GrandChair : conversational collection of grandparents' stories

Abstract

The act of sharing stories, which often characterizes the interactions between grandparents and grandchildren, exerts a profound influence on both the child listener and the grandparent teller. Unfortunately, opportunities for such sharing are rare for the many extended families who are geographically separated, and the stories go untold. Simple methods such as tape recorders or memory books can be difficult to work with, as they do not provide the powerful feedback that an active and interested listener can give. Computer-based systems have the potential to model this feedback, but in order to be effective at evoking stories, the interface must move away from keyboard and monitor and must be grounded in an understanding of conversation. This work argues that an effective story-eliciting system for grandparents must be based on a model of conversational behavior, must provide a comfortable and story-evoking environment, and that the ideal interface is an autonomous animated character. I present GrandChair, a system which can elicit, record, index, and play back grandparents' stories within an interaction model based on face-to-face conversation, and couched in an environment designed to be comfortable and story-evoking. Tellers sit in a comfortable rocking chair and tell stories with the assistance of a conversational agent on a screen, who takes the form of a child, to help them tailor their stories to a child audience, and prompts them with stories, questions, and video clips from their previous interactions.