Theta Oscillations Underpin Syntactic Encoding in First and Second Language Sentence Production

Successful language communication relies on efficient syntactic encoding to produce well-structured sentences, yet the real-time neural dynamics of this process, particularly in second language (L2) learners, are poorly understood. Here, we used scalp EEG to investigate the neural dynamics of syntactic encoding in first language (L1) and L2 production of Mandarin Chinese. Participants prepared sentences of varying syntactic complexity during a cued-delay period, allowing us to isolate syntactic encoding from articulation. We found qualitatively distinct neural dynamics between groups. Time-resolved multivariate decoding revealed that L1 speakers formed robust and sustained complex syntactic representations from the onset of sentence planning, whereas L2 representations were delayed and transient. This divergence was underpinned by theta-band oscillations, which flexibly tracked the subjective syntactic computational load for each group. Furthermore, L1 speakers engaged a specialized left-lateralized frontotemporal network, while L2 speakers recruited a compensatory bilateral frontal-posterior network. These results initially provide a multi-level account of syntactic encoding, demonstrating that L1 and L2 sentence production rely on fundamentally different neural computations and establishing that theta-band phase coherence is a core mechanism for building hierarchical syntactic structure.

高效的句法编码是产出结构规整语句的核心依托，亦是实现成功语言沟通的必要前提，然而该过程的实时神经动态机制，尤其是针对第二语言（second language, L2）使用者的相关机制，目前学界仍知之甚少。本研究采用头皮脑电图（scalp EEG）技术，针对母语（first language, L1）与第二语言使用者的汉语语句生成任务，探究其句法编码的神经动态过程。实验中，参与者在线索延迟时段内准备不同句法复杂度的语句，这一设计使我们得以将句法编码过程与发音过程分离开来。研究结果表明，两组使用者的神经动态存在本质性差异。时间分辨多变量解码分析表明，母语使用者从语句规划启动阶段起，便已形成稳定且持久的复杂句法表征；而第二语言使用者的句法表征则出现延迟且转瞬即逝。该差异的神经基础为θ波段振荡，其可灵活适配并追踪两组使用者各自的主观句法计算负荷。此外，母语使用者会激活专门化的左侧偏侧化额颞网络，而第二语言使用者则会募集双侧额-后部网络以进行代偿。本研究首次从多维度阐释了句法编码机制，证实母语与第二语言的语句生成依赖于本质迥异的神经计算过程，并明确θ波段相位相干性是构建层级句法结构的核心机制。