Figure 2

The influence of the number together with the average AoA of learnt non-native languages on the rapid response modulation of novel words with non-native phonology in the ignore condition.

(a) The two predictors of the multiple regression model, i.e. the number of learnt non-native languages (open dots and dashed line) and the average AoA (solid dots and line), significantly predicted the response increase to the non-native novel word-forms in the ignore condition (top) registered at ~50 ms after the word divergence point (indicated by the y-axis). The more learnt languages and the earlier average AoA, the larger the response increase between early and late stages in exposure to non-native novel word-forms was. Waveforms and histograms representing average response early (grey) and late (black) during the exposure to the non-native novel word stimuli for each subgroup by quartile (Q1–Q4) are shown to illustrate the differences in response development (bottom); the subjects were subgrouped such that those with the lowest number of learnt languages with highest average AoA belong to the first quartile subgroup (Q1) and those with the highest number of learnt languages together with the earliest average AoA are in the subgroup of the fourth quartile (Q4). Error bars denote standard errors of mean. (b) The group average of the response to the non-native word-forms and the corresponding scalp topographies at early (grey) and late (black) stages of exposure. Waveforms are from a sensor ROI combining Fz and FCz where the responses were strongest at the ~50 ms, i.e. the latency at which the learning-related neural response increase for novel native word-forms at the group level was elicited (denoted by the circle). The consecutive peak at ~150 ms did not reflect rapid learning effects²⁵ and was not analysed further here. (c) Scalp distributions of the difference between early and late stages of exposures for the first and last quartiles and the subtraction of their distributions.