In this article I will discuss an issue which must be raised in connection with the research program underlying much interesting recent work by Akmajian, Steele and Wasow (cf. Akmajian, Steele and Wasow (1979), henceforth ASW, and Steele et al. (1981), henceforth SEA).1 The goal of the program is to provide a framework in which the question of the similarity of categories across the grammars of particular languages can be provided with a ‘substantive empirical basis’: the framework uses cross-linguistic comparison to identify as equivalent, categories which are motivated in the first instance solely by language internal analyses. The treatment of the category AUX is intended to exemplify the general approach.2