In this paper, we would like to begin exploring what an integration
between Generative Lexicon and SDRT
look like, what problems it would face and what advantages it
might offer.  We willl concentrate on developing the composition logic
and the account of complex types, in which we'll make use of various
SDRT principles.  As we are interested in composition of information
to build logical forms, we will build on the standard way of getting
logical forms---the lambda calculus in which (functional) types are
exploited.  By relating types in the lexicon we can give partial,
implicit definitions, which will help together with how the items
compose together determine inferences based on truth conditional
contents. Secondly, by developing a strongly typed theory of lexical
items and a theory of how such lexical items combine and interact in
the process of semantic composition and of discourse interpretation,
we can constrain the lexical semantics with predictions of
semantically well-formed or ill-formed predications and word
combinations.  We'll produce a new type calculus that captures one 
of the general ideas of the generative lexicon: providing a set of
techniques governing type shifting possibilities for various lexical
items so as to allow for the combination of lexical items in cases
where there is an apparent type mismatch.  These techniques themselves
should follow from the way the lexicon is organized and its underlying
logic. We provide a type logic for interpreting dot objects in
coercive contexts, as well as a strategy for dot-introduction during
composition.