EMNLP Findings2025
CLEAR: A Comprehensive Linguistic Evaluation of Argument Rewriting by Large Language Models
Abstract
While LLMs have been extensively studied on general text generation tasks, there is less research on text rewriting, a task related to general text generation, and particularly on the behavior of models on this task. In this paper we analyze what changes LLMs make in a text rewriting setting. We focus specifically on argumentative texts and their improvement, a task named Argument Improvement (ArgImp). We present CLEAR: an evaluation pipeline consisting of 57 metrics mapped to four linguistic levels: lexical, syntactic, semantic and pragmatic. This pipeline is used to examine the qualities of LLM-rewritten arguments on a broad set of argumentation corpora and compare the behavior of different LLMs on this task and analyze the behavior of different LLMs on this task in terms of linguistic levels. By taking all four linguistic levels into consideration, we find that the models perform ArgImp by shortening the texts while simultaneously increasing average word length and merging sentences. Overall we note an increase in the persuasion and coherence dimensions. original improved lexical syntactic semantic pragmatic How does the vocabulary change? How is the sentence structure modified? How does the meaning shift? How is the argument's effectiveness affected? Arg Rewrite Argument Annotated Essays Microtexts You are given an argument about the topic topic. Your task is to improve it. Respond only with the improved argument wrapped in @ symbols and nothing else. Here are some examples of improvements:
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
1 top-tier paper has cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- SELF-DISCOVER: Large Language Models Self-Compose Reasoning Structures cites Language Models are Few-Shot Learners
- Connecting Large Language Models with Evolutionary Algorithms Yields Powerful Prompt Optimizers cites Language Models are Few-Shot Learners
Cited by top-tier papers1
Builds on19
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Chatbot Arena: An Open Platform for Evaluating LLMs by Human PreferenceWei-Lin Chiang, Lianmin Zheng, Ying Sheng, Anastasios Nikolas Angelopoulos et al.ICML 2024 · 1,212 citations
- Crosslingual Generalization through Multitask FinetuningNiklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts et al.ACL 2023 · 319 citations
- Connecting Large Language Models with Evolutionary Algorithms Yields Powerful Prompt OptimizersQingyan Guo, Rui Wang, Junliang Guo, Bei Li et al.ICLR 2024 · 257 citations
- SELF-DISCOVER: Large Language Models Self-Compose Reasoning StructuresPei Zhou, Jay Pujara, Xiang Ren, Xinyun Chen et al.NeurIPS 2024 · 151 citations
Related papers
- RewriteLM: An Instruction-Tuned Large Language Model for Text RewritingLei Shu, Liangchen Luo, Jayakumar Hoskere, Yun Zhu et al.AAAI 2024 · 92 citations
- Learning to Rewrite: Generalized LLM-Generated Text DetectionWei Hao, Ran Li, Weiliang Zhao, Junfeng Yang et al.ACL 2025
- Shaping Human-AI Collaboration: Varied Scaffolding Levels in Co-writing with Language ModelsParamveer S. Dhillon, Somayeh Molaei, Jiaqi Li, Maximilian Golub et al.CHI 2024 · 102 citations
- Mind the Style Gap: Meta-Evaluation of Style and Attribute Transfer MetricsAmalie Brogaard Pauli, Isabelle Augenstein, Ira AssentEMNLP Findings 2025
- Discourse Realization of Generics in Human and LLM-generated TextsSøren Kirkegaard Fomsgaard, Martial Pastor, Gaël Dias, Nelleke OostdijkACL 2026
