Cross-series selective adaptation with vowels
Dawn L. Dutton, James R. Sawusch · The Journal of the Acoustical Society of America · 1988
Cross-series selective adaptation has been found for vowels [J.J. Godfrey, Percept. Psychophys. 28, 103–111 (1980)], but the stimulus attributes relevant to adaptation were not clearly inferable. The predictions of two models of vowel recognition, the relational model and the octave band model [D. L. Dutton, J. R. Sawusch, and K. C. Tsoi, J. Acoust. Soc. Am. Suppl. 1 81, S18 (1987)] were tested using selective adaptation. An EHAE series served as the base continuum. Five vowels were generated to serve as adaptors. The relationship between the test series and the adaptors could be represented in three ways, depending on the stimulus information assumed to be relevant for pattern recognition. The two models above employ different frequency-based representations. A third model assumes adaptation is based on the pattern of movement of spectral peaks over time. Adaptors were constructed to be similar to one endpoint of the test series based on each of these three representations. Results will be discussed in terms of what cross-series adaptation can tell us about the representations that may be used during vowel recognition. [Work supported by NINCDS.]