Safety
Human-LLM Alignment in Language Attitudes Toward Non-Native Japanese
arXiv:2608.01629v1 Announce Type: new Abstract: Large language models (LLMs) increasingly evaluate human writing in high-stakes domains such as hiring and academic assessment, putting non-native speak
arXiv:2608.01629v1 Announce Type: new Abstract: Large language models (LLMs) increasingly evaluate human writing in high-stakes domains such as hiring and academic assessment, putting non-native speakers at particular risk. Drawing on the language attitudes framework, we compared human and LLM evaluations of parallel L1- and L2-written Japanese emails on three dimensions: fluency, status, and solidarity. Japanese raters rated L2 texts significantly lower on all three dimensions, with a fluency gap roughly twice the size of the status and solidarity gaps. Six LLM judges reproduced the direction of this bias, and five reproduced its ordering across dimensions. The models diverged from humans in two ways: all understated the solidarity gap, the most socially grounded dimension, and all differentiated among learner L1 backgrounds where humans did not. LLM judges thus reproduce native speakers' language attitudes in a structured yet attenuated form, and the language attitudes framework offers a ready-made yardstick for auditing them beyond English.
Related
- Beyond Marginal Distributions: A Framework to Evaluate the Representativeness of Demographic-Aligned LLMs
- C-Mining: Unsupervised Discovery of Seeds for Cultural Data Synthesis via Geometric Misalignment
- Alignment Data Map for Efficient Preference Data Selection and Diagnosis
Source: arXiv cs.CL | 2026-08-04