Computer-based alternative communication devices are widely used by nonverbal children with autism to make requests and independently communicate with caregivers. Despite advances in format and interactivity, all such devices are limited by the need to directly teach all requests, which is both time consuming and labor intensive. Providing a history of relational responding may facilitate emergent requests and functional language forms that are untrained. For instance, if a child learns to request items using pictures, he or she may then be taught to relate the picture and its dictated name (Name-Picture) and the dictated name and its corresponding text (Name-Text). If the child subsequently makes requests using text exchange, which was never directly taught, then derived requesting has occurred. The goal of this research is to adapt a new relational training and testing procedure, the Relational Completion Procedure (RCP), to facilitate emergent communication skills in nonverbal children with autism. In the RCP, stimuli are presented successively from left-to-right in order to mimic reading a sentence in English. The RCP allows the user to 'drag and drop' a picture or text stimulus to 'complete the sentence', before hearing it spoken and then confirming or cancelling each selection. After prescreening children's language, communication and social/cognitive ability, and identification of formal reinforcer preferences, two groups of nonverbal children with autism will be taught to request a preferred item via picture exchange. Conditional relations will then be established between the corresponding names-pictures and names-text. Derived requesting will be tested to see if children request a preferred item via text exchange. One group of children will receive conditional relation training with stimuli unrelated to the critical derived requesting posttest (text exchange) phase. It is predicted that the children in this control group will show below chance levels of derived requesting, while the remaining children will, with some between participant variability, pass the tests for derived requesting. The control group will subsequently receive the intervention. This study is relevant to Autism Speaks' priority research areas aimed at improving quality of life with behavioral interventions and of enhancing the diagnosis and treatment of underserved populations such as nonverbal children with autism.