mentat_kgs Wrote:I'll take some lifetimes to do the same with output.

It will. Fortunately, you don't have to. Output is, like you said before, an extremely small part. Going 95% input and 5% output is a good proportion, at least in the start. No matter how long you study, conversational output should always be of the smallest concern. I'm not saying one should spend 50% of their time on conversational output, I'm just saying that you won't converse fluently without ever training it.

mentat_kgs Wrote:And yes, my numbers are skewed.
It is not possible to account it perfetly, but probably 1% for speaking practice is already way too much.

For japanese I've read far more than 30000 sentences.
Right now, only my sentences deck account for more than 3000 sentences.

Also, I've listened far more japanese that I have read.

OK, when you put it in this perspective, I agree with you. There is many many many more instances of input if you consider every single sentence you read or hear to be input, then surely that's correct.

I was under the impression that you were talking about physically "practicing" input and output, which would be reading materials which I actually want to learn, as opposed to just flipping through some manga or something.

Tobberoth Wrote:Well, your counter to my first reason is valid even though I disagree. There might be other areas where training isn't needed, but output isn't one of them (my experience tells me) because of all the people I know who has been exposed to tons of Japanese yet speaks it badly because they aren't used to it.

Yes, and qualifying your assertion helps. It would just be quicker if you hadn't used the argument that it's logically impossible to do something, or saying "It really makes little sense to believe that a skill is randomly gained without training", to dismiss counter-arguments which relied on experience and specific situations. As you said somewhere, a discussion is to advance our comprehension, and we can do that if we discuss precisely how things work, not waving aside arguments with abstract impossibilities. I think we may end up being closer in our opinions now that we get down to it.

Quote:When you go keyword -> kanji, you see the keyword which makes you think of a story which makes you realize the bushu and thus you remember the kanji. When you see the kanji you see the radicals, thus remember your story and therefor the keyword. It only works because it's the exact same thing you've been training, just from different ends. Input and output is however completely different. You see a sentence and understand it, you don't create it. It's a different action all together. When you see a sentence in a book, you have quite some time to analyze it and understand it. You can reread parts you didn't get perfectly until you had the full picture. It's completely different when outputting, it has to be done really fast and there's no sentence already there to help you.

The example of Heisig and 'keywords <-> kanji' was meant to get rid of the logical dismissal. That being done I agree that there are differences, and you make your point about why those differences matter. Let me argue the opposite side. When you read, and read *a lot*, you encounter many situations where you identify with a character or at least recreate the situation in your head as vividly as possible. Maybe if this happens often enough, when you are in a conversation and encounter a similar "emotional" need for expressing something, an adequate sentence/expression/vocabulary will pop up, because of all the time you have spent linking situations with sentences of the target language. This is pure speculation, but now we are discussing what may be. My experience tells me it is possible and I'm trying to find plausible explanations which may be completely wrong, but I think neither of us can know for sure.

Quote:Training output helps in conversation. Did you write on forums etc while in France? That would certainly transfer to conversational ability (although not all the way, there's still a big difference in outputting fluently and fast in a conversation and outputting slowly with the ability to edit in a chat or on a forum).

I didn't have internet access until I was maybe 18, so no writing on forums for me. However, and to take up a similar point to what I said above, if I'm reading an opinion piece in a newspaper, I will think about the ideas and fact presented and maybe argue in my head if I disagree. If the newspaper is in english, it's quite likely that I will do this in english. I don't do this to train conversation, I don't think about the language when I do this, it just feels natural, as is creating a mental picture of what is happening when I read a novel.

Just to further the discussion, I'd like to ask: what do you do when you *train* conversation?