Methodological Problems in Quantitative Research on Environmental Effects in Phonology

Abstract

This paper engages with the quantitative methodology underlying studies proposing a link between environment and phonology by replicating three prominent studies on ejectives and altitude, vowels and humidity, and sonority and ambient temperature. It argues that there are several issues regarding the methodological footing of such correlational studies. Further, the paper finds that the problems of statistically analyzing environmental datasets in phonology run deeper than the focus on individual phonetic features suggests: there are several overarching patterns of correlations to be found in these datasets that, if not understood and accounted for, render mistaking spurious correlations for real effects inevitable. This paper further makes concrete suggestions for what is needed to move beyond pairwise correlational studies between environmental and phonological variables in future investigations.