Abstract:

Current models of selective adaptation for vowels implicate acoustic or auditory mechanisms rather than phonetic mechanisms. However, inconsistencies have been reported for /i/, /(small capital eye)/, and /(cursive beta)/ [Morse et al. (1976)] in that the middle vowel /(small capital eye)/ did not elicit significant adaptation effects. These inconsistencies may reflect category goodness effects [Miller et al. (1983); Samuel (1982)] in that adaptation is more robust for tokens further from the category boundary. In the present study, category goodness was determined precisely for /i/, /(small capital eye)/, and /(cursive beta)/ by obtaining within-subject rating data. Two continua were then constructed for each subject with these three vowels as the endpoints. Vowel adaptation was then assessed for the two continua using each subject's three best exemplars as the adapting stimuli. Despite the fact that each subject's adapting stimulus was, by definition, the ``best'' exemplar of that category, only /(small capital eye)/ and /(cursive beta)/ showed significant adaptation. The absence of adaptation for /i/ may reflect its status as a point vowel, suggesting that extreme vocal tract positions, and their acoustic correlates, are remarkably stable. Research in progress on the point vowel /a/ will reveal whether this hypothesis is correct. [Work supported by NSF SBR94-21064.]