Rules Reconsidered: A Statistical Reevaluation of Text-Critical Criteria
Jonathan Wilken · Apollo (University of Cambridge) · 2025
This thesis attempts to reevaluate several of the traditional canons of textual criticism based on their ability to reliably predict trends in scribal behavior. All alterations made by the scribes of a set of manuscripts were identified by comparing the text of those manuscripts to their extant exemplars. The text of all witnesses was digitized through a hybrid process of human and AI-powered OCR transcription. Treating the selected criteria as a series of hypotheses, their reasonableness was evaluated based on their ability to accurately predict statistically significant trends in the types of alterations produced by the sample of scribes. The primary criteria tested were the rules that the shorter reading should be preferred, that readings reflecting the vocabulary of the author should be preferred to readings that might have been more natural for later scribes, and that one should prefer the less harmonizing (in the local context) of two readings. Significant trends in error production were detected which contradicted the expectations of the shorter reading rule and the (local) harmonization rule. Further, the scribal bias toward omitting rather than adding was found to be stronger when dealing with longer variants. No significant patterns of behavior were detected regarding scribal preference for or against vocabulary more natural to an author’s style. This thesis concludes with recommendations for further research on and utilization of text-critical criteria and suggestions regarding methods for studying scribal habits.