Stimuli, Tasks and Task Order.

Our design had a 3 x 2 design with Structure (WordList, SentenceList, Story) and Speed as factors (Fast and Slow). We additionally had a chunk recognition task and a self-reported comprehension score.

Average behavioral performance per condition for MEG sample (n = 33).

Behavioral results for MEG experiment (n=33).

Comprehension scores and Task accuracy were higher in the Story condition than in the SentenceList condition (Accuracy: z = -3.32, p = .003; Comprehension: t = -13.33, p < .001), in the Story compared to the WordList condition (Accuracy: z = -8.75, p < .001; Comprehension: t = -37.41, p < .001) and in the SentenceList compared to the WordList condition (Accuracy: z = -5.47, p < .001, Comprehension: t = -24.08, p < .001). Reaction times were faster in both the Story (t = 17.314, p < .001) and SentenceList (t = 18.462, p < .001) conditions vs. the WordList condition. Significant differences between Fast and Slow presentation within one level of structure is indicated by a bolded asterix above the Slow condition. Asterix legend: . < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Comparative summary of timecourse and spatial extent of significant clusters across Slow and Fast presentation modes.

Slow results (in purple) have been placed above Fast results (in pink) on the x-axis.

Left hemisphere results for Slow and Fast conditions.

Fast results (in pink) have been placed to the right of their Slow presentation equivalent (in purple) where possible. a-d: Individual timecourses, mean amplitude across conditions and spatial extent of significant clusters. Asterix legend: . < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Right hemisphere results for Slow and Fast conditions.

Results for Fast presentation (in pink) have been placed to the right of their Slow presentation equivalent (in purple) where possible. a-e: Individual timecourses, mean amplitude across conditions and spatial extent of significant clusters. Asterix legend: . < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

GLM model main effects.

Trial# indexes block progression in all conditions but additionally tracks narrative progression only in the Story condition, which serves as the reference level in the model. A: Spatial summaries of β t-maps for each regressor.

GLM interaction results showing increases in SentenceList and WordList conditions but not in Story.

A: Predicted cluster amplitude for the Condition:Trial# interactions, averaged over significant vertices and timepoint for each cluster. 95% confidence intervals across subjects are indicated by a shaded bar. Interactions effects are indexed by changes in SentenceList and WordList relative to the Story baseline and the Story condition is plotted for reference. The same vertices and timepoints used to plot the conditions of interest (WordList, SentList) were selected to index the effect of Trial# in the Story condition. The significance of each interaction has been placed next to the condition label. If a condition had multiple clusters, the largest p-value has been indicated. B: Spatial summary of WordList:Trial# and SentenceList:Trial# effects.