News reporting plays an important role in setting the boundaries of public discourse. In order to appear neutral, mainstream news reporting generally favours the dominant values of the society in which the communicator and audience exist. In a patriarchal society, this often leads to the reinforcement of traditional gender roles and the cisnormative gender binary. The mechanisms by which social norms are reinforced can be studied via frame analysis – a type of content analysis involving the qualitative identification and interpretation of frame elements (e.g., problem definitions, cause and blame attributions, solutions) and other implicit characteristics of communication.

Frame analysis is a labour intensive task, and tends to be limited in scope by the intellectual and financial resources available. This project examines how Large Language Models (LLMs) can be used to inductively identify frames in large scale news corpora, by focusing on gendered framing patterns in the reporting of a range of contemporary issues. This project will produce a general-purpose methodological framework for LLM-assisted frame analysis, as well as a prototype software pipeline to support the valid use of LLMs as virtual researchers, contributing to broader discussions on the use of AI in social science research and the mechanisms of gender policing in news media, and examining the quality, accuracy, and ethical implications of LLM-assisted research.

Supervisors
Dr Katharina Esau
Professor Axel Bruns
Professor Nicolas Suzor
Professor Daniel Angus