[nodja]

This is actually NOT recommended for a beginner.

Writing and speaking are effective at establishing long term memories, it's why we do it for other things, but a language learning beginner has no idea if what they're writing makes sense or if there's any subtle mistakes in how they're pronouncing words or how they're putting them together, etc.

Language learning experts don't recommend you start speaking/writing unless you have a coach or have reached an intermediate level so that you can discern when something sounds native or not. That way you can self evaluate with recordings, etc.

It is an effective tool for learning, but for self-learning you're gonna be shooting yourself in the foot long term. You should only do it if you have, say, a partner that speaks the language and doesn't mind correcting you all the time.

For Japanese I recommend that you do learn how to write kana/kanji from the start, and even some vocab if you want. But stop there. Don't write sentences, don't try to talk to japanese people on those apps/discord etc. and wait until you're at an intermediate level to do it, otherwise you'll form some very bad habits that are very hard to undo.

[zakokor]

I think the concern about forming bad habits is real, but avoiding any writing until you’re “ready” can delay fluency.

Something that worked for me was limiting it: just 65 words a day in the target language. It forces you to think, but the risk of fossilizing mistakes is low because it’s short.

I even built a little site for this (65words.com) and it’s been fun seeing others use it. Curious if Japanese learners here think this approach makes sense.

[linehedonist]

Every language course I’ve taken has involved has some active production of language. Day 1 of my Japanese class in HS was introducing ourselves to one another. Language exams also require proof of correct and accurate production of language.

Do you have any citations for the idea that it’s better not to practice actually using the language while trying to learn it?

[carom]

This is the primary researcher behind the input hypothesis. [1]

1. https://en.m.wikipedia.org/wiki/Stephen_Krashen

[Karrot_Kream]

The Input Hypothesis is controversial and has little evidence backing it up. Krashen is a bit like a fitness bro of the language learning world.

[ntlk]

Making mistakes and seeing exactly how you’re being misunderstood is the best thing for improving your speaking skills. There’s absolutely no reason to wait before speaking, as wanting to be understood provides incredible motivation to correct issues as you discover them.

[charcircuit]

>has no idea if what they're writing makes sense or if there's any subtle mistakes

These days AI can tell you if it makes sense and the subtle mistakes you are making. I think this view point is outdated now that everyone has a personal language tutor in their pocket.

[nodja]

I've used several LLMs to do translations and they're very hit/miss, specially in very high context languages like japanese. I'm not sure recommending their usage for a beginner is good advice, it's better than nothing for sure, but still not a replacement for a human coach.

[Karrot_Kream]

I'm sorry but this is one of the most incorrect things I have ever read. If I could downvote it twice I would.

It's true that Japanese tend to be more strict about accurate production of phonology than many other language speakers but speaking and writing are huge enablers of becoming better. It's really not that hard to unlearn bad pronunciation especially in an immersive context. Also most Japanese have a tendency to gently correct a speaker if you use the wrong phrase, particle, or construction.

Obviously if you've been self-learning your first few conversations with real people are going to be rough and so maybe avoid dense topics like Japanese attitudes on the JSDF. But if you end up in a light conversation circle where you talk about your favorite foods you'll be fine.

[nodja]

Fixing a bad habit is very hard, and I clearly stated it that outputting is very helpful, but you need to be constantly corrected or you'll develop bad habits that are very hard to fix. I'm not a native english speaker and I'm in a community of immigrants in the US and most of them have developed very bad habits that are fixable, but would require major effort and time on their part. The main ones being that they do word-level translation from their mother language to english and keep the same sentence structure, or borrowed words that are common in both languages are pronounced in a half-half sense, i.e. they change how the words are pronounced to make them sound more english, but the vowels still sound spanish/portuguese/etc.

Also note that these are not barriers to being understood, but they are barriers to be fluent in the language. These people have lived in the US for 10+ years and communicate in english just fine, humans need very little language to communicate most things. But if they need to be taken seriously in jobs that require constant communication, becoming fluent should be a long term goal, and outputting early is bad. It's best to wait for 1-2 months until you get a grasp of sound and flow of the language.

[Karrot_Kream]

You do realize the Input Hypothesis is a bit of a crack hypothesis right? You can't claim that "language learning experts don't recommend" something when in fact that theory is controversial and has little evidence backing it up.

[grigri907]

What DO you recommend then between learning kana/kanji and full fluency?