A Computer-Assisted Stylometric Exploration of Early Modern Latin Genres
Šime Demo · 2024
Stlometry is a method of detecting similarities between texts by comparing their linguistic preferences. The advances in information technology have made it possible to perform such analyses at a large quantitative scale. One of the most popular ways of doing this is based on investigating each text’s Most Frequent Words (MFW). This kind of research has revealed that Latin texts tend to group neatly into authorial and chronological clusters, whilst the evidence about genre-based grouping has remained inconclusive. In the paper, I present a digitally supported stylometric MFW-based analysis of an extensive corpus of Latin literature from all periods and various genres, comprising 811 works and totalling almost 36 million words. The lists of MFW s were compared using the so-called consensus network analysis (as implemented in the “stylo” package of the R program environment), and their correlations have been visualised in the form of networks (with the help of the tool called Gephi). Finally, on the basis of the facts we know about the genres included in the analysis, I suggest that genre can indeed be singled out as one of the factors of MFW-based groupings, and point to some interesting trends revealed by the network visualisations.