Stephan Meylan & Roger Levy
Lecture 7 | Leipzig Lectures on Language—Combinatorics 2021
The Nature and Origins of Grammatical Productivity
How do children learn the combinatorial elements of language structure? In this talk, Dr. Roger Levy will show how a probabilistic generative model can be used to characterize combinatorial structure in language. Case studies include measuring children’s generalization when learning English determiners, as well as modeling adult preferences for ordering binomial expressions (“salt and pepper”). Dr. Stephan Meylan will present a related line of empirical work characterizing children’s acquisition of morphosyntax, focusing on the English plural. New eyetracking work reveals that children (24 - 36 months) can use agreement cues (“there **are two** wugs”) to identify referents before they can use phonemic distinctions alone (“Look at the wugs”). An analysis of plurals produced by these same children shows that they produce plural nouns—including novel ones—well before they demonstrate robust comprehension. These results suggest that grammatical generalizations are themselves an integral part of the language learning process, rather than its end product.
About the Speakers
Stephan Meylan is a Postdoctoral Fellow in the Computational Psycholinguistics Lab in the Department of Brain and Cognitive Sciences at MIT and a Postdoctoral Associate in the Bergelson Lab in the Department of Psychology and Neuroscience at Duke University. He studies the relationship of language processing and language development—with a focus on the emergence of combinatorial morphosyntax— using a combination of computational models, corpus studies, and in-lab experiments.
Roger Levy is an Associate Professor in the Department of Brain and Cognitive Sciences at the Massachusetts Institute of Technology (MIT). His research focuses on theoretical and applied questions in the processing and acquisition of natural language, employing computational modeling, psycholinguistic experimentation, and analysis of large naturalistic language datasets.
Keywords: language development; computational modeling; psycholinguistics; comprehension-production interface; eyetracking