Gender as a Variable in Natural-Language Processing: Ethical Considerations

By Brian N. Larson. 2017

115 citations recorded by the source corpus.

Researchers and practitioners in natural-language processing (NLP) and related fields should attend to ethical principles in study design, ascription of categories/variables to study participants, and reporting of findings or results. This paper discusses theoretical and ethical frameworks for using gender as a variable in NLP studies and proposes four guidelines for researchers and practitioners. The principles outlined here should guide practitioners, researchers, and peer reviewers, and they may be applicable to other social categories, such as race, applied to human beings connected to NLP research.

Subfields: Natural Language Processing, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

In Safety, Robustness & Interpretability

Citations in the corpus

Measurement and Fairness (2019, background): “arms are central to the literature on fairness for this reason. For recent general discussion about measurement and fairness in computer vision, natural language processing, and platforms, see, e.g., [27, 41, 50].) Inequality can become further entrenched by failing to consider the role of measureme”
BOLD: Dataset and Metrics for Measuring Biases in Open-Ended Language Generation (2021, background): “We note that the groups considered in this study do not cover an entire spectrum of the real-world diversity [21].”
Towards Understanding and Mitigating Social Biases in Language Models (2021, background): “We recognize that gender is non-binary and there are many ethical principles in the design, evaluation, and reporting of results in studying gender as a variable in NLP (Larson, 2017).”
Women also Snowboard: Overcoming Bias in Captioning Models (2018, methodology): “gender categories: male, female, and gender neutral (e.g., person) based on visual appearance. When designing gender classiﬁcation systems, it is important to understand where labels are sourced from [44]. We determine gender labels using a previously collected publicly released dataset in which ann”
Towards Debiasing Sentence Representations (2020, background): “Although we recognize that gender is non-binary and there are many important ethical principles in the design, ascription of categories/variables to study participants, and reporting of results in studying gender as a variable in NLP (Larson, 2017), for the purpose of this study, we follow existing ”
Artificial Intelligence (AI) Ethics: Ethics of AI and Ethical AI (2020, unknown): “ResearchonethicalissuesofAIfallsintothreecategories:featuresofAIthatmaygiveriseto ethicalproblems(Timmermansetal.,2010),humanfactorsthatcauseethicalrisks(Larson,2017), andsocialimpactofethicalAIissues.”
Gender and Representation Bias in GPT-3 Generated Stories (2021, background): “Gender inference raises many ethical considerations and carries a risk of harmful misgendering, so it is best to have individuals self-report their gender (Larson, 2017).”
Directional Bias Amplification (2021, background): “There are very few situations in which predicting sensitive attributes makes sense (Scheuerman et al., 2020; Larson, 2017), so we should carefully consider if this is strictly necessary for target applications.”
A Systematic Study of Bias Amplification (2022, background): “Moreover, the model developer may not always be able to measure the difficulty of recognizing group membership empirically as doing so may involve developing a model that predicts sensitive attributes—something that model developers may want to avoid [24, 27, 39, 43].”
It’s going to be okay: Measuring Access to Support in Online Communities (2018, unknown): “Following the guidelines of Larson (2017) for using gender in NLP, our use of gender is intentional and central to this study on gender disparities in received support.”
The ethics of algorithms: key problems and solutions (2020, unknown): “Consider, for example, how the outputs of translation and search engine algorithms are largely perceived as objective, yet frequently encode language in gendered ways (Larson 2017; Prates et al. 2019).”
Adhering, Steering, and Queering: Treatment of Gender in Natural Language Generation (2020, methodology): “Simultaneously, there is a growing body of work reﬂecting on the ethical principles and guidelines that can and should underpin the selection of category labels and assumptions in texts (e.g. [41,61]).”
A Survey on Gender Bias in Natural Language Processing (2021, methodology): “Using gender as a variable in NLP is an ethical issue, thus unreflectively assigning gender category labels may violate ethical frameworks that demand transparency and accountability from researchers [Larson 2017].”
Theories of “Gender” in NLP Bias Research (2022, background): “Larson provides four key guidelines for the responsible use of (author) gender as a variable in NLP, namely: i) make theory of gender explicit; ii) avoid using gender unless necessary; iii) make category assignment explicit; and iv) respect persons [37].”