Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Isolating Syntax with Frequency-Tagged Verbs: An Intracranial Study
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Adam Morgan1, Orrin Devinsky1, Werner Doyle1, Patricia Dugan1, Daniel Friedman1, Adeen Flinker1,2; 1NYU Grossman School of Medicine, 2NYU Tandon School of Engineering
Syntax, the abstract structure of language, enables humans to convey an unlimited number or complex thoughts. Despite its centrality, its neural instantiation remains poorly understood. In particular, there is ongoing debate regarding its localization, with some studies pointing to a hub in inferior frontal gyrus (IFG; e.g., Menenti et al. 2012; Zaccarella & Friederici 2015), others to posterior temporal regions (Matchin et al. 2020; Flick et al. 2020), and still others to a broadly distributed, non-localizable code (Ding et al. 2016; Blank et al. 2015). Progress has been hindered by two major obstacles: First, the trade-off between spatial and temporal resolution in non-invasive measures like fMRI and EEG. Second, is the difficulty of experimentally isolating syntax (see, e.g., Sprouse & Lau 2013; Murphy 2024). While concrete representations like phonemes or words can be directly activated via simple exposure, syntax is abstract — a property of multiple words — making direct exposure impossible. Studies must therefore manipulate syntax indirectly, using multiple words, as in a common paradigm that contrasts syntactically-structured strings of words (e.g., “red boat”) and unstructured word lists (“red, boat”). However, such manipulations typically introduce confounding processes that must be accounted for, such as word processing and combinatorial semantics. This project addresses these limitations by combining (1) intracranial sEEG recordings, which offer both high spatial and temporal resolution, and (2) an innovative paradigm that uses single words to target particular syntactic frames. Verbs are especially well-suited for this, as they are linked to specific argument structures (e.g., “dine (*bread)” vs. “devour *(bread)”). These abstract syntactic frames can be activated by presenting just the verb, a process which is apparently automatic as it has been documented even when verbs are presented subliminally (Trueswell & Kim 1998; Berkovitch & Dehaene 2019). Here, we exploit this property to elicit activation of three syntactic frames — intransitive, transitive, and clausal complement — by visually presenting neurosurgical patients with verbs selected for strong bias toward each frame. Verbs were shown at 1Hz in six 4-minute blocks. In odd-numbered blocks, every 3rd verb was intransitive (i.e., 1/3Hz), every 5th verb transitive (1/5Hz), and every 7th verb clausal (1/7Hz), whereas in odd blocks, these frequency tags were rotated (intransitives at 1/5Hz, transitives at 1/7Hz, and clauses at 1/3Hz). Thus, an electrode in a location that represents the transitive frame should show spike in spectral power at 1/3Hz in odd blocks but 1/5Hz in even blocks. At the time of submission, data from one patient have been collected (and we anticipate running at least three additional patients by SNL 2025). Preliminary analysis reveals the expected pattern of signifiant peaks in power (calculated from local field potentials) in one electrode in the left supramarginal gyrus — adjacent to posterior temporal areas implicated in hierarchical syntax by recent models (Matchin & Hickok, 2020). These results provide early evidence for frequency-tagged activation of abstract syntactic representations using single-word stimuli and high-resolution neural data.
Topic Areas: Syntax and Combinatorial Semantics,