Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Exploring neural correlates of discourse topic prediction via discourse markers
Poster Session C, Saturday, September 13, 11:00 am - 12:30 pm, Field House
Mercedes Villalobos Cardozo1, Hatice Zora2, Peter Hagoort2; 1University of Louvain, 2Max Planck Institute for Psycholinguistics
Discourse markers (DMs, henceforth) are widely accepted across frameworks as serving key interactional functions, among them topicality management in interaction, aiding speakers in navigating the conversational flow and discourse progression. However, the body of evidence supporting the claim that these devices serve as cues to signal maintenance of discourse continuity or topic shift to the interlocutor is primarily theoretical and corpus-based, providing only limited insights into the cognitive mechanisms involved. This study provides neural evidence for the processing of three turn-initial Spanish DM configurations: “y” (‘and’), perceived lengthened “y” (‘and’), and “bueno y” (‘well and’) and their alleged role as cues for continuity (the former) and discontinuity (the latter two). Participants (n = 30) of both Latin American and Peninsular varieties underwent EEG recording while passively listening to short three-turn dialogues (6 lists × 60 trials, Latin-square design), with critical utterances either continuing on the same topic or introducing a new topic, each preceded by one of three target DMs. Statistical analyses were conducted, focusing on the typical central-parietal topographical distribution of the N400 component in two time windows: 450–650 ms post-stimulus onset and 700–950 ms post-critical word onset. The first analysis, assessing core processing effects of DMs, revealed a significant main effect of DM (p < 0.001). Pairwise comparisons showed significant differences between “bueno y” and “y” (p < 0.001), and between “bueno y” and the perceived lengthened “y” (p < 0.001), but not between “y” and its perceived lengthened version (p = 1.000). The second analysis explored cognitive load in response to matching or mismatching expectations of discourse topicality set by the initial DM. No significant main effects of DM (p = 0.594), topic (p = 0.576) or DM × topic interaction (p = 0.814) were observed. Results from the first time window suggest that these devices encode specific pragmatic information and processing instructions, therefore generating distinct discourse expectations that translate into different processing efforts. Mean amplitudes of DMs revealed “y” and perceived lengthened “y” elicited similarly reduced N400s, aligned with continuity processing, whereas “bueno y” elicited larger negativity, suggesting greater processing demands, likely prompting the anticipation of topic shifts. Findings for the second time window suggest that topic shifts, while detected, may not consistently disrupt local semantic integration processes. One possible explanation is that the topic shifts presented were not disjunctive enough to elicit pronounced N400 responses when content contradicted expectations set by the markers. Overall, our results suggest that DM processing primarily modulates early predictive mechanisms, setting up initial expectations, while continuations are flexibly incorporated into the evolving discourse model without incurring costly processing efforts. This dissociation between early DM effects and later integration challenges the theoretical assumption that the studied DMs directly signal topic (dis)continuity in upcoming discourse.
Topic Areas: Meaning: Discourse and Pragmatics,