Towards a better understanding of the functional neural mechanisms underlying sentence-level auditory-oral bilingual processing : An EEG study

Abstract

Sentence-level auditory-oral online bilingual processing is cognitively taxing, yet the functional neural mechanism subserving such activity remains unclear, despite increasing efforts from the cognitive neuroscience field. The current thesis synthesized four bilingual processing tasks and analyzed their oscillation features using electroencephalography (EEG). Specifically, EEG experiments involving listening to an L2 passage (L2L), speaking in L1 (L1S), shadowing in L2 (L2SH) and backward interpreting (BI) were implemented on Chinese-English and German-English bilinguals respectively. Oscillation indexes including the static spectral power, time-frequency power and cross-frequency coupling were analyzed. The aim of the thesis was to provide exploratory findings concerning the oscillation features of the four above mentioned tasks, which has never been reported before. Study One consisted of 43 adult Chinese-English bilinguals (male = 22, female = 21) aged between 19 and 36 years (M = 26.05, SD = 4.55). Participants’ EEG was recorded while performing such tasks as L2L, L1S, L2SH and BI. Their oral output in the BI task was also analyzed as a behavioral index. There were four major findings revealed by Study One. First, alpha frequency (8-15 Hz) was the dominant oscillation band in the static spectral results across tasks, revealing an inhibition-related mechanism during bilingual processing. Second, motor tasks (i.e., L1S, L2SH and BI) incurred significantly higher time-frequency power values in the delta-theta (1-7 Hz) and gamma band (30 Hz+) compared to the sensory task (L2L). Third, a theta-gamma coupling (TGC) was consistently located in the right fronto-temporal lobe for BI compared to other tasks, indicating the unique functional synchronization of the task. Inferential statistics also showed that alpha band spectral power was mostly explained by participants’ exposure to the L2 environment. Study Two, using the identical materials and task procedures, consisted of 46 German-English bilinguals (male = 14, female = 32) aged between 19 and 46 years (M = 28.37, SD = 6.42), with the intention to replicate the protocol of Study One and examine if the oscillation patterns differ according to language pairs. Both participants’ EEG signals during the above-mentioned four tasks and their oral speech in the BI task were recorded. Findings similar to those of Study One were obtained. First, all tasks were dominated by alpha-band spectral power. Second, condition-wise comparisons revealed significantly higher delta-theta and gamma band time-frequency power for motor tasks compared to the sensory task, which was the same with the pattern in Study One. But the gamma band time-frequency power increases extended to the higher beta range (20-30 Hz), which was absent in Study One. Third, a robust right fronto-temporal TGC was found in BI compared to other tasks. Statistics revealed an association between alpha power and participants’ L2 exposure similar to that in Study One. The comparative analysis between Study One and Study Two showed that the Chinese-English and German-English bilinguals differed from each other in years of L2 learning, L2 proficiency and accuracy in the interpreting task, such that the former group spent significantly longer time learning English but achieved significantly lower scores in English proficiency and backward interpreting tasks. Such results lent support to the language distance effect. In terms of oscillation features, the language distance effect was seen in alpha band spectral power in the occipital lobe, and in beta-gamma range time-frequency power. Right fronto-temporal TGC was robust for the BI task across both groups. Taken together, the findings of the current thesis present a complex functional mechanism underlying sentence-level auditory oral bilingual processing. In auditory tasks the brain is mainly activated at alpha band in the occipital lobe to suppress the irrelevant language while storing the syntactic units in the verbal working memory. Such alpha activation is modulated by language distance, so that a distant language requires more cognitive resources to inhibit, thus eliciting higher alpha power. In motor tasks the brain shifts to a multi-tasking mode, with theta activation responsible for message retrieval and syllabic organization and gamma for meaning extraction, motor execution, and semantic inspection. If the motor task is semantic-heavy, the gamma band would extend to the higher beta band to assist semantic processing. If the task requires converting semantics from one language into another, a right fronto-temporal gamma coactivation component would appear to form a TGC network with another theta. Finally, a sentence-level auditory-oral bilingual processing model was proposed based on the findings of the current thesis. The potential application of the findings in interpreter training and aphasia treatment was also discussed.