Page 140 - Encyclopedia of Philosophy of Language

Truth and Meaning
science known as the Quine/Duhem thesis. The nati- vist account of learning one's first language sees such acquisition as a process of unconscious scientific theor- izing about the interpretation of the public language of one's adult teachers, the theorizing carried out in an innate language of thought, a 'machine code' in the brain. The Quine/Duhem thesis holds that only fairly comprehensive bodies of theoretical beliefs— hypotheses plus auxiliary hypotheses, plus data about boundary conditions etc.—entail empirical conse- quences. Applying such holism to linguistics, one might conclude that only interpretations of large frag- ments of a person's language have testable conse- quences. Hence only such comprehensive theories can be justified; only such theories could be known to be true. The nativist then concludes that only when the language learner has unconsciously formed an interpretative theory of most or all of the public language, does he or she have knowledge of that language.
A second, rather negative argument, is the anti- naturalist one: intentional, outward-directed psycho- logical states are fundamentally different from natural phenomena. In particular, humans, unlike stars or atomic particles, are rational agents, and this means explanation in the human sciences must be of a wholly different character from that in the natural: to do with empathetic understanding and interpretation rather than with capturing the phenomena in mathematical equations. According to Donald Davidson, the role of rationality in understanding agents necessitates an holistic account of mind and language (see Davidson, D.\
A third argument in favor of holism is also a nega- tive one: the alternatives to holism are unacceptable. Often the only alternative is thought to be an atomistic approach to mind and language, as found in behavior- ists such as Skinner. Holists have convincingly argued that cognitive states issue forth in behavior only as mediated by further states and this point demolishes any attempt to identify pieces of linguistic knowledge one-by-one with disconnected behavioral dispositions and, more generally, any view on language which does not take account of the interconnectedness of language understanding.
2. Counterarguments against Holism
The first argument for holism will carry little weight with those who find implausible the idea of an innate language of thought, or of young children engaging, even 'tacitly,' in complex linguistic theorizing. More generally, it will be dismissed by those who deny that knowledge of meaning is an example of knowledge that some proposition is correct. Similarly the second argument will be unpersuasive to anyone of a natu- ralistic bent unless it can be shown that a natural scientific explanation of agents is blocked by their purported rationality. The third argument has been
interpreted as providing an impossibility proof of this type. However, antinaturalist holism is not the only alternative to atomism. Many views on language and mind can be described as 'molecularistic.' On this approach, the mind is an hierarchically organized sys- tem of behavioral dispositions, with an extremely rich structure—perhaps too rich to be fully comprehended by the mind itself. The more complex elements of the hierarchy are dependent on some or all of the simpler, down to a bedrock of minimally complex proto-cog- nitive states whose nature is much as the behaviorist supposed all mental states were.
One example of a molecularist approach in the phil- osophy of language is Michael Dummett's according to which sentences stand in a hierarchy of complexity with grasp of all but the simplest sentences requiring grasp of others lower down in the hierarchy. For the holist, though, molecularism does not do sufficient
justice to the interconnectedness of the items of our linguistic knowledge. Consider semantic fields, such as the system of color concepts. It looks as if grasp of one requires grasp of all the others, so one gets a circle of interdependences with no winding down to a common primitive base.
3. Conclusion
Success in accounting for semantic fields, then, is one crucial test area where an adjudication between the molecular and holist approach to language may be found. Another might lie in compatibility with gen- erally accepted scientific approaches to mind. Molec- ularism, for instance, fits well with an information- processing approach to mind, whereas holism does not: e.g., if there is a circle of interdependencies in semantic fields then it looks as if the program which models our understanding of the elements of the field will loop, and so never issue in output to other parts of the system.
Failure to connect with a natural science approach is not, of course, a vice for militant antinaturalists. Even Davidson, however, feels the need to introduce rigorous theory into the interpretation of language via Tarski's method for defining truth for certain formal languages, a definition which incorporates an expla- nation of the truth conditions of complex sentences in terms of simpler. The Davidsonians seek to generalize this method to natural language and use it to generate, for each sentence of the language under study, a speci- fication of a state of affairs which we then interpret as the content of utterances of the sentence by the language users. This forms part of an overall attempt to make rational sense of their behavior and the linguistic interpretation is successful insofar as the overall attempt is (if the latter is not, one has, in accordance with the Quine/Duhem thesis, a choice of where to pin the blame). The suspicion arises, however, that this rigorous Tarskian theory is an idle cog in the holist interpretive exercise. Any old method
118