Computational Analysis of Alleged Homeric Texts

Abstract

Questions surrounding the Greek epic poet Homer have fascinated scholars for two and a half millennia. There has been much ink spilled on who he was, what he wrote, or whether he was even a single person, all backed up by hours of painstaking manual work. The types of tools that might allow a computational analysis, like a metrical scanner or dialect analyzer, are currently not accurate enough and do not provide enough feature extraction to be useful for this task. In order to use computational techniques to contribute to the discussion of these questions surrounding Homer, we create two new tools. The first, ᾠδικόν, scans dactylic hexameter and extracts relevant features about the meter of the line using two approaches, one inspired by modern students of ancient Greek and the other by ancient Greek speakers themselves. The second, τάμνον, uses a rules-based approach to analyze the dialect of ancient Greek words. We show that these tools are highly accurate and improve on previous tools with similar goals, and then use them to analyze a variety of allegedly Homeric and known Non-Homeric texts. These analyses provide support for some existing hypotheses about the relative dating of certain Homeric Hymns, insight into which features characterize the Iliad and the Odyssey, and evidence against common thinking on which of the books of these texts are unusual.