Presentation

Weiqi Wang¹, Charles Perfetti¹, Anne Helder²; ¹University of Pittsburgh, ²Leiden University

Readers build a situation model to represent text information, updating the model incrementally. Event structures—the temporal order, location, and causal relations of activities—are central to situation models and our focus here. ERP studies of word-to-text integration, linking the current word to the prior text, provide a window on these incremental processes. Both the N400 and P600 have been found to reflect these processes. A reduction in the N400 that depends on the presence of an antecedent has been interpreted as indicating the ease of integration. However, this interpretation is challenged by data showing that meaning priming, rather than integration, is responsible for the N400 reduction. These results are based on simple two-sentence texts with repetitive event structures and event-defining words, and strong priming situations (Delogu et al., 2019). Priming is powerful in the N400 and could mask effects of other aspects of text processing. Our study aims to examine whether the N400 can indeed reflect integration processing when longer story-like texts are used, and priming is reduced. To achieve this goal, we developed 4-5 sentence narrative texts and controlled the integration opportunity by manipulating the event structure in two ERP experiments. The second experiment extracts the two sentences from the story that convey the event structure in order to separate the effects of a longer, more engaging narrative from other effects produced only by whether the event structure is congruent or incongruent. We recorded EEGs of college students while they read passages presented as whole sentences (1st and final sentences) or word by word (two middle sentences). We compared ERP components on nouns (e.g., menu) in various positions in the second of these two middle sentences across three conditions: 1) the preceding sentence has an antecedent (e.g., restaurant), and event structure information is congruent; 2) the preceding sentence has an antecedent (e.g., restaurant), but event structure information is incongruent; 3) the preceding sentence describes event-unrelated activities. Critically, 1) and 2) contain exactly the same priming potential and differ only on whether the protagonist has entered (congruent) or left (incongruent) the restaurant when “opened the menu” is read. Thus, only if the N400 is sensitive to event structure congruence should a difference in N400 be observed. In the second experiment, different participants will read the middle two sentences of stories of the first experiment with ERPs recorded. Data collection is complete on the first experiment with analysis in progress. The second experiment will be completed this summer. If a robust N400 effect is found between the first two conditions describing related events (1 vs.2), we conclude that the N400 can reflect mental-model-based integration beyond mere meaning access. A failure to find an N400 difference between these two conditions (provided both differ from condition 3) coupled with a P600 difference will add evidence to the argument that only the latter reflects integration. Results from the current two ERP experiments will allow a more precise interpretation of ERP components during text reading.

Topic Areas: Reading, Meaning: Lexical Semantics