One of the bigger challenges instructors face is providing detailed, immediate feedback to individual learners. Julia Feldhaus, Senior Lecturer in Languages, Literatures, and Cultures (German) is working collaboratively with students to develop Artificial Intelligence (AI) literacy, provide differentiated feedback, and reflect together on the benefits and drawbacks of using Large Language Models (LLM) in learning.
What motivated you to include AI in your class(es)?
I initially began experimenting with AI because I was nervous it might become the next excuse for students not to learn world languages at UMass. I explored various Large Language Models (LLMs), such as ChatGPT, Claude, Gemini, and read extensively about both the potential and the drawbacks of AI for learning. I was concerned about cognitive offloading -- using AI to avoid deep engagement with learning -- but also intrigued by AI’s potential as a personalized learning coach. As an instructor of German working primarily with STEM students who bring a wide range of skill levels to my upper-level courses, I was curious whether AI could help support students as they navigate their individual language learning journeys.
What’s your approach?
My upper-level language courses include a self-learning project that addresses individual student abilities. The project challenges students to address areas for improvement in their language skills while aligning with their personal interests. I had already built a repository of authentic German learning materials and learning strategies, but some skills, such as speaking and writing, are especially difficult to develop on your own without timely feedback. That’s where AI became useful, serving as a conversation partner and writing consultant for students practicing outside of class. Advanced-level learners can practice speaking freely with the AI platform of their choice. The critical aspect is the feedback loop at the end of the conversation, where students review the transcript and ask AI to identify grammatical errors, suggest alternative vocabulary, or recommend language structures for future speaking practice. As for the writing practice, students upload passages from their own writing to an AI platform, and follow a similar process, asking for an analysis of grammar issues and explanations, and requesting synonyms for vocabulary and linguistic structures. In the future, we'll also be testing working with AI in a Socratic approach to strengthen students' critical thinking skills, by having AI flag passages in their writing that could be further developed, or to encourage students to consider aspects or perspectives that are missing in their answers. Some reflection prompts I use include:
- “Describe how working with AI on this task helped you improve a specific grammar point, your writing style, or your vocabulary.”
- “Share one or two concrete examples of changes you made that were influenced by the AI's suggestions.” (This reflection aims to help students distinguish between their own voice and the AI's input. Additionally, it helps reinforce what students have learned from this task that they can apply toward their future writing.)
I also designed structured writing assignments where students write their drafts first, and then use AI in the revision process. AI offers support by identifying grammar, spelling, and vocabulary issues.
The strength of AI feedback lies in its adaptability. It can be as concise or detailed as the student needs. For example, if a student overuses a particular vocabulary item, AI can point it out and suggest synonyms. Some of these may be familiar but not yet part of the student's active vocabulary. The student can then prompt the AI to provide example sentences using the new word, ask for a definition in the target language, or request a translation in their native language. AI can even generate practice sentences for the student to rephrase, supporting the kind of repetition that's key to language learning. This kind of just-in-time feedback is especially valuable for today's digital-native learners, who are used to receiving immediate feedback, which is something a human instructor cannot always offer outside of class or during take-home assignments. While human feedback remains essential, especially for developing real interpersonal communication skills, AI can serve as a personalized tutor that accelerates language acquisition. It helps students build the confidence and linguistic range they need to engage in more complex discussions and critical thinking earlier in their learning journey.
How do you prepare students to use AI?
My students and I begin the semester with open discussions about ethical AI use and academic integrity. In the first week, we do a gallery walk exercise where students evaluate different examples of AI use. They discuss which examples they find ethically appropriate, and which examples they would avoid. Many students also lack ideas on how to use AI support for their language learning, so I guide them through what AI is good at and where its limits lie. I create space for students to share their own experiences with AI, including what they have found useful and what made them uncomfortable. My goal is for them to see AI as a tool that supports their learning rather than as a shortcut.
What has been the impact of this strategy on student learning?
I’ve seen a significant increase in student agency throughout the self-learning project and a shift in attitudes toward writing. Students feel more confident experimenting with new vocabulary and language structures because they can receive immediate feedback. I’ve also noticed growth in their metacognitive skills. Students are better able to name their challenges, identify what they’re working on, and take ownership of their progress. AI’s ability to quickly differentiate tasks and learning materials has also helped me better support individual learners in my classes.
AI is an ever-evolving topic. What AI pedagogical approaches do you plan to try next?
I’m currently redesigning a third-semester course focused on sustainability in the German-speaking world. AI is a great help in this process, both for finding authentic materials and for creating interactive learning tasks. On the student side, I’m planning to develop custom chatbots tailored to beginner and intermediate learners. While tools like ChatGPT can simulate conversations, their language output is often too advanced for students just beginning their German language journey.
What considerations or tips do you have for other faculty interested in embedding AI in their courses?
A great place to start is C. Edward Watson and José Antonio Bowen’s book Teaching with AI. It offers helpful examples and a thoughtful framework for how AI might reshape education in the future. I recommend thinking first about how AI can support students’ thinking and learning, and modeling what ethical AI use looks like within your field. Be clear with your students about your expectations for AI use in your class and revisit these policies throughout the semester. Students are juggling multiple and sometimes conflicting AI policies across their courses, which can be confusing. Finally, leave space for exploration and reflection. Students are more likely to use AI ethically and effectively when they understand its role in their learning journey, when they trust their instructor, and when they feel like active participants in the learning process.
