Examining Gender Bias in Languages with Grammatical Gender

TL;DR

This paper investigates gender bias in word embeddings for gendered languages like Spanish and French, proposing new metrics and mitigation methods that effectively reduce bias while maintaining embedding utility.

Contribution

It introduces novel metrics for assessing gender bias in gendered languages and extends bias mitigation techniques to bilingual embeddings.

Findings

Gender bias exists in embeddings of Spanish and French.

Proposed methods effectively reduce gender bias.

Bias mitigation preserves embedding utility.

Abstract

Recent studies have shown that word embeddings exhibit gender bias inherited from the training corpora. However, most studies to date have focused on quantifying and mitigating such bias only in English. These analyses cannot be directly extended to languages that exhibit morphological agreement on gender, such as Spanish and French. In this paper, we propose new metrics for evaluating gender bias in word embeddings of these languages and further demonstrate evidence of gender bias in bilingual embeddings which align these languages with English. Finally, we extend an existing approach to mitigate gender bias in word embeddings under both monolingual and bilingual settings. Experiments on modified Word Embedding Association Test, word similarity, word translation, and word pair translation tasks show that the proposed approaches effectively reduce the gender bias while preserving the utility of the embeddings.