Assisting Social Conversation between Persons with Alzheimer's Disease and their Conversational Partners

The number of people with dementia of the Alzheimer's type (DAT) continues to grow. One of the significant impacts of this disease is a decline in the ability to communicate using natural language. This decline in language facility often results in decreased social interaction and life satisfaction for persons with DAT and their caregivers. One possible strategy to lessen the effects of this loss of language facility is for the unaffected conversational partner (Facilitator) to "co-construct" short autobiographical stories from the life of the DATaffected conversational partner (Storyteller). It has been observed that a skilled conversational partner can facilitate co-constructed narrative with individuals who have mild to moderate DAT. Developing a computational model of this type of co-constructed narrative would enable assistive technology to be developed that can monitor a conversation between a Storyteller and Facilitator. This technology could provide context-sensitive suggestions to an unskilled Facilitator to help maintain the flow of conversation. This paper describes a framework in which the necessary computational model of co-constructed narrative can be developed. An analysis of the fundamental elements of such a model will be