Current projects

Usage-based approaches to the acquisition of argument structure: artificial language learning studies (with Adele Goldberg)

Usage-based approaches to language development have long emphasized the important role played by statistical information in shaping the linguistic generalizations posited by learners. However, there is still much debate as to which aspects of the input matter. In this project, we notably consider the role played by alternations, i.e., whether and to what extent lexical items can be used in several different structures, in the acquisition of verb argument structure. We use artificial language experiments to investigate how usage impacts the acquisition of constructions (especially insofar as syntactic productivity is concerned) and how this interacts with other aspects of the language, such as the function of constructions.

Change in syntactic productivity in Late Modern English constructions

Syntactic productivity refers to the property of the slots of a syntactic construction to attract new lexical fillers. In diachrony, a construction comes to be used with a wider range of different items as its productivity increases over time. This project is concerned with the factors that influence the productivity of a construction, especially with respect to semantics. I use distributional vector-space semantic models to build semantic representations of the distribution of constructions, and I use this information to track their semantic development over time, drawing on data from diachronic corpora. This method allows to test hypotheses about the relation of semantics and usage to productivity in a fully data-driven way.