Frequently it is claimed that functional categories are different from lexical categories in having an abstract meaning. This holds both for functional categories in the way I have been discussing them in the previous chapters and for a special class of elements often analysed as functional, namely discourse markers. First, I will try to make the claim about the special semantic status of functional categories more precise and see whether it can be made testable, in the light of recent models of the syntax/semantics interface and interpretability. In the second part of the chapter, I will turn to discourse markers. If these should be treated as functional elements at all, they have a status and behaviour rather different from other categories. The discussion in the first section owes much to Cann (2000), the most lucid exposition of the semantic issues surrounding functional categories that I have encountered, even if I do not follow his distinction here between I- and E-language categories (cf. Chomsky 1986a).

Elements can be interpreted in various ways:

• through reference to a notion in the cognitive system

• through deixis

• through knowledge of the constructions they are part of

• through paradigms

• through discourse patterns.

In this chapter, I will deal with the last four possibilities for interpreting an element.