Accessing coarticulatory information

Abstract

The speech signal is comprised of coarticulatory cues. Here, we explore whether adults’ access to coarticulatory information depends upon the particular language task. Specifically, we tested adults’ sensitivity to coarticulation when segmenting new sequences versus remembering presegmented items. Forty undergraduates were familiarized to either a continuous string of appropriately coarticulated nonsense CV syllables where every third syllable was stressed to facilitate segmentation, or to the same CV syllables presegmented into trisyllabic units without any stress contour. During the test phase, adults were presented with familiar and novel sequences appropriately coarticulated, and to both sequence types with inappropriate coarticulation. Subjects rated the familiarity of items using a seven point scale. Both groups rated familiar test sequences as more familiar than novel sequences. However, only the group presented with presegmented items demonstrated sensitivity to coarticulation. These subjects, compared to the segmentation group, significantly preferred the coarticulated items (p<0.05). These results suggest that access to information depends on the task. In the case of word segmentation, adults play attenation to the most useful properties (in this case, stress). Whereas adults in the presegmented group were able to pick up all the details of the sequences since their task only required word recognition.