Code, Context, Canon: A Transferable Framework for Computational Canon Studies

This poster presents a transferable computational framework for studying canonisation processes through extracting and analysing canonical references in academic texts. Using a 16-discipline English corpus from JSTOR, we propose a three-phase methodology: code (developing Named-Entity Recognition methods for canonical reference extraction), context (visualising citation patterns), and canon (understanding formation processes). Our approach examines transferability at each stage: the code phase explores large language models over traditional keyword search methods, emphasising advantages of general over domain-specific models; the context phase demonstrates high transferability through established network analysis techniques; the canon phase reveals how the concept of canon itself transfers across disciplines. This iterative workflow contributes reusable and transferable Digital Humanities frameworks for studying cultural authority and textual transmission, with the objective of advancing understanding of canonisation dynamics across disciplinary contexts.