| Abstract #721, Date 2/17/99, Session U, Podium , 5:30p |
| The production and perception of minimal pair contrasts by pediatric cochlear implant users |
| *S.B. Chin, K.R. Finnegan (Indiana University School of Medicine) |
| The concept of "contrast" is central to the analysis of languages, for it is by means of the contrastive nature of speech sound segments that languages can form large numbers of words from relatively small repertoires of segments. A traditional type of evidence in linguistic analysis for a contrast between two segments is the existence of minimal pairs, that is, two words of distinct meaning that exhibit different segments at one point but identical segments at all other points (Trask, 1996). So, for instance, the minimal pair "pill" and "bill" provides evidence that [p] and [b] represent different "phonemes" (contrastive segments) in English. Minimal pairs can also be used to assess control over contrasts and speech intelligibility among clinical populations and those acquiring a language; in previous research, these have included dysarthrics, second language acquirers, and the hearing impaired. In this study, we examined the relations among minimal pair perception, minimal pair production, and the intelligibility of words in sentences in a group of 14 prelingually deaf children. All received CIS- or SPEAK-implementing cochlear implants before the age of six years and had used their implants for at least two years. Perception and production of minimal pair contrasts were each tested with a two-alternative forced-choice task (the first with the children as respondents, the second with normal-hearing adults as judges); speech intelligibility was tested in an open-set format with normal-hearing adults listening to sentences produced by the children. Percent correct scores were submitted to correlational analyses to determine relations among the three measures, as well as among consonant and vowel feature classes in the two minimal pairs tasks. Results indicated significant overall correlations among minimal pair perception, minimal pair production, and intelligibility. In general, however, intelligibility and perception feature classes tended to exhibit more variance than corresponding production classes, so that individual perception classes correlated more highly with intelligibility than did corresponding production classes. For the same reasons, although overall minimal pair perception correlated significantly with overall minimal pair production, there were no significant correlations between individual perception feature classes and the corresponding production feature classes. These results will be discussed in the contexts of general relations between perception and production of contrasts and of test specificity. We are grateful to David B. Pisoni and Karen Iler Kirk for comments and suggestions. |