# Grammatical gender associations outweigh topical gender bias in crosslinguistic word embeddings

arXiv: https://arxiv.org/abs/2005.08864 · HTML: https://mlsanity.com/paper/2005.08864/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 May 2020.

## Built on by (most cited first, 9 of 9)

- [RedditBias: A Real-World Resource for Bias Evaluation and Debiasing of Conversational Language Models](https://mlsanity.com/paper/2106.03521/) (2021)
- [Theories of "Gender" in NLP Bias Research](https://mlsanity.com/paper/2205.02526/) (2022)
- [Sociodemographic Bias in Language Models: A Survey and Forward Path](https://mlsanity.com/paper/2306.08158/) (2023)
- [DebIE: A Platform for Implicit and Explicit Debiasing of Word Embedding Spaces](https://mlsanity.com/paper/2103.06598/) (2021)
- [Measuring Gender Bias in Word Embeddings of Gendered Languages Requires Disentangling Grammatical Gender Signals](https://mlsanity.com/paper/2206.01691/) (2022)
- [Cross-lingual Transfer Can Worsen Bias in Sentiment Analysis](https://mlsanity.com/paper/2305.12709/) (2023)
- [Evaluating Short-Term Temporal Fluctuations of Social Biases in Social Media Data and Masked Language Models](https://mlsanity.com/paper/2406.13556/) (2024)
- [Data Bias According to Bipol: Men are Naturally Right and It is the Role of Women to Follow Their Lead](https://mlsanity.com/paper/2404.04838/) (2024)
- [Estimating Grammatical Gender Directions in Contextual Embeddings under Controlled and Natural Contexts](https://mlsanity.com/paper/2606.30152/) (2026)

## Abstract (arXiv)

Recent research has demonstrated that vector space models of semantics can reflect undesirable biases in human culture. Our investigation of crosslinguistic word embeddings reveals that topical gender bias interacts with, and is surpassed in magnitude by, the effect of grammatical gender associations, and both may be attenuated by corpus lemmatization. This finding has implications for downstream applications such as machine translation.