Typology, Areality, and Diffusion

Dunn et al. (2007) state that their typological comparisons do not demonstrate genealogical relatedness in the usual sense, but that the technique does accurately recapitulate trees established by the comparative method. We demonstrate that the signal picked up by their method is areal, rather than genealogical, and suggest that the method, when tested on known language families, will also show a high sensitivity to the effect of diffusion.