An important question about TCL remains open from the last chapter: what sort of general type constructors do we need for investigating the meaning shifts evident in coercion, aspect selection, and copredication? The simply typed lambda calculus distinguishes between simple types and functional types, and TCL requires the existence of bounded quantified types. But we need other complex types to model aspect selection and copredications involving aspect selection, which are the subject of this chapter.

A theory that invokes complex types has to say how these types get used in the analysis of the phenomena. And to do that, we need to provide a way of exploiting complex types and deciding what information flows from them— i.e., what are their effects on logical form and truth conditions. Underlying this system of type manipulation must be a semantics or model of what such complex types are. Given the nature of the type system and the separation between logical forms and types, two questions need to be answered: (1) How does the semantic conception of such complex types interact with the type hierarchy? And (2) how does information about types interact with logical form? In this chapter I investigate answers to these questions with respect to the complex type •, which has received the most scrutiny in Asher and Pustejovsky (2006) and Cooper (2005).

A type constructor for dual aspect nouns

Do we need a separate type constructor to model the predication of properties to selected aspects, or to model the behavior of dual aspect nouns?