2nd Semester 2016/17: Frame-based Bayesian Interpretation of Natural Language
- Henk Zeevat (ILLC and Heinrich-Heine-Universitataet, SFB991)
If you are interested in this project, please contact the instructor by email.
- The course develops one particular elaboration of Bayesian interpretation for natural language: starting from the assumption that humans do NL interpretation in a Bayesian way, it explores the consequences of that assumption for linguistics, computational linguistics, semantics and pragmatics from a theoretical viewpoint (linguistics, philosophy of linguistics, philosophy) without avoiding the details (computational linguistics subareas such as NL generation, pronouns resolution, discourse structure). The main thesis is that the human capacity of producing language is directly used in language comprehension.
In the second part, the course gives an introduction to frame semantics, a naturally emerging formalism for the description of --mainly-- lexical semantics and for which the case is being made that it has good psychological motivation or even a neurological basis. We will look at such motivation (and add some philosophical considerations), some standard applications and ponder the problem of extending it to a full proposal for NL semantics.
The backbone of the first part of the course is the lecturer's book Language Production and Interpretation: Linguistics meets Cognition, published by Brill in 2014. In the second half we will use a reader under compilation of mainly recent material.
- 1. Introduction 2. Optimality Theory 3. OT Production for Interpretation 4. Optional Marking Strategies and Linguistic Evolution 5. Interpretation 6. Frames 7. Applications 8. Logical and Stochastic extensions 9. Student presentations 10. Student presentationsThere will be 10 sessions of two hours in the first and third week of June, (subject to scheduling possibilities, including preferences of participants). There will be an additional date for preliminary presentation of the final assignments later on in June. The first session will be Monday 5 June at 3pm.
- The course will be hard without some knowledge in at least some of the relevant areas: computational linguistics, linguistics, AI, cognitive science, NL semantics and pragmatics, logic and mental representation. But there are no specific prerequisites.
- There will be a set of assignments and individual projects to be completed by 30 June. Some of these can be elaborations of the presentations in the second week.