Effects of talker variability on spoken word recognition for adult cochlear implant users

Abstract

Cochlear implants (CIs) have been highly successful in providing a restored sense of hearing to profoundly deaf individuals. However, CI users must rely on a signal that is highly degraded in acoustic-phonetic details, conveying limited talker information. Normal-hearing (NH) listeners devote processing resources to accommodating talkers’ voices, and show less accurate and slower word recognition in multiple-talker compared to single-talker conditions. CI users may display a different strategy in adapting to new talkers due to limitations in talker discrimination. The current study examined the effects of talker variability on spoken word recognition by experienced, post-lingually deafened adult CI users, and the extent to which talker variability effects are mediated by limitations in talker discrimination. Word recognition accuracy for lexically easy and hard words were obtained in single-talker and multiple-talker conditions, varying by gender (female, male, mixed). Results demonstrated that single-talker performance was highly variable, but a detrimental effect of talker variability was observed in mixed conditions, when talker differences were best discriminated, particularly for hard lexical items. Taken together, findings suggest that while CI users may be affected less by talker variability overall, they may display a similar strategy as NH listeners for talker adaptation when talker differences are discriminable.