between the analytic and synthetic: things can only be thought to be known analytically by reason alone given background assumptions about the meanings of terms. Those meanings can only be learned empirically, and are thus always open to revision—there is no true analyticity.
Stephen Priest December 2023 3 utterances. How else would we learn language than by being “a student of [one’s] neighbor’s behavior”? “The learner has no data to work with but the overt behavior of other speakers.”4 Linguistic behaviorism leads Quine to his underdetermination and indeterminacy theses about translation, which are best introduced in the same way Quine himself does: the context of “radical translation,” wherein a linguist makes contact with a people group that speaks a completely foreign, unknown language. The linguist must learn the foreigners’ language through observation of how they use it, the goal being to construct a translation manual from the foreign language into the linguist’s own language (say, English). Doing so is essentially building a theory of how best to translate the foreign language; theory-building progresses by the formation, testing, and revision of hypotheses. The holistic view of theories that Quine developed in his “Two Dogmas of Empiricism” means that any theory built this way will be underdetermined by the available evidence.5 One of the main examples given concerns instances when the foreigner seems to use a word (e.g., “gavagai”) in reference to rabbits. But how do we determine whether to translate “gavagai” as “rabbit” or as “undetached rabbit part”? Whenever one of these is present, so is the other; you cannot point to one without also accidentally pointing to the other. This may, at first, seem trivial: we might devise a way to test the difference by asking the foreigner something like, “‘Is this gavagai the same as that one?’ while making the appropriate ostentations.”6 However, this strategy would only work if we really knew that a phrase in the foreign language means “is the same as”—something that we could only know by the same indeterminate processes that we are using to translate gavagai. For all we know, we could have 4 Quine, “Ontological Relativity,” 186-7. 5 Specifically: the meaning of any given statement in a theory relies on all the other definitions and assumptions that are held constant in the theory. Thus, whenever new data is encountered that seems to contradict the theory one has built, there are any number of ways to account for this new data: the background assumptions can be altered, including dropping certain beliefs or changing certain definitions, all to make the new evidence fit. 6 This quote, and the line of explanation in this example, follows Quine, “Ontological Relativity,” 189-90.
Stephen Priest December 2023 4 actually asked, “‘Does this gavagai belong with that?’” The same uncertainty would apply to all terms in the foreign language. The end result is the UT thesis: it is possible for multiple translation manuals to each be compatible with the observed (behavioral) evidence of the foreign language but diverge from each other in the translations they give into one’s own language.7 Lycan gives a more transparent statement of Quine’s theses, and notices two versions of UT—a strong and weak, both of which Quine believes. The weak version claims underdetermination given all evidence a field linguist actually has and uses; the strong version claims underdetermination even given “all possible data.”8 Quine holds both and takes UT to imply the further IT thesis, which Lycan characterizes as stating that “no translation manual is ever true or correct or right, simpliciter. A single translation is "correct" only relative to some antecedently constructed and chosen translation manual.”9 It is not merely that we do not know which translation is correct (UT), but that there is no fact of the matter to be known (IT). If all the different contending translation manuals equally fit the observable facts about behavior (and UT says there can be many such manuals), then none is more right than the others. Lycan divides Quine’s arguments for IT into two classes: the “from below” and the “from above.”10 The former involve all the examples Quine gives to motivate IT, such as the indeterminacy between rabbit, undetached rabbit part discussed above. However, Lycan and Quine both admit that these cases do not so much argue for IT as demonstrate its plausibility.11 7 Quine, “Word and Object,” 27: “manuals for translating one language into another can be set up in divergent ways, all compatible with the totality of speech dispositions, yet incompatible with one another.” When explaining UT, Lycan simply quotes this passage directly. 8 Lycan, “Logical Form in Natural Language,” 223. 9 Ibid. The part about, “only relative to some antecedently... chosen translation manual” is exemplified by the gavagai example: you may be able to nail down that gavagai means “rabbit” and not “undetached rabbit part” if you assume some other phrase means “is the same as” in the foreign language. But this assumption is a choice. 10 Ibid., 224-6. 11 Ibid., 231 and 319.