Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Readers extract syntactic and semantic information during Rapid Parallel Visual Presentation (RPVP) sentence reading: Evidence from event-related potentials (ERPs)
Poster Session D, Saturday, September 13, 5:00 - 6:30 pm, Field House
Sofia E. Ortega1, Jonathan Grainger2, Katherine J. Midgley1, Karen Emmorey1, Phillip J. Holcomb1; 1San Diego State University, 2CNRS and Aix-Marseille Université
The extent to which readers can extract semantic and syntactic information in parallel from short sentences presented as a single stimulus (Rapid Parallel Visual Presentation or RPVP) is a hotly debated topic. Most evidence both supporting and opposing parallel processing comes from behavioral studies where participants make retrospective judgments about well-formed sentences vs. the same words in a non-canonical order. Both response accuracy and speed differ by sentence type which is interpreted as evidence that readers rapidly extract and use high-level structural information to influence their judgments. However, Staub (2024) argues that such results may reflect sophisticated guessing strategies, whereby participants use elements of the sentence context to make educated guesses rather than engaging in parallel processing. This explanation is plausible given that sentence judgements occur long after stimulus presentation, giving participants time to at least partially buffer and then process the words serially. To examine whether parallel processing is reflected by online neural processing, we combined event-related potentials (ERPs) and RPVP sentence reading. The temporal precision of ERPs allows us to determine whether parallel processing occurs early (within a few 100 milliseconds after sentence presentation) or later during the decision stage. Well-formed sentences (“The surfer rides a wave.” or “The zombie eats the brains.”) were compared to those containing one of four types of violations: (1) a syntactic verb agreement violation ( “The surfer ride a wave.”), (2) a syntactic word order violation (“The surfer a rides wave.”), (3) a semantic verb anomaly (“The zombie sows the brains.”), or (4) a final word semantic anomaly (“The zombie eats the escape.”). Twenty-four English-speaking participants viewed five-word sentences presented for 200ms and performed an acceptability judgment task (Is this a good or bad sentence?). Single-trial ERP responses were analyzed in four consecutive 200ms windows between 400–1200ms post-stimulus, encompassing the N400 and P600 time windows. N400 and P600 mean amplitudes were analyzed using separate LMER models to examine the anomaly effects for each sentence type. Additionally, scores from spelling and reading comprehension tests were included as continuous variables. While the N400 and P600 ERP components have been shown to be sensitive to semantic and syntactic anomalies in numerous rapid serial visual presentation (RSVP) studies, they have not yet been examined in the RPVP paradigm. We predicted that if either component showed differential responses to anomalous versus control sentences, this result would be incompatible with a serial guessing strategy given that the type and location of the single-word change varied across the sentence conditions. We found robust N400 effects for BOTH middle and final word semantic anomalies, as well as word order anomalies. We also found P600 effects for verb agreement and word order violations. Both syntactic anomaly effects were modulated by reading ability, with better readers showing larger N400 and P600 effects. Since participants could not anticipate anomaly locations, these findings are incompatible with a simple guessing strategy and suggest that readers can extract and process semantic and syntactic information in parallel during RPVP sentence reading.
Topic Areas: Reading, Methods