Unforgetting the Families Who Built Special Education: Computational Analysis of IEP Language and Family Voice

• In Event: Parenting as Praxis: Families, Schooling, and the Quest for Racial, Linguistic, and Disability Justice

Abstract

The Individualized Education Program (IEP) is a legal tool for planning student supports but often functions as a site of exclusion for Black, Latine, Indigenous, and multilingual families. This study uses large language model (LLM) analysis to examine 36 anonymized IEPs and applies a novel Family Mattering framework. Findings show IEPs for Black students contain more jargon (26.3 vs. 19.8), passive voice (22.1 vs. 15.2), and distancing language (15.6 vs. 9.7) than those for White students. A GPT-4 guided revision improved readability and inclusivity. Rather than automate compliance, this study shows how LLMs can support documentation repair and calls for a return to equity-centered, collaborative partnerships between families and schools.

Author

• Antoinette Banks, University of California - Davis