Basic auditory capabilities and resolving power for phonemes

Efforts to relate performance on speech test to general auditory abilities have yielded conflicting evidence. One possible reason is that the resolving power for individual psychoacoustic dimensions is related more closely to the ability to resolve acoustic details of individual phonemes, than to global speech perception as assessed by the overall percent of tokens correctly identified on a speech test. Auditory capabilities were measured with a battery of psychoacoustic tests [Watson et al., J. Acoust. Soc. Am. Suppl. 1 71, S73 (1982)]. Speech perception ability was studied with seven subtests of the CUNY Nonsense Syllable Test, NST (presented in quiet, 33 dB SPL). Indices of speech perception were derived for each individual phoneme, overall for each subtest, and overall for the seven NST subtests. Results from 34 normal listeners showed: (1) large differences in identification accuracy of phonemes within a subtest, which are not reflected in the overall measures; (2) strong biases towards specific phonemes within a subtest; and (3) differences in identification of an individual phoneme depending on the specific subtest in which it occurs. Thus the overall measures of speech processing are poor indicators of the ability to resolve specific phonemes. This result, together with the fact that phonetic distinctions are cued by multiple (redundant) cues, suggests that correlations between overall speech measures and auditory capabilities may be expected to be moderate or low. [Supported by NIH and AFOSR.]