Use of multiple training paradigms for the comparison of tactile speech displays

Abstract

Past investigations comparing the relative merits of vibrotactile speech displays may have been unduly biased by the particular training and testing methods used. In order to minimize methodological dependence, two distinct training methods have been employed in a 2 × 2 (display method × training method) design. The two training methods have been specifically designed to model strategies used in teaching and learning language, and are used to train separate groups of subjects. One method is called the “building block” strategy, a step-by-step approach by which a subject first learns to identify small, meaningless tactile patterns and progresses to larger and meaningful structures. The other, the “holistic” strategy, immediately immerses the subject in the final step of the “building block” process. The two displays employed were a vocoder type and a flowed spectrographic type. Building block subjects learn to identify individual tactile phonemes and words before learning phrases. But preliminary results indicate that holistic subjects can learn to identify the phrases without necessarily knowing the constituent tactile words (out of context) or even having an awareness of the phonemic structure of the words. Conclusions are drawn concerning the use of these two different training strategies for the comparison of tactile speech displays. [Work supported by NSF and NIH training grant 5T32GM07057.]