Level discrimination of stop‐consonant noise bursts: Performance in single‐ and two‐interval tasks with single and multiple standards.

Discrimination of increments in the rms level of either the initial or final consonant noise burst of the word (pæt/) was measured for seven highly trained, normal‐hearing listeners. The stimulus was a single token of /pæt/ produced by a female talker and digitized at 20 kHz. The noise bursts’ sample points were multiplied by (1+k), 0.025≤k≤4.1, to obtain rms‐level increments ranging from 0.2–14.0 dB. Experiment I (n=4) compared discrimination performance for noise bursts presented either ‘‘in isolation’’ (i.e., excerpted from the word) or in word context. A constant‐stimulus 2IFC procedure was employed. Level discrimination thresholds expressed as 20 log (1+k) were 2.0–3.0 dB for both the /p/ and /t/ bursts ‘‘in isolation.’’ Thresholds ‘‘in context’’ were 2.0 dB for /t/ and 5.0–7.0 dB for /p/. Experiment II (n=3) compared level‐discrimination performance measured in four different tasks for /p/ or /t/ bursts presented ‘‘in context.’’ The tasks were (a) 2IFC with a single standard (/pæt/); (b) 2IFC with three standards (/kæt, pæk, pæt/); (c) single interval, single standard; and (d) single interval, three standards. A strong context effect was again obtained with the /p/ burst, but the effects of task and number of standards were small. [Supported by OCAST Grant HSO‐005 and Presbyterian Health Foundation.]