Automating authenticity
Artificial intelligence assists in learning heritage languages.
by Sarah Park
As a second-generation Korean American with the equivalent of a second grader’s spelling and vocabulary knowledge of Korean, writing is a painstaking process for me. One of the few times I write in Korean is on my parents’ birthdays.
I try to turn my English stream of consciousness into meaningful Korean sentences, resorting to translation devices only to look up each word I do not know how to spell.
My siblings had copied my tradition in years past, so I expected to find my brother handwriting his letter to my mother. I was shocked when I found him using artificial intelligence, typing English into ChatGPT and requesting a translation into Korean.
If my 13-year-old brother could use AI for a deeply personal letter, I couldn’t imagine how easy it would be for language learners to do the same for their classes.
Associate professor of instruction of Asian Languages and Cultures Chin-Hung Chang had similar worries. In the Chinese language classes she teaches, she was able to implement a solution for students to utilize AI as a learning tool.
Chang’s colleague, associate professor of instruction of Asian Languages and Cultures Eun Hee Kim, first approached her with the idea of a chatbot to help students practice their conversational skills. After working together to apply for a grant, the two won NU’s Alumnae Award for Curriculum Innovation, jumpstarting a collaboration between them and the Weinberg College Media and Design Studio to make Jack, a chatbot on ChatGPT’s 4.1 model.
In higher-level language classes, AI may reduce originality with assignments like essay writing, Chang says. However, Jack, as a technical language learning tool, supplements students’ learning instead of doing it for them — like offering speaking practice, she explains.
“After class, I ask my students to chat with Jack because they may not have the resources to chat with a real language speaker or they don’t want to do the in-person conversation,” Chang says. “Then, they are able to have real conversations that match the natural flow outside of the class.”
Students’ reactions to Jack have been positive, with 91% saying that they agree or strongly agree that the AI chatbot increased opportunities to practice the target language outside the classroom, according to a survey Chang gave out to her classes.
Weinberg first-year Marisa Lin used Jack to practice conversational skills about vocabulary and set study goals in her Chinese class.
“When you’re speaking in class with a partner, you’re not necessarily getting feedback on grammar,” Lin says. “With the chat bot conversations, it’s going to catch all of your errors. It’s pretty convenient.”
Students across the world also see the benefits of using AI chatbots for language learning, according to a 2024 study by researchers Jinming Du and Ben Kei Daniel.
After examining 24 studies published on the use of AI chatbots, Du and Daniel found that the chatbots could alleviate speaking anxiety, correct pronunciation, promote fluency and enhance confidence. These conversations eliminate the fear of making errors, making students more willing to practice.
Although McCormick second-year Rishabh Balachandran sees the merits in using AI for language learning, he chooses to familiarize himself with languages through “comprehensible input.” By using this method, Balachandran has studied Mandarin, Japanese, Portuguese and Italian.
Rather than systematically memorizing vocabulary, Balachandran believes in the comprehensible input approach, which uses context to develop listening and reading language skills. This format has made him more comfortable speaking with people.
“I would listen to something that I understood 85% of, and 15% of the words I didn’t understand,” Balachandran says. “That helped me improve a lot. Comprehensible input is just the most natural way to learn a language. If you’re copying patterns of a native speaker, you’re [going to] sound more like a native speaker.”
Maintaining knowledge of a language and bonds with native language speakers are deeply intertwined for Balachandran. While studying at National Taiwan University, he practiced his Mandarin with his host family. He keeps in touch with them by texting and calling once or twice a month.
“My host family is almost like my family,” Balachandran says. “I call them all ‘mom, dad, sister.’ They speak some English, but it’s much easier for me to speak in Chinese. If I do one day forget my Chinese, then I also lose a connection with a lot of people. My host family will still know me, but it will be completely different.”
Balachandran also uses a language learning app called Pimsleur to analyze his pronunciation.
Still, he encourages students to speak to people to grow connections. Although AI is a helpful tool and can simulate talking to someone, it takes away from what makes learning a language special, he explains.
“Language learning is very intimate,” Balachandran says. “You have to talk to native speakers and the people in your class. That always makes it a personal thing.”
Similarly, Weinberg first-year Marni Kim is taking Korean classes as an opportunity to improve communication with her relatives in Korea. The language barrier often made it difficult for her to speak in Korean without translation, she says.
Although Kim had taken French throughout high school, she made it a priority to study Korean in college.
“I probably could have placed into a higher level for French than Korean, but I wanted to start from scratch because it is really important to me and my family,” Kim says. “I’m really excited to be able to talk to my grandparents more, especially because they’re getting older.”
Medill first-year Gavin Lin first started learning Chinese as a high school freshman. His trip to China during the summer only cemented his resolve to continue with it during college.
“Being Taiwanese is what motivated me to start learning Chinese in the first place,” Lin says. “Since I’m only half Asian, I get to see multiple cultures and perspectives, Taiwanese being one of them. I tried to involve myself as much as I could.”
As for the future of AI in her classroom, Chang is interested in how it can serve as advisers for students. Chamille Lescott, assistant professor of instruction of Biomedical Engineering, and Rachel Goc, Learning Engineer in the IT Services and Support department, initiated this idea, Chang says.
For example, if a student wanted to identify why they are performing poorly on tests, they could ask an AI adviser what their weaknesses may be, Chang explains.
“[Students could] reflect on what happened and [AI could] provide a constructive suggestion for their future study habits,” Chang says. “This process of reflection can help students understand themselves and their learning styles.”
However, for now, Chang views AI as just a technological learning tool. Although it may be helpful for speaking exercises, it cannot compete with formal instruction, she says.
Learning with AI would be self-paced with no direct curriculum, Chang says. Formal instruction provides students with professional techniques in vocabulary and repetition that AI could not.
“I’m confident that no one can replace me in my career,” Chang says. “[AI] is just an enhancement.”
This distinction only further validates the importance of human interaction in language learning. Although AI churned out a grammatically perfect letter for my brother, it took the emotion from his natural voice. Learning languages often boils down to fostering interpersonal connection. It would be a shame to lose that human touch.
