Skip to ContentSkip to Navigation
Research The Groningen Research Institute for the Study of Culture (ICOG) Research Research centres Research Centre for the Study of Democratic Cultures and Politics Events & Colloquium

EPS seminar series- STEFAN MÜLLER (University College Dublin): 'Multilingual Stance Detection in Political Speech: Interpretation, Applications, and Limitations'

When:We 01-05-2024 16:00 - 17:00
Where:Room 1315.0048, Harmonie building

This event is part of the Seminar Series of the theme group European Politics and Society.

Abstract

Measuring the stance of political actors on specific policies provides valuable insights for understanding policy-making, changes in political preferences, and party competition. In this paper, we fine-tune various multilingual transformer machine learning models based on annotated texts of stances in over 53,000 comments on Twitter and more than 67,000 comments to 150 political questions in German, English, and Italian. We validate and compare various multilingual transformer frameworks. After identifying the most suitable fine-tuned model, we compare the stance classification with human-coded positions of politicians’ support for the annual budget, positions and roll-call votes during energy policy debates, and stances on specific policies, such as abortion and immigration. We discuss the scope and limitations of stance classification compared to existing approaches, such as unsupervised and supervised text scaling, sentiment analysis, human judgement, and generative AI. Drawing from our systematic comparison and validation of various methods, we provide recommendations for researchers aiming to apply stance detection to political texts.