Cochlear implants in a tone language, Mandarin Chinese
Cochlear implantation, as a means of restoring hearing to profoundly and totally deaf people, has now become a routine clinical procedure. Implant users can perform remarkably well in many aspects of speech perception. However, current implant devices do not provide all speech information equally well. One major limitation is in providing voice fundamental frequency (fO), known to be problematic even for non-tonal languages in which intonation plays an important role in speech communication. This causes even more difficulty in tonal languages, such as Mandarin Chinese or Thai, in which pitch variations are used to convey lexical meanings. This thesis is mainly concerned with how implant users perceive and use tonal information. Studies were first conducted in normal-hearing listeners to investigate the nature of tone and the importance of voice fO in understanding running speech. Three acoustic cues (fO, amplitude envelope, and duration) were examined for their contributions to tonal perception in syllables. The results clearly demonstrated voice fO to be the dominant cue. To determine the effect of explicit fO in sentence recognition, vocoded stimuli with various degrees of spectral information were presented to four age groups of listeners (aged 6, 9, 12, and 20). Information about natural fO variations enhanced sentence recognition significantly even when spectral information was severely degraded, and the effect was strong across all ages. The investigation in implanted children first examined tone recognition performance and the acoustic cues used for recognising tonal contrasts, especially the use of amplitude envelope. These implanted children appeared to make some use of amplitude changes in recognising tonal contrasts, though the overall effect was rather small. They also showed some evidence for the use of duration and temporal pitch information. For the effect of fO in sentences, no significant difference was found in performance on sentences with their original fO contours and those with uninformative fO contours. This indicates that the voice pitch information provided by current implant devices is too limited to allow listeners to take advantage of the presence of natural fO. In the light of the significance of fO both in signalling tonal identity in syllables and in perceiving sentences, it is likely that implant users will further benefit with a better representation of voice pitch.