Abstract:
Previous ``say/stay'' perception studies using synthetic stimuli posited F1-onset frequency and gap duration as the relevant cues. Children needed shorter gaps than adults to respond ``stay'' when F1 onset weakly indicated ``stay,'' suggesting that children weight that formant transition more. Replicating these experiments with natural speech produced unexpected findings. Acoustic analysis of natural ``say'' and ``stay'' tokens showed that: (1) F1 onset did not vary for ``say'' and ``stay;'' (2) F2/F3 onsets did vary; and (3) a burst was present in ``stay.'' Perceptual stimuli, therefore, consisted of natural, burstless vocalic portions from ``say'' and ``stay,'' and these same portions with a burst added. Each portion was combined with a natural ``s'' noise at seven gap durations. The burst effect was stronger for adults than children, and F2/F3 onset had an effect only for burstless stimuli. A second experiment was designed to examine the effects of F2/F3 in burstless stimuli, but adults failed to hear any stimuli as ``stay.'' Children had no trouble. It is concluded that: (1) F1 onset is not a cue to ``say/stay;'' (2) adults have learned to use the burst as a cue to ``stay'' identity; and (3) previous results were not due to the use of synthetic speech. [Supported by NIDCD grant R01 DC 00633.]