Extended Data Fig. 5: Voice Onset Time Encoding Across Languages.

A. VOT values of unvoiced stop-consonants in Spanish (dark green) and English (light green) show systematic differences in duration. This systematic difference along with prior research on the neural encoding of voice onset time (VOT)²⁶ suggest that language experience may influence how sounds are processed in the STG. B. Electrodes are sensitive to different VOT values (0:10:60 ms). The top-right inset shows the VOT tuning function for this electrode, where each scatter point corresponds to the peak HFA response to the VOT bin. C. Consistent with prior research, average single electrode tuning curves either showed positive (V-, preferring unvoiced) or negative (V + , preferring voiced) tuning direction. We found no significant difference between tuning curves across VOT sensitive electrodes in Spanish and English speakers (two-tailed Wilcoxon rank sum test, V + n = [57, 32], V- n = [20, 21]). Error bars indicate standard error of the mean across electrodes within each VOT bin for each speaker group. D. Proportion of V+ electrodes was significantly higher (p < 0.05) in Spanish speakers as opposed to English speakers. Size of each scatter corresponds to the number of electrodes within each participant that show significant tuning for VOT. P-value corresponds to language background in a linear mixed effect model (t(128) = 2.11, SE = 0.0176, p = 0.0367). Box plots show the median (center line) and the 25th to 75th percentiles (box limits). E. Intermixed V+ and V- electrodes in English and Spanish speakers. Both left and right hemispheres showed a higher proportion of V+ electrodes in Spanish speakers, suggesting that language experience effects may arise at the population level²⁵.