Listeners compensate for coarticulatory influences of one speech sound on another. We examined whether lipread information penetrates this perceptual compensation mechanism. Experiment 1 replicated that when an /as/ or /a• / sound preceded a /ta/-/ka/ continuum, more velar stops were perceived in the context of /as/ ([1]). Experiments 2 and 3 investigated whether the same phoneme boundary shift would be obtained when the context was lipread instead of heard. An ambiguous sound between /as/ and /a• / was dubbed on the video of a speaker articulating /as/ or /a• /. Subjects relied on the lipread information when identifying the ambiguous fricative sound, but there was no boundary shift in the following /ta/-/ka/ continuum. These results indicate that biasing of the fricative and compensation for coarticulation can be dissociated.