Randomization tests in language typology

Abstract

Two of the major assumptions that common statistical tests make about random sampling and distribution of the data are not tenable for most typological data. We suggest to use randomization tests, which avoid these assumptions. Randomization is applicable to frequency data, rank data, scalar measurements, and ratings, so most typological data can be analyzed with the same tools. We provided a free computer program, which also includes routines that help determine the degree to which a statistical conclusion is reliable or dependent on a few languages in the sample. © Walter de Gruyter 2006.