Newest papers

Investigating the Impact of Learner Errors on Social Bias in Large Language Models

Federico Marcuzzi, Martin Christoph, Iryna Gurevych, Alla Rozovskays · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29896 · cited by 0 works (OpenAlex) · open-access PDF

Indonesian

Untuk mengatasi kesenjangan ini, kami meneliti bagaimana kesalahan tata bahasa oleh penulis non-native mempengaruhi bias sosial. Kami mengembangkan kerangka kerja terkontrol yang menyuntikkan kesalahan belajar umum ke dalam prompts dan menerapkannya pada tiga benchmark bias yang mencakup penyesuaian stereotip dan preferensi demografis di antara gender, ras, dan agama. Kami menemukan bahwa kesalahan tata bahasa meningkatkan bias, dan efeknya meningkat dengan tingkat kesalahan. Di antara empat jenis kesalahan umum yang kami pertimbangkan, kesalahan preposition menghasilkan peningkatan terbesar dan paling konsisten dalam penyesuaian stereotip. Analisis kami menunjukkan bahwa kesalahan meningkatkan bias ketika mereka lebih cenderung mengganggu pemahaman. Akhirnya, penyesuaian halus pada teks yang salah mengurangi bias yang mungkin meningkat oleh pengguna bias yang dipromosikan dalam bahasa Inggris. penilaian bias kami berdasarkan penilaian secara eksklusif menunjukkan

Machine translation (Meta M2M100); the English original follows.

Abstract

Existing social bias benchmarks evaluate large language models mostly on fluent, well-formed English, reflecting native speaker usage and leaving non-native writers unrepresented. To address this gap, we investigate how grammatical errors by non-native writers affect social bias. We develop a controlled framework that injects common learner errors into prompts and apply it to three bias benchmarks covering stereotype alignment and demographic preference across gender, race, and religion. We find that grammatical errors increase bias, and the effect grows with the error rate. Among the four common error types we consider, preposition errors produce the largest and most consistent increase in stereotype alignment. Our analysis suggests that errors increase bias when they are more likely to interfere with comprehension. Finally, fine-tuning on erroneous texts reduces the bias increase caused by grammatical errors. Our findings show that bias evaluations based exclusively on well-formed prompts may underestimate the bias experienced by users whose English contains grammatical errors.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Stereotype (UML), Affect (linguistics), Psychology, Preference, Gender bias, Computer science

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange