Segmenting Visual Narratives: Evidence for Constituent Structure in Comics

We have proposed that visual narratives in comics are organized with a hierarchic narrative “grammar.” Inspired by classic “click” studies of syntax, we inserted blank “disruption” panels Before, At, or After the constituent boundaries of comic strips. In self-paced viewing, Experiment 1 found that blanks After the boundary were viewed slower than Before or At the boundary. Panels immediately following blanks were slower than corresponding panels in sequences without blank panels, but only when placed Before or After the boundary. Three ordinal panel positions following the boundary, panels following blanks At the boundary or with No-Blanks were viewed faster than panels following blanks After or Before the boundary. This supports constituency because disruptions had greater impact within, as opposed to between, constituents. Using ERPs, Experiment 2 found a larger anterior negativity to blanks within constituents (Before/After) than between constituents (At). This indicates disruptions of constituents are recognized before reaching a subsequent panel. A larger P600 appeared to blanks After the boundary than in the first constituent (Before/At). This positivity may reflect a reanalysis reflecting the inability to integrate all prior panels into a single constituent, since they are divided by the constituent boundary.