Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Pause duration and sentence position modulate ERP responses during spoken language processing
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
Annika Andersson1, Hanna Lindfors1, John E. Drury2, Eric Pakulak3, Kristina Hansson4; 1Linnaeus University, 2Independent researcher, 3Stockholm University, 4Lund University
Pauses are a pervasive feature of spoken language and typically appear at boundaries at the sentence, phrase, and word level. The present ERP study investigated how pause duration and position within a sentence modulate neural responses during spoken sentence processing. Native speakers (N = 20, aged 18–35 years) listened to 500 Swedish spoken sentences presented as connected speech while watching 5 claymation movies. Sentences were recorded by an actress with a storytelling voice. She was instructed to speak with natural prosody and natural pauses at phrase boundaries. Half of the sentences contained a manipulation of semantics, morphosyntax, or pause length. Here we present analyses of the pause manipulations. Of the 500 sentences, 150 experimental sentences contained a natural pause (corrected to 400 ms) and a prolonged pause (1600 ms) inserted either at a clause boundary or within the first or the second constituent. While all pauses were inserted at phrase boundaries, within-constituent pauses occurred in either sentence-initial main clauses (WC1) or in sentence-final subordinate clauses (WC2). Between-constituent (BC) pauses occurred at the boundary between main and subordinate clauses. Based on previous research on processing of comic strips with inserted blank panels at the corresponding three positions (Cohn et al., 2014), pauses within constituents were compared to pauses inserted at BC, and to each other. Responses to pauses were time-locked to pause onset using a -200 to 0 ms baseline. Natural pauses (400 ms) at all three sentence positions (WC1/BC/WC2) elicited a positivity consistent with a Closure Positive Shift (CPS; Steinhauer et al., 1999; 500-700 ms, Pause: F(2,38) = 5.704, p = .009, ηp2 = .23). The responses elicited for pauses within constituents did not differ from each other but were larger when at the clause boundary, i.e., at the BC position (500-700 ms: BC vs WC1, Pause: F(1,19) = 4.764, p = .042, ηp2 = .20; BC vs WC2, Pause: F(1,19) = 13.204, p = .002, ηp2 = .41; WC1 vs WC2 all ps > .463). Importantly, sentence position also modulated responses to prolonged pauses (1600 ms). Specifically, prolonged pauses in the WC1 position elicited a sustained fronto-medial negativity, while a sustained centro-parietal positivity was elicited at the BC and WC2 positions (700-900 ms: Pause x AntPost, p = .045, ηp2 = .13; 900-1100 ms, Pause x AntPost, p = .028, ηp2 = .14; 1100-1300 ms, Pause, p = .008, ηp2 = .23, Pause x AntPost, p = .012, ηp2 = .16; 1300-1500 ms, p < .001, ηp2 = .22; 1500-1600, Pause x AntPost, p = .001, ηp2 = .21). Consistent with other prior work, our findings show that listeners are sensitive not only to phrase boundaries coinciding with the presence of pauses in spoken language, but also to pause duration and position within the sentence structure. Pauses thus appear to dynamically engage boundary detection and predictive mechanisms during spoken language processing.
Topic Areas: Speech Perception, Syntax and Combinatorial Semantics