Course Description
The course addresses the practical and theoretical relationship between contemporary weighted-constraint phonological theories (predominantly Maximum Entropy grammars), and statistical methods that are often employed in analysis of phonological data (log-linear regression models). The course focuses on the mathematical, conceptual, and theoretical intersections of these two analytical methods that are often misconstrued as disparate, examining the ways in which they are more deeply linked than often is acknowledged.
The first portion of the class introduces the basics of weighted constraint and regression analysis, and explores the broad intellectual history of these types of models throughout linguistics and cognitive science. The class then turns to the parallel ways in which these models (both phonological and statistical) are deployed in contemporary phonological analysis, particularly in issues of cumulativity and morpho-/lexically-specific phonological behaviors.
Finally, the class explores the new horizons opened by seeing log-linear models as special cases of a broader class of (Bayesian) cognitive models. This move provides us with a flexible phonological framework that blends information from multiple sources into the same model, can deal with variation at multiple levels of structure, produces theoretically meaningful constraint weights and parameter values, and offers prescribed and domain-general methods for model comparison. At the end of the course, students will be able to connect empirical method and linguistic theory, as applied to natural language phonological data.
Area Tags: Phonology, Statistics, Variation, Linguistic Frameworks, Computational Linguistics, Cognitive Science
(Session 1) Monday/Thursday 10:30-11:50
Location: ILC S231
Instructors: Canaan Breiss
Canaan Breiss completed his PhD at the Department of Linguistics at UCLA in 2021, and he is currently a postdoc in the Computational Psycholinguistics Lab at MIT in the Department of Brain and Cognitive Sciences. From Fall 2023, Breiss will be faculty at the University of Southern California in the Department of Linguistics. His research is in theoretical, computational, and experimental phonology, with particular interest in learning/acquisition, representation of overlapping and interacting phonological processes, and phonology’s interfaces with (morpho)syntax and the lexicon. He uses a diverse methodological toolkit, including computational and statistical modeling, corpus methods, and behavioral experiments.
