Talks in Linguistics: Lexical statistics and the grammar

The linguistic generalizations that we implement in our grammars can often be traced back to existing patterns in our lexicon. As a result, the patterns that emerge from our grammar are at least in part sensitive to statistical distributions in our mental dictionary. In a typical scenario, the isomorphic relationship between the grammar and the lexicon means that statistical patterns in one’s lexicon can be used to predict grammatical generalizations and vice-versa. But this relationship is not always straightforward. In this talk, we will explore situations where native speakers’ (or learners’) generalizations (i) match lexical patterns, (ii) fail to mirror lexical patterns, and (iii) directly contradict lexical patterns. I will then argue that all three scenarios are predictable based on whether patterns are prosodically natural (in the case of native speakers) and on whether patterns can be unambiguously detected given limited input (in the case of second language learners).

Guilherme D. Garcia is an Assistant Professor of English at Ball State University. His research interests are Phonology, (Second) Language Acquisition, and quantitative data analysis. He is especially interested in using data analysis to uncover patterns that help us better assess representational and theoretical assumptions in phonology.

https://uictil.uic.edu/