The influence of color and vowel context on the visual perception of /p,b,m/

Over the past 70 years, there have been few innovative strategies to enhance speechreading accuracy and efficiency. This study was designed to determine whether color and vowel context were significant influences on the speechreading ability of normal-hearing subjects. Traditional approaches to aural rehabilitation have emphasized strategies using visual aids, repetition, the manipulation of film and video production. However, there are no definitive approaches to speechreading training that optimize accuracy.; The /$p,b,m$/ viseme-cluster has been known to cause confusion in both normal-hearing and hearing-impaired groups. Several variables can influence the outcome of speechreading accuracy: (1) talker differences, (2) phonological factors, and (3) speechreading training.; In the case of speechreading, visual information from key regions of the face (lips, cheeks, eyes) must be directed to the color-sensitive foveal region of the eye for closer examination and for decision-making. Up until now, it has not been established whether viseme-cluster recognition could improve with the interaction of several independent factors—for example, color-vision conditions, vowel context, and the /$p,b,m$/ viseme-cluster. Erber (1974b) reported that confusion of /p/, /b/, and /m/ occurred for the vowels /$i$/, / α /, and /$u$/. This investigator examined the hypothesis that the interaction of opponent colors and the vowels /$i$/ and / α / are the sources of variation for the accurate visual recognition of phonemes /$p,b,m$/.; Based on opponent color theory, it is physiologically impossible for humans (with normal color-vision) to perceive simultaneously the opponent colors red and green, blue and yellow. To produce opponent lip-color conditions, the lips of an actor were cosmetically treated to create four opponent-color and one natural experiment condition.; There were two experiments with the phonemes/$p,b,m$/: one for vowel-consonant-vowel (VCV) disyllables in the vowel context of /$i$/ and one for VCVs in the vowel context of / α /. Six subjects, naïve to speechreading, viewed a video production of one talker speaking disyllables in five color combinations. The subjects marked their perceived selection for each phoneme in a three-choice multiple-choice format for a total of 2,160 observations per experiment.; Based on logistic regression analysis, the results indicated that the null hypothesis was not rejected for either Experiment I or II. However, for some individuals, color was an important variable for visual recognition. The findings warrant more study of this different approach for aural rehabilitation. An established cross-modal perception between color-vision and hearing could be key for some individuals to maximize visual-speech perception.