Oliver Hellwig
Title: Dating Vedic texts with computational models: Algorithmic considerations and data selection
Text: In spite of over 150 years of scholarly research, the chronology of the Vedic corpus is still far from being well understood as external historical evidence is largely missing and post-Rigvedic Sanskrit shows only minor developments on the levels of phonetics and morph-syntax.
This presentation discusses mathematical models that can be used for dating (Vedic) texts based on the linguistic evidence they provide. It also addresses the important questions of how to select expressive linguistic features, i.e. those whose distribution is coupled with the time of composition; and how to interpret the parameters of the resulting models in a linguistic context. The discussions are exemplified by a corpus of classical and medieval Latin texts which show comparable linguistic developments, but can, in contrast, be dated exactly, thereby facilitating model evaluation.