Filled pauses engage attention but, uh, do not seem to affect structural processing

This study asks whether filled pauses actively engage attention and whether such engagement influences structural processing, measured through structural alignment. Participants orally described a series of pictures, then listened to a short informational passage and answered questions about it, and then described another series of pictures. Participants completed this twice: The first passage had double-object targets, and the second passage had passive targets. Critically, one of the passages contained filled pauses (before the verb in target sentences), and the other did not. We measured participants’ pupil diameter in target sentences, an index of attention engagement, and pupil size synchrony, an index of moment-to-moment shared attention across observers. For speech with filled pauses compared to without, we found that pupil size was larger overall, peaking after the filled pause appeared, and higher synchrony across listeners. We interpret this as evidence that filled pauses actively engage attention. However, we did not detect a structural alignment effect in any condition, and thus found no evidence for an influence of filled pauses on structural processing (or, in another analysis, comprehension accuracy). These results suggest that filled pauses before the verb attract attention in the moment but do not influence downstream structural processing and general memory for content.