Taxonomizing Representational Harms using Speech Act Theory

2025-04-01

AI Summary

This paper addresses the conceptual ambiguity and coarse-grained categorization of “representational harm” in generative language systems. Drawing on Austin’s speech act theory, it reconceptualizes representational harm as illocutionary acts that produce harmful perlocutionary effects, thereby refining definitions of stereotyping, denigration, and erasure. It introduces speech act theory to AI fairness research for the first time, establishing the first fine-grained taxonomy of illocutionary acts—surpassing prior high-level classifications. Integrating insights from linguistic anthropology, sociolinguistics, and conceptual case studies, the framework clarifies conceptual boundaries and demonstrates practical utility in both conceptual analysis and the development of empirically verifiable measurement tools. The core innovation lies in using illocutionary acts as a theoretical and methodological bridge, unifying conceptual modeling and empirical measurement of representational harm.

Technology Category

• Knowledge Representation and Reasoning: Argumentation
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Representational harms are widely recognized among fairness-related harms caused by generative language systems. However, their definitions are commonly under-specified. We present a framework, grounded in speech act theory (Austin, 1962), that conceptualizes representational harms caused by generative language systems as the perlocutionary effects (i.e., real-world impacts) of particular types of illocutionary acts (i.e., system behaviors). Building on this argument and drawing on relevant literature from linguistic anthropology and sociolinguistics, we provide new definitions stereotyping, demeaning, and erasure. We then use our framework to develop a granular taxonomy of illocutionary acts that cause representational harms, going beyond the high-level taxonomies presented in previous work. We also discuss the ways that our framework and taxonomy can support the development of valid measurement instruments. Finally, we demonstrate the utility of our framework and taxonomy via a case study that engages with recent conceptual debates about what constitutes a representational harm and how such harms should be measured.

Problem

• Classifying representational harms in generative language systems
• Defining stereotyping, demeaning, and erasure using speech act theory
• Developing a taxonomy for measuring representational harms effectively

Innovation

• Framework based on speech act theory
• New definitions for stereotyping, demeaning, erasure
• Granular taxonomy of harmful illocutionary acts