Exploring the frontiers of phylolinguistics – X-PhyLs

Historical linguistics have recently witnessed a new quantitative turn with the development of computational models and methods that complement the traditional methods for reconstructing language histories. Such methods have have been successfully applied to many language families, including Austronesian, Bantu, Indo-European and Sino-Tibetan.
These methods have successfully reconstructed previously known facts, but have also provided answers to previously unresolved questions on the deep relations between languages, on the dating of ancestral languages, and on the processes which drive language change. These works rely on Bayesian statistical inference and represent language relations as a binary tree, although recent works propose more complex models that take into account lateral transfer events resulting from contact.
Our project will bring together linguists and mathematicians to further explore the possibilities and push the frontiers of computational language phylogenies. We shall tackle several related but distinct problems about the methods of both deep and shallow linguistic histories.
First, we will address two general outstanding issues: What is the time depth after which most of the phylogenetic signal originally present in languages vanishes, making linguistic relationships undetectable?
How can our computational models of linguistic change be made more realistic, in particular by incorporating lateral transfer due to contact? We will also look at two case studies where phylogenetic methods can bring answers to interesting questions, one about the shallow Japonic family where relationships are obscured due to contact, and the other deep linguistic relationships in East Asia and the Pacific. We will also tackle the issue of modelling the evolution of paradigmatic morphology using the example of person indexing in Sino-Tibetan in order to go beyond the current frontiers of phylolinguistics, which is limited to the lexicon and structural features.