“Biblicality” of Early Medieval Canon Law through the Lens of Language Modeling

Using language modeling and the information-theoretic concept of perplexity, this study explores the influence of the Bible on early medieval canon law. We demonstrate that calculating perplexity of a corpus of canon law under a language model trained exclusively on Scripture can serve as a reliable proxy for accessing the overall linguistic similarity—stylistic, semantic, and syntactic—of canon law to the Bible. The paper presents the measured “biblicality” of various canon law texts, explores its chronological development, and delves into the linguistic and stylistic meaning of higher or lower “biblicality” by observing correlations between perplexity under the biblical language model and various linguistic features of canon law texts which can be extracted from rich morpho-styntactic annotation. We hypothesize that changes in the levels of “biblicality”, clearly observable across the chronological subdivisions of the corpus, suggest that the imitation of scriptural language may have been a deliberate strategy reflecting evolving views on the role and place of Scripture in legislation. Further research is needed to trace in more detail the connection between the authoritative status of canon law collections and the use of the Bible.