Research Summary: Human-written and AI-generated revisions: an appraisal analysis of dialogic positioning in argumentative writing

A study comparing human-written and AI-generated revisions in argumentative writing found that AI-generated revisions (from ChatGPT-4) utilized significantly more 'appraisal resources' than peer-feedback revisions, which in turn used more than initial drafts. This suggests a quantifiable difference in the dialogic positioning and persuasive language incorporated by AI compared to human revision processes, specifically within academic writing.

Why it matters

This research provides empirical evidence of how AI-driven text generation differs from human revision in terms of persuasive language use. Understanding these distinctions is critical for organizations developing or deploying AI tools for content creation and revision, as it influences the perceived authority, tone, and engagement of AI-generated communications.

Key insights

• The study systematically compared appraisal resource use in human-written and AI-generated texts.
• It analyzed first drafts, peer-feedback revisions, and AI-generated revisions of argumentative essays.
• AI-generated revisions were produced by ChatGPT-4 based on student first drafts using a standardized prompt.
• A corpus of 168 essays from Chinese undergraduate English majors across three draft types was used.
• AI-generated revisions employed significantly more appraisal resources than peer-feedback revisions.
• Peer-feedback revisions used more appraisal resources than the initial first drafts.

Source

Frontiers in Education — https://www.frontiersin.org/articles/10.3389/feduc.2026.1904852