A Longitudinal Natural Language Processing (NLP) Analysis of Language and Agency in Science Classrooms (Poster 25)

• In Event: AERA Poster Session Saturday 5:10 p.m.
  In Poster Session: Advances in STEM Learning and Educational Psychology

Abstract

This study aims to explore the discursive resources and distribution of agency within urban middle school science classrooms using natural language processing (NLP) techniques. By analyzing classroom discourse data, we identify the integration of students’ everyday and academic languages and examine how these elements contribute to equitable student engagement in science learning. This research combines quantitative NLP methods with qualitative analyses to contextualize findings. Results highlight the presence of hybrid discourse, showcasing how teachers and students coordinate social and academic language. Additionally, we investigate the distribution of agency through metrics such as turn-taking and clout scores. The study offers valuable insights for educators, providing actionable data to enhance instructional practices and support inclusive science education.

Authors

• Christine Lee Bae, Virginia Commonwealth University

• Kamil Hankour, Virginia Commonwealth University

• Kimberly Williamson, Lehigh University

• Morgan DeBusk-Lane, Gallup

• Rachel Niemira, Virginia Commonwealth University

• Singith Nuwanga Perera, Virginia Commonwealth University

• Ryan LeVault, Virginia Commonwealth University