Research

Paper

TESTING March 16, 2026

Which stylistic features fool ChatGPT research evaluations?

Authors

Kayvan Kousha, Mike Thelwall

Abstract

Large Language Models (LLMs) have the potential to be used to support research evaluation and have a moderate capability to estimate the research quality of a journal article from its title and abstract. This paper assesses whether there are language-related factors unrelated to the quality of the research that influence ChatGPT's scores. Using a dataset of 99,277 journal articles submitted to the UK-wide Research Excellence Framework (REF) 2021 assessments, we calculated several readability indicators from abstracts and correlated them with ChatGPT scores and departmental REF scores. From the results, linguistic complexity and length were more strongly associated with ChatGPT research quality scores than with REF expert scores in many subject areas. Although cause-and-effect was not tested, these results suggest that ChatGPT may be more likely than human experts to reward linguistic complexity, with a potential bias towards longer and less readable abstracts in many fields. The apparent preference of LLMs for complex language is an undesirable feature for practical applications of LLMs for research quality evaluation, unless solutions can be found.

Metadata

arXiv ID: 2603.14919
Provider: ARXIV
Primary Category: cs.DL
Published: 2026-03-16
Fetched: 2026-03-17 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.14919v1</id>\n    <title>Which stylistic features fool ChatGPT research evaluations?</title>\n    <updated>2026-03-16T07:25:46Z</updated>\n    <link href='https://arxiv.org/abs/2603.14919v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.14919v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Large Language Models (LLMs) have the potential to be used to support research evaluation and have a moderate capability to estimate the research quality of a journal article from its title and abstract. This paper assesses whether there are language-related factors unrelated to the quality of the research that influence ChatGPT's scores. Using a dataset of 99,277 journal articles submitted to the UK-wide Research Excellence Framework (REF) 2021 assessments, we calculated several readability indicators from abstracts and correlated them with ChatGPT scores and departmental REF scores. From the results, linguistic complexity and length were more strongly associated with ChatGPT research quality scores than with REF expert scores in many subject areas. Although cause-and-effect was not tested, these results suggest that ChatGPT may be more likely than human experts to reward linguistic complexity, with a potential bias towards longer and less readable abstracts in many fields. The apparent preference of LLMs for complex language is an undesirable feature for practical applications of LLMs for research quality evaluation, unless solutions can be found.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.DL'/>\n    <published>2026-03-16T07:25:46Z</published>\n    <arxiv:primary_category term='cs.DL'/>\n    <author>\n      <name>Kayvan Kousha</name>\n    </author>\n    <author>\n      <name>Mike Thelwall</name>\n    </author>\n  </entry>"
}