Prosodic and Segmental Aspects of Speech Perception with the House/3M Single-Channel Implant

Abstract

Four adult users of the House/3M single-channel cochlear implant were tested for their ability to label question and statement intonation contours (by auditory means alone) and to identify a set of 12 intervocalic consonants (with and without lipreading). Nineteen of 20 scores obtained on the question/statement task were significantly better than chance. Simplifying the stimulating waveform so as to signal fundamental frequency alone sometimes led to an improvement in performance. In consonant identification, lipreading alone scores were always far inferior to those obtained by lipreading with the implant. Phonetic feature analyses showed that the major effect of using the implant was to increase the transmission of voicing information, although improvements in the appropriate labelling of manner distinctions were also found. Place of articulation was poorly identified from the auditory signal alone. These results are best explained by supposing that subjects can use the relatively gross temporal information found in the stimulating waveforms (periodicity, randomness and silence) in a linguistic fashion. Amplitude envelope cues are of significant, but secondary, importance. By providing information that is relatively invisible, the House/3M device can thus serve as an important aid to lipreading, even though it relies primarily on the temporal structure of the stimulating waveform. All implant systems, including multi-channel ones, might benefit from the appropriate exploitation of such temporal features.