Mirachat

Research

Why text chat is the most underrated way to learn a language

Chat is slow enough to read and write, fast enough to stay engaging, full of the vocabulary you actually need, and it leaves a transcript you can study. Few other methods manage all four.

The MiraChat team

Hands typing a message on a phone at a wooden cafe table
Photo by Swello on Unsplash

Ask people how they plan to learn a language and you'll hear the usual list: an app streak, a class, a tutor, maybe a trip abroad. Text chat rarely makes the list. If it comes up at all, it's as a side activity, something you do with a language exchange partner once you're "good enough."

We think that's backwards. Chatting by text sits in an unusual sweet spot. It's slow enough that you can actually read and write, fast and dynamic enough to keep you engaged, practical, because it runs on the everyday words and grammar people really use, and studyable, because everything you say and read stays on the screen.

Most methods get one or two of those right. Chat gets all four.

1. Slow enough to read and write

Spoken conversation is fast, much faster than most learners realize. Across languages, the typical gap between one person finishing a turn and the next person starting is roughly 200 milliseconds.1 That means speakers start planning their reply while the other person is still talking. For a learner who's still decoding the sentence they just heard, there's simply no time.

Text chat changes the timing completely. A message stays on screen as long as you need. You can reread it, look up a word, check a conjugation, write a reply, delete it, and write a better one. Nobody expects an instant answer. When we measured about 400,000 replies in a corpus of real Japanese text chats, native speakers took a median of 20.8 seconds to reply, roughly 100 times the gap in speech.2

That extra time is where learning happens. You get to notice the structure of what you read, try out the structure in what you write, and fix it before anyone sees it. In speech, all of that has to happen in real time or not at all.

Reading and writing are also more forgiving than listening and speaking because you set the pace. Fluent adults read silently at around 240 words per minute,3 and learners often read at a quarter to half of that. That would be hopeless for keeping up with speech, but it's plenty for messages that average a sentence or two. On the writing side, people type on phones at around 36 words per minute,4 slow enough to plan each sentence as you write it.

2. Fast and dynamic enough to stay engaging

If slowness were all that mattered, the best method would be reading textbooks. It isn't, because textbooks don't talk back.

Chat is a real conversation. Someone asks you a question and is waiting for the answer. They react to what you said, change the subject, send a photo, make a joke. You can't predict the next message, so you have to understand it on its own terms. That small social pull, someone is waiting for me, is a powerful motivator, and it's exactly what's missing from most self-study.

It's also why chat practice tends to be dense. In text chat, you produce language constantly instead of waiting for your turn in a classroom. Studies comparing classroom chat with face-to-face discussion have found that students produce more language, and participate more evenly, in chat.5,6

3. The most practical vocabulary and grammar

The language of chat is the language of everyday life: plans, food, work, weather, feelings, small complaints, good news. That's not a limitation. It's the most useful register you can learn.

Everyday conversation leans heavily on a relatively small core vocabulary. Estimates suggest understanding spoken English takes around 6,000–7,000 word families, versus 8,000–9,000 for written text like newspapers and novels.7 One-on-one chat has the same profile: in our analysis, the 2,000 most common English words covered about 88% of chat text, versus 74% of Wikipedia. In Japanese it was 86% versus 70%. Every chat session is a review session for the words you need most, without you having to plan it.

Grammar works the same way. Chat sentences are short (six to nine words on average in the English chat corpora we analyzed, against fifteen to twenty in news and Wikipedia), and they're full of questions, I, and you. Chat is where you see how people actually talk to each other: contractions, sentence-ending particles in Japanese, dropped pronouns in Spanish, and the shift between polite and casual speech as people get to know each other. You also learn which structures matter most, simply because they keep showing up.

We dug into the numbers, including where chat doesn't come out ahead, in Chat-based language learning, by the numbers.

4. A transcript you can study

Spoken conversation evaporates. A chat leaves behind a complete record: every word you looked up, every sentence you wrote, every correction, every phrase your partner used that you want to borrow.

That transcript is the best study material you'll ever have, because it's personal. Each word you save comes with a memory of the conversation it came from, and a good chance it'll show up again in your next chat. Turn those moments into spaced-repetition review and you get two of the most reliable effects in learning research, retrieval practice and spacing, applied to vocabulary you've already shown you need.8,9

We wrote a full guide to that in Your chat history is the best study guide you'll ever have.

What about speaking?

The obvious objection: if your goal is to speak, shouldn't you practice speaking?

Eventually, yes. But chat isn't a detour. It builds the vocabulary, phrases, and grammar you'll speak with, and it does so in the same interactive, turn-by-turn format as spoken conversation. There's also research suggesting the transfer is real. In one small study, learners who spent half their class time in chat sessions scored higher on oral proficiency than learners taught entirely face to face.10 Meta-analyses comparing text chat with face-to-face interaction overall find them roughly equally effective.11 That's the point: chat isn't a magic shortcut, but it's as good as in-person practice, and far easier to get more of.

Think of chat as a slow-motion version of conversation. You learn the moves at a pace you can handle, then you speed up.

Why we built MiraChat around chat

The hard part of chat-based learning has never been the method. It's the logistics. Language exchange partners get busy. Time zones don't line up. Friends are too polite to correct you. And the great phrases you see in a chat scroll away and get forgotten.

MiraChat is our answer to that: AI language partners who chat about their day whenever you want, and Mira beside the conversation for translations, lookups, corrections, and review that comes straight from your own chats. Our founder, James Nicolas, got fluent in Japanese mostly through chat, the slow, messy, human way, and wrote about it here. We want to make that path available to everyone.

If you're new to all this, start with How to start texting in a language you barely speak.

Sources

  1. Stivers, T., et al. (2009). Universals and cultural variation in turn-taking in conversation. PNAS, 106(26), 10587–10592. https://doi.org/10.1073/pnas.0903616106 ↩
  2. Our analysis of RealPersonaChat (Yamashita et al., 2023), about 13,600 timestamped Japanese chats. Details in Chat-based language learning, by the numbers. ↩
  3. Brysbaert, M. (2019). How many words do we read per minute? A review and meta-analysis of reading rate. Journal of Memory and Language, 109, 104047. https://doi.org/10.1016/j.jml.2019.104047 ↩
  4. Palin, K., Feit, A. M., Kim, S., Kristensson, P. O., & Oulasvirta, A. (2019). How do people type on mobile devices? Observations from a study with 37,000 volunteers. MobileHCI '19. https://doi.org/10.1145/3338286.3340120 ↩
  5. Kern, R. G. (1995). Restructuring classroom interaction with networked computers: Effects on quantity and characteristics of language production. The Modern Language Journal, 79(4), 457–476. https://doi.org/10.1111/j.1540-4781.1995.tb05445.x ↩
  6. Warschauer, M. (1996). Comparing face-to-face and electronic discussion in the second language classroom. CALICO Journal, 13(2–3), 7–26. https://doi.org/10.1558/cj.v13i2-3.7-26 ↩
  7. Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? Canadian Modern Language Review, 63(1), 59–82. https://doi.org/10.3138/cmlr.63.1.59 ↩
  8. Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x ↩
  9. Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380. https://doi.org/10.1037/0033-2909.132.3.354 ↩
  10. Payne, J. S., & Whitney, P. J. (2002). Developing L2 oral proficiency through synchronous CMC: Output, working memory, and interlanguage development. CALICO Journal, 20(1), 7–32. https://doi.org/10.1558/cj.v20i1.7-32 ↩
  11. Ziegler, N. (2016). Synchronous computer-mediated communication and interaction: A meta-analysis. Studies in Second Language Acquisition, 38(3), 553–586. https://doi.org/10.1017/S027226311500025X ↩