Recent Developments in the Analysis of Legislative Speeches

Abstract

Speech-making activity’s essential role in representative democracies has consistently been recognised in theoretical work on parliamentary systems but has only recently become the focus of empirical legislative studies. The literature on the subject has rapidly expanded, starting with Proksch and Slapin’s seminal contribution “The Politics of Parliamentary Debate: Parties, Rebels and Representation” (2015). The availability of large text corpora of legislative speeches, methodological advances in computer-assisted content analysis and the improved quality of translation tools have fueled current research. For example, the ParlSpeech and ParlSpeechV2 data collections include full-text vectors and meta-data of more than 6.3 million parliamentary speeches delivered in the legislative chambers of several liberal democracies, covering periods between 21 and 32 years until recently. On the methodological side, dictionary-based approaches and supervised and unsupervised machine learning techniques have enriched the toolkit for automated content analysis, ranging from classifying documents—either into existing or yet-to-be-determined categories—to scaling political actors in a policy space. One of the challenges in this context is the growing gap between work in informatics (e.g., computational social sciences) and computational linguistics on the one hand and theory-driven work in political science on the other. Moreover, there is a need to identify opportunities for coordinated data collection across countries, standardising formats and metadata, linking textual data with other political science datasets, and creating open access to such data for the broader political science community. The proposed panel features the latest contributions to the study of legislative speeches and aims to: ▪️ Explore and evaluate the extent to which the texts of legislative speeches can provide valid and reliable data that can be converted into theoretically meaningful variables to test and develop established and new theories of legislative behaviour, inter-party competition, polarisation, intra-party politics and policy-making. ▪️ Critically assess the extent to which methodological advances in data exploration in computer science can be utilised to provide more valid operationalisations of theoretical concepts in political science as well as the ability to reliably identify meaningful latent patterns in speech data. ▪️ Identify opportunities for the collaborative collection of theoretically meaningful contextual data in a larger number of countries than has been the case in the past, including the use of existing data and the improvement of data linkage methods.

Title	Details
Party Competition in Times of Crisis	View Paper Details
Contagiously Eurosceptic? The Spillover Effect of the M5S’ Eurosceptic Stance on Other Parties’ Attitude towards the EU	View Paper Details
Interruptions in German Parliamentary Speeches: Sentiment Expressed by the Alternative for Germany	View Paper Details