Just to be able to read the majority of relatively average Japanese media. I know there's a ton of kanji, but I think I've heard that not too many (relative to the total) are actually needed? Is that true?
I think 常用漢字 (2,136 kanji) is non-negotiable. Depending on your definition of "able to read" and "average", you might need a few hundred more. I know about 1300 and depending on what I'm reading, I frequently run into kanji I don't know / can't read.
For example, 少年漫画 or Social media posts are mostly fine, but give me a Wikipedia article and I suddenly feel illiterate...
I've always heard kids learn about 2000ish up through high school and maybe another 1000 or so by the end of college, but I don't have a source on that. I imagine people also pick up some uncommon kanji through osmosis.
So as far as conscious study goes, probably about 2000-3000.
I think you might have picked up some information on frequency of kanji usage, which can be misleading.
I do not have the numbers on hand, but it is most likely something along the lines of "30% of kanji make up 80% of all usage". The issue, though, is that the majority of kanji appear in compounds of 2 or more terms, so even if the first character in a ton of terms is 学, for example, the variation with the rarer terms can come in the 2nd character, etc.
~150 kanji are the top 50% of kanji used across various corpora
~900 kanji makes 90%
~1250 kanji makes 95%
~2000 kanji makes 99%
~4000 kanji makes 99.99%
(These numbers are averages and will of course vary depending on what type of content you look at; e.g. newspapers require fewer kanji to read with good fluency than literature.)
I would say the question is based on thinking of kanji like an alphabet (with a fixed number of characters you need to read things, and then you can just stop learning the characters), when for counting purposes they're much more like words (with the number you learn being potentially hundreds more than even most native speakers learn, because depending on what you read, there will always be contexts that you will encounter new rare ones in).
I disagree with 2136 Jouyou kanji being non-negotiable, especially if your goal is to read average Japanese media. The Jouyou list contains a lot of rare kanji that are only included because they appear in historical documents or the constitution, and will likely not be meaningful for the average learner (looking at you, 穀), whereas there are many non-Jouyou kanji that are comparatively common or you can expect to see often in the names of people and places.
If you're interested in taking advantage of frequency and getting away with as few kanji as possible, my recommendation is to first set yourself a goal for say 1000 kanji (~90-95% coverage). You can get those 1000 from whatever study order you prefer, and then once you reach that point and feel you have a solid foundation, switch to studying based purely on frequency in whatever corpora you're interested in. Once you start to feel diminishing returns from studying that way, and you're getting lots of kanji that you haven't and don't feel likely to encounter, switch to just picking up new kanji as you encounter them in the wild. At a certain point, it will be up to you whether you continue to actively study new kanji, or just look them up in a dictionary whenever you find a new one and move on, but that point is different for everyone.
@ポールおじちゃん、that link lead to the word rice but the 2nd kanji #穀 in it means cereal and grains. But the first one #米 means Rice usa and meter but I can't find anything about constitution unless it is pronounced the same with different kanji
Edit #米国(べいこく) means USA possibly being considered a constitution and pronounced the same as rice 米穀(べいこく)
I think to me, the main point is that as you continue to read, you will continue to encounter new kanji. there are after all tens of thousands of them. So you will never actually stop learning kanji, as long as you keep on reading japanese.
For me, i'm continuing to learn kanji individually, i don't think i've reached the point of diminishing returns yet. But i'm taking that nice and slow. At the same time, as i'm encountering new words, if i find them interesting or relevant or anything, i add them to my interesting words schedule. And i learn the kanji just in the limited use it has there (the specific reading in this word). then if i encounter the kanji more often (in other words) i learn that as well. and if at some point it appears in my kanji schedule, all the better, i already have vocab with it, which makes it much easier to remember.
So i end up knowing kanji on a sliding scale from "i know the meaning, all the readings, even obscure ones or ones only used in names. can write it by hand, etc." to "i kinda know one word which uses this kanji with this reading, and i usually recognise it when it appears in this context."
Of course if you're learning for an exam or something this will be different, you might have a list of kanji you need to know. But for reading manga, i find i can enjoy them easily with where i'm at (about 800 kanji i know well). while still having to look up a word or two every few pages (or more often if it's a topic i don't know much about yet. But i fully expect that number to increase steadily for a couple more years.
"Just to be able to read the majority of relatively average Japanese media."
I think you should rethink that goal and what it will allow you to do once you reach it. A lot of people think that after learning a certain number of common kanji or words (JLPT list, jouyou kanji, etc for example), they will be able to read without trouble on average. That never happen. Just ask to some N1 holders and see how many still struggle with reading.
And the reason for that is simple. As soon as you try to read a specific topic, you're bound to find specific kanji, kanji reading. or words. And just 1 of them can prevent you to understand a entire sentences. That's why a lot of intermediate and advanced learners still feel frustrated. Because they only focused on common kanji and words. And every time they try to read, there are a few specific few specific words/kanji that appear and are like the most important part to understand like 60% of the text (random number but still, 60% is not an exaggeration).
Basically, you could study 4000 kanji or 10 000 words and still struggle to read on average. But that's fine. Because you don't need to be able on average.
That "average" is something people often underestimate. It's a statistical average for the whole society including every specific topic that you may struggle with, even in your own mothertongue. For example, newpapers about recent research, math/economy article, law, etc. I don't how to really explain it in a simple way but in practice that 90% or even 99% doesn't mean that when you read a random, you will understand 99% of it. Most of the time, it will be like more 70%. Or if you're unlucky, it could drop to 40%. And the only reason why you hear number like 90% is because of statistic (common words raise "artificialy" that number).
So yeah, it's a trap. But the good news, on the other hand, is you don't NEED even 2000 kanji or 10 000 words to read most specific topic. You probably don't even need 500 kanji or 2000 words for most of them. What you will need sometimes is just a few hundred common words/kanji + a few dozens specific words/kanji. And you can achieve that goal faster than you think if you focus on those specific topic.
TLDR: Think of common kanji/words list as a gamble that will help you on average. But sometimes even the 10 000 words that you did study for 3 years will be mostly useless for a specific text or conversation. If you want to read something specific in the future, then start reading it and studying it. You will be able to read what you read. If I take the most extreme example, there are words and expressions, that you won't even find in dictionnaries. So, you can study common words/kanji as much as you want, there are things you will miss if you don't focus on those specific topic. And those are necessary to enjoy the topic that you like. And if you start right now, you will reach that goal faster you think.
But if you don't, and focused only on common words/kanji, then you will feel frustrated later when you start to read. You like historical drama/novels/anime/game? Well, you see all those words with "historical" or "archaism" on renshuu? You will encounter them a lot. And even natives struggle to understand them sometimes. You're more on the Sci-Fi side? Same things but with scientific words which are also uncommon. And natives can also struggle a lot on those. But the ones that often consume those? Well, that's how they learned all of those uncommon words/kanji and are able to enjoy it now. Start studying what you like right now and you will be as "imperfect" as a native. Because they are no native in any language that can read everything perfectly without any trouble or the need to use a dictionary. That's why dictionnaries exist even for natives.
PS: Despite what I said, I still think it's important to focus on common kanji. And I also agree with Gillian, a 1000 kanji is where the most of the kanji. Not because of frequency but because of "compoundability" (the ability of kanji to form compound words). Basically, the first thousand (like the kyouiku kanji) can unlock a hundred words on average (they are found in at least a hundred words). So it's a "gamble" where you win a lot. But past that... it's not as good as before. I'm currently studying the middle school kanji and readings, and it's already annoying. Like I already studied a few readings that even natives don't remember except in specific circumstances. There are a lot of things that become far less useful especially for a foreigner (historical reading with only one important words, reading with almost only technical scientific words, etc). So, past that point, I also recommend to almost fully switch to "targeted" topic instead. It's still somewhat useful to study them but you will gain far more by focusing on what you like and want to talk about (or read).
99% sounds like a lot, and it is—for one-off events. But for continuing processes, like reading or computing, it’s nowhere near adequate. You may have a 0.99 chance of getting each character right, but over 99 characters, 0.99^99 works out to be 0.36973.
That’s why reliability engineers speak of “five nines” (99.999%) of uptime. That’s the equivalent of fluent reading.
Yeah but even with that, it's not really equivalent in my opinion.
Like a engineer will say that for one specific process. And it's quite accurate in this case. But reading is more like a thousand different of process. There are sub-cluster that don't even shared 50% in common. No engineer will dare to claim 99% when the component are completely different each time. The scale, and claim behind it, are completely different. It's more like common words will be helpful 99% of the time. That doesn't mean that they will be enough 99% of the time.
Like the engineer, it's more useful to talk about specific things. In linguistic, there are the concept of breath and depth of vocabulary knowledge to evaluate reading ability. And every combination of different breath and depth, as well as topic, can create different types of fluency. Well, "types" is inaccurate since it's more like a quantitative perspective. You can talk a bit more about fluency in this case but it the exact opposite of "general fluency". It's more like "Broad fluency" + a few "domain specific fluency". No natives are considered "fluent" in this case and every native end up with different profile. The "Broad fluency" part is also lower than the "fluency" most people think of when they talk about language learning (like, you often hear 10 000 words in general but in those research, it's more like 2000 to 6000 words). And after that threshold, they start diverging a lot in term of depth and topic they "mastered" (whatever mastered means here btw).
In other words, that 99.999% is more like a level that is far beyond even the most educated natives. Even by spending your whole life, you won't be able to reach it. Fortunatly, like natives, you don't need too.
Hello? Um Shamugan I just want to say I am impressed and dizzy with how much you typed in that reply. But I think that means that you really want to help people and I thank and applaud you for that!👏🙏
At the same time, it's also a bit on purpose. First, I don't think most beginners should read me. Because they should only focus on one thing: Not giving up and finding their own rythm. But on the internet, you can hear lots and lots of potentially good advises. The problem is... It can be overwhelming to try to apply all of that. They think should apply everything they hear and if they don't, they will learn poorly, inefficiently. That's not true and also, it's useless if you get overwhelmed and just give up even before being able to apply all of those "good" methods. It's better to just ignore all of that and just focus on making your daily routine into a habit first. So if beginners feel dizzy and forget about what I say, or just ignore what I say, it's a good thing :3
Second, there are a lot of "good" methods out there. But there are also sometimes very different, sometimes contradictory and will also work for different people. And to find your "best" method, you need to know a bit of everything. Some method work well only with another method for example. But when you hear or read advises online, you only scratch the surface of how those people study. So I also talk a lot because I think people should know a bit more than the surface. When they know more than the surface, they can take an informed decision and decided for themselves if they need or want to use what I said. Even if that's mean doing the exact opposite of what I said. Honnestly, I saw a lot of different type of learners, doing completely opposite things and... they're fine. They have fun too. So whatever xD
So same for you. If it's too much, feel free to ignore it, you will be fine either way =p You don't need any of that, I'm just of the type that like that much detail. But some people don't and among those people, there are some that progress faster than me because they don't waste time on that kind of knowledge. So yeah x) And if you can find something useful, then good :D
Yeah but even with that, it's not really equivalent in my opinion.
Like a engineer will say that for one specific process. And it's quite accurate in this case. But reading is more like a thousand different of process. There are sub-cluster that don't even shared 50% in common. No engineer will dare to claim 99% when the component are completely different each time. The scale, and claim behind it, are completely different. It's more like common words will be helpful 99% of the time. That doesn't mean that they will be enough 99% of the time.
Like the engineer, it's more useful to talk about specific things. In linguistic, there are the concept of breath and depth of vocabulary knowledge to evaluate reading ability. And every combination of different breath and depth, as well as topic, can create different types of fluency. Well, "types" is inaccurate since it's more like a quantitative perspective. You can talk a bit more about fluency in this case but it the exact opposite of "general fluency". It's more like "Broad fluency" + a few "domain specific fluency". No natives are considered "fluent" in this case and every native end up with different profile. The "Broad fluency" part is also lower than the "fluency" most people think of when they talk about language learning (like, you often hear 10 000 words in general but in those research, it's more like 2000 to 6000 words). And after that threshold, they start diverging a lot in term of depth and topic they "mastered" (whatever mastered means here btw).
In other words, that 99.999% is more like a level that is far beyond even the most educated natives. Even by spending your whole life, you won't be able to reach it. Fortunatly, like natives, you don't need too.
The joke is funny though.
Learning kanji's good for you,bruh.