Nonmonotonic discrimination functions for temporal gaps in noise‐buzz sequences. II: Filtering affects the discrimination peaks

Abstract

It has been reported [P. L. Divenyi and W. F. Danner, J. Acoust. Soc. Am. Suppl. 1 55, S31 (1975) and J. D. Miller et al., J. Acoust. Soc. Am. 60, 410–417 (1976)] that discrimination of the gap bounded by a brief, filtered noise burst and a longer,/a/‐like buzz is a nonmonotonic function of the gap duration, in a way similar to the well‐known nonmonotonicity of the VOT discrimination functions [A. S. Abramson and L. Lisker, Proc. 6th Int. Congr. Phonet. Sci., 569–573 (1970)]. Earlier, we reported that discriminability of the gap in such sequences is affected by passing the stimulus through 1/3‐octave filters and that the discrimination thresholds are correlated with the difference between noise and buzz intensity within the particular bands—i.e., With one of the main attributes of temporal masking. The present experiments extend the previous ones, in that the continuum of “buzz‐onset‐times” was explored in finer steps (10, 15, 20, 25, 30, 35, 40, 55, and 70 ms from noise onset to buzz onset). In addition to determining gap discrimination thresholds for the original, unfiltered stimulus, thresholds were also obtained at three filter settings (780, 1230, and 2550 Hz center frequencies). Three trained listeners participated. While the smallest thresholds at the 780 and 1230‐Hz filter settings occurred at the same gap duration as for the unfiltered stimulus (15 ms), the threshold minimum at the 2550‐Hz filter setting occurred at longer gap durations (20 and 25 ms). Implications of these results for variations in voicing category boundaries as well as for the possible role of listening bands in the perception of speech will be discussed. [Work supported by NIH Grant NS 16019 and by the Veterans Administration.]