Exploring the Attention Distribution Around Perceptual Boundaries of English Continuous Speech

In everyday life, people tend to segment real-life ongoing experience into discrete events. The same is true for perceptual segmentation of language. However, little research has examined how attention is distributed across perceptual event boundaries, specifically at the three distinct preboundary, on-boundary, and postboundary points. This study aimed to explore the distribution of attention around the perceptual boundaries of continuous speech.

A total of 26 native English speakers (16 women, 10 men; ages ranged from 19 to 29 years) were instructed to listen to and remember a series of isolated English spoken sentences where an attention (syllable) probe "ba" was embedded at preboundary, on-boundary, and postboundary points. Meanwhile, they were asked to press the key as soon as possible whenever they heard an incidental syllable "ba." A linear mixed-effects model was applied to compare response times (RTs) of "ba" at different points.

Participants showed faster RTs at postboundary points than at both on-boundary and preboundary points. That is, they allocated more attention at postboundary points (as indicated by negative correlations between RTs and attention) than either on-boundary or preboundary points, showing a low-low-high attention pattern.

Enhanced attention at postboundary points implies that event model updating might occur at these points. Thus, the well-established event boundary advantage effect in prior studies may be more closely related to intensified attention at postboundary points. Additionally, the low-low-high attention pattern has the potential to serve as an indicator of normal perceptual segmentation. This finding provides implications for future research on diagnosing atypical populations, such as schizophrenia, obsessive-compulsive disorder, Parkinson's disease, lesions of the prefrontal cortex, and Alzheimer's disease, as these populations often exhibit impaired segmentation abilities.