Abstract
Recently, the role of emotions in defining or influencing behaviour, including political behaviour, has been acknowledged, and research is increasingly addressing how affective processes shape our attitudes, actions, and decision-making. Policy studies have also started to analyse how emotions are reflected in policy discourses and how they influence policy change and support for policies. Most of these studies use qualitative methods. Our paper seeks to contribute to the field by conducting quantitative, text-as-data analysis to identify the emotional content of policy discourses. The aim is to give a descriptive analysis of which emotions are mobilised by different policy fields, which emotions are used by the government and the opposition when framing policies, and how the emotional patterns of policy discourses have changed over time. The parliamentary speech databases of the Hungarian Comparative Agendas Project are analysed using stateof-the-art large language models fine-tuned for emotion analysis. The time frame of the project covers the period 1998-2022. Preliminary findings of the computational analysis confirm the tendency of emotionalisation: the manifestations of emotions increase over time, which is especially true for joy and fear.
Keywords
Emotion Analysis, Large Language Models, Public Policy