2024
By Justin Chun-ting Ho & Chung-hong Chan
The study examines the challenges and strategies for analyzing nationalist frames in multilingual and multinational texts, with a special focus on linguistic and contextual transferability. Measuring nationalism from text is a challenging task. While pioneers have made efforts to capture nationalist frames and discourse using computational methods, these attempts are limited to single country and single language applications. Due to the nuanced and context sensitive nature of nationalism, approaches that work in one case are not guaranteed to work in another social setting. This project aims to address these limitations by developing a comprehensive framework for identifying nationalist frames across diverse languages and social contexts.
The project uses political advertisements from 90 countries, manual translation, and crowd-annotation. The performance of three large language models (LLMs)—Multilingual BERT, LLaMA, and Mixtral—will be evaluated for their ability to adapt across languages and contexts. Findings will enhance computational frame identification and contribute to understanding nationalism globally.