The Syntax/Semantics Interface: Compositionality Issues
Lecturer: Pauline Jacobson
The broad objective of this course is to focus on some theories of how the syntax and semantics of natural language interact and the consequences of these for the notion that the semantics of natural language is 'compositional'. Particular emphasis will be given to the view that the syntax is seen as a system proving expressions well-formed while the semantics works 'in tandem' with this to provide an interpretation for each syntactic expression as it is 'built' (proven well-formed) in the syntax. (This view was well-known especially in work in so-called 'classical Montague Grammar'.) We will, however, be also looking at some alternative views of the syntax/semantics interface which maintain that the semantics is compositional, but works from a level of representation (Logical Form) which is more abstract than the pronounced version of an expression. The data will be drawn entirely from English, but the major goal is to introduce the issues and the tools used to model the relevant data.
The different theories - and what is at stake - will be elucidated by looking at a number of phenomena which at times have been thought to present challenges to a strong view of compositionality. I will begin with two very classic phenomena - the analysis of quantified NPs (or 'DPs') and the analysis of coordinated expressions (and the interaction of these two). After looking at some classic results on these - and on how these results bear on the issue of the syntax/semantics interface and compositionality - we will turn to some questions raised by the binding and distribution of pronouns, by the interpretation of 'missing' material in ellipsis constructions, and on the interactions of all of these. This is a huge and open-ended domain; so the goal will be to pick a few interesting cases to show what is at stake in analyzing them, and what kinds of tools are motivated by these constructions.
Download: course handout
