AI and English Written Communication in Academic Email Correspondence among University Faculty in Higher Education
This mixed-methods study of 407 university faculty reveals that while ChatGPT use enhances perceived authorial control and self-efficacy in L2 academic email writing, it simultaneously increases writing dependence and leads to a measurable decline in actual discourse quality, creating a critical divergence between perceived and pragmatic competence.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the modern university, a quiet revolution is reshaping how scholars communicate. For decades, writing academic emails, grant proposals, and research papers in English has been a significant hurdle for faculty members whose first language is not English. They have long relied on their own skills, years of practice, and the slow, often difficult process of drafting and revising to ensure their ideas were heard clearly and respectfully. Recently, a new tool has entered this landscape: artificial intelligence. These systems can generate fluent, grammatically perfect text in seconds, promising to level the playing field for non-native speakers. But a critical question remains unanswered: when a professor uses a machine to write their words, does the professor still own the thought? Does the machine help them learn and grow, or does it quietly take over the very skills they need to maintain their professional identity? This is not just about checking for grammar; it is about the nature of authorship, the confidence writers feel in their own abilities, and the subtle social rules that govern how experts speak to one another.
A recent study set out to investigate this exact tension among university faculty members. The researchers focused on how the use of generative AI, specifically a tool called ChatGPT, changes the way professors write their English emails. They wanted to know if using these tools made writers feel more in control and confident, or if it actually made them more dependent on the machine, potentially weakening their ability to write independently. To find the truth, the team did not just ask people what they thought; they asked them to write real emails. They gathered data from 407 university faculty members who regularly write in English. These participants filled out surveys about their habits and feelings regarding AI use, and a smaller group of 33 professors was asked to write actual emails for realistic scenarios, such as requesting time off for a conference or asking for an extension on a deadline. Two expert linguists then read these emails, rating them not just on grammar, but on how well they sounded like a real person speaking to a colleague, how polite they were, and whether they fit the specific situation.
The results revealed a surprising and somewhat troubling disconnect between what the writers felt and what the experts saw. The study found that faculty members who used ChatGPT frequently reported feeling a strong sense of control over their writing. They believed they were the author of the text and felt more confident in their ability to communicate effectively. In fact, the data showed a very strong link between using the AI and feeling a high level of self-assurance. However, this feeling of control was an illusion. When the experts analyzed the actual emails written by the heavy AI users, they found the opposite of what the writers believed. These texts, while grammatically correct, often lacked the specific, human touch required for professional communication. They sounded generic, overly formal, and sometimes failed to address the specific nuances of the relationship between the sender and the recipient. The more a professor relied on the AI to draft their messages, the less authentic and contextually appropriate the final email became.
This phenomenon creates a paradox. The study showed that as faculty members leaned more heavily on AI to write their emails, their actual ability to write independently began to erode. They became dependent on the tool to produce text they felt they could not create on their own. Yet, because the AI produced such polished surface-level text, the writers felt more capable than ever. It is as if a person who uses a calculator to do simple math begins to feel like a math genius, even though they have forgotten how to do the math themselves. The researchers observed that the heavy users were essentially outsourcing their thinking. They would type a prompt, accept the first draft the machine generated, and send it off with minimal changes. In doing so, they lost the opportunity to engage with the difficult but necessary work of finding the right words, which is where true language learning and professional growth happen.
The study also highlighted a specific danger for the academic community. When professors use AI to write, they risk producing text that is "too perfect" to be real. The emails from the high-use group were often filled with standard phrases and polite formulas that sounded like a robot trying to be a human, rather than a human speaking to another human. Experts could easily spot these messages because they lacked the specific details and the genuine voice that come from a person struggling to express a complex idea. In contrast, the professors who used AI less often, or not at all, wrote emails that were sometimes less polished in terms of grammar, but they were far more authentic. These messages showed a clear understanding of the situation, the right level of politeness, and a genuine connection to the recipient. The writers who struggled a bit more with the language actually demonstrated a deeper understanding of the social rules of the academic world.
Ultimately, the research suggests that while AI tools can make writing easier and faster, they may be doing so at a high cost. The study found that the more a faculty member relied on the tool, the more their actual writing skills and their ability to navigate complex social situations in writing declined. The feeling of confidence they experienced was not a sign of improved skill, but a sign of increased dependence. The researchers concluded that there is a measurable gap between how faculty members perceive their own writing abilities when using AI and the actual quality of the discourse they produce. For universities and the scholars who work within them, this means that the widespread use of these tools could be quietly reshaping the way academic knowledge is shared, potentially replacing genuine human connection with a smooth, but hollow, simulation of it. The challenge for the future is to find a way to use these powerful tools without losing the very human voice that makes academic communication meaningful.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.