Even though we know that external memory aids support communication in Alzheimer's disease, the components of the communication aids for individuals with Alzheimer's disease have not been studied systematically. The goal of these two pilot experiments was to examine differences in conversational performance of adults with Alzheimer's disease related to the presence and absence of an aid, the type of symbol embedded in the aid, and the presence or absence of voice output. In Experiment 1, 30 adults with moderate-to-severe Alzheimer's disease participated in 10-min conversations with and without personalized AAC boards. There was no effect of AAC, regardless of symbol type, and a deleterious effect of voice output. In Experiment 2, modified spaced-retrieval training preceded conversations, standardized prompts were presented, and semantically-based dependent variables were examined. For the 11 participants in the second experiment, there was a significant effect of AAC, showing that the presence of AAC was associated with greater use of targeted words during personal conversations. We discuss new information about the contribution of AAC for persons with Alzheimer's disease, and demonstrate how the applied research process evolves over the course of a long-term commitment to a scientific investigation.