Gender-Neutral Machine Translation Strategies in Practice

2025-06-18

AI Summary

Machine translation (MT) systems often enforce binary gender assignments in grammatical-gender languages, thereby erasing source-language gender ambiguity and perpetuating representational harm. Method: We introduce the first systematic framework to evaluate gender neutrality in MT, employing a manually curated gender-ambiguous test suite to conduct comparative evaluation, error-pattern analysis, and strategy categorization across 21 state-of-the-art MT systems. Contribution/Results: We empirically identify and classify spontaneously emerging gender-neutral strategies—including pronoun omission, substitution with gender-neutral nouns, and passivization—demonstrating their distribution correlates significantly with target-language typology and sociocultural gender stereotypes. Results show that most systems lack robust gender-neutral capabilities; only a few exhibit effective deployment of such strategies in language pairs like English→French and English→German. This work establishes a reproducible methodology and empirical benchmark for developing fairer, more inclusive MT systems.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Evaluation and Analysis
• Multiagent Systems: Agreement, Argumentation & Negotiation

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Gender-inclusive machine translation (MT) should preserve gender ambiguity in the source to avoid misgendering and representational harms. While gender ambiguity often occurs naturally in notional gender languages such as English, maintaining that gender neutrality in grammatical gender languages is a challenge. Here we assess the sensitivity of 21 MT systems to the need for gender neutrality in response to gender ambiguity in three translation directions of varying difficulty. The specific gender-neutral strategies that are observed in practice are categorized and discussed. Additionally, we examine the effect of binary gender stereotypes on the use of gender-neutral translation. In general, we report a disappointing absence of gender-neutral translations in response to gender ambiguity. However, we observe a small handful of MT systems that switch to gender neutral translation using specific strategies, depending on the target language.

Problem

• Assessing MT systems' sensitivity to gender ambiguity preservation
• Categorizing gender-neutral strategies in machine translation practice
• Examining binary gender stereotypes' impact on neutral translations

Innovation

• Assessing MT systems for gender neutrality sensitivity
• Categorizing observed gender-neutral translation strategies
• Examining binary gender stereotypes in translation