We quantify the dimensions of high quality feedback apply our automated measures and characterize the written feedback of expert teachers and LLMs. Can we adapt LLMs to do better?
We develop measures for language practices in multilingual students' math conversations. What do these measures tell us about student sense-making and collaboration?
We measure the punitive and restorative attitudes reflected in teachers' classroom management language at scale. What do the patterns tell us about classroom identities and dynamics?