Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
How delaying the target verb resolves the N400 semantic illusion
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.
Poster C46 in Poster Session C, Wednesday, October 25, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Kate Stone1, Milena Rabovsky1; 1University of Potsdam
Unexpected words within a context elicit large N400 brain potentials. However, sometimes N400s at unexpected words are small when stereotypical agent and patient roles are reversed, such as at ‘arrested’ in ‘the cop that the thief arrested’ (“N400 semantic illusion”). The illusion can be resolved if the presentation of ‘arrested’ is delayed temporally or with neutral information (‘that evening’) [1,2], but this delay effect was only recently observed. We tested a novel manipulation of the delay’s content to probe whether an existing account of the illusion could accommodate the delay effect: In the Sentence Gestalt (SG) model, the illusion results from uncertainty between event knowledge and syntactic cues like word order [3]. Semantic association is the stronger cue and ‘arrested’ is perceived as plausible. Delaying the verb may resolve the illusion by allowing syntactic cues more time to constrain interpretation. If so, this would predict that a delay containing semantically associated information may instead sustain the illusion. We tested these predictions by manipulating the semantic association of the delay. In contrast, other accounts of the initial illusion would predict that the delay effect should not replicate at all [4:6]. Methods. Materials: German sentence pairs such as “Everyone in the train saw which.ACC/NOM fare evader the.NOM/ACC ticket inspector caught”; case marking on the determiners was swapped to reverse the stereotypical agent and patient. The verb was either not delayed, delayed with neutral words (“further up”: syntactically consistent, no semantic association), or delayed with semantically informative words (“without a ticket”: consistent + semantic association). Participants: 74 native German speakers. Analysis: Bayesian linear mixed effects model of mean amplitude from 300-500 ms after target verb onset (caught) across centro-posterior electrodes. Predictors: role order (canonical/reversed) × delay (none/neutral/informative). Predictions: An interaction of role order and delay: N400 should be larger for reversed sentences after the neutral but not the informative delay where stronger semantic association should sustain the illusion. Results. Despite a main effect of role order, estimates were not consistent with an interaction of role order and delay at the group level. However, individual estimates were surprisingly uniform: In the neutral delay condition, every reader demonstrated a small N400 increase for reversed sentences relative to the no delay condition. In the informative delay condition, most (but not all) readers showed an N400 decrease in reversed sentences relative to the neutral delay condition. Conclusions. The small but consistent effect of the neutral but not the informative delay across readers supports the SG model’s account that the N400 semantic illusion results from initial semantic-based interpretation that can be constrained by syntax over time [3]. Competition from additional semantic cues appears to outweigh the benefit provided by additional syntactic constraint. The small interaction effect size informs future experiments as large samples are needed to provide evidence for interactions [7]. References. [1] Chow et al. (2018) [2] Momma et al. (2015) [3] Rabovsky et al (2018) [4] Bornkessel-Schlesewsky & Schlesewsky (2013) [5] Brouwer et al. (2017) [6] Kuperberg et al. (2007) [7] Gelman et al. (2020)
Topic Areas: Syntax and Combinatorial Semantics, Meaning: Lexical Semantics