Psychophysical tuning curves for phonetically relevant acoustic information

Abstract

In a series of experiments we have begun mapping psychophysical tuning functions for acoustic information that is used in specifying segmental distinctions. Our procedure is a modified selective adaptation technique, whereby we systematically vary the value of the adaptor along the dimension of interest and measure the magnitude of shift in the phonetic boundary. To date, we have found, as might be expected from previous research, that the magnitude of the adaptation effect increases as the distance between the adaptor and the phonetic boundary increases. More importantly, we have found that the shape of the tuning curve beyond the asymptotic value systematically differs as a function of the particular acoustic dimension under test. For example, for a voice‐onset‐time (VOT) continuum specifying a syllable‐initial voicing distinction, the effect remains at maximal level across a wide range of values (at least up to + 175 ms VOT); whereas, for a continuum of second‐formant transitions specifying a syllable‐initial place distinction, the curve falls after asymptote, more closely resembling a typical psychoacoustical tuning curve. The significance of these findings for describing both the effective information for specific phonetic distinctions, and the operation of the mechanisms that process that information, will be discussed. [Supported by NIH.]