Statistical bias control in typology

Abstract In this paper, we propose two new statistical controls for genealogical and areal bias in typological samples. Our test case being the effect of VO-order on affix position (prefixation vs. suffixation), show how modeling including a phylogenetic regression term (phylogenetic control) two-dimensional Gaussian Process (areal can be used to capture effects large but unbalanced sample. We find that, once these biases are controlled for, has no position. Another important finding, which is line with previous studies, that as effects, emphasizing importance or contact control studies built language On other hand, also strict probability sampling not required propose, long sample variety enough cover different areas families. This crucial practical consequence it allows us include much available information possible, without need artificially restrict potentially lose otherwise information.