Poster C65, Sunday, March 26, 5:00 – 7:00 pm, Pacific Concourse
Reading naturalistic text alters the information processing timeline: Evidence from concurrent self-paced reading and electroencephalography
Shannon McKnight1, Albert Kim1; 1University of Colorado, Boulder
We recorded brain activity during word-by-word reading of short factual narratives on various science related topics in a self-paced reading paradigm. This approach departs from most previous EEG research on language processing, which has focused on the processing of individual sentences or single word recognition, with particular design constructions, at controlled presentation rates. However, EEG and ERP are especially useful in evaluating the neural processes underlying reading in more naturalistic paradigms due to millisecond-by-millisecond recording of scalp potentials over a number of different locations. By allowing people to control the rate of presentation, we created a naturalistic reading setting. Based on previous work in sentence processing and word recognition, we expected to observe a progression in word-onset-locked ERP sensitivity from low-level visual characteristics to high-level semantic and contextual characteristics over time. For example, we expected to see correlations between P1 activity, length, and visual complexity, and correlations between N400 activity and lexical frequency. We did observe correlations between P1/N170 activity and length. However, we observed correlations between lexical frequency and the earlier P2 component with only a marginal effect of frequency at N400. We hypothesize that the temporal dynamics of word recognition are altered in more natural reading settings. This hypothesis is based on the idea that top down contextual information increases in short stories relative to individual sentences or words. More work is needed in order to leverage language models of context to better examine the impact context can have at individual word recognition during reading.
Topic Area: LANGUAGE: Other