Handwritten Text Recognition and Epistemic Change in Philological Practice

Aglae Pizzone (SDU)

Handwritten Text Recognition and Epistemic Change in Philological Practice: A Case Study from Byzantine Rhetorical Manuscripts

This paper examines how the integration of Handwritten Text Recognition (HTR) technologies is reshaping the epistemic foundations of philological work, using a corpus of Byzantine rhetorical manuscripts as a case study. 

While HTR promises unprecedented access to large quantities of transcribed material, its adoption also reconfigures the balance between heuristic discovery, traditionally grounded in close reading and palaeographic expertise, and critical‑analytical evaluation, increasingly required to interpret machine‑generated outputs. 

Drawing on experiments with training and adapting HTR models to 14th‑century informal scholarly hands, the study highlights how automated transcription alters our workflows, affects the identification of textual relationships, and challenges assumptions about scribal variation or similarities. The case study shows that HTR, while accelerating existing practices, also introduces new epistemic tasks: the need to assess algorithmic uncertainty, to filter over‑generated parallels, and to contextualize results within broader codicological and rhetorical traditions. Ultimately, the talk argues that HTR catalyzes a shift in philological practice and invites to reconsider what counts as evidence, expertise, and hermeneutic in the study of ancient documents.