Over the last two chapters, I have been fleshing out the schema for dependent case assignment in (1) and exploring its parametric possibilities for accounting for crosslinguistic variation.

(1) If a category XP bears c-command relationship R to another category ZP in domain W, then assign Case C to XP.

So far we have investigated the range of relevant c-command relationships R, and the types of domains W, the latter being essentially the spell out domains implied by the theory of phases. The third major dimension to consider is exactly what XP (the case receiver) and ZP (the case competitor) must be in order for them to interact case-theoretically. The core notion that I have assumed throughout is that XP and ZP are overt nominals – NPs and DPs. These are the quintessential argumental categories in natural languages. Moreover, thinking in terms of Comrie's (1978: 181, 1981a) notion of the discriminating role of structural case, they are the main things that a clause often has more than one of, such that it is useful to have morphological marks that distinguish them. It therefore stands to reason that dependent case marking applies to DPs and NPs in a language, if it applies to anything at all. But what exactly is a “nominal,” and is it the same thing in all languages? There is some vagueness here, and with it some room for crosslinguistic variation.