Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Theories of "Sexuality" in Natural Language Processing Bias Research

Jacob Hobbs

TL;DR

The paper investigates how sexuality is theorized and operationalized in NLP bias research, revealing a widespread lack of explicit theoretical grounding and a tendency to equate sexuality with gender or to default to heterosexual norms. By surveying 55 papers across ACM and ACL venues, it shows heavy reliance on identity word lists, limited use of intersectionality, and prevalent heteronormative biases in methodology. The authors synthesize queer/feminist theory to critique current practices and provide actionable recommendations—explicit theory, intersectional analyses, power-dynamics-aware datasets, and engagement with queer methodologies—to improve validity and inclusivity in sexuality-bias analyses. This work advances sociotechnical critique in NLP and offers a framework to align bias research with lived queer experiences and interdisciplinary scholarship.

Abstract

In recent years, significant advancements in the field of Natural Language Processing (NLP) have positioned commercialized language models as wide-reaching, highly useful tools. In tandem, there has been an explosion of multidisciplinary research examining how NLP tasks reflect, perpetuate, and amplify social biases such as gender and racial bias. A significant gap in this scholarship is a detailed analysis of how queer sexualities are encoded and (mis)represented by both NLP systems and practitioners. Following previous work in the field of AI fairness, we document how sexuality is defined and operationalized via a survey and analysis of 55 articles that quantify sexuality-based NLP bias. We find that sexuality is not clearly defined in a majority of the literature surveyed, indicating a reliance on assumed or normative conceptions of sexual/romantic practices and identities. Further, we find that methods for extracting biased outputs from NLP technologies often conflate gender and sexual identities, leading to monolithic conceptions of queerness and thus improper quantifications of bias. With the goal of improving sexuality-based NLP bias analyses, we conclude with recommendations that encourage more thorough engagement with both queer communities and interdisciplinary literature.

Theories of "Sexuality" in Natural Language Processing Bias Research

TL;DR

The paper investigates how sexuality is theorized and operationalized in NLP bias research, revealing a widespread lack of explicit theoretical grounding and a tendency to equate sexuality with gender or to default to heterosexual norms. By surveying 55 papers across ACM and ACL venues, it shows heavy reliance on identity word lists, limited use of intersectionality, and prevalent heteronormative biases in methodology. The authors synthesize queer/feminist theory to critique current practices and provide actionable recommendations—explicit theory, intersectional analyses, power-dynamics-aware datasets, and engagement with queer methodologies—to improve validity and inclusivity in sexuality-bias analyses. This work advances sociotechnical critique in NLP and offers a framework to align bias research with lived queer experiences and interdisciplinary scholarship.

Abstract

In recent years, significant advancements in the field of Natural Language Processing (NLP) have positioned commercialized language models as wide-reaching, highly useful tools. In tandem, there has been an explosion of multidisciplinary research examining how NLP tasks reflect, perpetuate, and amplify social biases such as gender and racial bias. A significant gap in this scholarship is a detailed analysis of how queer sexualities are encoded and (mis)represented by both NLP systems and practitioners. Following previous work in the field of AI fairness, we document how sexuality is defined and operationalized via a survey and analysis of 55 articles that quantify sexuality-based NLP bias. We find that sexuality is not clearly defined in a majority of the literature surveyed, indicating a reliance on assumed or normative conceptions of sexual/romantic practices and identities. Further, we find that methods for extracting biased outputs from NLP technologies often conflate gender and sexual identities, leading to monolithic conceptions of queerness and thus improper quantifications of bias. With the goal of improving sexuality-based NLP bias analyses, we conclude with recommendations that encourage more thorough engagement with both queer communities and interdisciplinary literature.

Paper Structure

This paper contains 35 sections, 1 figure, 6 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Related Work
  4. Sexuality, Gender, and Social Theory
  5. Sexuality Spectra and Identities.
  6. Gender.
  7. Intersectionality.
  8. Queer and Feminist Theories.
  9. Survey
  10. Method
  11. Results
  12. Sexuality Theory.
  13. Sexuality Proxy.
  14. Sexuality Bias.
  15. Discussion
  16. ...and 20 more sections

Figures (1)

  • Figure 1: Additional paper classifications. Intersectional: Is sexuality bias measured together with other oppressions? The paper must consider multiple oppressed identities simultaneously. Sexuality Focus: Is measuring sexuality-based bias in NLP systems the primary focus of the article? Beyond Duality: Does the article go beyond a simple heterosexual/LGBTQ+ binary comparison structure?