Prosody of Dutch cochlear implant users
Profoundly deaf persons may receive a cochlear implant (CI). CI users receive adequate information on the spectral envelope of speech sounds and the way the spectral envelope changes as a function of time but do not get much information on the tonality of speech. This is due to the fact that the acoustic signal processed by the device's speech analysis component is divided into a small number of frequency bands, thus removing the fine-structure of the spectrum. Consequently, CI users are quite well capable of discriminating vowels and consonants and intensity changes, but have difficulties in the processing of prosodic structure (word tones, stress and rhythm, accentuation and intonation). Prosody is essential to language and speech, marking not only the syntax and semantics of utterance but also the type of sentence (e.g., statement, question), the emotional status (angry, afraid, happy) and the attitude of the speaker (remorseful, ironic). Our program aims to determine in detail the prosodic structures of CI users’ speech and the way they are acquired. Good hearing is indispensable to good speaking and therefore we will compare the speech characteristics of CI users (and normally hearing controls) to their discriminatory abilities. Previous research suggests that the discriminabilitory skills of CI users as to tonal speech variations is poor relative to those of normally hearing people. This is expected to be the source of any abnormalities in CI users' speech. Correlations between perception and production deficits will yield recommendations on how to improve the speech processing capabilities of cochlear implants. A second part of the program is dedicated to vocoders. Vocoders simulate what CI users hear. We will study to what extent these simulations are realistic by comparing the result of vocoder experiments with the results of perception experiments using normal CI users. Realistic vocoders would make the research on the perception of CI users much easier, because they allow for larger and more homogeneous subject groups.
For this group of research questions, our project is divided into four parts: (1) the assessment of the prosody (mainly intonation) of the spontaneous speech of young children wearing a CI and (2) of normally hearing children, (3) the assessment of CI users skills in discriminating prosodic variations, and (4) the assessment of prosody by normally hearing individuals listening through a vocoder. Recordings of normally hearing and CI wearing Dutch children’s spontaneous speech are already available, the other data will be acquired in new experiments.
In sum, the objective of our program is to gain more knowledge about what prosodic features CI users actually hear and how this develops in children. Implementation of the recommendations about the improvement of the prosodic processing of CIs will both render speech more comprehensible for CI users and make CI users’ speech more intelligible. This will upgrade their communicative abilities and perhaps even there emotional well-being.