Cross-modal statistical learning and its impacts on language learning

Abstract

In the 21st Century, picking up a new language is a favourite pastime of many. However, there are many challenges when embarking on this journey into a new language. In this study, we investigate the feasibility of using known cross-modal correspondences in line with the foundations of statistical learning to improve second language acquisition. We tested whether participants will be able to learn pseudowords better when they are presented with visual stimuli containing congruent cross-modal correspondences as compared to when the pseudowords are presented with visual stimuli containing incongruent cross-modal correspondences. 45 participants were given a 3-minute training sequence of visual (Gabor patches) and auditory streams (nonsense language) that were yoked at the syllable and presented simultaneously. After training, participants were tested on their ability to identify six pseudowords that were embedded in the nonsense language. Our results show that participants were not able to extract the statistical patterns in the nonsense language as their scores were not significantly above the chance level of 50%. Furthermore, our results seem to indicate that participants perform better with pseudowords that are in the incongruent condition, but this effect may be due to peculiarities of the individual stimuli used in the test, rather than a general mechanism.