Linguistic competence in the frequency domain

Can rhythmic neural activity be driven to reveal purely internal linguistic competence? To answer this question, I record human electroencephalographic data as participants listen to isochronously delivered word stimuli composing chains of sentences, varying in phrasal complexity and grammaticality. I show that the perceived repetition of abstract syntactic categories - sentence and phrasal nodes, together - internally generates a harmonic structure of their period, independently of stimulus rate (Fugure 1). Neural harmonic structures extracted from regular trials predicted participants' grammatical sensitivity better than harmonics extracted from irregular trials, suggesting a direct modulation of grammatical competence. I also study whether this effect requires attention. Preliminary work suggests that attention is not needed for this neural signature to emerge, making it a potentially novel and robust tool for the investigation of language development in children and language competence in patient populations.

Finally, I look beyond simple sentences and attempt to ‘frequency tag’ two sentences joined by a coordinative conjunction, the most frequent discourse marker. I use complex spoken sentences of the type “Mary has a lamb and John is too tall” (nine words) and “Mary had a new lamb and John has one old friend” (eleven words). I find significant peaks for each complex sentence, with its own harmonic structure. This proves, in principle, principle that also spoken discourse perception can be studied easily and efficiently in the frequency domain.