Quantifying Gender Bias Towards Politicians in Cross-Lingual Language Models

TL;DR

This study investigates gender bias in multilingual language models by analyzing adjective and verb usage around politicians' names across seven languages, revealing language-dependent biases and challenging assumptions about model size and bias.

Contribution

Introduces a multilingual probing method and a large dataset to quantify gender bias towards politicians in various language models, highlighting language-specific biases and size-related findings.

Findings

Bias varies significantly across languages.

Certain words are gender-specificly associated with politicians.

Model size does not correlate with increased bias.

Abstract

Recent research has demonstrated that large pre-trained language models reflect societal biases expressed in natural language. The present paper introduces a simple method for probing language models to conduct a multilingual study of gender bias towards politicians. We quantify the usage of adjectives and verbs generated by language models surrounding the names of politicians as a function of their gender. To this end, we curate a dataset of 250k politicians worldwide, including their names and gender. Our study is conducted in seven languages across six different language modeling architectures. The results demonstrate that pre-trained language models' stance towards politicians varies strongly across analyzed languages. We find that while some words such as dead, and designated are associated with both male and female politicians, a few specific words such as beautiful and divorced are predominantly associated with female politicians. Finally, and contrary to previous findings, our study suggests that larger language models do not tend to be significantly more gender-biased than smaller ones.