Hesitations appear to make listeners more likely to detect the, …er, change: Evidence for lasting effects of hesitations on listeners from change-detection experiments
Osterrieth MA 2011.doc (366.5Kb)
Item statusRestricted Access
MetadataShow full item record
Everyday speech is littered with disfluencies like filled pauses (e.g. ‘er’, ‘um’), silent pauses, repetitions or repairs. Experimental evidence indicates that even though listeners may not consciously perceive disfluencies (Lickley, 1995), they affect comprehension, both in the short and in the long term (e.g. Collard, 2009; Corley, 2010; Corley, MacGregor & Donaldson, 2007; Fox Tree, 2001). The present study aimed to explore the effects of three different types of hesitations (‘er’, ‘like’, silent pauses) on listener’s representation of the word following a hesitation using a change-detection paradigm. Participants listened to 12-20 second long passages of uttered speech, some of which contained a single hesitation in front of a critical content word. After a 500 millisecond interval they were presented with a text version of the same passage, which was either unchanged or it contained a change of the critical word to a frequency-matched semantically similar word (e.g. Graveyard changed to Cemetery). The participants were then asked to judge whether a change had occurred or not. Using counterbalanced within-subjects designs, we found that disfluencies caused people to be more likely to report changes. Consistent with numerous studies that I review which have shown that disfluencies affect processing and cause better memory for the post-disfluent word (e.g. Collard et al., 2008; Fox Tree, 2001; MacGregor, 2008), it is fair to assume that the observed disfluency effect was caused by changes in processing of the post-disfluent word due to changes in attention. Furthermore, the disfluency effect was found in all three experiments. When comparing change-detection likelihood across the three experiments, no effect of disfluency type was found, suggesting that there are no significant differences between the types of hesitations in how they affect listener’s likelihood of reporting changes. The presented findings are discussed with regard to the roles of attention and prediction in language processing.