Perceptual confusions of 16 consonant-vowel (CV) syllables were studied with normal-hearing adults under conditions of auditory-visual presentation at various signal-to-noise (S/N) ratios, as well as under auditory-only and visual-only conditions in quiet. An articulatory feature classification system was used to analyze responses with regard to percentage correct intelligibility and relative information transmission. In the auditory conditions, features of voicing and nasality were least affected by noise, while place of articulation showed the greatest reduction in intelligibility. The auditory-visual confusions indicated that the visual channel in bisensory presentations reduced errors when phonemes differed by place of articulation, with the greatest visual complement occurring at the poorer S/N ratios. Responses from the visual-only condition indicated that the subjects were able to categorize the phonemes into discrete homophenous groups. Part-whole reliability for the visual-only condition was high, indicating that a short form of the test could be used for screening aural-rehabilitation clients. Finally, since the subjects' ability to visually recognize five places of articulation was nearly perfect, it was suggested that emphasis in aural rehabilitation be placed on auditory training or use of contextual cues, rather than on lipreading.

The speech of a five-year-old boy who suffered a profound hearing loss following meningitis was sampled at two-week intervals for nine months. Speech samples were subjected to phonetic transcription, spectrographic analysis, and intelligibility testing. Immediately post-trauma, the child displayed slightly slower, F0 elevated, acoustically intense speech in which phonemic distortion and syllabification of consonants occurred occasionally; single word intelligibility was depressed below normal between 20-30%. By the 18th week, a sudden decline in intelligibility, increasing monotony of pitch, and a pattern of strongly emphatic, prolonged, aspirated, syllabified, and increasingly distorted consonants were manifest. At year's end, the child's speech bore some resemblance to the speech of the deaf in terms of suprasegmentals, intonation, and intelligibility, but differed because the child rarely, if ever deleted speech sounds or diphthongized vowels strongly. It is speculated that phonetic processes such as diphthongization, syllabification, and prolonged duration may be strategies for enhancing feedback during speech.

A sample of 110 middle-aged and geriatric subjects (40 to 87 years) with normal hearing and vision was drawn from the general population in order to compare visual performance for consonant-vowel (CV) syllables and sentences. Results of this investigation revealed that, above 70, age was a factor affecting visual perception of syllables. Individuals above age 70 received the poorest speechreading scores and were inconsistent in viseme categorization. Results of a comparison of speechreading scores for sentences and syllables revealed a greater number of differences among sentences. Only individuals between 40 and 60 years of age received statistically similar mean scores when presented with common sentences. Finally, using a linear regression model, it was found that sentence speechreading performance could be accurately predicted from the CV syllable score within a range of accuracy of +/- 9.7%.

A lipreading screening test consisting of 100 consonant-vowel (CV) syllables was prepared on videotape and presented to subjected with normal hearing and vision. The results of visual consonant confusions revealed nine homophenous categories with high test-retest reliability. Based upon these categories, obtained under ideal viewing conditions, criterion levels of performance were specified in an attempt to determine the need for place of articulation instruction during lipreading training. Examples of confusion patterns from hearing-impaired observers are shown and implications for visual training are presented.

This study was concerned with the extraction, description, and verification of visual perceptual features underlying vowel lipreading performance. Ten viewers with normal hearing rated the visual similarity of pairs of 15 vowels and diphthongs presented in an /h_g/ context by four speakers. Multidimensional scaling techniques were used to extract potential perceptual features which were then labeled by the experimenters. The resulting perceptual dimensions were correlated with physical measurements of lip shape to evaluate the adequacy of the feature labels. The results indicated that the traditional extended-rounded vowel feature and a vertical lip separation feature were the characteristics most prominent in judging the stimuli. In addition, a feature related to overall area of maximum lip opening and two features unique to diphthong perception were tentatively identified.