Search Abstracts | Symposia | Slide Sessions | Poster Sessions
The N400 is not blind to truth-value violations in predictable negated contexts
Poster Session B, Wednesday, September 30, 4:30 - 6:30 pm, Wangari Maathai
Sara Farshchi1; 1Lund University
Negation has long been argued to involve delayed or non-incremental processing, with some accounts proposing that comprehenders initially represent the affirmative meaning before integrating negation. Recent ERP evidence, however, suggests that negation can be processed incrementally in highly constraining contexts. In Farshchi and Paradis (2026), predictable negated information elicited reduced N400 amplitudes, indicating facilitation for highly expected continuations in negated sentences. Nevertheless, because the anomalous continuations in that study were not truth-value violations, it remained possible to argue that the observed effects were driven primarily by semantic association rather than by the integration of negation itself. The present EEG study directly addresses this issue by using antonym pairs to manipulate truth value while minimizing semantic similarity differences between conditions. The study used 84 sets of materials adapted from Farshchi and Paradis (2026). Only items in which the predictable negated concept had a clear antonym were included. Three conditions were created: (1) a true continuation in a highly constraining negated context (e.g. awake), (2) a false continuation in the same highly constraining negated context created using the antonym of the expected adjective (e.g., asleep), and (3) a true continuation in a low constraining negated context (e.g., awake). For example, in a context describing a patient deeply asleep after surgery, “the patient was not awake” constituted the true condition, whereas “the patient was not asleep” constituted the false condition. The low constraining condition was included to examine whether reduced contextual predictability would modulate N400 amplitudes for true negated continuations. By using antonym pairs, the study manipulated truth value while keeping semantic similarity between conditions to a minimum. This is because antonyms map onto the same conceptual domain and this way, the semantic content associated with the two antonym adjectives is the same. This design allowed us to test more directly whether readers detect truth-value violations in negated contexts. The stimuli were distributed across three experimental lists containing 126 trials each. EEG data was collected from 32 channels using Neuroscan Easycap. Preliminary ERP results from 15 participants so far show a larger N400 for false compared to true continuations in highly constraining negated contexts. This provides evidence for the integration of negation, as indicated by the detection of truth-value violations. Crucially, because the true and false continuations consisted of antonymic pairs with closely matched semantic similarity, the N400 difference between them is difficult to attribute solely to semantic association effects. In addition, true continuations in low constraining contexts elicited larger N400 amplitudes than true continuations in highly constraining contexts, suggesting facilitated processing for predictable negated information and increased semantic integration costs when the negated concept is less strongly supported by context. These findings provide preliminary direct evidence that comprehenders integrate negation incrementally in highly constraining contexts and are sensitive to truth-value violations under negation. By manipulating truth value through antonymic oppositions, the present study offers a stronger test of negation integration than previous ERP studies and contributes to ongoing debates regarding the relationship between semantic association, predictability, and truth-value computation during negation processing.
Topic Areas: Meaning: Discourse and Pragmatics, Meaning: Lexical Semantics