There is an old saying that “a problem well posed is half solved.” With this saying goes a correlate: “to set up a problem that does not grow out of an actual situation is to start on a course of dead work,” one that “has the semblance but not the substance of scientific activity,” in that the efforts are “merely excuses for seeming to do something intellectual”.
Quine opens his The Roots of Reference with the following remark: “Relatively little mystery enshrouds the ways in which we learn to utter observation sentences, and to assent to them or dissent from them when asked.” Who knew? Apparently for Quine, how an infant goes from hearing (what is to it) an undifferentiated stream of modulating sound in adult speech to parsing that speech stream into recognizable and repeatable phonemes holds no mystery. So the stage from babbling to canonical babbling is not even observed. Or how an infant goes from recognizing these phonemes to detecting their formation into differentiated words. Nope. Or how it goes from these differentiated words to simple declarations of intent or wish using those words? Nada. Or how it goes from these merely implicitly grammatical declarations to grammatical observation sentences negotiated in dialogue, with the shared intention of establishing common facts or features in a common environment—well, only here does the-mystery-that-is-no-mystery for Quine even begin; that’s where Quine starts, well past the use-by date. But prior to this stage of language use rests the entire problem of language acquisition. It is a shroud of mysteries that Quine not only fails to recognize. He obliterates it.
If infant researchers had Quine’s insouciance they would not know what is now known about the acquisition of speech by infants, much less its use once acquired. And worse than that: if infant researchers started with the preconceptions Quine starts with in The Roots of Reference, not only would they not know what is now known; they wouldn’t have even been able to pose the right questions to discover it. This makes those preconceptions worse than useless. Surely pre-conceptions are unavoidable in scientific inquiry, but equally unavoidable is starting with preconceptions specific to the problem set up to be resolved. What, specifically, are Quine’s?
First, what problem is Quine trying to address? Simple enough: how are the elements of a theoretical language derived “given only the evidence of our senses”? To simplify this issue, he boils it down to one underlying issue: “the referential aspect, the acquisition of the apparatus for speaking of objects”. In other words, the acquisition of language is the key. But there is a problem here. The above points already suggest that Quine, if he is really going to solve the problem he wants to solve, might be starting too high up the conceptual food chain. That is, he may in be starting with the use of language once it is relatively acquired, not its acquisition. But grant the starting point. Assume that the problem of reference—and therefore that the origin of theoretical language (and “entities”) from sense experience–is independent of the acquisition of language, which is instead a separable problem of its own. This is the only conceptual move that can save Quine’s account in The Roots of Reference because we already know his assumption of language acquisition doesn’t really address how language is acquired. What’s worse, the very issue gets mis-specified, so using his tools it can’t even be addressed.
But a problem—an insurmountable problem—arises. One can’t start where Quine needs to start to make his account workable at all because starting there presupposes exactly what it seeks to explain, reference. Observation sentences, to even be formed, require reference. To exchange observations in observation sentences about objects requires reference. In short, to have or recognize observation sentences at all requires an acquired language—and therefore reference. By implication then, Quine does need to start where he thinks the starting point is, namely the acquisition of language, not its use. But he starts at its earlier uses, not its acquisition.
Since Quine actually needs to start in a place conceptually and temporally prior to the one where he starts—at the acquisition of language prior to the ability to recognize or use observation sentences—no remarks are in order on whether or not the rest of his preconceptions are appropriate for trying to solve the problem he thinks he is solving—beyond pointing out again that they are not because they mis-specify the underlying problem. This mis-specification is all the clearer in his own account of how “behavioristic psychology” “blooms” in the “bright domain” of “infant research”. There Quine declares: “language is learned ostensively”. No, language is used ostensively; it can’t also be learned ostensively because that would presuppose having words and other elements of language prior to its acquisition. How infants acquire these words and other elements of language, and how their accomplishments “ratchet” on themselves as they develop it, was, when The Roots of Reference was written, a new but burgeoning field. It has grown tremendously since. The problem is: Quine started with an already, at that time, outmoded and discredited approach to language acquisition that was discredited precisely because it 1) describes virtually nothing observable about how infants actually learn a language, and 2) presupposes exactly what it is trying to explain, i.e. the acquired language. One can’t reinforce words to establish reference until words and basic forms of use are acquired, and no parent has ever listened to an infant babbling, selected out phonemes from it, re-formed those phonemes into recognizable words, then rewarded infants when they isolated and repeated those formed words in the midst of their babbling. Besides, the phonemes infants start with are non-canonical anyway, and they learn canonic ones by some process that involves, minimally, imitation—an anathema to reinforcement (learning through it was one of the swords that killed behaviorism as a catch-all explanation for learning). What we observe about infants learning language just doesn’t happen the way Quine needs it to happen for his account to make any sense. In short order, verbal behaviorism confused “use” with “acquisition” and fell as a theory, with props to Hume, dead born from the press. It has guided nothing about infant research since. The field that has given us what is now known about language acquisition has done so by respecting the mystery of it, not obliterating it.
Confusing “use” with “acquisition” is fatal to the entire execution of The Roots of Reference, and probably to most of Quine’s work that follows from this “problem” (the anus may take that up later). The problem Quine raises—how reference determines the derivation of theoretical entities from sense experience—may be the right one (the anus is pretty sure it’s not), but the dead work he proposes makes the problem, as posed, unsolvable.