Research on the efficacy of using generative artificial intelligence tools for English-language development is nascent. But initial examinations of AI’s role suggest the technology has the potential to improve instruction.
There are challenges, of course. Teachers are worried about students using AI tools to cheat and that the technology can generate inaccurate or biased responses. Plus, some English learners might be too reverent about AI responses, rather than trusting their own thinking or writing.
Despite those challenges, some teachers of English learners have been experimenting with generative artificial intelligence tools since they became easily accessible two years ago. For instance, some teachers are using ChatGPT to reduce Lexile levels—the measure for how difficult a text is—for their students.
More broadly, one third of school and district administrators say they are using artificial intelligence technologies in programs serving multilingual learners, according to a nationally representative EdWeek Research Center survey of 1,135 educators conducted in September and October. Another 40 percent said they’re either “considering,” “exploring,” or “piloting” these tools.
With more than a decade of experience studying multilingual learners, Avary Carhill-Poza, an associate professor of applied linguistics at the University of Massachusetts Boston, spoke with Education Week about ways to use AI with English learners.
This interview has been edited for brevity and clarity.
What research is already out there about using AI with English learners?
There are some emerging studies and findings about using things like ChatGPT in classrooms, but that’s a pretty narrow definition of AI. We might want to think a little more broadly and go back to things like Google Translate and Grammarly and some of the work that people have been doing over the last decade or two [thinking] about technologies that use AI and how schools and communities have really adapted those different technologies to their classrooms. The idea of using AI is not actually so new in our classrooms.
What can we learn from previous research into other AI-driven tools?
One of the things that I would say to think about is—[in] that initial moment when we become aware of something new and it feels disruptive or threatening—the way that a conversation can turn from thinking about a [tool] to thinking about a practice, thinking about an app to thinking about a practice, thinking about AI to thinking about practice.
What does that mean? Usually what that means is learning from parents and students. What are they already doing with this new technology—with AI in this case—and how does that align with or challenge the kinds of things that we’re doing in school right now?
A lot of what goes into district thinking sometimes is like, “Can we get some PD for this?” And while that’s really helpful, I think we want to start with what our students already know, what their families and communities already know, and move toward what we can learn from and with them about the ways that they’re already using AI [and] bring that back into our classroom and our learning practices.
Probably the biggest struggle for individual teachers and districts as a whole is how to engage with, how to really learn with families and communities.
It's really taking a step back and trying to be open to processes that might support language learning and content learning in what you might consider unorthodox.
What are some ways English learners are already using AI?
I had a conversation very recently with a mom who was going to have her daughter pick some things up on her way home from school, so she was texting her this list of ingredients. The high school student [who is still learning English] was doing some quick translations [using an AI-powered translation app] into English so that she’d know what to look for and get as she was running this errand on her way home.
I have a student who’s doing a dissertation project where she’s interacting with a lot of students [face-to-face in their homes], and a lot of them use AI to summarize articles that they’re reading for class before they take a deep dive to find details that they need for certain things. She [commented] on how efficient she saw the students working when they were at home and using AI to do this.
Some teachers are worried AI could be a crutch for students. What would you say to that?
For multilingual learners, the key point is to really think about practices that let the students bring what they know to whatever task a teacher has, whatever learning objectives a teacher has.
If what they know could encompass an overview of an article that they’re going to read, would that be helpful for the students? Would that be a way to support their learning, a part of a new or developing set of practices? If there’s a way to think about what students are doing and what they’re bringing to learning, it can lessen the fear that students aren’t doing the work or aren’t meeting expectations.
If you spend a whole class period laboriously translating your way through an article, is that the best strategy for leveraging [background] knowledge and experience? Is there a way that the AI technologies that students are already using could support learning and maybe allow them to center more the kinds of knowledge and experience that they already have?
It’s really taking a step back and trying to be open to processes that might support language learning and content learning in what you might consider unorthodox, and moving past the initial sense that the technology is going to be disruptive just because it’s not what you’ve been doing.
Do you have other advice for educators of English learners?
If you are generating some sort of text with AI, you still need to go back and edit it and clarify and proofread and make sure it’s saying things that you’re interested in having it say. Students who are multilingual learners can feel, sometimes, that those texts are maybe better than what they would be able to say themselves or more correct. [It’s important to get] them to take it for what it is and feel comfortable using it when it’s most beneficial but not feeling bound up or reverent of the texts that are generated.
Data analysis for this article was provided by the EdWeek Research Center. Learn more about the center’s work.
