Conversational Agents for Young Learners

PI: Mark Warschauer (School of Education, University of California, Irvine)

Co-PI: Andres Bustamante (School of Education, University of California, Irvine)

Investigator:
Ying Xu (School of Education, University of California, Irvine)

Summary

Young children learn best – whether in daily life, reading books, or watching television – when they socially interact with an interested, caring, and knowledgeable adult. Given that learning is profoundly social in nature, we would expect limits to what children can learn from digital media. However, advances in speech recognition now make possible a way out of the conundrum, as evidenced by children’s natural interaction with Siri, Alexa, and other conversational agents (CAs). We are exploring two educational applications of conversational agents, including i) audio storybooks to promote early language and literacy skills and ii) science videos to foster scientific knowledge and curiosity.

CA for Early Language and Literacy skills
In this project, we designed CA-based audio stories using Google’s voice-driven interface (Google Assistant). The CA pauses at particular points in the story and prompts children to answer an open-ended question. The CA gives feedback on the children’s responses, explaining why the answer is correct or incorrect. In cases where children fail to produce comprehensible answers (due to fuzzy pronunciation or a lack of comprehension of the prompt), the CA rephrases the original question in a multiple-choice format.
A small-scale pilot study has suggested that children enjoyed the interactive story and conversed naturally with the CA. Findings from the pilot study guides the revision. A larger-scale randomized control trial will be carried out in Summer 2019. We will compare children’s reading experiences with the revised CA versus those with a human partner in four aspects, including verbal interaction, engagement, perceptions, and learning outcomes. The findings of this study can offer design implications for dialogic systems for young children’s informal learning.

CA for Scientific Knowledge and Curiosity
In this project, we propose to develop and test an interactive app based on children’s science-focused television programming, which incorporates a CA — an animated character — to socially interact with children, asking children questions and following with appropriate response-specific conversation, with the goal of enhancing children’s engagement and learning.

This project consists of three phases. In phase one, we will start with an extant science video episode targeted at young children (ages 4-6) made available by a partner to the research. We will write questions and build these questions into the app. In phase two, field testing of the app will then be conducted with a convenience sample of about two dozen children age 4-6, In phase three, a more formal pilot test will be conducted. A total of 150 children age 4-6 will be randomly assigned to one of two groups–a conversational app group (which will watch the video with interspersed virtual conversation) and a non-conversational app group (which will watch the video without virtual conversation). This will allow us to compare the impact of the virtual conversation on children’s engagement and learning, as well as to examine any heterogeneous impact by age, gender, socioeconomic status, or prior level of cognitive or linguistic development.

Publications

Ying Xu and Mark Warschauer. Young Children’s Reading and Learning with Conversational Agents. In CHI Conference on Human Factors in Computing Systems Extended Abstracts (CHI’19 Extended Abstracts), May 4–9, 2019, Glagsow, Scotland, UK. ACM. https://doi.org/10.1145/3290607.3299035