Towards a probabilistic model of dialect classification: a case study from Ancient Greek.
The regional dialects of Ancient Greek (attested ca. 700 – 350 BCE) have a complex geographical distribution. These dialects, however, can be classified into four principal subgroupings: Attic-Ionic, Doric, Aeolic, and Arcado-Cypriot. The Attic-Ionic, Aeolic, and Arcado-Cypriot subgroups are normally considered to be genetic subgroupings, that is they are believed to form distinct clades descended from a putative Proto-Greek ancestral to all the local dialects. Some recent scholarship has argued against the genetic subgrouping of the Aeolic and Arcado-Cypriot dialect subgroupings using a hypercritical methodology. In this talk I will argue that these views are methodologically flawed. To counter these hypercritical proposals I propose that the traditional classification may be re-affirmed through quantitative analysis of the dialect features, and will attempt to demonstrate this case through the development of a probabilistic model of dialectal subgrouping. To do this, I make an application of this probabilistic methodology to the Arcado-Cypriot subgrouping, by applying these methods to the Arcadian and Cypriot dialects of Ancient Greek.
Matthew Scarborough is currently a fourth-year Ph.D. Candidate in the Faculty of the Classics (E Caucus: Philology and Linguistics). His Ph.D. dissertation is entitled "The Aeolic Dialects of Ancient Greek: A Study in Historical Dialectology and Linguistic Classification" and is due to be submitted in 2015. He holds B.A. (1st Class Honours) and M.A. degrees in Classical Languages from the University of Alberta, Canada, and has published articles and reviews on Ancient Greek linguistics and epigraphy. His principal research interests are in Ancient Greek dialectology, historical and comparative linguistics (primarily of but not restricted to the Indo-European language family), language contact and sociolinguistics in antiquity, and the history of writing in the Ancient Mediterranean.