Lavi-Rotbain, O. & Arnon, I.
The Hebrew University of Jerusalem
Many studies show humans can use statistical information to segment auditory speech. However, real language includes multimodal cues (both visual and auditory) to more aspects of language, including word-object associations. Recent work by Thiessen (2010) examined how visual cues affect segmentation and found that adults show better segmentation when 'words' are regularly mapped to visual objects (e.g., 'dukame' always appears with blue star), but not when the mapping is irregular. In contrast, segmentation in 8-month-old infants was not facilitated by neither visual cue. The facilitation was thought to reflect adults' understanding that objects have labels. If so, then children should also show such facilitation. Here, we ask whether children's segmentation abilities are similarly affected by visual cues. We used the same design as in the previous study to compare segmentation in Hebrew-speaking children (mean age=9.5, N=63) and adults (mean age=24, N=58) in three conditions: only auditory, regular visual cue, irregular visual cue. As in Thiessen-2010, only the regular condition facilitated segmentation for adults. Children, in contrast, showed a different pattern: both the regular and irregular conditions facilitated segmentation. These findings show that children can benefit from multimodal cues, and suggest that the facilitation was not merely due to object-label understanding.