Neural mechanisms of structural inference: An EEG investigation of linguistic phrase structure categorization

A key component of language comprehension is structural inference, which involves computations that determine the category of phrases, such as noun phrases (NP) and verb phrases (VP). Although this mechanism is crucial in the interpretation of linguistic structures by conferring distinct semantic and distributional properties, identifying its neural correlates is still challenging, as modulations of phrasal category usually alter lexical content. Here, we approached this topic by exploiting the grammar of Mandarin. We collected scalp electroencephalography (EEG) data while participants read Mandarin NPs (e.g., 'the wall painted white') and VPs (e.g., 'paint the wall white'), which differ prominently in their 'headedness'. We discovered significant robust theta and alpha power increase at left centro-parietal sites in NPs relative to VPs around the presentation (theta: ∼ -70 to 190 ms; alpha: -470 to -50 ms) of the nouns (e.g, 'wall'), where only the nouns in NPs project a head. Due to the timing of these effects and hypotheses that we leverage from an existing neurocomputational framework for structure-building (ROSE), we interpret these oscillatory signatures as reflecting structure-related computations that contribute to the inferences concerning the two linguistic structures. These effects may also reflect some semantic computations that contribute to the interpretations of NPs and VPs at the syntax-semantics interface. In addition, both NPs and VPs relative to one-word phrases exhibited spatiotemporally consistent low-frequency oscillatory modulations, consistent with previous studies. Overall, our findings offer novel constraints on neurocomputational accounts of structure-building through these previously undocumented signatures of structure-related computations.