Multi-Voxel Pattern Analysis Applied to the Language Switch in the Bilingual Brain--An fMRI Study

Abstract

Previous fMRI studies about bilingual speakers proved that the left caudate and the anterior cingulate cortex (ACC) play the important role in intralanguage task switching (Crinion et al., 2006; Abutalebi et al., 2008). Our research based on Multi-voxel pattern analysis (MVPA) suggests that these brain regions are involved in the language control not only, but also in semantic processing which encapsulates a covert translation between the first/second native languages. Using a set of stimuli which was composed of two semantic categories, Korean-Chinese bilinguals were requested to do a property generation task in Korean for a stimulus word in Chinese, vice versa. Machine learning methods applied to the fMRI datasets enabled us to create, even from the aforementioned regions of interest (relatively small with only about 1000 voxels), significant classification models with the identification accuracy of more than 60% (above chance) for predicting the semantic category of each target.


Back to Table of Contents