Speech prosody perception in cochlear implant users with and without residual hearing.

Abstract

The detection of fundamental frequency (F0) variations plays a prominent role in the perception of intonation. Cochlear implant (CI) users with residual hearing might have access to these F0 cues. The objective was to study if and how residual hearing facilitates speech prosody perception in CI users. The authors compared F0 difference limen (F0DL) and question/statement discrimination performance for 15 normal-hearing subjects (NHS) and two distinct groups of CI subjects, according to the presence or absence of acoustic residual hearing: one "combined group" (n = 11) with residual hearing and one CI-only group (n = 10) without any residual hearing. To assess the relative contribution of the different acoustic cues for intonation perception, the sensitivity index d' was calculated for three distinct auditory conditions: one condition with original recordings, one condition with a constant F0, and one with equalized duration and amplitude. In the original condition, combined subjects showed better question/statement discrimination than CI-only subjects, d' 2.44 (SE 0.3) and 0.91 (SE 0.25), respectively. Mean d' score of NHS was 3.3 (SE 0.06). When F0 variations were removed, the scores decreased significantly for combined subjects (d' = 0.66, SE 0.51) and NHS (d' = 0.4, SE 0.09). Duration and amplitude equalization affected the scores of CI-only subjects (mean d' = 0.34, SE 0.28) but did not influence the scores of combined subjects (d' = 2.7, SE 0.02) or NHS (d' = 3.3, SE 0.33). Mean F0DL was poorer in CI-only subjects (34%, SE 15) compared with combined subjects (8.8%, SE 1.4) and NHS (2.4%, SE 0.05). In CI subjects with residual hearing, intonation d' score was correlated with mean residual hearing level (r = -0.86, n = 11, p < 0.001) and mean F0DL (r = 0.84, n = 11, p < 0.001). Where CI subjects with residual hearing had thresholds better than 60 dB HL in the low frequencies, they displayed near-normal question/statement discrimination abilities. Normal listeners mainly relied on F0 variations which were the most effective prosodic cue. In comparison, CI subjects without any residual hearing had poorer F0 discrimination and showed a strong deficit in speech prosody perception. However, this CI-only group appeared to be able to make some use of amplitude and duration cues for statement/question discrimination.