arXiv cs.CL· Naho Orita, Hayato Ogawa, Daisuke Kawahara·· 14 小时前AI 评分12
研究人类与 LLM 对非母语日语的语言态度对齐的论文被撤稿
Human-LLM Alignment in Language Attitudes Toward Non-Native Japanese
AI 导读
论文《Human-LLM Alignment in Language Attitudes Toward Non-Native Japanese》已被作者 Naho Orita 撤稿，原因是所谓 L2 日语学习者邮件实为母语者用各自语言所写邮件的日语直译（I-JAS FOLAS 语料），并非 L2 写作，标题与关于学习者写作的结论无法成立。
正文 · 原文
This paper has been withdrawn by Naho Orita
No PDF available, click to view other formats
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.
| Comments: | Withdrawn: our stimuli, described as emails written by L2 Japanese learners, are in fact Japanese literal translations (I-JAS FOLAS corpus) of emails native speakers wrote in their own languages. The stimuli are not L2 writing, so the paper's title and conclusions about learner writing cannot be sustained. We apologize to readers |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2608.01629 [cs.CL] |
| (or arXiv:2608.01629v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2608.01629 arXiv-issued DOI via DataCite |
Submission history
From: Naho Orita [view email]
[v1]
Mon, 3 Aug 2026 02:59:59 UTC (1,031 KB)
[v2]
Sun, 25 Oct 2026 07:42:40 UTC (1 KB) (withdrawn)
来源：arXiv cs.CL · arxiv.org