Hansard DB: A Relational Database of Australian Parliamentary Speech

Legislative chambers are central institutions of democratic governance, where representatives debate policy, justify decisions, and are held accountable. In Australia, the principal record of what is said and done in parliament is the official parliamentary transcript, commonly known in Westminster systems as Hansard. Yet while these records are publicly accessible, they are difficult to use at scale because they are not readily queryable, computationally integrated, or suitable for systematic longitudinal analysis. This paper addresses that problem by introducing Hansard DB, a relational database of Australian federal parliamentary speech spanning Federation to the present. Hansard DB integrates speeches, questions, answers, and interjections with parliamentarian metadata from the Parliamentary Handbook. Built through a multi-stage parsing and validation pipeline, the database supports flexible querying across text and metadata for large-scale and longitudinal analysis. The paper also discusses the epistemic limitations of Hansard research. The official transcripts are edited and incomplete records of political speech, shaped by institutional rules, editorial practices, and omissions of tone, gesture, and context. Hansard DB therefore contributes not only a usable database, but also guidance for interpreting parliamentary transcripts acconting for these limitations.