Polysemy in Semantics with Contextualized Language Models - (Language and Computers) by Hanjung Lee (Hardcover)

Book Synopsis

What makes two meanings of the same word feel closely related, while others seem almost independent? Why do people and AI systems confuse some senses but easily distinguish others? This book tackles these questions through a comprehensive study of verbal polysemy. Combining corpus analysis, human experiments, and contextualized language models, it investigates how polysemous senses are distributed, perceived, represented, and interpreted in context. The findings show that sense boundaries are not fixed categories but form a continuum shaped by semantic connectivity among meanings. Building on these results, the book introduces the Generative Activation Package Model (GAPM), a new framework for explaining semantic multiplicity and sense interaction. Offering both a methodological bridge between linguistics and AI and a new perspective on lexical meaning, this book will be essential reading for researchers in semantics, pragmatics, cognitive linguistics, corpus linguistics, and natural language processing.

About the Author

Hanjung Lee, Ph.D. (2001), Stanford University is Professor of English Linguistics at Sungkyunkwan University. Her research focuses on semantics and computational modelling. She has published in numerous journals, including Language (2024) and Journal of Linguistics (2022, 2016).