From Novelty to Necessity: A Survey on Second Language Learning in the Age of AI
By Hossam Elsherbiny, PhD., Associate Director of The Center for Languages & Intercultural Communication, Rice University, and Dean Toumajian, Rice University
.
DOI: https://www.doi.org/10.69732/PWER3384
From the fictional artificial intelligence computer HAL 9000’s menacing evolution in “2001: A Space Odyssey” to Star Trek’s seamless universal translator, science fiction has long presented visions of artificial intelligence (AI) that flirted with absurdity. Now, as if torn from the pages of these futuristic tales, AI tools are materializing in the real world at an alarming speed. What was once the product of the Hollywood imagination is rapidly becoming our everyday reality. It’s not surprising that billionaires and technology giants, such as Elon Musk, feel compelled to sign an open letter requesting a pause on the training of new super-powerful systems for six months and citing “profound risks to society and humanity”. But how are we—mere mortals—supposed to feel?
The broader implications of AI advancements appear to be causing global anxiety, and the impact on education couldn’t be more apparent, with institutions scrambling to regulate their use. Take, for instance, the recent decision by the New York City public school system to block access to ChatGPT on school computers and networks due to “concerns about negative impacts on student learning, and concerns regarding the safety and accuracy of content” or the Group of Eight Leading Universities (Go8) in Australia deciding to return to supervised pen and paper exams to prevent cheating. Many voices, however, view these bans as a futile effort—a game of cat and mouse between students and teachers.
Admittedly, the advent of technology has often been met with apprehension and reluctance in educational circles. If we think back to the 1970s when calculators were invented and introduced into mathematics classrooms, the initial reaction was one of alarm—is this the death of mental arithmetic? Fast forward a few decades, calculators are now a staple of classes: pen, paper, and calculator. It appears that we have a tendency to catastrophize new technologies that we don’t fully understand, but humans are plastic; we eventually adapt. The current hype (or dread, depending on your perspective) surrounding AI in (second language) education feels eerily familiar —is this a repeat of what happened with other technological inventions, or is this time truly different?
A 2018 study entitled “‘My robot is an idiot!’ – Students’ perceptions of AI in the L2 classroom” offers a fascinating glimpse into how learners viewed AI tools in language education just a few years ago. The study indicated that Japanese English learners perceived the chatbot that was used in class as a novelty, rather than a viable language-learning tool. Since then, advancements in second language acquisition technology have undergone a remarkable transformation. All of this prompts us to wonder if we’re on the cusp of a revolution in language learning technology, or if we’re simply experiencing fascination with a new technology tool.
To have a better understanding of how AI is changing the landscape of second language acquisition, we, the authors of this article—a language educator and an undergraduate student—have joined forces and surveyed over 100 individuals, between language learners and instructors, across several higher education institutions in the United States. The survey aims to capture a vignette of the current impact of AI tools on language learners and educators. More specifically, we seek to understand how frequently these tools are being used, the motivations behind their use, and their perceived (in)effectiveness. By gathering data from both learners and instructors, our goal is to draw a somewhat comprehensive picture of AI’s role in today’s language learning landscape.
METHODOLOGY | Taking Attendance: Surveying the Room
What seat does AI sit in during classroom instruction? Is it front and center, does AI wear a hoodie and sit in the back, or maybe it skips class altogether? To find this out, as aforementioned, we developed (yes, we developed it—not ChatGPT) a brief, anonymous survey. This was distributed to both language learners and instructors to assess their relationship with AI tools both inside and outside the classroom. The survey was distributed to the following universities: Rice, Cornell, the University of Chicago, and University of Hawai’i at Mānoa. If you would like, you can look at a copy of our survey instrument.
Our survey consisted of four paths respondents could take: students who did not incorporate AI tools in their language learning, students who did, instructors who did not incorporate AI tools in their language instruction, and instructors who did. By filtering respondents into one of these four paths, we could ask questions tailored to their personal experiences. At the beginning of the survey, respondents were asked what language(s) they have taught or studied and how often they use AI tools to support their instruction/learning.
Once the respondents were filtered into their appropriate groups, they were asked questions that would range from explaining their rationale for using AI tools (or not using them) to describing how comfortable they are with AI tools being incorporated into classroom settings. Each group was eventually filtered back into the same path to answer broader questions that did not rely on their use, or lack thereof, of AI. The breadth and depth of responses were important to us, so we gave many opportunities for our respondents to express their nuanced views on this topic. Not to toot our own horns, but, in our survey design, our logic was undeniable (Get it? If not, we recommend that you watch “I, Robot”).
SUMMARY | Survey says
Upon collecting responses from 105 participants in our survey, comprising 16 language instructors and 89 students, we felt that we had a decent profile on the zeitgeist of modern attitudes towards AI utilization in second language learning. The data reveals that a significant majority (62%) of instructors do not employ AI tools to supplement their instruction at all. Interestingly, none of the AI-abstainers indicated the reason(s) for which they do not employ AI tools. Nevertheless, among those who do integrate AI, its application spans various facets of pedagogy, with particular emphasis on developing class materials that target specific phonological or grammatical features. Additionally, instructors leverage AI as a way to save time when creating lesson plans and assignments.
However, the use of these tools is not void of challenges. Some instructors indicated that they felt there were insufficient customization options. Others cited prohibitive costs that make traditional textbooks appear economically prudent by comparison. Many mirrored the sentiment that AI tools have “low-quality outputs” (exclamation mark omitted). Furthermore, there appears to be a substantial learning curve associated with these tools, irrespective of the degree of implementation. Many respondents denounce a lack of training or understanding of the underlying mechanisms of these tools. A few respondents indicated that AI exhibits certain linguistic limitations, such as an inexplicable predilection for male pronouns (even when specifically instructed otherwise—a case of artificial selective hearing, perhaps?) and difficulties in addressing grammatical inquiries that pertain to non-Roman character languages. Interestingly, while a subset of instructors embrace these tools, there is a pervasive concern regarding student utilization of similar technologies and the potential implications on academic integrity.
“Beginning-level students are using AI to produce texts that don’t correspond with the knowledge they have and contribute very little to their learning.” -Second language instructor
Nonetheless, instructors express confidence in their ability to authenticate students’ work by relying on their intimate knowledge of individual students’ capabilities. This, however, isn’t always the case when instructors are not familiar with the students whose work they evaluate, as emerged in a 2023 study by Frances and Zimotti.
In contrast to their pedagogical counterparts, a majority of students (55%) report utilizing AI to support their language learning to some degree. The rationale behind AI absention among students appears to be multifaceted rather than dominated by one single reason. A plurality (44%) of students believe that AI is outright not helpful; 42% harbor reservations regarding academic integrity (either deeming it unethical or fearing accusations thereof), while a quarter (25%) have adopted a wait-and-see approach regarding how this technology progresses.
Notably, students who avoid using AI tools seem to possess not merely hesitation, but a certain disdain towards the technology. Some students articulated “moral” and “ethical” objections to its utilization, as one second language learner eloquently pointed out:
“AI is environmentally harmful, and not conducive to my or anyone’s learning or writing. I see learning and producing as a creative outlet, and AI is a danger to creatives.” -Second Language Learner
However, in a plot twist worthy of a telenovela, a whopping 82% of students indicated they would be willing to use AI moving forward—not so moralistic now, are we?
Regarding their non-Luddite peers (i.e., those who do leverage AI tools to assist their language learning), there is a unanimous consensus that these tools have had a positive impact on their learning. This favorable perception likely stems from the multifarious applications of these tools: a significant majority (80%) rely on AI for checking grammar, while 67% utilize it for direct translation, and 64% use it for vocabulary practice, among other things. While reading and writing skills appear to have the lion’s share of AI use, speaking and listening skills exhibit markedly lower utilization rates, with only 11% of students using AI for speaking practice and 18% for listening comprehension.
Intriguingly, most students reported using AI tools at a higher rate to learn languages than in other disciplines (so much for the notion that foreign language classes are “easy,” right?). Students typically reported using AI two or three times as much for their foreign language classes than for other classes, with some students saying they use AI ten or thirty times more.
The majority of students (70%) believe their AI use aligns with principles of academic integrity, while only 15% acknowledge potential academic dishonesty in their practices. The remaining 15% opted to not disclose, which, in the realm of academia, is equivalent to pleading the Fifth Amendment.
However, it is crucial to note that language learners appear to be sensitive to the bias in AI-generated content. Additionally, AI does not appear to be used in lieu of traditional learning methods. In fact, several students described AI usage in their responses as merely supplemental. One student went so far as to say that “AI [is a] last resort.”
“For the algorithms underlying language learning AI tools to support the learning and thriving of all humanity, developers must consider culturally diverse approaches and the cross-cultural impacts of these technologies so as not to unintentionally perpetuate unjust structures and perceptions of people and the world.” –Second language learner
CONCLUSION AND DISCUSSION | Growing pAIns of AI
The results of this survey clearly show that AI has already made significant inroads into our language classrooms, bringing opportunities and challenges with it. We are at a critical juncture that will mark the way second languages are learned and taught over the next few decades. The stark contrast between instructors and students in their perception and adoption of AI tools is noteworthy. The majority of instructors (62%) do not use AI tools in their teaching, while most students (55%) report using AI to support their language learning to some degree. This disparity might be indicative of a generational gap in technology adoption, with instructors being more reluctant to incorporate novel technologies. The hesitancy to adopt AI tools seems to stem from concerns about academic integrity, the quality of AI-generated output, or simply the lack of training on how to use these tools.
On the other hand, students appear to be embracing AI tools with open arms, with the vast majority reporting that it has a positive impact on their learning. Despite the multifaceted use of these tools, from grammar checks and translations to vocabulary practice, students largely view AI as a supplement to traditional learning methods rather than a replacement thereof— thankfully, students aren’t throwing their textbooks out of the window just yet.
The ethical dilemma is real, folks! While most students believe their AI use is in alignment with principles of academic integrity, a significant minority either acknowledge potential academic dishonesty or prefer not to disclose their stance. This gray area will likely become increasingly complex as AI tools continue to advance. The pervasive use of AI tools by second language learners may require an added workload for instructors who now have to use plagiarism detection software, ask students to submit drafts of their work, or check for originality. This 2023 article presents a comprehensive list of strategies to verify the authenticity of students’ work and, quite frankly, it’s exhausting, especially when we consider how underpaid and overworked most second language instructors are.
A significant limitation of AI has to do with its innate cultural bias, since it draws from an English corpus. This particular limitation seems to defeat the very purpose of learning a second language, which is to challenge stereotypes, not perpetuate them. The lack of linguistic accuracy of AI output, especially as it pertains to gendered languages, is another limitation. The latter, however, should perhaps be celebrated as it highlights the complexity and nuance of human languages that AI has yet to fully grasp. This limitation also serves as a reminder of the irreplaceable role of human instructors and the importance of cultural context in language learning.
As we sit here, pondering the results of our survey, we cannot help but feel a mix of excitement and trepidation about the future of language learning. In a little over half a decade, AI has gone from an intriguing technological “novelty” act to a potential mainstay in the classroom. While the potential that AI offers is exciting, the limitations and ethical implications cannot be understated. The second language learning landscape is undoubtedly changing because of AI. All we can do is to keep an open mind and a sense of humor. After all, if HAL 9000 can learn to sing “Daisy Bell,” surely we can learn to coexist and harmonize with our AI assistants in the language classroom.
References
Frances, C., & Zimotti, G. (2023). ChatGPT and second language writing: Perspectives from writing instructors and students. The FLTMAG. Retrieved from https://fltmag.com/chatgpt-second-language-writing/
Group of Eight. (2023, January 10). Universities to return to pen and paper exams after students caught using AI to write essays. The Guardian. Retrieved from https://www.theguardian.com/australia-news/2023/jan/10/universities-to-return-to-pen-and-paper-exams-after-students-caught-using-ai-to-write-essays
Musk, E., et al. (2023, March 22). Pause giant AI experiments: An open letter. Retrieved from https://time.com/6266679/musk-ai-open-letter/
New York City Department of Education. (2023, January 5). New York City public schools ban ChatGPT over fears of cheating and misinformation. CNN. Retrieved from https://www.cnn.com/2023/01/05/tech/chatgpt-nyc-school-ban/index.html
Rasmussen, J., & Lund, A. (2018). “My robot is an idiot!” – Students’ perceptions of AI in the L2 classroom. ReCALL, 30(3), 337-354. Retrieved from https://www.academia.edu/85493022/_My_robot_is_an_idiot_Students_perceptions_of_AI_in_the_L2_classroom
Zhao, Y. (2023). How to live with ChatGPT: Seven strategies for teachers. Innovations in Education and Teaching International. https://doi.org/10.1080/14703297.2023.2190148
Excellent survey of how university students are looking at AI’s use in language learning. Granted, this will change as AI develops, but it’s always good to have some baseline information on what the students think to start from. I hope you’ll continue this research into the future to see how the students’ perceptions change and what concerns they will continue to have. Thanks!