An Analysis of Social Biases Present in BERT Variants Across Multiple Languages

Parishad BehnamGhader, Aristides Milios

Keywords: ML, bias, social biases, NLP, BERT, MLM

TL;DR: This paper examines social biases present in monolingual BERT models for the English, Greek, and Persian languages, testing several metrics as well as visualizing the biases and drawing conclusions based on the data the models were trained on.

Abstract: Although large pre-trained language models have achieved great success in many NLP tasks, it has been shown that they reflect human biases from their pre-training corpora. This bias may lead to undesirable outcomes when these models are applied in real-world settings. In this paper, we investigate the bias present in monolingual BERT models across a diverse set of languages (English, Greek, and Persian). While recent research has mostly focused on gender-related biases, we analyze religious and ethnic biases as well and propose a template-based method to measure any kind of bias, based on sentence pseudo-likelihood, that can handle morphologically complex languages with gender-based adjective declensions. We analyze each monolingual model via this method and visualize cultural similarities and differences across different dimensions of bias. Ultimately, we conclude that current methods of probing for bias are highly language-dependent, necessitating cultural insights regarding the unique ways bias is expressed in each language and culture (e.g. through coded language, synecdoche, and other similar linguistic concepts). We also hypothesize that higher measured social biases in the non-English BERT models correlate with user-generated content in their training.