Simulating Challenging Social Situations with Intelligent Agents to Support Novice Computer Science Teachers in Self- and Social- Regulation Strategies

Massachusetts Institute Of Technology, Cambridge MA

Investigators

Abstract

Effective educators must often engage in difficult conversations that require emotional self-regulation (strategies of managing oneself) and social-regulation (strategies of managing the interaction between oneself and others) to have successful outcomes. In dealing with angry parents, confused students, or frustrated colleagues, educators must address stressful situations while remaining calm, attending to student needs, and simultaneously deploying effective teaching practice. When these conversations engage issues of identity, self-esteem, or sense of belonging, the consequences of educators' words and actions can make a significant impact--for good or for ill--on the lives of young people. In this project, the team builds upon an existing mobile web application for digital simulations, Teacher Moments. Novice teachers will be immersed in vignettes of classroom life presented in text, animation, and video. Participants provide improvised text and audio responses to scenarios involving students, parents or other school personnel. The project will develop intelligent agents within the system to analyze participant audio data for emotional and sociolinguistic cues to help assess whether novice teachers are practicing effective self-regulation and social-regulation strategies that may lower the level of tension within the interaction. These "regulation sensors" will be combined with intelligent, personalized interventions to support novice teachers through scaffolded reflection and feedback, while supporting teacher educators with data for coaching. If successful, the models can be adapted to other fields--such as social work, pastoral care, law enforcement, and medicine--where practitioners must effectively manage difficult conversations. This research project is implemented as a research-practice partnership combining the expertise of learning scientists, computational linguists, and teacher educators. The team will recruit a cohort of ten Fellows, teacher educators who train and support computer science teachers. Regarding specific activities for the learning and educational research, the team will: (1) co-design and field test new scenarios in Teacher Moments that reflect high-leverage, high-equity computer science practices; (2) determine the baseline affective experience of and conversational strategies employed by preservice computer science teachers in responding to difficult teaching conversations in a digital simulation; (3) use technology to support preservice computer science teachers in affective self-regulation with the aim of a positive impact on students; (4) support preservice teachers' use of effective conversational strategies during difficult teaching conversations in the simulation; and (5) foster metacommunicative awareness in preservice computer science teachers. With respect to technology research and development, the project will: (1) develop sensors based on natural language processing that can identify more and less effective emotional and social regulation; (2) develop instructional scaffolds that can support novice teachers in emotional regulation, social regulation, and teaching practice; (3) personalize the deployment of appropriate scaffolds--directly to novice teachers and through teacher educators--based on data collected by regulation sensors; (4) evaluate how digital clinical simulation systems with regulation sensors and instructional scaffolds need to be adapted to diverse teacher education contexts; and (5) support technical foundations for an intelligent teacher feedback system for novice educators with potential for applications in other fields. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.