Why We Struggle with Second Languages and How AI Can Help | Steven Toy | TEDxNiendorf

Why We Struggle with Second Languages and How AI Can Help | Steven Toy | TEDxNiendorf

Like most of you, I was taught a second language in school. In my three years of French, I learned vocabulary, verb conjugation, and the grammar rules that allowed me to pass a lot of rather painful tests certifying that I knew French. Then I went to France. After navigating customs, catching a cab to the city, and checking into my hotel, my personal narrative of being bilingual, backed by all those tests, was instantly replaced by a narrative many of you might recognize: I'm just bad at languages.

I spent the next five days in France with people who couldn't say hello in English, so French was the language du jour all day every day. During that time, I used more French than I did in three years of school. What's more, the French I learned in France was not the French I was taught in school. As I left France, my personal narrative was rewritten once again by the realization that to learn a language, you've got to use the language. Or as my granny would creatively say when I was a bit frustrated and struggling with high school French, "Sweetie, the best way to learn a language is between the sheets."

Now, if we examine our instincts on the matter, we all intuitively know this grandmotherly wisdom to be true. To learn to swim, you’ve got to swim; no amount of reading will get you there. To learn to ride a bike, you’ve got to ride a bike; knowing how the gears work isn’t going to help you stay upright. To learn another language, you’ve got to use it, something we are forced to do when we are learning a language in a country that speaks the language, which is why we learn so quickly when we do.

Finding ways to learn grammar and the definition of words is relatively easy and often gamified with jewels and badges, which keep us going with the promise of that 520-day streak or a hit of dopamine. Finding ways to practice hearing and using the words we have memorized is much more difficult. Let me explain, starting with videos. There are almost infinite videos in the world that can help us recognize the words that we are learning in real life, but they can’t be filtered by our individual language level. As a result, any repository of videos is overwhelming and frustrating. There’s just no way to filter those videos to something that you have a chance of understanding.

Videos made for language learning are mind-numbingly boring and reflect the words found in a dictionary, not in the street. They’re usually behind a paywall or part of a course where you have to learn words in a certain order in order for those videos to make sense at all. When you go to a country to learn a language, the words come at you so fast it’s like drinking from a fire hose. There’s just no way to stop, rewind, filter, and repeat the real world to get your head above water.

And those are just the problems in hearing the words that we have learned. When we seek to use the words we have learned, it usually involves teachers or tutors or conversation partners who often cost money, always require scheduling, and suffer from the limits of human patience that encumber us all. I mean, let’s be honest: how long can you handle listening to someone butcher your mother tongue as they tell you about the green turtle on your red sweater and the blue chair? All of these challenges pale in comparison to our innate fear of being judged by others. I can’t tell you how many times I’ve been in online conversation classes where the majority of participants have their cameras off and their mics muted for the entirety of the session out of fear of making a mistake in a room full of newbies.

Fortunately, AI, in particular large language models or LLMs, has allowed us to address these fatal flaws by giving us the ability to do two very important things. First, LLMs let us break down the world's content into its component words and phrases. With this capability, we can take videos created by native speakers—including all the associated accents and truncations that come in the real world—and filter them by the words and phrases we have already learned. Presto! No overwhelm from a tsunami of videos that you have no chance of understanding.

Furthermore, by ingesting the world content for this purpose, we can overcome the quantity and quality problems that come with educator-produced boring videos. Now we can practice listening with the TV shows, music, and sports videos that we are already watching on YouTube and TikTok. It gets better: when we know the words and phrases in the videos that people want to understand, we can determine the words and phrases we need to teach on a user-by-user basis. This is particularly useful in teaching for corporate environments where the focus might need to be on terminology that is specific to a given industry like banking or technology.

The second thing that LLMs let us create are conversation partners that can focus on particular topics, like travel, and groups of words, like the words that we have learned in class or in our app. Targeted conversation partners that are available 24/7 with infinite patience and not a hint of the human judgment that stops many language learners from even saying hello, hola, or privet.

AI and LLMs are changing the landscape of language learning. They can get you using the words you are learning from the very beginning, in much the same way we did as babies learning our mother tongue. Now, beyond my personal experience or Granny’s strategy of learning a language in the linens, there is a common thread that runs through all language learning theories in existence today. Whether it be Nativist Theory, most closely associated with Noam Chomsky, or Stephen Krashen’s comprehensible input theory and beyond, that common thread is: you have to use the language to learn the language.

And here’s why you should consider learning a second or third language. First and foremost, increased earning potential. A second language means more money. Expanded opportunities: more jobs are open to you. Better social connections with your neighbors, your co-workers, and beyond. Certainly more enjoyable travel experiences or just understanding one another a bit better—something we can use in the world today. No matter what might drive you to learn another language, you must use it to succeed. It used to be hard to find opportunities to hear and use a new language, but now you can do it for free with the device you carry with you everywhere you go.

As for learning a language Granny’s way, well, that’s a different app.

Keywords

• second language
• French
• learning
• vocabulary
• grammar
• living abroad
• language practice
• AI
• large language models (LLMs)
• language learning theories
• Noam Chomsky
• Stephen Krashen
• comprehensible input
• increased earning potential
• social connections

FAQ

Q: Why did the author feel inadequate in French after traveling to France despite studying it for three years? A: The author realized that the practical use of French in real-life settings was much more effective for learning than the theoretical knowledge gained from classroom education.

Q: How does AI help in language learning? A: AI, specifically large language models (LLMs), breaks down content into words and phrases, allowing learners to practice with real-life videos and creating conversation partners that are available 24/7 without judgment.

Q: What are some benefits of learning a second language according to the article? A: Learning a second language can increase earning potential, expand job opportunities, enhance social connections, and improve travel experiences.

Q: What major challenge do learners face when trying to use the words they've learned? A: Learners often struggle with the fear of being judged, the cost and scheduling of tutors, and the patience required by human conversation partners.

Q: What common thread runs through all language learning theories? A: The common thread is that you have to use the language to learn the language effectively.