arXiv cs.CL· Evelina Leivada, Raquel Montero, Paolo Morosi, Natalia Moskvina, Tamara Serrano, Nikoleta Pantelidou, Marcel Aguilar, Fritz Guenther·· 12 小时前AI 评分38
研究：大语言模型的概率无法可靠区分可能语言与不可能语言
Large Language Model probabilities cannot reliably distinguish between possible and impossible language
AI 导读
一项新基准对 8 个模型提取对数概率并计算最小对困惑度差，发现不合语法句并未呈现独特的困惑度特征：语用异常句的困惑度与之相当，语义异常句的困惑度反而持续更高。研究表明概率不能作为语法性的可靠代理，仅凭概率差异无法推断模型内部的句法知识表征，关于模型能否区分可能语言与不可能语言的主张需用其他方法验证。
正文 · 原文
Abstract:A controversial test for Large Language Models concerns the ability to discern possible from impossible language. While some evidence attests to the models sensitivity to what crosses the limits of grammatically impossible language, this evidence has been contested on the grounds of the soundness of the testing material. We use model-internal representations as a direct means of investigating the way Large Language Models represent the grammatical-ungrammatical distinction. In a novel benchmark, we elicit sum log probabilities from 8 models and compute minimal-pair surprisal differences, juxtaposing probabilities assigned to grammatical sentences to probabilities assigned to (i) lower frequency grammatical sentences, (ii) ungrammatical sentences, (iii) semantically odd sentences, and (iv) pragmatically odd sentences. The prediction is that if string-probabilities can function as proxies for the limits of grammar, the ungrammatical condition will stand out among the conditions that involve linguistic violations, showing a spike in the surprisal rates. Our results do not reveal a unique surprisal signature for ungrammatical prompts: across models, we observe comparable surprisal for the pragmatically odd condition, and the semantically odd condition consistently shows even higher surprisal. We thus demonstrate that probabilities do not constitute reliable proxies for grammaticality, showing that probability differences alone cannot be used to infer model-internal representations of syntactic knowledge. Consequently, claims about models being able to distinguish possible from impossible language need verification through a different methodology.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2509.15114 [cs.CL] |
| (or arXiv:2509.15114v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2509.15114 arXiv-issued DOI via DataCite |
Submission history
From: Evelina Leivada [view email]
[v1]
Thu, 18 Sep 2025 16:17:48 UTC (695 KB)
[v2]
Mon, 26 Oct 2026 08:49:23 UTC (338 KB)
来源：arXiv cs.CL · arxiv.org