Background: Many of our conversations occur in non-ideal situations, from the hum of a car to the babble of a cocktail party. In conversation, listeners are required to switch their attention between multiple talkers, which places demands on both auditory and cognitive processes. Speech understanding in cocktail party situations appears to be particularly demanding for older hearing-impaired listeners. As such, this study examined the relation of age and hearing ability on performance in a speech in noise talker switching task.

Methods: Older adults (65-81 years) with a range of hearing abilities, and younger adults (21-30 years) with normal hearing, took part in an online speech recall task using the coordinate response measure (CRM) corpus. For each trial, two utterances were presented one after the other, analogous to a conversational turn switch. The first target sentence was presented in quiet, and the second target sentence was masked either by noise (steady-state speech shaped noise) or speech (another CRM sentence). The two target sentences were either spoken by the same voice or different voices.

Results: Relative to conditions in which the target talker remained the same between sentences, participants were less accurate when the target talker changed, particularly when the original target talker became a masker for sentence 2. Listeners with poorer speech-in-noise reception thresholds (assessed via a digit triplet test) were less accurate at recalling target information in the second sentence in both noise and speech masked trials, and made more masker confusions in speech masking trials (i.e., erroneously reporting masker information instead of target information). An interaction between switch type (i.e., same or different talker) and speech-in-noise thresholds revealed that participants with poorer hearing in noise received less benefit from the presence of a consistent speaker. Additionally, worse ability to distinguish between talkers (lower accuracy on Bangor Voice Matching Test) was associated with decreased accuracy and increased masker confusions in speech masking trials.

Conclusions: Our findings replicate those reported by Lin and Carlile (2019, doi:10.1038/s41598-019-44560-1) regarding the cost of following a switch in talker, extending these earlier findings to older adults with a range of hearing abilities. Furthermore, we demonstrate that greater difficulty in speech recall after a turn change relates both to reduced audibility (i.e., speech reception threshold) and reduced ability to distinguish between competing talkers (i.e., voice perception). This provides evidence in support of anecdotal reports of difficulty following conversational turns by people with hearing impairment.