Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Quick, Don't Move!: Wh-Movement and Wh-In-Situ Structures in Rapid Parallel Reading – EEG studies in English, Urdu, and Mandarin Chinese
Poster Session B, Friday, October 25, 10:00 - 11:30 am, Great Hall 3 and 4
Hareem Khokhar1,2, Jill McLendon1, Donald Dunagan1, Zahin Hoque1, Dustin Chacón1,2; 1University of Georgia, 2University of California Santa Cruz
[INTRODUCTION] A major question in the neurobiology of language is how the brain represents syntactic structure. This is challenging, because surface properties of languages may obscure underlying similarities. Linguistic theory posits that wh-in-situ constructions (Mandarin Chinese, ~'she read which book?') and wh-movement constructions (English, 'which book did she read?') share abstract properties, despite superficial differences (Huang 1982), and engage similar memory retrieval mechanisms (Santi & Grodzinsky 2007; Xiang et al. 2015). Here, we present EEG data collected in two wh-in-situ languages, Urdu and Mandarin Chinese (MC) and one filler-gap language, English. Contrasting with previous studies, short sentences were presented in parallel, centered in the fovea and read in one fixation, partially mitigating memory demands of word-by-word reading. Some grammatical features are distinguished by EEG/MEG responses in parallel presentation (Snell & Grainger 2019; Flower & Pylkkänen 2024; Krogh & Pylkkänen 2024). Our results show that brain activity distinguishes wh- vs. NP-object constructions in all languages within 800ms, but we failed to find a clear cross-language uniform response to wh- vs. NP-structures. [METHODS] Participants (N = 35 English; 33 Mandarin; 27 Urdu) were given a sentence-matching task (Flower & Pylkkänen 2024); they saw a target sentence followed by a memory probe test sentence, with brain activity continuously recorded using a 64-channel EEG. Target sentences were presented for 200ms followed by 600ms of blank screen. Participants decided whether they matched via key press. Stimuli were 64 sets of questions, manipulating Subject Type (Wh, NP) and Object Type (Wh, NP). [RESULTS] [Behavioral] In all languages, memory probe task performance was lower for wh- compared to NP-object constructions (ps < 0.05), but observed no effect of wh- vs. NP-subjects. Memory probe task performance was lower for wh-subject, wh-object (ps < 0.05), suggesting an additional cost for 'multiple wh-' structures. [EEG Sensor Data] Analyses were conducted independently per language. Spatio-temporal cluster-based permutation tests (Maris & Oostenveld 2007) were conducted with ANOVAs Subject Type × Object Type, from 200–800ms post-sentence onset, with p<0.05 for English and MC p<0.2 for Urdu, 20ms and 3-sensor minimum clustering parameters. For English, wh-objects elicited a positivity 380–486ms and 500–557ms over right posterior sensors (ps < 0.05); for Urdu, wh-objects elicited a negativity 558–800ms over right fronto-parietal sensors and frontal midline sensors; for MC, we observe a bipolar response, with positivity 200–266ms in bilateral fronto-parietal and anterior midline sensors with a negativity 200–227ms in right posterior sensors, followed by a 311–389ms positivity over anterior midline sensors. Wh-subjects only elicited a reliable response in English: a negativity in left lateral sensors 294–320ms. [CONCLUSION] We sought to identify the neural bases of wh-constructions, abstracting away from superficial features of individual languages (wh-in-situ vs. wh-movement) and the processes deployed in word-by-word reading. In English, Urdu, and MC, neural activity was sensitive to wh- vs. NP-objects, although the topographic distribution, polarity, and timing of the effects were not uniform. Nonetheless, some patterns emerge – right posterior sensors respond to wh-objects in English and MC, and midline anterior sensors to wh-objects in Urdu, MC.
Topic Areas: Syntax and Combinatorial Semantics, Reading