Historical Linguistics: Processes, Inference, and Reconstruction

This is an exciting time for studying historical linguistics using formal statistical and machine learning methods.  The traditional approaches used in this field have been based on rules derived painstakingly by linguists by extracting regularities amongst related languages.  Such regularities are, however, rarely perfect: histories of individual linguistic features or of individual languages often display idiosyncratic variation.  A lot of historical linguistics is focused on these instances and their interesting peculiarities, whereas the major questions in history often depend on the general patterns obscured by such details. Molecular phylogenetic, and bioinformatics, has been quite successful in developing and applying statistical methods that deliberately focus on the dominating regularities in large genomic data set. Biological evolution is grounded in mechanisms different from those at work in cultural, including linguistic, evolution, nevertheless many of the observed regularities are the same or at least similar. This working group will bring together scientists working on developing methods that can discover and test the underlying linguistic patterns and histories from a corpus of linguistic data, providing a default expectation against which to judge the exceptions.  The discovered patterns can then be used to inform, or be informed by, population dynamics and movements inferred from other branches of the historical sciences.