| Abstract: | The acoustic characteristics of voiceless velar and alveolar stop consonants were investigated for normally articulating and phonologically disordered children using spectral moments. All the disordered children were perceived to produce /t/ for /k/, with /k/ being absent from their phonetic inventories. Approximately 82% of the normally articulating children's consonants were classified correctly by discriminant function analysis, on the basis of the mean (first moment), skewness (third moment) and kurtosis (fourth moment) derived from the first 40 ms of the VOT interval. When the discriminant function developed for the normally articulating children was applied to the speech of the phonologically disordered group of children, no distinction was made between the velar and alveolar stops. Application of the model to the speech of individual children in the disordered group revealed that one child produced distinct markings to the velar-alveolar contrast. Variability measures of target /t/ and /k/ utterances indicated greater variability in this disordered child's productions compared with the normally articulating children. Phonological analysis of this child's speech after treatment, in which the velar-alveolar contrast was not treated, revealed target appropriate productions of both /t/ and /k/. By contrast, the other three phonologically disordered children, for whom no acoustic distinction was found between target /t/ and target /k/, did not evidence any knowledge of the contrast after treatment with other target phonemes. |
