The Computer-Aided Speech Perception Assessment (CASPA) is a clinical measure of speech recognition that uses 10-item, isophonemic word lists to derive performance intensity (PI) functions for adult listeners. Because CASPA was developed for adults, the ability to obtain PI functions in children has not been evaluated directly. This study sought to evaluate PI functions for adults and four age groups of children with normal hearing to compare speech recognition as a function of age using CASPA. Comparisons between age groups for scoring by words and phonemes correct were made to determine the relative benefits of available scoring methods in CASPA.
Speech recognition using CASPA was completed with 12 adults and four age groups of children (5- to 6-, 7- to 8-, 9- to 10-, and 11- to 12-yr olds), each with 12 participants. Results were scored by the percentage of words, phonemes, consonants, and vowels correct. All participants had normal hearing and age-appropriate speech production skills.
Differences in speech recognition were significant across all age groups when responses were scored by the percentage of words correct. However, only differences between adults and the two youngest groups of children were significant when results were scored by the number of phonemes correct. Speech recognition scores decreased as a function of signal to noise ratio for both children and adults. However, the magnitude of degradation at poorer signal to noise ratios did not vary between adults and children, suggesting that mean differences could not be explained by interference from noise.
Obtaining PI functions in noise using CASPA is feasible with children as young as 5 yrs. Statistically significant differences in speech recognition were observed between adults and the two youngest age groups of children when scored by the percentage of words correct. When results were scored by the percentage of phonemes correct, however, the only significant difference was between the youngest group of children and the adults. These results suggest that phoneme scoring may help to minimize differences between recognition scores of adults and children because children may be more likely to provide responses that are phonemic approximations when words are outside their lexicon.