Laura Korte
In this talk I hope to emphasize the importance of interdisciplinary research by introducing a deep type system for categorial grammar. Deep types were originally designed for the impure functional programming language Standard ML, to narrow the type information gap that arises when imperative features (with side effects) are added to a pure functional language. This narrowing is accomplished by defining a type system which does not only specify a program's functional (or shallow) type, but its side effect behaviour (or deep type) as well.
In categorial grammar, we are confronted with a similar situation: in trying to design a type system for natural language, shallow types quite often prove to be insufficient for a parser to exhibit a desired behaviour. Especially in cases which involve linguistic counterparts of side effects like intensionality, variable binding, quantification, interrogatives, focus or presupposition as described by Shan in his paper on Linguistic Side Effects. In this talk, I propose to extend the shallow CG type system with deep types, and will demonstrate that the resulting type system is able deal with examples of linguistic side effects in a surprisingly intuitive way.