Applications of Large Language Models (LLMs) in Political Science Research: A Text-Annotation Approach to Measuring Obstructive Speech in Legislatures

Obstruction — the intentional delay of government business using dilatory tactics such as filibustering, amendments, etc. — occurs in every legislature, but is difficult to measure. Taking advantage of the ability of Large Language Models (LLMs) such as OpenAI’s GPT4 to provide structured annotations of long sequences of text, Mitchell Bosley, a PhD candidate at the University of Michigan, presents a new text-annotation strategy to measure obstructive speech in the Canadian House of Commons. He will also discuss the potential for future applications of LLMs in political science research.