Infant speech‐discrimination testing: Effect of stimulus intensity and methodological model on estimates of performance

Robert J. Nozza

DOI: https://doi.org/10.1121/1.2023275

1986-05-01

The Journal of the Acoustical Society of America

Abstract:Infant speech-sound discrimination data have typically been obtained at only a single stimulus intensity level, with no consideration given to the bias in outcome that can result when full performance—intensity (P-I) functions are not determined. Another potential for bias lies in the assumption made about the underlying methodological model. Data from infant head-turn discrimination studies are treated as if from a single-interval psychophysical procedure, but the infant is operating with high temporal uncertainty. The task is better described as vigilance. This investigation was done to measure performance versus intensity (50 to 70 dB SPL) in a phoneme discrimination task (/ba/versus/da/) by infants and to determine if the choice of underlying methodological model influences results. Results indicate that performance varies with intensity and suggest caution in interpreting results obtained at a single stimulus intensity level. The P-I functions have similar form under both models, but the finding of a response decrement, typical of vigilance rather than fixed-interval tasks, supports the appropriateness of the vigilance model. Analysis of response density over time, as suggested by Watson and Nichols [J. Acoust. Soc. Am. 59, 655–668 (1976)] for vigilance-type data, provides limited support for the use of the single-interval model with the VRISD (undefined trials) procedure. [Supported by Pennsylvania Lions Hearing Research Foundation.]

acoustics,audiology & speech-language pathology