Profiling individual differences in speech perception outcomes by CI users

Background
Cochlear implant (CI) users show great individual variability in speech comprehension outcomes, and in the amount of effort engaged while processing speech. This variability is a challenge for hearing rehabilitation, and for the understanding of factors, such as ageing and cognition in listeners’ post-implantation performance. In order to bridge objective
measures with performance we study individual differences amongst CI users through a combined analysis of electrophysiological measures (N400 and electro-acoustic coherence)
with online measures of speech processing (gaze tracking) and processing effort (pupil dilation) when participants are listening to continuous speech.

Methods
34 experienced CI users, and 34 normal-hearing (NH) age-matched controls were tested in experiments that capture, within participant, the aspects of processing effort, timing of
lexical mapping, predictive processing of speech, and active attentional engagement. Participants’ ocular responses (gaze fixations and pupil dilation) were recorded when they
were listening to running speech and disambiguating onset overlapping words (e.g. hamster versus ham). Participant’s active and predictive processing of speech was recorded in EEG as
their ability to use sentential context to anticipate sentence-final words, based on preceding rhythmic and semantic information.

Results
The combined ocular and EEG responses for NH listeners show homogenous timing and recruitment of attentional control, as well as active predictive processing of speech with
targeted engagement of attention. CI listeners showed a fast integration of sentence rhythm information, but we see individual differences in their engagement of attention during
lexical disambiguation, and in their active use of preceding information to anticipate words. The combination of EEG and eye-tracking measures provides information that allows
clustering participants into groups of different processing fluency.

Conclusions
Profiles of speech processing for individual CI users will be discussed, as they appear to bear potential for individualised hearing rehabilitation, when used to establish baselines to
capture listeners’ performance when processing running speech.

Acknowledgments
This work was supported by a by a Med-El research grant and by a Marie Curie Intra-European Fellowship (FP7-PEOPLE-2012-IEF 332402).