Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Endogenous Cortical Rhythms of Hierarchical Structure Building in Language Production
Poster C45 in Poster Session C, Wednesday, October 25, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
This poster is part of the Sandbox Series.
Jiachen Yao1, Ashley G. Lewis1,3, Vitória Piai2; 1Radboud University, Donders Institute for Brain, Cognition and Behavior, Centre for Cognitive Neuroimaging, 2Radboud University, Donders Institute for Brain, Cognition and Behavior, Centre for Cognition, 3Max Planck Institute for Psycholinguistics
Previous research has demonstrated the neural oscillations' ability to track quasi-rhythmic patterns (syllables, words, and phrases) during speech comprehension (Ding et al., 2016; Kaufeld, 2020; Meyer, 2018). Since syntactic phrases are not directly encoded in speech signals, it was proposed that an internally generated rhythm might feed the acoustic speech entrainment with syntactic knowledge. Yet the correspondence between this rhythm and prosodic boundaries (physically marked in speech), has been questioned (Glushko et al., 2022). Moreover, previous studies focused on linear word groups without embedding, raising concerns about capturing nested syntactic structures with a single oscillatory band (Kazanina & Tavano, 2023). Our study aims to test these ideas in language production, where hierarchical structure building may be imposed on the production output in a top-down manner, given the absence of speech features for entrainment. We recorded participants’ EEG responses while they typed sentences of different embedding depths, following specific picture configurations on a computer monitor. Typing was chosen over speaking to avoid potential entrainment confounds to self-monitoring during speaking, while still requiring typical production stages prior to motor execution (Rumelhart & Norman, 1982). We contrasted between Dutch subordinate (e.g., Dutch equivalent of [[the woman] [yells [that [[the man] cries]]]]) and coordinate (e.g., [[[the woman] yells] [and [[the man] cries]]]) clauses. Although they share a similar linear order of input, subordinates involve additional syntactic depth of embedding (indicated by bracketing). We hypothesized that producing subordinates results in increased neural entrainment at the rate of phrase-structure building compared to coordinates. Participants were instructed to start typing immediately after picture presentation to minimize variability in planning time prior to typing onset. Keystrokes were time-logged for synchronization with EEG recordings. Data from 18 participants have been collected. Typed strings were annotated in two ways. Firstly, each keystroke opening a phrasal bracket was coded as 1, while the rest as 0 (bracket presence), reflecting linear concatenating. Secondly, each keystroke was coded based on the number of phrases it opened (bracket count), reflecting hierarchical embedding. Two phrase rates were obtained by dividing annotation values by the duration of each trial, and ± 1 standard deviation of each mean rate was taken to form the phrase bands. After preprocessing and artifact removal/attenuation, the EEG signal was band-pass filtered separately through each band, and then Hilbert-transformed to extract instantaneous phase and power. We computed Gaussian Copula Mutual Information (MI) (Ince, et al., 2017) between the resultant time series and the annotated time series, separately for each electrode, participant, and condition to quantify the degree of phrase rate tracking. Our hypothesis predicted higher MI values for subordinates at the phase rate reflecting embedding (bracket count) if the brain employs rhythmic patterns to encode nested relationships during structure building in language production. Conversely, if structure building relies solely on linear concatenation of phrases, we expected higher MI values for coordinates at the phrase rate of bracket presence. Significance of MI values will be determined with cluster-based permutation tests (Marisand & Oostenveld, 2007). Preliminary analysis has confirmed such trend in our data.
Topic Areas: Syntax and Combinatorial Semantics, Language Production