EEG-based brain-computer interface (BCI) dataset for directional word recognition.

We present an EEG dataset recorded from 22 neurologically healthy volunteers (12 native Russian speakers and 10 native Spanish speakers) during overt and covert articulation of six spatial-direction words. Monopolar EEG signals were acquired from 38 electrodes positioned according to the international 10-10 system using a Neurovisor-BMM-52 (NVX) amplifier at 500 Hz. In a subset of participants, electromyography (EMG) was simultaneously recorded from the masseter muscle and laryngeal region to exploratorily characterize articulatory muscle activation. Exploratory spectral and coherence analyses, together with classification using standard machine learning methods (Random Forest, SVM, LDA), confirm the presence of condition-specific neural activity distinguishable by standard classifiers (best accuracy 78 ± 4%). The dataset is intended to support the development and benchmarking of algorithms for inner speech recognition in brain-computer interface applications.