week of january 28

On Monday, I introduced the main features of the course and distributed a questionnaire (PDF) for you to fill out and return to me. On Wednesday, we talked about compositionality as a design feature of human languages. Here are the slides (PDF) that formed the background for the lecture. You can access all materials with your UMail user name and password. We discussed two rather different methodological 'styles' in approaching the topic of compositionality: Krifka's encyclopedia article starts out with a common characterization of what the human ability to compose the meanings of complex expressions from the meanings of their parts may consist in. This ability might correspond to a function (literally!) that computes a unique denotation for any node in a phrase structure tree if given the denotations of its daughter nodes. It already emerges from Krifka's short discussion that this way of characterizing compositionality runs the danger of becoming trivial unless we impose further constraints on the properties of composition functions. Jackendoff's chapter approaches the topic of compositionality in a very different way. He identifies certain constructions in human languages that show no compositionality at all (expressives like ouch or oops), or only limited compositionality, like N-N compounds. Looking at the kind of compositionality we find in N-N compounds may not only support certain speculations about the evolution of the language faculty that Jackendoff is interested in, but may also help us develop an understanding of the much more elaborate type of compositionality that comes with phrasal syntax. For that reason, most of Friday's lecture was dedicated to N-N compounds. The slides for Friday's lecture are here.

The two different ways of approaching the topic of compositionality (trying to understand it as a formal property of functions versus comparing different instantiations of it in human languages) illustrate nicely the different roots and current superegos in linguistic semantics. When we think about formal properties of semantic composition functions, semantics looks a little like a branch of applied mathematics. When we are comparing various instantiations of compositionality in natural languages, semantics looks more like a branch of theoretical psychology. We thus arrive at the perspective of Partee at the beginning of the chapter you read: "The perspective of the author of this chapter is that of "formal semantics" with roots in logic, philosophy of language, and linguistics, developed in an environment in which linguistic and cognitive science questions have been at the forefront, with logic and the philosophy of language providing important tools and foundations." (Partee, op. cit., p. 112).

Food for a technical project if you have a background in math, logic or computer science and want to think deeply and technically about what it might take to save the Principle of Compositionality from triviality: Zoltán Gendler Szabó's article "Compositionality" in the Stanford Encyclopedia of Philosophy. For clarification of some of the issues discussed in Szabó's overview article, read Westerståhl (1998) or Kazmi and Pelletier (1998), both from Linguistics and Philosophy.