Johann-Mattis List:
Chances and Challenges for Computational Comparative Linguistics in the 21st Century

Abstract:
The quantitative turn at the beginning of the 21st century has drastically changed the field of comparative linguistics. Had individual genious and expert insights dominated historical linguistics in the past, we now find many studies by interdisciplinary teams who use complex computational techniques to investigate the history of invididual language families based on large amounts of data. Had the identification of linguistic universals in hand-crafted language samples  dominated linguistic typology for a long time, scholars now use large cross-linguistic databases to investigate dependencies among linguistic and non-linguistic variables with the help of complex statistical models.

However, despite a period of more than two decades in which quantitative approaches have been increasingly used in comparative linguistics, gaining constantly more popularity even among predominantly qualitatively oriented linguists, we still find many problems, which have only sporadically been addressed. In the talk, I will present three of these so far unsolved problems, which I find particularly important for the future of the field of comparative linguistics. These are: (1) the problem of modeling and comparing sound change patterns across the languages of the world; (2) the problem of identifying cross-linguistic patterns of semantic change, and (3) the problem of estimating the borrowability of linguistic traits across languages and times.

While none of these problems has been solved so far, I will argue that substantial progress on their solution can be made by improving the integration of cross-linguistic data and by developing dedicated problem-solving strategies in computational linguistics which take the specifics of cross-linguistic data and language evolution into account.