On Defining Erasure Harms for NLP

2026-06-14

AI Summary

This study addresses the challenge of identifying and measuring erasure harms—a form of representational harm in natural language processing (NLP) systems—that remains poorly understood due to the absence of a clear, unified conceptual framework. To tackle this issue, the paper develops a structured and operational definition of erasure harms through conceptual analysis and ethical reasoning, delineating its core components and criteria for judgment. The proposed framework overcomes limitations of existing definitions, which are either overly broad or confined to specific contexts, thereby offering a robust theoretical foundation and practical guidance for systematically recognizing, evaluating, and mitigating erasure harms across diverse NLP applications.

Abstract

The deployment of NLP systems has raised concerns about harms they might produce, including representational harms. Recent literature has begun to conceptualize and measure one such harm, the harm of erasure. Nevertheless, the field lacks a clear and cohesive conceptual foundation for identifying and measuring erasure. Existing conceptualizations of erasure are often broad -- making it difficult to identify what is needed to establish and measure erasure -- or else specific to particular settings -- facilitating measurement for those settings but potentially challenging to adapt to other settings. To address this gap, we develop and propose a structured definition of erasure that clarifies what components are necessary for establishing whether erasure has occurred, which practitioners need to explicitly articulate and operationalize in order to measure erasure.

Problem

• erasure harm
• representational harm
• NLP
• conceptual foundation
• measurement

Innovation

• erasure harm
• structured definition
• representational harm
• NLP ethics
• measurement framework