Providing young children with opportunities to develop early literacy skills is important to their success in school, their success in learning to read, and their success in life. This project focuses on the creation of a new interactive reading primer technology on tablet computers that will foster early literacy skills and shared parent-child reading through the use of a targeted discussion-topic suggestion system aimed at the adult participant. The Cloud Primer will crowdsource the interactions and discussions of parent-child dyads across a community of readers. It will then leverage this information in combination with a common sense knowledge base to develop computational models of the interactions. These models will then be used to provide context-sensitive discussion topic suggestions to parents during the shared reading activity with young children. The work will be evaluated in week-long at-home studies.
Intellectual merit: The project will make fundamental theoretical contributions to models of human-human and human-computer interaction, and their use in fostering engagement and learning. The effort will also produce new insights into how common sense reasoning can be integrated with large-scale data collection to develop interactive technologies that deal gracefully with inconsistencies and noise to provide diverse and semantically meaningful responses in unconstrained, real-world environments.
Broader impacts: Research shows that one in three children in the United States enter kindergarten unprepared, and the majority of children who start behind typically stay behind. The Cloud Primer will counteract this trend by leveraging a community of readers to define a set of common discussion topics, actively exposing parents to these topics, and, through a simple touch interface, providing children of pre-reading age a mechanism for engaging adults in discussion. The new context and common sense aware interactive reading primer will be a fundamental advance over current digital reading technologies, which neither effectively achieve educational goals when used alone, nor support joint reading and adult engagement. The project will further contribute to education through undergraduate research, graduate thesis research, graduate course development, and outreach programs to women and under-represented minorities. To promote research in this area, the project will make all parent-child interaction data captured and annotated in the process of this research freely available to the research community, together with the Cloud Primer software. Furthermore, the computational methods developed through the course of this research will have applications in interactive domains beyond early literacy, such as foreign language learning.
Providing young children with opportunities to develop early literacy skills is important to their success in school, their success in learning to read, and their success in life. This project focused on the creation of a new interactive reading primer technology for tablet computers designed to foster early literacy skills and shared parent-child reading, focusing particularly on the goal of increasing parental engagement in the interaction. Based on recorded interactions of parent-child dyads across several populations, we used semantic networks to model topics of discussion and to automatically identify a set of related topics beyond the storybookâ€™s scope that were then used to generate discussion topic suggestions. Our hypothesis was that providing recommendations for additional thematically appropriate topics of conversation over the course of the story would result in increased measures of engagement. We tested our hypothesis through the development of the CloudPrimer, which consists of modular software packages designed for tablet devices that 1) record the interactions and discussions of parent-child dyads engaged in a reading activity, 2) leverage semantic reasoning to identify topics of discussion that occur during reading, 3) generate discussion topic suggestions based on relationships between the topics being discussed, and 4) presents topic suggestions, in the form of questions, to readers in order to foster parent-child interaction. Contributions of the above work also include the GlobalLit system for running and managing tablet-based data collection and user studies, the Tinkrbook story app, and the Semantic Similarity Engine. We evaluated our system by recruiting parent-child participants who took a tablet home and used our app over the course of a week. Our study compared three conditions, the original version of the story without discussion prompts, the story that uses discussion prompts written by an expert in dialogic reading, and the story that uses the semantically-generated discussion prompts. Our findings show that the use of a discussion prompt significantly improves a wide range of engagement measures, including time spent reading, count of words spoken by parent and child, count of novel words and lexical diversity. The expert-generated prompts are more effective at increasing engagement, suggesting that human experts should be relied on when possible. However, the semantically generated prompts also show statistically significant improvement over traditional reading, suggesting that an automated system can successfully improve engagement and parent-child reading practices in the absence of an expert tutor.