Accent detection is a slippery slope: Direction and rate of F0 change drives listeners' comprehension

Abstract

The present study tests whether listeners use F0, duration, or some combination of the two to identify the presence of an accented word in a short discourse. Participants' eye movements to previously mentioned and new objects were monitored as participants listened to instructions to move objects in a display. The name of the target object on critical trials was resynthesised from naturally produced utterances so that it had either high or low F0 and either long or short duration. Fixations to the new object were highest when there was a steep rise in F0. Fixations to the previously mentioned object were highest when there was a steep drop in F0. These results suggest that listeners use F0 slope to make decisions about the presence of an accent, and that F0 and duration by themselves do not solely determine accent interpretation.

Keywords:

• Prosody
• Discourse
• Pitch accent
• Comprehension

Acknowledgements

This work was supported by NIH Grant R01DC008774-01A1. The first author was supported by NIH Grant T32HD055272. We would like to thank our undergraduate assistants Kathleen Eberhardt and Ahmad Parvinian for help with stimulus preparation and data collection.

Notes

¹F0 values were measured across the entire critical word including both mono- and disyllabic words. Visual comparisons of the F0 contours collapsed across mono- and disyllabic words separately reveal no difference in contour shape. Separate analyses of the fixation data for mono- and disyllabic words show the same pattern of results, so they are collapsed in the results section.

²Although we find a new bias in this data, replicating Dahan et al. (2002), Arnold (2008) did not find this bias using a slightly different task. This suggests that this bias may be due to task demands rather than a general listener preference to fixate the new item within a contrast set.