Page 263 - Encyclopedia of Philosophy of Language

5. Naturalizing Reference
The ambitious program is in trouble if reference turns out to be inexplicable. Quine, as noted earlier, has long been sceptical that any account of meaning can be had. In arguing against the analytic/synthetic dis- tinction he claimed that sameness of meaning, hence meaning, can be explained only by other semantic notions. There is no escape from the circle of semantic concepts. So these notions have no place in real science, however attractive they may be. Quine is right in demanding that semantic properties be explained or abandoned; defenders of causal theories hope how- ever to show he is too pessimistic about the prospects of explaining meaning.
The reductive task is formidable. Late-twentieth- century attempts rely on the notion that rep- resentation is a causal relation. As we have seen, the root idea is that the 'tiger'-tiger connection depends in some way on the causal connections between 'tigers' and tigers. But the causal relations in question are crucially ambiguous. Consider the relations between Max's uses of 'tiger' and that set of tigers Max has encountered. His 'tiger's, surely, apply to these tigers and to all the others. But how could that be so? He has no causal connection with the rest. Somehow the term's range of application must be generalized beyond the grounding set, and by just the right amount. It must apply to all tigers, but not more: not to all large carnivores, all felines, all tigerish looking things, or to everything that is either a tiger or will be born next year. It must include all tigers; not all except those born in the New York Zoo. Even the notion of the grounding set itself is problematic. For consider the causal chain. It goes via the sensory impressions of tigers, tiger surfaces, tigers, the causal history of there being tigers near Max. So why aren't Max's uses of 'tiger' about tigerish retinal projections or tiger surfaces rather than tigers? Ambiguity problems seem to arise in one way or another for all attempts to explain representation.
5.7 The Appeal to Psychology
The natural way to resolve these indeterminacies is to appeal to the psychology of the language speaker, to what the speaker intended or believed. Max refers to tigers, not their characteristic retinal projections, because he intends to speak of a kind of animal. Responses along these lines threaten naturalistic sem- antics with circularity. For the response seems to be presupposing just the same capacities in thought that are to be explained in talk; causal theories of reference totter on the edge of being buck-passing theories after all. Psychological processes can appropriately be appealed to in giving a theory of meaning for natural language. But it is not much of an advance to pre- suppose the very properties of mental representation that are to be explained in linguistic representation. Moreover, in appealing to a speaker's knowledge, a
more cognitive theory threatens to return causal theories to the descriptive theories from which they were in flight.
Names pose the same problem. On the Kripkean view, names are introduced into the language in for- mal or informal naming ceremonies. Max acquires a
jet black kitten with fierce yellow eyes and proposes to his friends that they call her 'Satan.' Thus Satan is named, a name for the cat, not a cat surface or a temporal slice of a cat. Those present, in virtue of their interaction with the cat and the name, acquire the semantic ability to designate that cat by that name. This story is plausible, but it clearly presupposes that sophisticated perceptual, intentional, and linguistic capacities are already in place.
Naturalist semantics thus faces a dilemma. Rep- resentation seems to be a species of causal relation. Yet causal relations do not seem sufficiently deter- minate. They leave too many candidates as the content of any given term. Winnowing these candidates requires one to move back inside the mind. In turn, this threatens the project withcircularity.
5.2 'Semantic Bootstrapping'
There is a natural strategy for escaping this dilemma, which might be called 'semantic bootstrapping.' Per- haps there is a hierarchy of terms: from pure causal terms through descriptive-causal terms to descriptive terms. A pure causal term is one for which no descrip- tive knowledge of the referent is required. It can be acquired atomically, hence no threat of circularity arises. This base could play a role in the explanation of the reference of some nonbasic ones, names, perhaps, or natural kind terms. The descriptive knowl- edge required for their acquisition might require only base level concepts. This larger class could then play a role in our story for others, and so on. Successful bootstrapping would simultaneouslyexplain both the nature of the referential relation, and explain how terms with those referential properties could be acquired.
Tough problems confront this proposal. Can the less basic be explained by the more basic in some way that does not presuppose that nonbasic terms are defined? There is very good reason to deny that many concepts are definable. So descriptive elements must be built into a causal theory of reference in ways compatible with its original demonstration that our capacity to refer to the world's furniture does not in general depend on identifying knowledge of that to which we refer.
Still more serious is the problem of actually finding basic terms. Sensory terms at first sight seem candi- dates, but even for them the ambiguity problem arises. Does 'red' (when first acquired by some child) name a color or a shade of that color, or even an intensity level of light? Concepts for which no problem of ambi- guity arise look decidedly thin on the ground. So our
Reference: Philosophical Issues
241