Evoking and Contesting Expertise in Parliaments: An AI-supported Analysis of Climate Change Debates in Seven Countries

The Paris Agreement of the United Nations Framework Convention on Climate Change (UNFCCC) put nation-states front-and-centre in the politics of climate change mitigation. Several governments enshrined the pledges made in Paris into domestic legislation and exposed themselves to more systematic domestic scrutiny through parliaments and courts than ever before (Falkner 2016). The proliferation of climate change legislation and climate litigation in recent years suggests that legislatures and courts have taken to their new role, and climate policy is becoming a key concern for parliaments. More and more parliaments seek advice on climate policy from outside traditional political and expert forums and have strengthened their reach on channels of external and in-house climate expertise that are independent of government influence (Averchenkova et al. 2021). Set before this background, the paper brings together three fields of research related to climate change which are rarely interlinked: legislative studies, the study of climate policy and quantitative text analysis (Sebők et al. 2023). We examine seven European country cases of the ParlLawSpeech dataset to 1, identify debates on climate policy and 2, to explore how expertise is evoked and contested in these debates. We do this by using AI-assigned policy topic codes related to climate policy and deep learning-based large language models to tease out the sentiments associated with references to expertise. We identify temporal trends and differentiate between different party families regarding frequency, argumentation and reliance on external expertise. The paper contributes to the evolving interdisciplinary agenda analyzing the politics of climate change. Keywords: climate change, parliamentary debates, expertise, text mining, sentiment analysis, large language models