Tutors are vital to language instruction, providing nuanced feedback to writing; creating prompts and exercises tailored to areas that need improvement; and engaging in oral conversations. But they can’t always be available at all times that are convenient for the student.
That’s where recent advances in artificial intelligence and large language models can come in, according to Attia Youseif, director of tutoring for the Arabic Flagship Program at the Indiana University Hamilton Lugar School of Global and International Studies.
Youseif, who taught Arabic at IU since 2017, is leading an international conversation about how to harness the power of tools like ChatGPT or Google’s Gemini to enhance language instruction and learning. He said these large language models can improve students’ skills in all areas of language learning: reading, writing, listening and speaking.
In his role as director of tutoring for the Arabic Flagship Program — an intensive language program designed for students in any major to develop professional proficiency in Arabic — Youseif recruits, trains and manages a team of native Arabic speakers who work as language tutors. Every flagship student spends two hours each week speaking one-on-one with their tutor. While this time is incredibly valuable for students, Youseif said it’s nowhere near enough.
“Additional training must happen outside of the four hours students spend in my classroom each week, and these new tools make it more convenient and playful for students to do that,” Youseif said. “A student can be in the car driving or walking from class to class and also engaging in conversation.”
Youseif also encourages students to solicit feedback from large language models on first drafts of written assignments and rewrite based on the response. While earlier versions had limited Arabic capabilities, the tools now can provide sufficient edits that extend beyond typos and grammar to include critiques of word choice, sentence structure and other content. Youseif said this approach has not only further engaged this digital generation of students, but it’s also made grading papers easier.
Streamlining grading is just one way these AI-powered tools can improve language instruction. Using more advanced versions of the tools, Youseif said instructors can input desired learning outcomes and standards and receive prompts and exercises tailored to help students meet them. Because the large language model has access to more information than any instructor can analyze, these exercises are likely to be more efficient and effective.
Currently, Youseif is conducting research on which large language models are most useful to engage with when practicing writing, reading, listening and speaking. He and several colleagues are also interested in creating a digital Arabic textbook that incorporates activities within the text that leverage AI tools to give students feedback in real time.
While much of the conversation around AI has centered on the fear that these tools will ultimately eliminate jobs done by humans, Youseif said he believes large language models could never replace himself and his colleagues.
“When we teach languages, we teach humanity,” Youseif said. “And that’s something that large language models will never be able to do. Instructors are similar to a coach or trainer, and these tools simply enhance students’ abilities to practice what we teach them on their own.”
Youseif has been working with the Language Flagship Innovation Center at the University of Hawaii to help other leading language programs in the U.S. utilize AI as a comprehensive resource for tutors across various languages. The initiative also offers guidelines for evaluating the output of AI-generated prompts, supplemented with illustrative examples.
He has also done extensive international travel to present his findings on the use of AI in language learning to other academics around the world, including lectures at Cambridge University, University of Manchester, University of Exeter, University of Leeds and Scotland’s Al-Maktoum College of Higher Education in the United Kingdom; University of Monastir, University of Tunis and Carthage University in Tunisia; The American University in Cairo in Egypt; and the Arab League Educational, Cultural and Scientific Organization, which is also headquartered in Tunisia.
IU’s place at the forefront of incorporating these transformative tools is just one example of the university’s national leadership in world language instruction. IU Bloomington offers more languages than any other university in the U.S. Even amidst national cuts, the university boasts three Language Flagship programs — an initiative of the Defense Language and National Security Education Office within the U.S. Department of Defense — the most in the nation.
“IU is known for languages, and it’s the reason many students decide to come here,” Youseif said. “We owe it to them to use the most cutting-edge approaches to language learning.
“The Hamilton Lugar School motto is ‘globally ready.’ Being globally ready starts with learning languages, and the key to mastering this is leveraging AI and LLMs in language instruction. By integrating these advanced technologies into our learning processes, we can unlock new levels of proficiency and preparedness.”