Comparing languages and comparing typological databases

In this online talk, I report on some of the conceptual challenges that arise in creating typological databases such as the World Atlas of Language Structures (WALS) or the Atlas of Pidgin and Creole Language Structures (APiCS). While the empirical challenges (getting relevant data from many small languages around the world) are obvious, the conceptual challenges have been understood in different ways: (i) finding the true natural categories for comparison; or (ii) making sure that the comparisons are valid in the absence of natural kinds. I argue that comparative concepts (as in Haspelmath 2018) provide the right methodological tool. In a next step, we would like to compare different typological databases with each other, so we want to make the comparative concepts commensurable across projects. I will argue that the Concepticon (a set of lexical comparison meanings) provides a good model that one can emulate also for grammatical databases.