Training paradigms for multichannel tactile aids

Abstract

Because connected speech presented through a multichannel tactile speech aid results in a complex and rapidly changing signal, it has been assumed that subjects having no prior tactile aid experience would require considerable amounts of “analytic” tactile training with phoneme and word‐level single‐item identification tasks, before they could usefully interpret connected speech. While it is the case that subjects who receive substantial amounts of analytic training with tactile aids perform better than subjects with no training, it may be the amount of prior experience, rather than the specific training tasks employed, that determines asymptotic performance levels. In the present study, groups of subjects were trained with a multichannel tactile aid in either an “analytic” condition (a graduated series of single‐item tasks) or a “nonanalytic” condition (connected speech tracking only). Each group received the same number of hours of tactile aid experience. During the final 20 h of the study, all subjects were tested with connected speech tracking. Results comparing these 20 h of tracking for the two groups showed no significant differences in performance. However, substantial intersubject differences indicated possible influences of other variables, including talker effects and subject familiarity with phonetics. These issues were addressed in a second experiment. [Work supported by NIH.]