Six state-of-the-art LLMs systematically prefer Standard American English over AAE continuations, and a training-free activation steering method reduces this bias 5-20x more than prompting while preserving fluency.
Multi- VALUE : A Framework for Cross-Dialectal E nglish NLP
3 Pith papers cite this work, alongside 17 external citations. Polarity classification is still indexing.
3
Pith papers citing it
17
external citations · OpenAlex
fields
cs.CL 3years
2026 3representative citing papers
Dialectal robustness and generation are dissociated in LLMs: benchmarks are driven by pretraining and SFT while alignment reshapes generation invisibly to benchmarks, and the method maximizing dialectal reward is least preferred by human evaluators.
Random forests on string similarity features outperform LLMs for German dialect lexicon induction and boost dialect information retrieval by up to 50% in recall.
citing papers explorer
-
LLMs Silently Correct African American English: Auditing and Mitigating Dialect Bias via Activation Steering
Six state-of-the-art LLMs systematically prefer Standard American English over AAE continuations, and a training-free activation steering method reduces this bias 5-20x more than prompting while preserving fluency.
-
DiaLLM: An Investigation into the Robustness-Generation Gap in English Dialect Adaptation
Dialectal robustness and generation are dissociated in LLMs: benchmarks are driven by pretraining and SFT while alignment reshapes generation invisibly to benchmarks, and the method maximizing dialectal reward is least preferred by human evaluators.
-
Resource-Lean Lexicon Induction for German Dialects
Random forests on string similarity features outperform LLMs for German dialect lexicon induction and boost dialect information retrieval by up to 50% in recall.