Using hyperscanning electroencephalography to measure speech neural tracking during spontaneous parent–child conversation

This exploratory study investigates auditory and lexical neural tracking during bilingual parent–child interactions using hyperscanning EEG and acoustic analyses. EEG and spontaneous speech were recorded simultaneously from pairs of typically developing children (ages 7–10) and their parents while they performed an image description task using Diapix’ speech elicitation materials, both in English and Spanish (approximately 12 min per language). Speech was recorded via condenser microphones while the parent and child seated in separate silent rooms and communicated with each other via electrically shielded headphones. A Biosemi hyperscanning setup was used to acquire EEG from both individuals, whereas the speech acoustic envelope and lexical factors (e.g., word frequency and surprisal) were extracted from their voice recordings using a combination of language-specific corpora, automatized transcription and large language models (GPT-4). The relationship between the speech acoustic and lexical measures and the EEG signals was modeled using subject-dependent multivariate temporal response function (mTRF) analysis. Preliminary results indicate that it is possible to successfully measure acoustic and lexical neural tracking from spontaneous speech, increasing the ecological validity of neural tracking measures.