Chinese Semantic-Syntactic Violation ERP

Purpose & Design: This dataset is designed to investigate the neural correlates of language processing, specifically the N400 (semantic) and P600 (syntactic) event-related potential components elicited by violations in Chinese sentences. It serves as a benchmark for developing and evaluating brain-computer interfaces (BCIs) and computational models grounded in language cognition. Participants: Thirteen healthy, right-handed native Chinese speakers (11 male, 2 female), aged 22–26 years, all with undergraduate education, were included in the final analysis. Experimental Paradigm & Conditions: Sentences were presented in a word-by-word manner (400 ms per word, 100 ms blank interval). Each participant completed trials under three conditions: (1) Correct sentences; (2) Sentences with local syntactic violations; (3) Sentences with semantic violations. Each condition comprised 64 trials per participant. Data Acquisition: EEG System: Brain Products actiCHamp Plus amplifier with an actiCAP slim/snap 64-channel electrode cap. Montage: Electrodes were placed according to the international 10–20 system. Sampling Rate: 500 Hz. Event Markers: Stimulus onset: S1 (correct), S2 (semantic violation), S4 (syntactic violation). Behavioral response: S7 (correct button press), S8 (incorrect button press). Preprocessing Pipeline (Recommended & Applied for Analyzed Data): The released data is provided in a raw format. For the results presented in this study, the following standard preprocessing pipeline was applied using EEGLAB: a.Band-pass filtering between 0.1 and 60 Hz. b.Ocular and muscular artifact removal via Independent Component Analysis. c.Re-referencing to the average of TP9 and TP10. d.Epoch extraction from -150 ms to 1000 ms relative to critical word onset, with baseline correction using the [-150, 0] ms window. Data Structure & Usage Notes: The dataset is organized by participant ID. Researchers are advised to exclude the data from the participant with noted hardware issues (Subject ID: 250102) prior to analysis. The provided preprocessing steps are recommended for reproducibility. This dataset is suited for research on cross-subject ERP analysis, domain generalization in BCIs, and as a testbed for neural decoding algorithms.