Abstract
Words in semantically predictable sentences are more intelligible than words in less predictable or semantically anomalous sentences. However, the intelligibility benefit of semantically predictable words is reduced in noise and for non-native listeners, suggesting that semantic context contributes less to intelligibility under difficult listening conditions. The goal of the current study was to explore the effect of dialect variation on the semantic predictability benefit. Larger semantic predictability benefits were observed for more familiar dialects than less familiar dialects. In addition, more dialect differences in the size of the semantic predictability benefit were observed in conditions in which perceptual normalisation was more difficult. These results are consistent with a cue-weighting model in which dialect variation is comparable to energetic noise masking. Listeners attend less to semantic cues when perceptual normalisation for dialect variation is difficult and more to semantic cues when perceptual normalisation is easy.
Acknowledgments
I would like to thank John Pate and Haley Schram for research assistance.
Notes
1In the American variationist sociolinguistic tradition, the term “dialect” is used to denote language varieties that differ segmentally (e.g., Labov, Ash, & Boberg, Citation2006), as well as those that differ morphologically, lexically, and syntactically. In keeping with this tradition, and to avoid potential confusion with non-native or foreign accents, the regional varieties in this study are referred to as dialects.
2A close inspection of Figures 2 and 3 suggests that the semantic context effect may be reduced for the General American and Northern dialects in the blocked and control conditions relative to the unblocked condition. However, posthoc Bonferroni-corrected t-tests did not reveal any significant differences in the difference scores across blocking conditions within talker dialects.