Many teachers I’ve been speaking with believe that large language models (LLMs) like ChatGPT should complement rather than replace traditional language instruction. Happily for society, though perhaps not for us, the reverse is true.
Unlike practising guitar or shooting hoops, most people don’t learn languages for enjoyment; they endure the process for the language. You rarely hear someone say, “I can’t wait to drill verb conjugations tonight!” the way they might eagerly anticipate a jam session or a pickup game. For the vast majority, language learning is a chore, a necessary evil on the path to other goals—be it landing a job, navigating a new country, or accessing education. It’s not the practice itself that’s rewarding, but the eventual outcome. And if that outcome—effective communication—can be achieved more efficiently through technology, most will jump at the chance to skip the gruelling memorization of vocabulary and the embarrassment of mangled pronunciation.
If you’ve experimented with large language models yourself, you’ve likely encountered their impressive capabilities alongside their current limitations. Yes, these systems can make mistakes, sometimes spectacularly stupid ones. But consider this: I spent a decade living in Japan, immersed in the language, and while I can speak Japanese pretty well, I’m nowhere close to matching the breadth of vocabulary, grammatical accuracy, or cultural knowledge of a well-trained LLM. These AI systems, despite their imperfections, can already switch between hundreds of languages, maintaining context and nuance in ways that far surpass most human polyglots. They don’t get tired, they don’t forget, and their knowledge doesn’t fossilize the way human language skills often do. Perhaps most importantly, they’re improving at a rate that outpaces human learning by orders of magnitude.
If you haven’t tried these tools yet, you might be skeptical of their capabilities. It’s natural to wonder how a machine could navigate the nuances of human language. But these aren’t simple translation programs. They’re systems that thrive on context, adapt to different communication styles, and even deliver cultural subtleties. While not perfect, their ability to facilitate communication across language barriers is unprecedented. They offer a level of linguistic flexibility and knowledge that would take most humans decades to achieve in just one additional language, let alone dozens. I’ve spent hours each day for months using ChatGPT, Claude, Gemini, and various other such tools, and it’s clear to me that the gulf between their cross-linguistics capabilities and those of the average human is widening.
It’s probably not accurate to say they understand, but they don’t have to. They mimic the output of a person who understands, and for interacting in a language that you don’t, that mimicry is far better than good enough. It’s the difference between spending years struggling with textbooks and flashcards, and having instant access to fluent, nuanced communication in any language. For most people, the choice is clear: years of struggle or instant fluency, painful practice or effortless communication.
To see why, let’s consider the global impact of language barriers and the resources devoted to overcoming them. Conservatively, if we assume that the average person spends just 500 hours in their lifetime studying a foreign language, with about 1 billion people learning a new language at any given time, that’s 500 billion hours globally. At an average global hourly wage of $5, we’re looking at $2.5 trillion in annual opportunity cost—time that could be spent on other pursuits, from scientific research to artistic creation to spending time with family and friends.
Consider the costs of miscommunication in business deals. Consider the misunderstandings in international diplomacy. Consider the errors in healthcare settings, the literature you can’t read, the friends you never made. All due to language barriers. Then consider the vast body of knowledge inaccessible to various populations due to linguistic barriers. If just 1% of the world’s valuable information remains untranslated and inaccessible at any given time, we’re potentially losing out on trillions in economic value, alongside untold cultural losses.
These numbers, while extremely rough, illustrate the enormous potential impact of removing language barriers. And that’s precisely what LLMs promise to do.
Consider a concrete example: an ambitious 20-year-old Anglophone student from Ontario who wants to study aerospace engineering at the Warsaw University of Technology. Currently, this student would need to spend at least a year, likely more, intensively studying Polish to reach the proficiency required for university-level studies. That’s a year of their life, a year of lost wages, and tens of thousands of dollars in tuition and living expenses, not to mention the stress and cognitive load of learning a new language while trying to prepare for a challenging degree program.
Now imagine a near future where this student could use LLMs and other technology to attend lectures, participate in discussions, and even write papers in English, with real-time, highly accurate translations. The student could focus entirely on learning aerospace engineering, potentially shaving vast chunks of time off their studies and immersing themselves more deeply in their chosen field from day one.
This scenario isn’t just about individual convenience. It’s about unlocking global talent and fostering international collaboration. How many brilliant minds are we missing out on because of language barriers? How many potential breakthroughs in science, technology, or the arts are delayed or never realized because knowledge is trapped behind linguistic walls?
Will that student be able to make Polish friends? This question touches on a crucial aspect of language learning that we often tout: cultural integration. But the reality of international student experiences suggests that language barriers are not the only obstacle to cross-cultural friendships. In Ontario, for instance, international students are often socially segregated from their Anglophone peers, regardless of English proficiency. Many socialize primarily with co-nationals, a trend that goes both ways.
A shift away from learning English towards a reliance on LLMs might exacerbate this segregation. But it could also lower initial barriers to crosscultural interactions, potentially creating connections that might otherwise never form.
Consider how technology already mediates many of our relationships: A significant portion of our daily communication occurs through cell phones and social media. While this is not ideal, would real-time, face-to-face LLMmediated communication through earbuds be fundamentally worse? It might even serve as a stepping stone, encouraging more people to engage across linguistic divides and forging more human connections.
Moreover, LLMs may offer an unexpected benefit for linguistic diversity. As these tools become more sophisticated in handling a wide array of languages, including minority and endangered ones, they could reduce the pressure on speakers of these languages to abandon them in favour of majority languages. Traditionally, economic and social pressures have forced many to prioritize learning dominant languages like English, often at the expense of their native tongues. LLMs could allow individuals to maintain their linguistic heritage while still accessing global opportunities, potentially slowing the rate of language death and preserving cultural diversity. This technological bridge could be particularly crucial for indigenous communities and other groups whose languages are at risk of extinction.
For TESL-Ontario members, this shift presents an existential challenge. Our profession is likely to contract significantly as LLM-driven alternatives become more sophisticated and accessible. Universities and colleges may maintain language requirements and ESL programs for a while, but eventually, even they will adapt. The study of English for non-native speakers may become more akin to studying Latin or Old English—pursued for scholarly interest rather than practical necessity.
This change will be painful. Many in precarious positions may find themselves out of work, while those in stable positions will need to adapt, potentially retraining or shifting focus. But from a broader perspective, this shift is overwhelmingly positive. Instant, accurate cross-language communication has the potential to promote greater global understanding, facilitate international cooperation, and open up opportunities for millions.
As language teachers, we’ve long believed that learning a language is essential for unlocking new worlds. For some, it still will be. But for many, likely most, functional communication suffices. If technology can provide that more efficiently than years of study, that’s the route they will choose.
This doesn’t spell the end of language teaching, but it does signal a fundamental shift. We may focus more on cultural aspects, advanced communication skills, or aesthetic appreciation of language and literature. There will probably be many fewer of us, and our classrooms may be smaller but filled with more passionate learners.
In the end, our goal as language teachers has always been to support communication between people of different linguistic backgrounds. If technology can achieve that more effectively, we should embrace it, even as we re-imagine our roles. The era of traditional language learning may be waning, but learning through languages—and the human flourishing it brings —remains as crucial as ever.
Acknowledgements
I acknowledge the use of Claude 3.5 Sonnet (Anthropic 2024), in drafting and editing this essay.
References
Anthropic. 2024. Claude 3.5 Sonnet model card addendum. Tech. rep. San Francisco: Anthropic. https://paperswithcode.com/paper/claude-3-5-sonnet-model-card-addendum
Author Bio
Brett Reynolds is Contact magazine’s editor emeritus, a professor of EAP and TESL at Humber College, and an adjunct professor of linguistics at the University of Toronto. He co-authored the second edition of A Student’s Introduction to English Grammar, and his recent research focuses on the micro-syntax of English words. Currently, he is writing Language Landscapes: The ESL Teacher’s Guide to How English Works, a textbook that demystifies English structure for TESL students and teachers, and (Un)grammatical, a popular book on grammar’s place in everyday thought, as well as editing a new edition of Jespersen’s Negation in English and Other Languages.