Measurements have been made on the discrimination of speech contrasts
on the basis of single-channel vibrotactile presentation of a variety
of speech-derived signals, coded as amplitude- and frequency-modulated
pulse trains. Stimulation was at the index fingertip. The signals cho
sen for tactile presentation were the speech amplitude envelope, the v
oice fundamental frequency FO and the zero-crossing frequency in the 1
.3-6.6 kHz band. 'Two-feature' codings, which present two of these sig
nals simultaneously (one coded as stimulus frequency and one coded as
stimulus amplitude), were found to be no more effective than 'single f
eature' codings which present only one signal (coded as both amplitude
and frequency). Scores for consonant discrimination were highest for
the single-feature coding of zero-crossing frequency, although differe
nces between the codings were not, in general, significant. Scores for
emphatic-stress discrimination were highest for the single-feature co
ding of FO, and this coding produced best results overall. A practical
wrist worn device, whose design is influenced by these experimental r
esults, is briefly described.