Statistics / Linguistics Seminar - Rob Voigt

Linguistics Statistics Seminar

Event Date

Social Sciences & Humanities 2203

Speaker: Rob Voigt, Assistant Professor of Linguistics, Northwestern University

Title: "Linguistic Mechanisms: Data Science for the Social World"

AbstractThe structures of our social world are ultimately the aggregation of many person-to-person interactions, and language is often the primary medium of these interactions. My research program builds upon this insight by developing and applying methods from statistical natural language processing to examine the on-the-ground mechanisms of social phenomena. In this talk I aim to show how computational linguistics can be productively combined with methods from statistics and data science to ask large-scale questions about the social world while maintaining a connection to the deeply human linguistic behaviors that underlie and constitute it. To illustrate the possibilities therein I will discuss three areas of research and the corresponding approaches I have employed targeting social understanding and social impact: police-community interaction, by analyzing body camera footage including randomized controlled trial designs and evaluation of behavioral interventions; polarization surrounding immigration, by applying large language models to the congressional record; and race in media representations of gun violence, by linking large corpora of TV and print news to incident metadata at scale. In this work I argue that integrating the theoretical toolkit of linguistics into data science enables interpretable methods for quantifying real-world language use and actionable findings on its accumulation into patterns of social bias, polarization, and intergroup conflict.

Speaker's website (links to Northwestern Univ.):

Seminar Date/Time: Monday March 4th, 2:00pm, at SSH 2203