Data:Reisinger et al. (2015) and White et al. (2020) on collecting corpus annotations of the proto-role properties proposed by Dowty (1991). We will use the Universal Decompositional Semantics (UDS) dataset (v2.0 Gantt, Glass, and White 2022), which is packaged with the decomp toolkit, available here.
Theory:Levin and Rappaport Hovav (2005, Ch. 2) on the explanatory role of generalized thematic roles.
In this fourth and final module of the course, we are going to focus on introducing nontrivial structure into the representations we learn from annotated corpus data while retaining the benefits of mixed effects models. As a case study, we’re going to be interested in the question–discussed by Levin and Rappaport Hovav (2005, Ch. 2)–of what sorts of representations determine the relationship between individual thematic roles and syntactic positions.
Gantt, William, Lelia Glass, and Aaron Steven White. 2022. “Decomposing and RecomposingEventStructure.”Transactions of the Association for Computational Linguistics 10 (January): 17–34. https://doi.org/10.1162/tacl_a_00445.
Levin, Beth, and Malka Rappaport Hovav. 2005. Argument Realization. Cambridge: Cambridge University Press.
Reisinger, Dee Ann, Rachel Rudinger, Francis Ferraro, Craig Harman, Kyle Rawlins, and Benjamin Van Durme. 2015. “Semantic Proto-Roles.”Transactions of the Association for Computational Linguistics 3: 475–88. https://doi.org/10.1162/tacl_a_00152.
White, Aaron Steven, Elias Stengel-Eskin, Siddharth Vashishtha, Venkata Subrahmanyan Govindarajan, Dee Ann Reisinger, Tim Vieira, Keisuke Sakaguchi, et al. 2020. “The Universal Decompositional Semantics Dataset and Decomp Toolkit.” In Proceedings of the Twelfth Language Resources and Evaluation Conference, 5698–5707. Marseille, France: European Language Resources Association. https://aclanthology.org/2020.lrec-1.699.