Model design in R: Testing complex semantic theories with simple statistical tools
Alexandre Cremers (ILLC – Universiteit van Amsterdam, Netherlands)
Developments in experimental semantics and pragmatics have led to experiments testing increasingly subtle theoretical predictions. Unfortunately, simple mappings from relevant semantic factors to experimental factors are not always possible, making interpretation of the experimental results often indirect. This issue is somewhat specific to semantics and pragmatics, and is rarely discussed in the methodological literature. In this course I will introduce a method to build statistical models in R that immediately encode semantic theories, and thereby allow direct interpretation of the results.
In the last session, we will discuss the possibilities offered by more powerful methods and how they can further close the gap between theory and experimental results. While a background in statistics and knowledge of R might be necessary to fully implement the ideas discussed in class, the course itself will not presuppose such knowledge. The discussion will be illustrated with concrete examples and data sets from real studies.