Dept. of Linguistics, UCLA, Los Angeles, CA 90025-1543
The results of experiments on trading relations have been interpreted as evidence for theories that speech perception is discrete, such as the motor theory [e.g., Libermann and Mattingly, 1985]; however, the experimental stimuli have not tested the predictions of theories of continuous perception [e.g., the FLMP, Massaro, 1987]. The present study addressed this issue using a pair of experiments that use a ``slit--split'' stimulus continuum varying in /p/ burst amplitude, rather than formant transitions, along the spectral dimension and varying in preburst silence duration along the temporal dimension. Experiment 1 was a 2IAX task in which subjects responded ``same'' or ``different'' to ten blocks of stimuli that were presented in pairs with a 300 ms ISI and a 1000-ms inter-block interval. Experiment 2 was a forced choice identification task in which subjects responded ``slit'' or ``split'' to ten blocks of stimuli that were presented singly with a 1000 ms ISI. A trading relation occurred between the amplitude of the burst and the duration of the silence. This finding is incompatible with a theory in which a burst is absent or present.