This is the final version, dated 8/98, of a paper to be published in R. Borsley (ed.) The Nature and Function of Syntactic Categories (Academic Press). It replaces:
It originated as a paper presented to the Bangor International Workshop on Syntactic Categories in 7/96.
The paper considers the notion `functional category' and concludes that, at least as far as overt words are concerned, the notion is ill-founded. First, none of the definitions that have been offered (in terms of function words, closed classes or non-thematicity) are satisfactory, because they either define a continuum when we need a sharp binary distinction, or they conflict with the standard examples. And second, the two most commonly quoted examples of word classes that are functional categories cannot even be justified as word classes. Complementizers have no distinctive and shared characteristic, and Determiners are all pronouns which are distinguished only by taking a common noun as complement - a distinction which is better handled in terms of lexical valency than in terms of a word class.