Computational authorship attribution in medieval Latin corpora: the case of the Monk of Lido (ca. 1101–08) and Gallus Anonymous (ca. 1113–17)

Abstract

This paper applies computational methods of authorship attribution to shed light on a still open question concerning two Latin works of the twelfth century: are the anonymous authors of the Translatio s. Nicolai (ca. 1101–1108) and the Gesta principum polonorum (ca. 1113–1117) one and the same person? The Translatio was written by the so-called Monk of Lido and describes Venice’s role in the First Crusade. The Gesta were written by the so-called Gallus Anonymous and contain a panegyric of the contemporary Polish ruler, Boleslaw III the Wry-Mouthed (r. 1102–1138). This study attributes authorship to these works within four corpora of Latin texts composed between the tenth and twelfth centuries, each with between 39 and 116 texts written by between 15 and 22 different authors. The goal of including four corpora is to see how robust the similarity between the target texts is to changes in text length, genre, and class balance in the corpora. In each corpus, nine different distance metrics and one machine-learning algorithm are used to classify the authors of the Translatio and Gesta. I conclude that it is highly likely that Gallus and Monk were indeed one and same anonymous author, and highlight the effectiveness of the Bray–Curtis distance and logistic regression as methods of attribution.