Revising a system of beliefs › Why a dictionary cannot finish the argument

The natural answer is that a dictionary defines the relevant word. In §2, Quine asks what such a definition records. A lexicographer normally reports established usage. If a dictionary correctly explains a word by another expression, the explanation is supposed to fit a linguistic practice that already exists. The dictionary can be excellent evidence about that practice without creating its underlying relation of synonymy.

This is a question of explanatory direction. Imagine checking whether two measuring instruments agree by reading a report that says they agree. The report may be trustworthy, but it cannot also explain what agreement between the instruments consists in. Similarly, citing a dictionary entry does not by itself analyze the relation that makes the entry correct.

Quine recognizes an important exception: introducing a new notation expressly as an abbreviation. Suppose a policy document introduces "eligible visitor" to abbreviate "registered visitor aged eighteen or older." Within that document, the stipulated equivalence is transparent. Nothing about the argument requires us to stop making such definitions. Quine, §2.

The exception limits an overbroad reading of his criticism. His objection is not that definitions never establish synonymy. It is that this special case does not explain the whole range of relations that ordinary definitions presuppose. A successful local stipulation cannot automatically provide an account of meaning throughout a natural language.

The policy example also reveals two questions that can come apart. We can ask whether a visitor satisfies the stated condition, and we can ask whether the condition is a sensible one for the policy's purpose. The definition answers the first kind of question only after relevant facts are supplied. It does not make the policy immune to criticism.