The role of listening effort during degraded speech recognition: A comparison of the dual-task and pupillometry paradigms

Abstract

Listening effort is an increasingly important factor in understanding how people recognize speech in adverse conditions, especially for listeners with hearing impairment (Ohlenforst et al., 2017). However, the link between this basic construct and commonly used measures of listening effort is unclear. We compare two measures: dual-task performance and pupillometry, to investigate the relationship between different techniques for measuring effort and performance recognizing vocoded speech. In the dual-task experiment, participants (n = 26) completed a non-linguistic task (pattern matching) either with or without a concurrent speech task. The same participants completed a pupil task, which tracked changes in pupil size while participants recognized words. Both tasks used natural and noise-vocoded speech at two levels of degradation to vary participants’ ability to perform accurately. We found no relationship between dual-task performance and task-evoked changes in pupil size. However, the level of degradation uniquely predicted effort in the dual task, while accuracy uniquely predicted effort in the pupil task. This suggests that different measures of listening effort are sensitive to different underlying factors, including listening conditions and individual ability. Researchers should thus be aware of the sensitivities of a given task when measuring listening effort.