Speaker Series: Canaan Breiss, Univ. of Southern California (theoretical/computational/experimental phonology)

Apr 26, 2024 at - | Fisher Bennett 401. (In Person)

Title: Token frequency in the grammar: evidence from Japanese voiced velar nasalization

Abstract: This talk examines how token frequency influences an optional paradigm uniformity effect in Japanese voiced velar nasalization. We report two wug-tests demonstrating the frequency-conditioning observed by Breiss et al. (2021b) in corpus data is reproduced in existing compounds and novel compounds, demonstrating that an adequate model of the phonological grammar needs to be able to integrate information about the lexical frequency of the morphemes that it manipulates. We propose a psycholinguistically-informed formal phonological model that allows the token frequency of surface forms to scale the severity of faithfulness violations that reference them, extending a model which was originally proposed to model lexical and phonological conditions on word-formation in Lexical Conservatism (Steriade 1997).