AI writing tools may not close the gap between native and non-native English writers as neatly as hoped. In an online experiment, non-native speakers drafting a job cover letter leaned toward simple AI-generated wording, while native speakers more readily adopted advanced, sophisticated phrasing.
The study, by Inyoung Shin of Yale University, Hyesun Choung of Purdue University and Mina Choi of Kyung Hee University, appears in the April 2026 issue of Computers in Human Behavior (volume 177) after being published online on December 24, 2025. Its title is “Bridging or Broadening Gaps? AI-Assisted Professional Writing among Native and Non-Native English Writers”.
A cover letter with AI help
The researchers recruited 327 native and non-native English writers in the United States and asked each to write a cover letter for a hypothetical job. Participants were then randomly given AI-generated content written at either a simple or an advanced vocabulary level and told they could revise their letter however they liked, including by borrowing from the AI text.
Using natural language processing, the team measured how much of the AI content ended up in each final letter. They also recorded participants’ writing self-efficacy, or confidence in their own writing, and asked whether they judged the AI text to be better than their own.
Confidence shaped the gap
Non-native writers favored the simpler AI content, while native writers tended to incorporate the advanced version. The difference was most pronounced among people with lower writing self-efficacy.
Perceived superiority of the AI text did not explain the pattern, the authors report. They conclude that while AI can support lower-skilled groups on specific tasks, sociolinguistic gaps remain even with AI assistance, and they call for AI training that builds skills while addressing both proficiency gaps and social constraints.
Study Details:
- Title: Bridging or Broadening Gaps? AI-Assisted Professional Writing among Native and Non-Native English Writers
- Authors: Inyoung Shin, Hyesun Choung, Mina Choi
- Journal: Computers in Human Behavior
- Publication Date: December 24, 2025
- DOI: 10.1016/j.chb.2025.108897
