Old Texts, New Tools: A Data Science Approach to Authorship Attribution for Early Italian Poetry

Abstract

This article focuses, from a multidisciplinary perspective, on an ad hoc corpus of medieval Italian literature defined by geographical (Tuscany) and chronological criteria (13th and 14th centuries). The goals are both methodological and operational: first, we discuss the contribution of computational linguistics techniques in attributive philology; second, we use a philological approach combined with some newer artificial-intelligence (AI) approaches to promote or reconsider a variety of proposals of attribution for unknown texts selected as part of Antonio Pucci’s corpus of disputed authorship. It was considered interesting to also include in such corpus some representative texts in Dante studies, such as the Fiore, the Detto d’Amore and the sonnet Quando ’l consiglio degli ucce’ si tenne – whose authorship, still in doubt, is disputed with Pucci – without any pretense of providing a solution, but with the purpose of both testing the reliability of these methods in identifying the most probable candidate authors and adding new elements to the debate over these case studies.