Reading Supreme Courts from afar: Topic modelling judgements of the Supreme Courts of Sri Lanka and the United Kingdom

In legal scholarship, court judgments are pivotal in shaping jurisprudence. Situated within the field of digital humanities as it applies to legal texts, this article takes a closer look at the underlying themes in Supreme Court judgments by applying topic modelling to the judgments delivered by the Supreme Court of Sri Lanka (LKSC) and the United Kingdom Supreme Court (UKSC). Using two custom datasets curated by (web) scraping the respective LKSC and UKSC websites, this article employs Latent Dirichlet Allocation (LDA) with the Machine Learning for Language Toolkit (MALLET), a commonly used tool for topic modelling by digital humanists, to identify topics (themes) that represent the main areas of law in each jurisdiction primarily dealt with by the respective courts. 25 was selected as the number of topics after experimentation, and the topics identified in each jurisdiction were manually labelled. The results reveal the composition and evolution of judicial workloads, the shifting socio-political priorities in each jurisdiction, as well as the similarities and differences between the two courts. The findings have several implications for legal research and practice. These suggest that topic modelling can be used as a tool to organize and categorize judgments based on their themes, which can facilitate access and retrieval of relevant cases, and identify priority areas for judicial and legal training. They also challenge conventional legal taxonomies and classifications, and demonstrate the potential of computational methods for enhancing the understanding and analysis of law.