Acoustic evaluation of vowels produced by hearing-impaired children
Marios S. Fourakis, Ann E. Geers, Mary Russo, Emily A. Tobey · The Journal of the Acoustical Society of America · 1990
Three hearing-impaired children were recorded as they attempted to produce monophthongal vowels of American English. Children were matched for age, hearing, and speech and language ability, yet differed in prosthetic devices (a cochlear implant, a tactile aid, or a conventional hearing aid). Children produced vowels with their respective prosthetic aid on and off during four recording sessions over a period of 15 months. Four repetitions of each vowel, in each aid condition, at each recording date were digitized at 20 kHz with 16-bit precision, analyzed using ILS, and formant tracks obtained_ Geometric means of F1 and F2 were used to represent each utterance in Koenig space. Mean Euclidean distance from the Peterson and Barney equivalent vowel for children [J. Acoust. Soc. Am. 24, 1562–1567 (1952)] was computed, along with the standard deviation, for each vowel set. It was predicted that, if a child's productions improved, either longitudinally or as a function of prosthetic device feedback, these distances, and standard deviations, would decrease. Preliminary results for the child with a cochlear implant indicate that these predictions were true for some vowels but not for others. These results will be compared to data from the children with tactile and conventional hearing aids. Conventional measures of variability, such as the ratio of standard deviation to mean formant frequency, will also be discussed and compared with the distance measures. [Work supported by NIDCD.]