Native speakers are often the worst guide to their own language. But understanding why can help us enormously.

In a recent comment, one of our readers told us what a native Japanese speaker said about the word いぬ, meaning “dog”:

as I was told by a Japanese person, 犬 is usually used for a wild dog, イヌ to refer to a dog in a more biological context and いぬ for a pet dog.

While one feels hesitant to disagree with a native speaker, I am 95% sure this is not correct. I have seen pet dogs called 犬 so many more times than I have seen it written any other way. But not relying on my own memory and limited experience I tried an experiment:

I tried searching “うちのいぬ” and “うちの犬” (“our dog”, therefore definitely a pet) on Google (using quotes to specify the exact form) and got around 70,000 entries for うちのいぬ and over half a million for うちの犬.

How surprising is this kind of misinformation coming from native speakers? Actually not very surprising at all. One of the reasons I shy away from sites like Lang8 (where native speakers correct one’s writing) is that I saw a large amount of pure misinformation coming from native English speakers about English usage.

It is a commonplace in linguistics research that if you want to know how someone pronounces a word, you must never ask them how they pronounce it. You must steer the conversation so that they say it naturally.

Why? Because how people think they pronounce a word and how they actually pronounce it are often two different things. If you ask them they will think about it and tell you what they think they say. If you maneuver them into saying it they will simply pronounce it as they always do.

Similarly with rules and usages. An ordinary untrained speaker gets natural usage of her native language right most of the time. But when she stops to think about what that usage actually is, and what the “rules” are, she gets it wrong as often as not. Just as when you start thinking too hard about the spelling of a word you often become unsure of it, even though you know it well.

Using my old analogy of touch-typing, it is just the same as the fact that when I am typing I hit all the keys accurately and at speed, but if you ask me where the V-key is, suddenly I can’t even find it.

The part of the brain that processes immediate, automatic actions is different from the part that processes conscious, thought-out actions: and it seems that those parts often can’t even communicate with each other very well, and even obstruct each other.

I am not saying that native input and correction is not useful, though it is more useful when the native speaker is actually present and says “Oh, that isn’t how we say it”, because her first reaction is usually correct, and the more she thinks about it and refines it the more likely it is that misinformation will creep in.

However, my point isn’t really about native correction at all. It is about the automaticity of language and how it works both for us and against us. It works for us when we speak a language natively or pick it up by immersion.

Language is an automatic process. The more we reduce it to rules and conscious processes, the further we get from natural language. This is not to say that we shouldn’t learn basic grammar. And I do spend time explaining how Japanese actually does work in a way that most textbooks never do.

So am I contradicting myself? Not really. As I have always said, learning rules and grammar is a shortcut. Native speakers often can’t articulate these things very well themselves, but they use them mostly unerringly.

If we want to speak and understand Japanese naturally we have to get to the point where these academic abstractions are mostly forgotten. If I want to know a “conjugation” or whether a ru-ending verb is ichidan or godan, I almost always have to reverse-engineer it by first remembering how I would say a particular thing and then seeing what that tells me about the verb or “conjugation”.

I am not working from the rule to the language, I am working from the language to the rule, which is how we work in our native language. This is not to say that my Japanese is anywhere near perfect. But what I do know tends to work the natural way around.

How did that happen? By immersion and massive input.

I have done some study, but I would say that for every hour of actual “study” I have done, I have done at least 20 hours of reading novels, watching anime and other shows, listening to stories, playing text-heavy games and daily conversation.

I would say that a 10 : 1 immersion-to-study ratio is the very minimum one should have (except in the very early stages). Learning Japanese primarily by study keeps it in the area of the brain dedicated to conscious, deliberate knowledge, and that is not the area in which real, natural language lives.

In language, conscious study is a good servant, but a very bad master.