A composite randomization procedure for measuring spectral shape discrimination.

In studies of auditory profile analysis [D. M. Green, Profile Analysis: Auditory Intensity Discrimination (Oxford U. P., New York, 1988)], the sounds are presented at random levels to discourage the listener from basing the discrimination on a difference in absolute level rather than on a difference in the shape of the spectrum. A difference in absolute level, however, can still provide an effective discrimination cue if the difference is comparable to the range of randomization. Using enormous ranges of random levels is not desirable, because it is distracting to normal listeners and may exceed the dynamic range of hearing for listeners with hearing loss. This article describes a new experimental procedure which permits the experimenter to greatly reduce the range of level randomization in roving-level tasks.