Profoundly deaf persons hear only time and intensity variations in amplified speech and must rely on lipreading to perceive speech segments [N. Erber, J. Acoust. Soc. Am.51, 1224–1227 (1972)]. These persons may perceive prosodic features of sentences by hearing. It is not known whether linguistically significant variations in duration and intensity of speech segments are perceived by reference to the inherent duration of those segments and thus, are dependent upon intelligibility. Subjects responded to three sentence tests wherein the sentences differed in either phrase‐structure, major constituent boundary (MCB), or contrastive stress. Responses were by forced choice to the prosodic features; subjects were not required to identify words. Stimuli were presented under audition‐alone, vision‐alone, and audiovisual conditions. The acoustic signal was noise, amplitude modulated by the sentence materials. The audiovisual condition resulted in highest mean scores. Identifying phrase‐structure, audition‐alone scores were near chance levels. Vision‐alone and audiovisual scores were close in value, suggesting that time and intensity cues supplemented visual cues to a small extent. Identifying MCB location, vision‐alone scores were near chance levels. Audition‐alone scores were equivalent to audiovisual scores, suggesting that time and intensity variations provided primary cues. Identifying contrastive stress location, score data suggested that the acoustic signal allowed detection of contrastive stress and the exact word was located with visual cues. These results indicate that an unintelligible speech signal may supplement lipreading by providing certain prosodic information, increasing perceptual strategies available to profoundly deaf persons.