Responsibility Framing under the Magnifying Lens of NLP: The Case of Gender-based Violence and Traffic Danger

Abstract

We introduce a framework for the computational analysis of how responsibility is framed in the reporting of two types of socially relevant events: gender-based violence (specifically, femicides in the Italian press), and traffic danger (specifically, traffic crashes in Dutch and Flemish news reports). We advocate for the parallel analysis of these two phenomena under the same theoretical framework, which draws on Frame Semantics, Critical Discourse Analysis and Natural Language Processing. Reusing two existing event-text datasets we show how computational experiments and the resulting analyses can be run. This work supports the testing and development of tools for NLP practitioners, as well as large-scale linguistic analyses for activists and journalists, in the context of socially impacting events.