Effect of set size and method on speech reception thresholds in noise

The threshold for speech is known to improve as signal (Miller, Heise, & Lichten, 1951) or response (Pollack, 1959) uncertainty is decreased. The definition of threshold as the signal level or signal to noise ratio (S/N) at which a fixed percent-correct score is obtained, therefore, becomes problematic when set size, M, is varied through the range for which threshold varies systematically (M = 2 to M = 16) because the meaning of that fixed percent changes as M is increased or decreased. The goal of the present study is to examine the effect of set size, M, on speech reception thresholds (SRTs) under two testing strategies.

SRTs were obtained in the presence of 80 dB SPL white noise for sets containing 2, 4, 8, and 16 words using two different procedures, one in which threshold was based on a fixed percent correct (American Speech-Language-Hearing Association, 1988) and one in which threshold was defined in terms of d' = 1.00. The subjects were 12 young women with normal hearing and little or no experience with audiologic testing procedures.

When threshold was based on a fixed percent correct, S/N at threshold was found to be dependent on set size (F = 3.333; df = 3, 33; p = 0.031). When threshold was defined in terms of d' = 1.00, S/N at threshold was found to be independent of set size.

If a smaller set size than that recommended by the American Speech-Language-Hearing Association (1988) guidelines is to be used for obtaining SRTs in a clinical setting, thresholds should be based on a criterion-free measure that is independent of the size of the set of words being tested, if possible.