A note of caution on tone language advantages for music

Liu et al.¹ reported recently in Current Biology a large-scale citizen science replication of the tone-language advantage for musical pitch perception. Their 493,100 volunteers spoke 54 languages, including 19 tone languages, those in which a word's pitch pattern contributes to its meaning. For example, Mandarin ma with high pitch means 'mother', while ma with dipping pitch means 'horse'. Previous studies reported tone language advantages, but with far smaller and less linguistically diverse samples (mostly East Asian tone languages). I applaud the authors' exploration of diverse tone languages with disparate tonal properties, including varying numbers, types, and linguistic uses of tones, plus varying cultural factors. While I do not dispute the overall tone language advantage, I take issue with Liu et al.'s¹ inference that the effect is consistent across the varied tone languages tested: because of the properties of the models they used to estimate individual-language effects, their more-diverse tone languages spuriously appear to pattern consistently, when in actuality the data are too noisy to draw strong conclusions about tone languages as a unified group.