Linguistic Focus Promotes the Ease of Discourse Integration Processes in Reading Comprehension: Evidence From Event-Related Potentials

Abstract

Psycholinguistic studies of focus processing have yielded varying results regarding how focus affects language processing. We report the results of an event-related potential (ERP) experiment that used question-answer pairs in a discourse to manipulate whether a target word was contextually focused, contrastively focused, contextually defocused, or contextually neutral. We found a negative-going waveform that was sustained in the time-course (250-800 ms after the target word onset) with a maximum over frontal-central scalp sites. As the structure of the discourse made the target word more focused, the negative-going deflection was systematically reduced. We also observed a frontal positive-going waveform that was larger for the focus-marked words relative to the neutral target word in an earlier time window (150-250 ms, P200), which may reflect increased attention allocated to the focused items. We propose that the reduced negative ERPs for the focused words reflects facilitation of meaning integration when focus functions to establish reference in the discourse representation. This can be attributed to extra attention paid to the focus-marked items that in turn promotes the prominence of focus-marked referent and prompts the contextual priming mechanism that facilitates the access of propositionally relevant items in text memory during reading.

Keywords: ERPs; P200; contrastive focus; focus processing; information structure; informational focus.

Figure 1

Grand average ERPs elicited by target words as a function of experimental conditions plotted for the midline electrode clusters (Fz, Cz, Pz). On the y-axis, positive amplitude values in μV are plotted upward.

Figure 2

Interaction of the Information × Electrode for the early P200 effect. Mean amplitudes of each condition were averaged across the midline and lateral clusters for each frontal, central, and parietal site from 150–250 ms after the target word onset. The y-axis indicates the amplitude values in μV, positivity upward. Error bars show the standard error of the mean.

Figure 3

Interaction of the Information × Electrode for the sustained negative ERPs of (A) the midline ANOVA and (B) the lateral ANOVA. The y-axis indicates the amplitude values in μV, positivity upward. Error bars show the standard error of the mean.