Our initial design sessions with StoryCarnival generated these questions: Could we use a voice agent instead of an adult to facilitate make-believe play? How would a voice user interface look like for preschool children?
My research group and I explored using the some of the characters from our stories to be the body of the agent. We built low-fidelity paper prototypes that encapsulated a small Bluetooth. We used Wizard-of-Oz, where a researcher controlled the speech of the agent in a computer using a program to send it to the speaker in real time.
The paper prototypes proved too fragile because children wanted to pick them up and incorporate them into their play settings. So, we had to develop a sturdier version made with foam and paper that was more resistant.
We also explored other activities throughout during 24 participatory design sessions with a group of nine preschool children, including using an app to allow them to control the voice agent speech during make-believe play. First, children choose what character they want to direct the speech, then they choose the topic of the speech (e.g. food, water, shelter, random), and finally they choose the kind of speech: events, facts or feelings. The adults supported children while they used the interface by prompting them to take turns, giving them tips, and holding the device.
Another exploration was using an animated version of the cat agent on the screen, to see how children would react. Children clearly preferred the tangible version of the agent, which was in line with our expectations.
Luiza Pantoja © 2018