What do you think about the input hypothesis and related ideas?

The most dubious idea I regularly hear is:

“If you are having difficulty with output, that invariably means you need more input. If you output too early, you will do harm to your Japanese.”

If the input hypothesis and related ideas were testable and had been empirically validated over many years with a high degree of academic rigor, I would not find them so dubious.

Another problem is attribution. Humans are extremely bad at accurately attributing their success to causes, so even the most successful learners often give terrible advice. Anecdotal evidence is weak.

The last problem is extrapolation. I think that, within the Japanese language learning community at least, it is incredibly common to build layers of assumptions and conclusions on top of these ideas that are already problematic.


I can see this possibly being true (still dubious about it in the long-term though) if you have no form of feedback loop on your output, but if you are working with a teacher or a tutor, etc. that can correct your output this doesn’t seem to hold up. This claim also appears to contradict how kids pick up and learn their language. They speak and babble all the time and yet they eventually learn to speak properly because of the feedback loop of other people in their lives. I don’t see how this is any different from a Japanese learner.


The 5 hypothesis as stated on wikipedia are a lot less sensational. There’s nothing there that seems unreasonable, apart from the acquisition–learning hypothesis, which is stated a bit too strongly for my liking.

I guess the problem is how the hell do you test it…I’m not sure any competitor theory in language learning has the academic rigor you’re asking for either.

@athomasm but then kids learn very differently to adults, and don’t have an existing language to help/hinder them. It doesn’t really say much about second language learning (if I had to choose I’d agree with you though, as long as the feedback is good quality and immediate).


I found a pretty comprehensive argument against the input hypothesis. It’s quite an interesting read, and gives citations to various studies. It’s on google books though, so there’s two missing pages.


You regularly hear this? In 2+ years of studying I don’t think I’ve heard this once.

1 Like

It’s the AJATT people and the people who listen to the AJATT people mainly (e.g. MattVsJapan, BritVsJapan, followers thereof, people on LearnJapanese subreddit, etc). Perfectionists who are dead set on sounding like native Japanese speakers are also common. This is not a direct criticism of these people or their goals. That’s just where this all comes from.


I will talk a lot about this in my level 60 post so I will make it simple here.

How do people(non-natives) become good at English? With a ridiculous amount of input. They don’t become good because they started talking with their friends in English. I mean, yes, that might happen when they play video games with a friend but the amount of input they get overall is tremendously high compared to the output.

I mean what do you want to say early on anyway? 私はペンです? It’s useless. Unless you want to be like one of those people on YouTube(I learned a language in 7 days), talking very early on isn’t gonna bring much to the table. The reason why people like the idea of starting to speak early is because it’s motivating. However, that doesn’t mean it’s efficient.

This brings me to the second part which is that your accent will heavily suffer. I am not here to debate whether AJATT is good or not but they are right when they say you need a lot of input first if you want to sound close to a native speaker. If you aren’t used to the different sounds first you will ruin yourself.

Imagine someone who never watched any Japanese show trying to learn the language. He uses some textbooks and starts speaking as he is learning. The problem is he doesn’t have the right sounds in his head yet. So what he is gonna do? Imitate/create the Japanese sounds using the sounds of his own language. This will result in having a very weird accent. Not only that but his pitch accent will be completely destroyed.

Yes, he will be able to speak. He might even impress some Japanese learners but he will sound funny to natives. I mean, it would be sad to ruin 3 years of Japanese studies because your accent is awful. It’s insanely hard to fix a bad accent.

So which one do you prefer? The short term win or the long term win?


Stephen Krashen is one of the big names on this topic in language acquisition theory. There has been a bit of research done in this field with a certain degree of peer-review, statistical analysis, sample-size testing, etc. So, I’d say there has been a fair amount of research. The response is mixed within the research community. The simple version of the story is: the input hypothesis is worthwhile to an extent, but it depends on how you use it…

The issue comes down to, how is it applied? It is true that input is required for output. But this is obvious in that we all need to learn おはよう [ございます] before we can say good morning. It is also true that i+1 is a nice, convenient way to express that we need comprehensible input and data that is just above our level… but that isn’t to say that we can’t start to learn content far beyond our level… and then work to fill-in the pieces. The i+1 framework is somewhat arbitrary, but I’d say that its application is more important for language instructors and curriculum designers than for language learners themselves.

Finally, there is some truth to the notion that language learners can develop some bad habits due to fossilization and repeated errors that eventually get ignored (e.g., second language speakers dropping the be-verb repeatedly and ignoring it while thinking they have achieved fluency and accuracy). Still, language is acquired through both input and output! Even reading is a skill where the reader is actively forming meaning together alongside the input from the text. And if you want to speak a language… you need to start speaking the language (obviously)! And mistakes are actually a very important part of the learning process! So waiting for “input to overflow” can be a bit silly. I think Krashen was just concerned that some learners might be required to do too much at once because bad instructors force them to make paragraphs before they can make a sentence, so to speak.


Is this all assuming you are learning solely on your own in a complete vacuum? With no feedback loop from a qualified teacher and/or native speaker? My native-speaker teachers so far have all worked to provide feedback on pronunciation specifically to make sure we are speaking correctly.

Is my experience that different where I don’t understand where statements like this come from? :man_shrugging:

I also gotta disagree that outputting the language isn’t helpful. Speaking and writing and getting lots of feedback has been nothing but immensely helpful to me even when it comes to applying it to inputting content.

Even when learning Spanish in high school and college all my teachers emphasized good pronunciation.

1 Like

Yes, I am talking about self-learners in general. The thing is most people don’t have someone to fix their mistakes. Like, a friend might fix one or two mistakes you make but at some point, he will get tired. So you don’t have any way to know how well you are doing unless you compare yourself to native speakers.

For example, in English, we don’t pronounce the “b” in “doubt”. Even the word “water” is pronounced with a weird “t”, more like a “d”. If you don’t hear these words many times you might pronounce them wrong and keep doing that because no one fixed that for you. Same thing when you talk about places, how am I supposed to know if I should use “in”, “at” or “on”?

For your example, I can’t say because I am not there with you but I doubt they will correct every mistake you make. Also, I have no idea how strict they are with you. If your pitch accent is not bad, I don’t think they will say anything. Take that “not bad” with a lot of other words and you will speak “not bad” overall. You won’t have a thick accent but you won’t sound close to a native speaker.

What I want to say is feedback on its own is not enough. It’s like you are telling me if I speak with a lot of italki teachers I will sound like a native. You can repeat what I said before and add the fact that the teacher will never speak at full speed because he is taking your level into consideration. You can sound fine saying ”私は(name)です” but how will you sound when you pronounce something like “そうだね気分転換に気楽なデュエルもいいかも”?

The input hypothesis works great if you want to have a great accent. If you just want an okay accent it might not be needed.


But output is absolutely necessary for this too, because sometimes speakers perceive their own utterances to be more accurate than they actually are, and it is important to train the tongue, which is what the input hypothesis tends to overlook.


Okay, then I guess I get the perspective more. Seems odd to me, though, to rely on no other person as a feedback loop when learning a foreign language.

Sure in the absolute beginning there isn’t a strict correction of everything, but my beginning-level teacher emphasized proper speaking of syllables. Even so far as to emphasize the proper way of saying ふ and the R-row. The more I’ve advanced, the more picky they’ve gotten. And, yes, I have had one teacher who was very particular about pronunciation and would have you repeat yourself multiple times until you spoke the way he deemed proper. He is also very adamant about writing a lot, providing feedback and then having us speak what we wrote.

How exactly do you confirm you have a ‘great’ accent if both never speak and have never spoken to anyone? What we sound like in our own heads is not going to be the way we sound to others.


Of course, I never denied that. I just said that it’s not what we should start with. I did practice speaking a lot with English to reach a level that I am satisfied with. However, I did that when I was already fluent.

Is it necessary to start that late? No.

The thing is if you get enough input(especially “active listening”(no subs)) you will reach a point where you will be able to tell if your accent is good or bad because you know the sounds so well and you know how the native speakers usually speak. You might overestimate yourself your accent a little bit but you will have a good idea about your level overall.

You don’t need anyone to learn a language. Do enough immersion and you will get there. Also, why do you think you need to hear people say “your accent is good” to know if your accent is good or not? I mean even if you speak two Japanese sentences people will tell you “日本語上手” anyway. It’s not like people will tell you “your accent sucks” to your face.


Note: This may not be relevant to beginners.

Just to provide some context on my end, I have done close to zero output since my beginning to learn Japanese. (I had to do a little output for one semester in college and for the 6 weeks or so in which I have been in Japan on vacation.) I am roughly N3 level in terms of traditional textbook learning (finishing up Tobira) with maybe 200 hours of listening practice outside of the classroom. That’s not a lot of listening overall. However, I also have no intention of practicing output in a vacuum. For the time being, I will be doing all speaking with professional teachers on Italki (without explicitly asking to be corrected) and all writing in Italki notebooks with community feedback.

I don’t care about sounding native level, and I don’t want to develop a thick accent. With the above strategy, I don’t think I will develop a thick accent, even if it’s generally easy to tell I’m foreign. In my position, I think, in terms of marginal utility, that it’s more optimal to start practicing output now than later.

I do see how practicing output in a vacuum from the beginning could result in a thick accent.

1 Like

I am glad to see that you came to this conclusion.

I can understand your problem with the method. As you may have guessed, it’s an extreme method for people who want extreme results. Not everyone wants to sound close to native(like polyglots for example) and not everyone has the dedication to do that kind of immersion.

I am not trying to prove that this is right way to do things. I just wanted to explain the goal behind it.

One thing I don’t like about these SUPER SERIAL INPUT ONLY YOU PLEB Japanese is that it ignores a very important set of Japanese learners; those that actually live in Japan. These kind of methods assume people are only learning for a hobby and that simply isn’t true.

I actually met one of these types before. She came to Okinawa for a few days before going to Akita university. She could read a heck of a lot better than me but when it came to conversation it was robotic stiff and just… bad. Apparently it was the first time she was speaking Japanese. This girl was supposedly N2+ and could barely hold a conversation. It was unreal.

In the end you need both. That’s just a fact of life. You need lots and lots of input and lots and lots of output. Anyone telling you otherwise is honestly trying to sell you something.

You want to know how you learn Japanese?! You just stick with it and don’t give up. That’s it. There isn’t any secret methods. You put in the work and grind. Pick whatever way you want just stick with it.

Sorry if my reply sounds a bit aggressive but this kind of snooty nonsense about there only being ONE TRUE PATH is just silly and all it does it make people feel bad.


But I never sold anything :sob:

I completely agree with you on this one. However, the method is just looking for the most effective way to fluency/sounding like a native. I don’t think it’s being mentioned as the go-to for everyone, in every circumstance(maybe I am wrong).

That’s the risk that you take to develop a great accent. That being said, N2 isn’t a good measure for speaking. People who focus on JLPT are usually weak in listening. And if you want to sound good, your listening needs to be at least as good as your reading skills. So it’s no surprise that she sounded like a robot.

She didn’t specifically mention studying for the N2 I was just using it as a reference. Also her accent was decent but it was strange. At the same time, despite outputting since I’ve studied Japanese I’m told I have an understandable and clear accent. Not just coming from friends but Japanese teachers as well. Are the blowing smoke up my butt? Maybe but I never had a problem being understood.

I just think that kind of thinking that it’s a little too armchair linguist to make these claims about being the most effective way to be fluent/native.

Plus it has a touch of “icky foreigner” nuance to it as well. As if having an accent is such a terrible thing in the grand scheme of things.


Not to mention these trends come and go. Before it was all about output getting people to speak as soon as possible. Other times it’s all about writing. I remember when anki was the coolest kid on the block. The next trend will be all about grammar or shadowing… who knows.

But what is the truth is that through all these trends we have people becoming effective and fluent Japanese speakers. From the days of Hepburn to Heisig to Breen. There will always be the BEST way to learn the most EFFECTIVE way to sound like a native.

I’ll just echo what I said. The one cool trick to learning Japanese is simply to not stop.


I fundamentally agree that input is a very important part of the equation, and one that is often underlooked by people who think textbook + srs is the way to go.

Anecdotally, I’ve been in Japan the past 6 weeks, barely speaking but having the TV on all day playing whatever garbage NHK program is on, a nice benefit being that Japanese daytime TV has subtitles on basically everything. I actively watch only about 30-60 minutes a day, but I still have Japanese background noise the rest of the time. It has helped tremendously. Despite mostly pausing my grammar studies in this time and slowing down on WaniKani significantly (max 50 apprentice items, max 10 lessons/day), I feel like I’ve progressed more than in any other prior 6 week period.

The AJATT type of people are too extreme in their views IMO, and textbooks, SRSs, etc certainly have their place. But I think it’s important not to discount their ideas completely as I do believe that most people err on the side of too little input.