arXiv cs.CL· Zandi Eberstadt·· 12 小时前AI 评分23
指令微调模型比人类产生更多句法重叠，但条件复用更少
Instruction-Tuned Models Locally Produce More Syntactic Overlap -- But Not More Conditional Reuse -- Than Humans Do
AI 导读
研究基于 Blevins 等人的替换范式数据，在 16 个开源 Llama 和 Gemma 模型（1B-70B，含预训练与指令微调版本）上各测 1,901 个匹配位置，衡量相邻轮次对上下文无关文法（CFG）规则的复用。
正文 · 原文
Abstract:Syntactic convergence - the tendency of speakers to adapt in language towards the grammatical profiles of their interlocutors - is a well-documented, largely implicit feature of human dialogue. Whether large language models exhibit analogous syntactic convergence towards human users relative to human baselines (including across a broad range of syntactic constructions) remains an open question. Using substitution-paradigm data from Blevins et al. [2026] in which model generations replace one speaker's turns in pre-existing human dialogues, this study measures turn-adjacent reuse of context-free grammar (CFG) rules across sixteen open-weight Llama and Gemma models (1B-70B, pretrained and instruction-tuned) at 1,901 matched positions per model. Every model showed greater CFG-rule overlap with the preceding human turn than with a sampled unrelated human prime (a context-associated effect that does not isolate syntactic priming from topical continuity), and this actual-versus-random difference was larger for lower-frequency rules. Each instruction-tuned model also showed greater natural-output overlap with the actual prime than the human response it replaced, and all eight matched architecture pairs exhibited greater actual-prime overlap after instruction tuning. However, relative to pretrained variants, instruction-tuned outputs overlapped more with unrelated primes, showed a smaller actual-versus-random increment, and had lower conditional rule-reuse odds once target rule-set size was held constant. In exploratory analyses, each model exhibited greater mean lexical and semantic similarity to the preceding turn than matched human responses did. Instruction-tuned models additionally produced responses with greater mean semantic similarity than their pretrained counterparts in all eight architecture pairs, whereas the lexical similarity results were more heterogeneous.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2607.26015 [cs.CL] |
| (or arXiv:2607.26015v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.26015 arXiv-issued DOI via DataCite |
Submission history
From: Zandi Eberstadt [view email]
[v1]
Tue, 28 Jul 2026 17:27:47 UTC (47 KB)
[v2]
Sun, 1 Nov 2026 00:18:39 UTC (47 KB)
来源：arXiv cs.CL · arxiv.org