Abstract
Cochlear implants (CI) can be surgically implanted to restore hearing in individuals who can no longer benefit from hearing aids. However, the loss of spectral richness brought about by cell death and tonotopic mismatch from implantation makes processing speech more difficult – users often experience fatigue during everyday social situations, despite scoring well on listening tasks post-surgery. Listening is, therefore, differentially more effortful for CI users than normal-hearing individuals. The loss of spectral richness also causes degradation of pitch perception, which has caused current literature to claim that CI users have a prosody deficit. However, there is more to prosody than just pitch variation. In this thesis I use pupillometry as a physiological index of effort employed by 22 CI users during listening tasks in which I manipulated the prosodic markings of clause boundaries. I found that CI users are, in fact, able to perceive some features of prosody and to use prosody to facilitate recall of sentences when the prosody coincides with a syntactic clause boundary and is detrimental to recall when prosody and syntax conflict, thus countering claims that CI users are insensitive to prosody. I also found that word recognition ability predicted cognitive effort, but only for those with a word recognition score above 45% correct. I report two additional experiments with normal-hearing listeners designed to determine which features of prosody CI users might be utilizing to detect clause boundaries in comprehension of globally ambiguous sentences. This was done by selectively removing pitch, amplitude, and timing cues, using pupillometry to assess increases in processing effort as each prosodic feature was removed. Preliminary results show that individuals exert the most effort when timing cues (clause-final pauses and word-lengthening) are removed. Results are considered in the context of the FUEL model of effortful listening.