Analysing New York Times Coverage of Sexual Abuse Over Time Through Natural Language Processing

Abstract

Sexual abuse is a prevalent issue in modern society, having acquired prominence during the #MeToo movement in the early twenty-first century. Previous studies note that the media, as a major source of information, plays a significant role in reflecting and shaping societal perceptions of sexual abuse incidents. As a result, analyzing language use in media is crucial in forming an understanding of how inherent biases are conveyed in text. This work aims to investigate the media framing of sexual abuse using NLP techniques such as Valence Arousal Dominance (VAD) analysis and analysis of power and agency scores of verbs. This research investigates the implications of media framing of sexual abuse by incorporating computational modelling and sociolinguistic research. This approach can demonstrate how conscious and unconscious prejudices materialize in text, impacting how specific groups—in this research, victims and perpetrators—are portrayed in the media. Our results show that although the increase in coverage surrounding sexual abuse contributed to an increasingly positive view of female victims from the early 20th century—signaled by an increase in power and agency over time—they were still portrayed as having considerably less power and agency than their male perpetrators. Additionally, girl victims were portrayed with consistently less agency than women victims.