The central hypothesis is that preschoolers receiving AAC Generative Language Intervention will create longer, more grammatically complete sentences compared with the Standard of Care condition. Specific Aim 1 compares these interventions for preschoolers with severe speech impairments and typical receptive language, and Specific Aim 2 does the same for children with Down syndrome. Specific Aim 3 tests for possible moderation of the intervention effect by age, dynamic assessment scores, sentence type, receptive language abilities, prior AAC use, sex, mental age, and disability. Aims 1 and 2 will be accomplished using randomized controlled trials.