Side-by-side Comparison Amplifies Dialect Bias in Language Models
For NLP practitioners and fairness researchers, this work reveals that existing evaluations underestimate dialect bias severity in contrastive settings, highlighting a critical gap in current mitigation strategies.
This paper quantifies covert dialect bias in language models, finding that bias against African-American Vernacular English (AAVE) is significantly exacerbated when tweets are compared side-by-side with Standard American English (SAE), a setting common in high-stakes decision-making. Counterfactual fairness finetuning reduces bias in isolated evaluations but not consistently in side-by-side comparisons.
Language models (LMs) can exhibit systematic biases against speakers based on variations in their dialects, even in the absence of a dialect label, a behavior known as covert dialect bias. In this work, we quantify covert dialect bias in online discourse by evaluating how LMs associate stereotypical traits (derived from social psychology research on racial bias) with intent-equivalent tweets in Standard American English (SAE) and African-American Vernacular English (AAVE). While prior work shows that LMs associate more negative stereotypes with AAVE when evaluating tweets in isolation, we are surprised to find that this bias is significantly exacerbated when SAE / AAVE tweet pairs are compared side by side, a setting that more closely reflects high-impact decision making contexts in which models are used to rank candidates. The bias only worsens when dialect labels are explicitly specified. This is striking, given the extensive efforts from commercial developers to mitigate bias in their LMs. Encouragingly, we show that counterfactual fairness finetuning can mitigate covert dialect bias for some stereotypical traits, reducing average disparities when evaluating tweets in isolation, however, these improvements do not consistently hold across traits when evaluating SAE / AAVE tweets side by side. Our findings show that existing evaluation settings for covert dialect bias may underestimate its severity, specifically in contrastive settings. Additionally, overt dialect bias remains pronounced even after safety aligned finetuning, indicating that it remains an unresolved problem, and motivates the need for more robust evaluation and mitigation frameworks.