Fig. 1. Methods.

a Pitch contours (M and SD) of the four Mandarin Chinese tones across syllables and female speakers included in the study. b To estimate the categories that would be easier (Tone 1 and Tone 3) and harder (Tone 2 and Tone 4) to learn, we examined an Aggregate dataset of 678 Mandarin tone learners collected across eight published training studies. Left. Individual and mean percent correct responses across learners and tone categories. Right. Mean percent correct responses (99% CI) across learners and categories for easier- and harder-to-learn categories. c Left. Categorization trial structure and categories paired with stimulation in each participant group. Right. tVNS-stimulus alignment in one example trial. d Before the training task, we conducted a perceptual identification task to rule out group differences in perceptual identification skills. Left. Participants were asked to categorize as “rising” or “level” a perceptual continuum of Mandarin tones ranging from high-level (Tone 1) to low-rising (Tone 2) pitch. Right. The slope of the perceptual identification curve was used as a metric of perceptual acuity. e Left. To assess the effects of tVNS on the sensory encoding of stimulus pitch, we collected frequency-following responses (FFRs) to Mandarin tones before and after the training task. Right. To assess neural pitch encoding quality, we correlated neural (FFR) and stimulus pitch.