The structural uniqueness of languages and the value of comparison for language description

This paper shows why it is not a contradiction to say that each language is structurally unique and must be described with its own categories, but language description profits enormously from typological knowledge. It has sometimes been suggested that the Boasian imperative ("each language should be described in its own terms") leads to uninsightful analyses, and that language description should instead be "typologically informed". But the Boasian imperative is not at all incompatible with an intimate connection between description and comparison: Comparative (or typological) knowledge is highly valuable both for making our descriptions transparent and comprehensible, and for helping describers to ask a wide range of questions that would not have occurred to them otherwise. Since we do not know whether any of the building blocks of languages are innate and universal for this reason, we cannot rely on general frameworks (of the generative type) for our descriptions, but we can use typological questionnaires and other kinds of comparative information as a scaffold. Such scaffolds are not theoretical components of the description, but are important methodological tools.