The purpose of this study is to analyze the characteristics of EEG responses to native-language auditory stimuli, non-native-language auditory stimuli, and a resting condition with eyes closed, using publicly available auditory EEG datasets. This stud...
The purpose of this study is to analyze the characteristics of EEG responses to native-language auditory stimuli, non-native-language auditory stimuli, and a resting condition with eyes closed, using publicly available auditory EEG datasets. This study is a basic research project that reanalyzes publicly available data from non-clinical, general volunteers; it is not a study involving a clinical patient cohort. Therefore, the focus of the study is not on verifying the diagnostic accuracy for MCI or dementia, but rather on identifying single-channel frequency responses and network connectivity patterns in healthy subjects in response to auditory language stimuli.
The analysis utilized the Auditory Evoked Potential EEG-Biometric dataset from PhysioNet. This dataset consists of 2-minute EEG recordings collected from 20 participants and includes both rest and auditory stimulation conditions. To control for differences in stimulus delivery pathways, this study analyzed only the in-ear condition, limiting the analysis to the Native, Non-Native, and EC conditions. The EEG channels used were T7, F8, Cz, and P4, and the frequency bands were divided into Delta (1 4 Hz), Theta (4–8 Hz), Alpha (8–13 Hz), Beta (13–30 Hz), and Gamma (30–40 Hz). After calculating the average power per frequency band based on PSD for each condition, differences between conditions were compared using a paired t-test. Additionally, the connectivity structures between channel groups for each condition were examined through BNC analysis and full-frequency network analysis.
The results showed that, within individual frequency bands, a significant increase in alpha-band power was observed in both the Native and NonNative conditions compared to the EC condition. In the Native condition, T7 alpha and Cz alpha showed significant increases, while in the Non-Native condition, T7 alpha and P4 beta showed significant increases. Therefore, the results observed in this study were interpreted as EEG response characteristics observed in auditory language stimulus conditions relative to the EC, rather than as definitive evidence of differences between native and non-native language conditions.
In the BNC and whole-frequency network analyses, the alpha band —which was prominent in the PSD analysis—did not emerge as a key connectivity structure in either the Native or Non-Native conditions. In the Native condition, the number of strong connections decreased, but Cz-P4-centered delta and theta connectivity was maintained; in the Non-Native condition, multi-band connectivity, including T7, increased. Furthermore, in the full-frequency network analysis, the Native condition showed low-to-high-frequency connectivity between T7 and P4, as well as delta and theta connectivity between Cz and P4, whereas the Non-Native condition exhibited same-band connectivity between T7 and P4, along with delta and beta connectivity between Cz and P4. In summary, while the increase in alpha-band power during auditory language stimulation was consistent with some previous studies, the detailed network structures under native and non-native conditions differed. In particular, the increased T7-centered connectivity under the Non-Native condition suggests the possibility of additional processing burden or overload during the processing of non-native stimuli. Therefore, in the analysis of language-based EEG biomarkers, it is necessary to collect and interpret data by considering network connectivity structures in addition to simply examining active channels. However, this study has limitations, as it used data from healthy volunteers, had a limited sample size and number of channels, and did not account for educational level, measurement environment, language proficiency, hearing ability, or behavioral responses.