Learning from Machinery Misspeaks: Cognitive Basis of Ableism and Bias in AI-Generated Content

This study investigates whether AI’s representation of People With Disability (PWD) in different social contexts are biased. We performed a corpus-based cognitive-linguistic analysis of Arabic and English AI-generated corpora. The cognitive linguistic analysis revealed a judgmental framing of disabi...

Full description

Saved in:
Bibliographic Details
Published in:Corpus pragmatics : international journal of corpus linguistics and pragmatics Vol. 10; no. 1
Main Authors: Abdelzaher, Esra, Alshammari, Shahd
Format: Journal Article
Language:English
Published: Cham Springer International Publishing 01.12.2026
Subjects:
Computational Linguistics
Computer Appl. in Arts and Humanities
Linguistics
Pragmatics
Semantics
Social Sciences
Disability
Frame semantics
Ableism
Conceptual metaphor theory
Artificial intelligence
Corpus analysis
ISSN:2509-9507, 2509-9515
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study investigates whether AI’s representation of People With Disability (PWD) in different social contexts are biased. We performed a corpus-based cognitive-linguistic analysis of Arabic and English AI-generated corpora. The cognitive linguistic analysis revealed a judgmental framing of disability and an inherent inequality in the conceptual metaphorization of ability and physical strength, while the corpus-based analysis of AI-generated content showed a systematic use of linguistic markers that constructed identities of PWD characterized by vulnerability, which were generally negative and evocative of bodily and medical frames. We detected clusters of biases in the representations of PWD, which inclined towards positivity in the English texts describing males from American or European societies than in the Arabic texts describing females. We argue that despite socio-cultural differences in the depictions of PWD, the same cognitive-linguistic strategies, mainly conceptual metaphors and semantic frames, can be used to construct counter-narratives to the current biased ones.
ISSN:2509-9507
2509-9515
DOI:10.1007/s41701-025-00216-2