ComputationalDiscourseModeling

3 Credits · Level: Graduate

Recent years have seen an explosion of computational work on higher level discourse representations, such as entity recognition, mention and coreference resolution and (shallow) discourse parsing. At the same time, the theoretical status of the underlying categories is not well understood, and despite progress, these tasks remain very much unsolved in practice. This graduate level seminar will concentrate on theoretical and practical models representing how discourse unfolds across sentences as it grows. We will explore cohesion in text by means of discourse relations (e.g. expressing causality, contrastivity), the use of recurring referring expressions, such as mentions of people, things and events, and how these are coded during language processing. We will also study multiple levels of discourse processing in terms of information structure, discourse relations and theories about anaphora, including classic theories as Centering Theory and Alternative Semantics, and newly developing ones such as Question Under Discussion (QUD) trees. With these in mind we will look at computational implementations of systems for entity recognition, coreference resolution and discourse parsing and explore their relationship with linguistic theory and textual coherence, including symbolic, neural and prompt-based approaches using LLMs. Over the course of the semester, participants will implement their own project exploring some phenomenon within the domain of discourse processing. Intermediate programming skills (preferably in Python) are required, and a previous computational course such as Intro to NLP (LING-4400 or higher) or Computational Corpus Linguistics (LING-4427) is recommended.