This study was conducted in order to simultaneously compare the relative effectiveness of three different language training models (total communication sign training, nonverbal sign-alone training, oral (vocal) training), for teaching expressive language skills to three nonverbal children. A single subject, simultaneous treatment (multi-element) design with replication within and across subjects, was used to compare the rate of expressive word acquisition across training models. Results showed the total communication training model to be substantially superior to both oral and sign-alone" training models, and suggest that the use of physical prompts combined with multi-sensory inputs provide a basis for the demonstrated success.