Dr. Yongfang Zhang thinks artificial intelligence — with thoughtful integration — can be a powerful tool for learning a new language.

During July, Zhang, associate professor of Chinese and coordinator of the Chinese Studies program, led Armala Miller ’27 and Anna Kerr Reavis ’27 in a research project to determine how AI can aid writing processes for second-language learners and deepen their understanding of grammar, sentence structure and vocabulary.

“As AI becomes more prevalent, it is important for educators to use it efficiently to help students learn,” says Zhang, who studied the topic in June at the “Empowering Language Teachers to Leverage AI for Student Success”, one of the six summer institutes hosted online by The Center for Advanced Research on Language Acquisition at the University of Minnesota. Zhang was one of 10 recipients of the 2026 Priority Teacher Professional Development Scholarship from CARLA to attend a summer institute.

“The institute helped prepare me for collaborative research with the students as I experimented with a wide variety of AI tools for developing instructional and learning materials,” she says.

With Zhang’s guidance, Miller and Reavis began the project by reviewing literature on AI as a teaching tool.

“We saw consistent themes — students frequently use AI to reiterate topics and generate deeper explanations, but teachers don’t have good outcomes in using AI to teach language because they lack training on how to engage with it,” says Miller, a Chinese major from Greenville, South Carolina.

To find the best formula for AI-generated feedback to assist student learning, the researchers put three platforms to the test. They plugged a rough paragraph in Chinese into ChatGPT, Microsoft Copilot and Google Gemini, asking each platform to correct errors and explain the changes. The students rated each answer for clarity and helpfulness.

At first, the generated responses were vague and suggested phrases that were too advanced for intermediate-level writing. However, as the team refined its prompts — focusing on a single area like grammar or vocabulary, specifying learners’ proficiency and asking for charts to help visualize feedback — it received clear, accurate and proficiency-appropriate corrections and explanations. The researchers say the findings demonstrate that prompt design plays a critical role in the quality of GenAI-generated feedback. In addition to developing prompts that enable GenAI to support multiple stages of the second-langage writing process, the team compared feedback generated by different platforms, identified the strengths and limitations of each and developed recommendations for instructional use.

“This research has shown me that AI can be a great resource if you know how to feed it the right information,” says Miller, who plans to become a teacher. “Now, I have the tools to implement it in my future career and share these best practices with students.”

Reavis, a biology major and Chinese minor from Greenville, South Carolina, says the study also shows how AI can help people grow their language skills independently.

“AI can give you instant feedback when you don’t have other resources and can tailor answers and explanations to your specific needs,” she says. “It will help me maintain my Chinese proficiency after I graduate.”

For Zhang, close collaboration with students was especially important for this project.

“As a language teacher, I need to know what is most helpful for students,” she says. “With Armala’s and Anna’s help, I get the learner’s perspective, and the study has input from both sides.”

Zhang, Miller and Reavis plan to present their research at the Student Research Presentations and Reception during Family Weekend on Sept. 26 and at the Southern Conference Undergraduate Research Forum on Oct. 9-10.