New models of language dynamics: The role of cross-linguistic data
Colloquium | October 28 | 3:10-5 p.m. | 370 Dwinelle Hall
Damian Blasi, Radcliffe Institute for Advanced Study, Harvard University (US) Department of Linguistic and Cultural Evolution, Max Planck Institute for the Science of Human History (Germany)
The status of cross-linguistic data in theories of language change and evolution has varied substantially over the course of the history of the discipline. The comparative study of languages and language histories has been a classic testing ground for hypotheses on the subject, but in an influential recent line of work it has been systematically sidelined in favor of highly controllable experimental and computational approaches that are only weakly connected to the bulk of cross-linguistic knowledge. The reasons behind this shift are many, but they are often tied to the notion that observational data can only provide a weaker form of evidence in contrast to their alternatives such as experimentally obtained data. In this presentation I will provide a critical examination of these more recent models, focusing on their ecological validity and empirical coverage. I will illustrate my points by drawing across disciplines and topics, ranging from the emergence of creole languages and Bayesian learning to early Holocene linguistic diversity, sound-meaning associations and stochastic processes on phylogenies. I will conclude that naturally occurring linguistic data, while particularly challenging from a statistical inference point of view, has been unjustly underused in some current models of language dynamics, and I will conclude by outlining emerging trans-disciplinary efforts that address this situation.