Course Goals
LING257 - “Psycholinguistics and linguistic theory” - was a broad topical overview of issues in language processing and its interfaces with core areas of the grammar. LING258 takes a more in-depth and computational approach to a more restricted set of issues. Part of the task of a theory of language processing is to explain why some forms in language are (dis)preferred or difficult to produce, comprehend or acquire. Broadly speaking, these theories all incorporate some account of (i) resource limitations and (ii) how knowledge of language and experience with language processing combines with perceptual evidence. To characterize pieces (i) & (ii) and how they might interact, this course divides itself into three units: firstly, we read some important early versions of both the Garden Path theory and species of Constraint-based Interactionism; then, we consider the role expectation and prediction plays in language processing; finally, we address the issue of memory architecture and memory limitations We will learn in detail about some explicit computational models - hypotheses - postulates, like Competitive Attachment, the Entropy Reduction Hypothesis, Surprisal, and ACT-R. The readings will begin somewhat historically, but quickly become very contemporary.
How you will contribute
Course members will …
- be responsible for giving presentations on some readings - how many will depend in part on the size of the class, but it will not exceed two.
- engage in a extended research project, by choosing from one of the several proposed themes below and meeting regularly with the course instructor throughout the quarter to develop the project. This project will culminate in a seminar-length paper.
Themes
- The status of adjuncts
- Empty categories
- Levels of representation in making predictions
- Case and agreement