Year: 2025
Type: conference-paper
Abstract: As our awareness of autism and ableism continues to increase, so does our understanding of ableist language towards autistic people.Such language poses a significant challenge in NLP research due to its subtle and contextdependent nature.Yet, detecting anti-autistic ableist language remains underexplored, with existing NLP tools often failing to capture its nuanced expressions.We present AUTALIC, the first dataset dedicated to the detection of anti-autistic ableist language in context, addressing a significant gap in the field.AU-TALIC comprises 2,400 autism-related sentences collected from Reddit, accompanied by surrounding context, and annotated by trained experts with backgrounds in neurodiversity.Our comprehensive evaluation reveals that current language models, including state-of-the-art LLMs, struggle to reliably identify anti-autistic ableism and diverge from human judgments, underscoring their limitations in this domain.We publicly release our dataset 1 along with the individual annotations, providing an essential resource for developing more inclusive and context-aware NLP systems that better reflect diverse perspectives.Trigger warning: this paper contains ableist language including explicit slurs and references to violence.
Authors: Naba Rizvi, Harper Strickland, Daniel Gitelman, Alexis Morales Flores, Tristan Cooper, Aekta Kallepalli, Akshat Alurkar, Haaset Owens, Saleha Ahmedi, Isha Khirwadkar, Imani N. S. Munyaka, Nedjma Ousidhoum
Institutions: University of California San Diego, Cardiff University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Reduced inequalities
Open Access status: bronze
OpenAlex ID: W4412944904