节奏
韵律
语音识别
计算机科学
语法
判决
语音处理
人工智能
声学
物理
作者
Johannes Gehrig,Christine Bergmann,Marie-Thérèse Forster,Christina Weismantel,Fan Bai,Marcus Czabanka,Andrea E. Martin,Antje S. Meyer,Christian A. Kell
标识
DOI:10.1523/jneurosci.2160-24.2025
摘要
The human brain must add information to the acoustic speech signal in order to understand language. Many accounts propose that the prosodic structure of utterances (including their syllabic rhythm and speech melody), in combination with stored lexical knowledge, cue and interact with higher order abstract semantic and syntactic information. While cortical rhythms, particularly in the delta and theta band, synchronize to quasi-rhythmic low-level acoustic speech features, it remains unclear how the human brain encodes abstract speech properties in neural rhythms in the absence of an acoustic signal, i.e. when speakers hold planned sentences in working memory. This study disentangles the contributions of prosodic and syntactic features in cortical rhythms during delayed sentence repetition. Using high-resolution ECoG during awake tumor surgery in the left perisylvian cortex in nine patients (five female), we show that the phase of neural rhythms with frequencies ranging from 1-48 Hz and the broadband gamma power envelope code both low-level acoustic and abstract syntactic speech features during sentence processing and retention. Syntax and prosody coding occurred in the same frequency bands, which argues against the assumption of different frequency channels for processing and representing these speech features. Our data suggest the brain leverages the phase of various neural rhythms to code both acoustic and abstract linguistic features. Significance statement Intracranial data demonstrate that both acoustic and abstract properties of speech, like prosody and syntax, are coded in the phase of neural rhythms in the left speech-relevant neocortex. The observed overlapping pattern in space, spectrum, and time suggests that neural encoding of prosody and syntax are intertwined, confirming proposals from psycholinguistics.
科研通智能强力驱动
Strongly Powered by AbleSci AI