Schedule of Events | Symposia

Probing Prediction-Related Processes in Language Using an EEG Word Stem Completion Paradigm

Poster Session E - Monday, March 31, 2025, 2:30 – 4:30 pm EDT, Back Bay Ballroom/Republic Ballroom

Hui-Sun Chiu1 (hschiu2@illinois.edu), Ryan J. Hubbard2, Kara D. Federmeier1; 1University of Illinois, 2University at Albany

Humans process words rapidly, with active prediction of upcoming words as a key mechanism. Event-related potential (ERP) sentence processing studies have delineated the consequences of prediction for later word processing, with accurate predictions facilitating semantic access (N400), and prediction violations engaging later brain mechanisms (anterior positivity). It has been harder, however, to probe prediction formation itself. To do this, we used a word stem completion task wherein participants saw word stems (e.g., bro_) that they were asked to mentally complete and were then presented with either (1) a probable completion (e.g., brother), (2) an improbable completion (e.g., bronze), or (3) a pseudoword (e.g., *brom). The word stems varied in their predictive strength towards a particular completion (constraint), measured by entropy. In Experiment 1, participants reported match/mismatch upon seeing completions, while in Experiment 2, they performed a lexical decision on the completion and then typed in their original prediction. In both experiments, we observed a reduced N400 for completions that matched or that were inflectionally or derivationally similar to the participant’s prediction. Although there was no anterior positivity effect for low entropy improbable completions in Experiment 1, a reliable anterior positivity effect emerged in Experiment 2, mirroring findings from sentences and potentially offering insights into the task sensitivity of the anterior positivity. Critically, in both experiments, ERPs measured at the word stem revealed a central-posterior negativity (600-850 ms) that was graded by entropy (more negative to high-entropy/low constraint stems) suggestive of an effect related to prediction formation.

Topic Area: LANGUAGE: Semantic

CNS Account Login

CNS2025-Logo_FNL_HZ-150_REV

March 29–April 1  |  2025

Latest from Twitter