Yonosa Wrote:Did anyone besides me notice that Mr. Igor practically just started this thread to argue? Dude, why are you asking people's opinions and then bashing them? Whether or not you want to use an SRS or not is up to you. But don't debate the factual evidence that it does help you remember what is in it, just the same as reviewing other ways does like you said you do. The thing is some people don't want to bother to try and "schedule" physical reviews as they are too busy or otherwise too distracted with other areas of interest perhaps even in the the said language itself. In other words we all agree that reviewing helps us too remember... so what the hell is this arguement about? You keep bringing up about the fact that you find SRS boring, great, don't do it. But don't assume everyone finds it boring, I personally enjoy successfully reading multitudes of sentences that contain a wide range of vocabulary each day and testing my ability in them, and then reading for enjoyment and speed on top of all that makes for an even more enjoyable experience for me. You don't enjoy SRSing, so just say that, no point arguing against it, rather you're only arguing against the idea of reviewing on a computer which a lot of people don't mind at all.

Are you talking about the catfight that Woodjor picked with me? I didn't want to argue about that. It wasn't even an argument about SRS. It was a stupid argument about him misrepresenting what I said. <---Which you have also just done. You're making the same exact mistake that he made. My point was never that SRS reviews are intrinsically boring. Ever. My point was that SRS reviews *IN COMPARISON TO READING* are boring. On the contrary, I often enjoy sentence reviews very much some days.

And yes, the point of this thread was to argue. What the flip is the point of a discussion forum if you can't argue/debate to explore an issue? I don't think you quite understand the difference between a flame-war and a genuine argument.

yudantaiteki Wrote:

Matthew Wrote:The idea that any native speaker would only know 1000 kanji is beyond absurd.

Are you just talking about highly educated native speakers, or all native speakers? There's no reason to believe that Japan's education system is light years ahead of the US's. Highly educated native speakers probably know more than 1000, but I don't know why you think that the average schlub that doesn't read very often and hasn't studied kanji in decades would know huge amounts. Of course the average native speaker is probably not a good target for your own studies.

Contrary to your experiences, I've seen native Japanese working on a PhD (in Japanese studies) have trouble with kanji, even ones on the Jouyou list.

Quote:I think the root of this erroneous way of thinking is likely just Westerners' tendency to overestimate their own Japanese ability, eg "My Japanese is REAL GOOD, so if something seems hard to me, the only possible conclusion is that is too obscure, even for Japanese people." I think that anyone who spends enough time on the road to learning Japanese will eventually realize even level 1 of the JLPT is still just pretty basic stuff.

Rather it stems from the eventual realization that you don't actually need an excessive amount of kanji to do quite a lot with the language as long as your vocab and grammar is good.

Ok, this 1000 kanji business is just beyond ridiculous. Devise whatever method you want to test any native Japanese on kanji, and the results will completely disprove this nonsense.

Recently I had a Japanese friend of mine try the kanji/vocab section of JLPT1. She went to high school in Japan, college in the US, and has lived out of Japan ever since. She always complains that she's forgetting Japanese because she never uses it anymore. Anyway, she finished the section in 1/4 the allotted time with only one question wrong. Her opinion was that it was so easy she couldn't understand why even I would have any problems with it.

And saying people "don't read" in Japan makes no sense. If you've ever even visited Japan, you would know that people are constantly surrounded by at least 3000 kanji on a daily basis. Just in commonplace things like reading a menu, renting an apartment, documents at work, labels at the drug store, names of people and places, newspapers, magazines, advertisements, etc. They don't need to study when there are constant reminders everywhere.

And again, you don't even need to take my word for it, go ahead and test any Japanese person in any manner you like, and you will find out what I am saying is accurate.

Matthew Wrote:Ok, this 1000 kanji business is just beyond ridiculous. Devise whatever method you want to test any native Japanese on kanji, and the results will completely disprove this nonsense.

There's no point trading anecdotes; it doesn't prove or disprove anything. Without literacy studies, which the Japanese government doesn't do, there's know way to know for sure.

Quote:And saying people "don't read" in Japan makes no sense.

A lot of people don't read all that much, just like in the US.

Quote:If you've ever even visited Japan, you would know that people are constantly surrounded by at least 3000 kanji on a daily basis.

Even if that's true, it doesn't say anything about whether every Japanese person can actually read all of those kanji. People in the US are constantly surrounded by written material everywhere they go, but it's an unwarranted assumption that all Americans can read 100% of any written material they pass on a daily basis.

Quote:And again, you don't even need to take my word for it, go ahead and test any Japanese person in any manner you like, and you will find out what I am saying is accurate.

On the contrary, as I said, I've had numerous experiences with highly educated native speakers having trouble with even the Jouyou kanji -- but as the saying goes, "the plural of anecdote is not data."

In any case, the root of this was your claim that you need 2500 kanji and 15000 words to be "fluent", and your implication that this cannot be done without some sort of SRS or the like. 2500 is a pretty arbitrary number that I think is higher than most people are going to need.

IceCream Wrote:ok. i totally reserve the right to change my mind on this in a few days / weeks / months...

As we all do!

IceCream Wrote:Over the last month or so, me and Albion have been trying to collect together a manga JLPT... every time i go to do a bit, just how many words we haven't encoutered in that time... Every new one seems hard, then by ep 7 its fairly easy. But even then, there's a ton of words that are only said once.
and, don't expect to encounter those same words often outside of a series.

There's an 80/20 application here. 80% of the words on the page you will encounter again, quite frequently. The other 20% aren't just rare, they are rare and drawn from a very large sample set of higher vocabulary. This is what I was talking about in the other thread with regards to long, and statistically imprecise tails. There is a core of the language that you can learn in order to get 80% comprehension. But that remaining 20% depends so much on subject matter and authorship that there's no way you can expect to systematically attack it in a reasonable amount of time, except by consuming a lot and learning from context. (Or, I suppose, looking up as you go and adding to an SRS. I've never claimed that this won't work, just that it is inefficient.)

BUT, the silver lining to this debate is the that 80% is all you need to start learning from context. Yes, it will be frustrating at first as you will have no idea what 1 out of 5 of the words on a page mean. That's what I mean by learning in context is not something that you just do, but rather a skill that has to be developed. And, if you start reading something to learn from context, I would recommend that you finish it. The author will continue to draw on the same vocabulary for the same subject matter, so it's really not until much later in the book/series that you start to get the payoff of having seen these words in multiple contexts.

If you're just starting out at that 80% level, I recommend manga, where the pictures provide plenty of context, or reading a book that you've already read in translation so you're familiar with the story and writing style. Or hey, a manga that you've already read in translation (or watched the anime adaptation) would be even better.



I'm sorry, I wanted to respond to your other points as well, but I have to go right now.

I used the Tensei Jingo column from the Asahi newspaper a lot when I was at that level:
http://www.asahi.com/paper/column.html

What I liked about them is that they're short, there's a new one every day, and they're on a wide variety of topics so it's easy to find 1 or 2 a week that are interesting. In addition, they can be copied and pasted into JWPce for kanji analysis or lookup, and you can print them out in useful layouts for study.

What I did was print out the column so that it took about 2/3 of the page, then I wrote the definitions of new words and kanji on the bottom part. I usually read through the column once without looking anything up, then looked everything up while reading it through more slowly. Then I would come back to it in a day and fold over the bottom of the paper so that I couldn't see the things I had written down, and try reading it again.

yudantaiteki Wrote:In any case, the root of this was your claim that you need 2500 kanji and 15000 words to be "fluent", and your implication that this cannot be done without some sort of SRS or the like. 2500 is a pretty arbitrary number that I think is higher than most people are going to need.

I think most people studying here plan to read a lot and will need more kanji and vocab than the theoretical average native who doesn't read much. At least in order to have some level of ease in reading.

I agree, which is why I probably should have just not mentioned that Jim Breen thing. The dilemma with kanji studies is that there's a very severe situation of diminish returns in studying them. I know these statistics are well worn, but the newspaper studies show that the most common 150 characters account for 50%, 500 for 80%, 1000 for 94.5%, 2000 for 99.7%, 2500 for 99.9213%, and 3000 for 99.9742%. Studies in Chinese also come up with almost the same numbers.

Given that, it's hard to say how many kanji you actually need to study. Going from 2000 to 2500 gives you an increase of .22% in kanji coverage, which isn't much. I think it's up to each individual learner to decide when they hit the point where the improvement they get out of kanji study is less than their interest in it. You're never going to hit a point where you can say "OK, I'm done, I have studied every kanji that I will ever see." Of course there are extremes -- 1000 is probably not enough, and 10,000 is most likely overdoing it. But the number of kanji, and which kanji, someone needs to know are also heavily dependent on exactly what they're going to be reading and how much they read. Taking any list, whether the Jouyou, or JIS level 1, or "2500 kanji", or RTK 1 + 3 and declaring that to be what you need for "fluency" is arbitrary.

(I also think that as your Japanese improves, there's actually a paradox where you don't need as many kanji because, like a native speaker, you can figure out the meaning of words from context even if they contain kanji you haven't seen before.)

igordesu Wrote:I know the guy in the following link (that Khatz mentioned once) talks about that: http://tesl-ej.org/ej45/tesl-ej.ej45.fr1.pdf

Kato Lomb was a woman, not a guy.

I haven't read your entire three page thread (and at some points flame war, or so I assume by the topic under discussion), but I pretty much just read and only use SRS for Heisig and it seems to work fine for me. I like smart.fm too but am pretty lazy about it.

The diminishing return has come into play for me for many aspects of my Japanese studies.

For Kanji, I stopped halfway into RTK3, but I'd argue that one can do just well with RTK Lite (either version) then add kanji as you get to them. At 2500, I've only added an additional 35 that I've come across "naturally" in studying. Basically, I spent too much time on kanji.

For Grammar, I did Tae Kim which seems reasonable on returns. I do come across items now and again not in Tae Kim, but not so much. I think I've spent the right amount of time with these.

For Vocabulary, I did iKnow 2000 then went to adding as I came to them. I'm still adding a lot of words. Probably was a few more "common words" that could be needed. Thing is, it's not that I did not spend enough time, I just probably used the wrong source. Based on other testimony, KO2001 may be the best source to learn 3500 basic (and useful) vocabulary fast.

There may be a "sweet spot" of Kanji, Vocabulary and Grammar concepts to learn but that's a hard spot to find. Probably you'll realize it after the fact. Plus, it'll be different for each person.

Either way, one does need to get to native material proper and try learning off those. The sweet spot is just meant to get you going with the native material at the ideal pace. I think we're finding what can work best on average.

Yeah, I used the above with an SRS. They're basic and soon you'll see them soon enough in native material. But in the build-up, the SRS prevents memory decay so you're not reproducing effort later.

I'm going to side with Matthew here. It's silly to think that a normal, functioning Japanese adult would only know a thousand kanji. That's what a beginning Japanese learner would know after having completed a book like Kanji Odyssey 2001. I think the whole argument rests on a misunderstanding of what it means to know a character. If you define knowledge of a character as the ability to draw it perfectly, to know all of its readings or to identify it out of context then perhaps you might find that Japanese know no more than a thousand kanji. But that's for all practical purposes a meaningless figure. What is certain is that Japanese can recognize and read much more than 1000 characters when used as part of a text. You would probably have to look hard to find even a high school student who could not recognize 稀, let alone　滅　or 更 when placed in context (or even out of context I would say).

Just for fun, I replaced the four characters that you have identified with question marks. That's how the text would look to someone who only knows a thousand characters.

方向指示器（ほうこうしじき）とは、自動車、オートバイ等に付ける保安部品で、右左折や進路変?の際に、その方向を周囲に示すための装置である。方向を灯火の点?で示すことから、日本では通常、ウインカー（英語: winker“まばたきするもの”）と言うが、現在英語?において winker と言うことは?である。

Out of curiosity, would you say that the average adult American knows all the vocabulary from the SAT?

I would not. But the SAT is a test aimed at aspiring university students. 1000 kanji is the standard expected out of Grade 6 students.

Well, to be fair, I find the 1000 number a little surprising as well and I wonder whether the original scholar meant that the average Japanese person would only be able to know around 1000 kanji out of context (and presumably a fair amount more in context), which I would find easier to accept. (3000 still seems like a very high number for an average native speaker, though -- I think saying that an average Japanese native speaker knows 3000 kanji would be like someone saying that the average American knows all the words on the SAT.)

I'd say the average native speaker would know around 2000 with knowledge of kanji beyond that being present but possibly incomplete if you get what I mean.

I think jinmeiyou kanji is a good example, I'm sure Japanese know lots of most it or whatever but lots of the time names are displayed with furigana so ppl know how to read them and such. Place names too perhaps.

You can't learn Japanese without using SRS, just like you can't learn kanji without using stories, via the Heisig method. 'Nuff said, ruiner out.

Well, I'm going to side with Matthew here.

"the plural of anecdote is not data." So here are some facts which show that the standard for the average native speaker in Japan is higher than you might expect:
1. Reading capabilities of 15 year old Japanese is in all (OECD) Pisa studies conducted between 2000 and 2006 above the US or Germany.
2. The rate of Japanese pupils going to high school is frightening high (>95%)
3. Of these almost 50% go to University or Junior college

I am not going to trade any anecdotes, but knowing 3.000 kanji is just quite plain standard for a huge group of the Japanese population.

Transparent_Aluminium rightly pointed out that "knowing" is a very relative term. So yes, Japanese make mistakes, sometimes even very simple ones. But here is the explanation for these mistakes which fits nicely to the topic of this thread:
Japanese do not use SRS for the kanji!

yudantaiteki Wrote:A lot of people don't read all that much, just like in the US.

I'm sure a lot of people don't, but the book industry does way better in Japan than in the US, as do the newspapers.

Matthias Wrote:Well, I'm going to side with Matthew here.

"the plural of anecdote is not data." So here are some facts which show that the standard for the average native speaker in Japan is higher than you might expect:
1. Reading capabilities of 15 year old Japanese is in all (OECD) Pisa studies conducted between 2000 and 2006 above the US or Germany.
2. The rate of Japanese pupils going to high school is frightening high (>95%)
3. Of these almost 50% go to University or Junior college

I am not going to trade any anecdotes, but knowing 3.000 kanji is just quite plain standard for a huge group of the Japanese population.

Transparent_Aluminium rightly pointed out that "knowing" is a very relative term. So yes, Japanese make mistakes, sometimes even very simple ones. But here is the explanation for these mistakes which fits nicely to the topic of this thread:
Japanese do not use SRS for the kanji!

Finally, some solid hard evidence.

Matthias Wrote:1. Reading capabilities of 15 year old Japanese is in all (OECD) Pisa studies conducted between 2000 and 2006 above the US or Germany.

"On the other hand, the low internal variation in results implies, despite strong average
performance, that Japan is not exceptional in the numbers of highly proficient readers – only an average proportion of 10% reach the highest literacy level, at which students are able to understand complex texts, evaluate information and build hypotheses, and draw on specialised knowledge (see Table 2.1a). In other countries with a similar level of average performance, such as Australia, New Zealand or the United Kingdom, the proportion of top-performers is larger (between 15% and 19%) mainly because the gap
between the lowest and highest performing students in these countries is larger. In mathematical and scientific literacy, on the other hand, the score above which the top 10% of students perform is higher in Japan than for any other country, and the same is true for the top 25%."

Also remember with that 95% of students going to high school, that includes agricultural and technical high schools, which typically have no entrance exams (or easy ones).

In any case, going to high school and college still doesn't really say anything about this magic 3000 number; where is that number coming from anyway?

It's a quote from the Pisa study that Matthias mentioned; it indicates that Japan's literacy is similar to other countries at least among 15 year olds. I'm certainly not disagreeing that there are a number of highly literate Japanese who can probably handle 3000 characters, and maybe more -- what I don't agree with is the idea that 3000 is a standard that any Japanese person can be expected to know, and more relevantly, that 3000 kanji (or 2500) is a standard that learners must aim at if they want to be "fluent".

IceCream Wrote:The amount of kanji that you need to be fluent is the amount you can read comfortably with.

And that's pretty much what I'm saying. Basically my view is that in the beginning, you have to do almost exclusively "study" with not much "natural" reading. As you move on, you can decrease the amount of study that you do and increase the amount of reading, depending on your interest. If you don't like using flash cards, SRS, or textbooks, you can abandon it earlier than someone who does like it. Because everyone is going to be reading different things (and different amounts), it's impossible to come up with a fixed list of kanji or amount of kanji that everyone *must* know to be fluent.

I'm not a native speaker, but I understand enough kanji that I can read novels and such--and I don't know every word, but I know what I need to know.

Looking at the kanji frequency table in my kanji dictionary, 1600-1615 are:

顕、鈍、隅、凶、糾、塾
誓、酬、冗、紺、漂、桃
伯、庶、晶、尻

I guess that 晶 has become more common since my dictionary was compiled, with the rise of liquid crystal displays. 鈍、隅、誓、漂、紺、桃 are ones I seem to come across pretty frequently in my reading. You couldn't live in Japan without coming across 塾 all the time. 冗談 is common enough to be worth knowing but I don't know if I've ever seen 冗 in any other compounds. Depending on your choice of reading material 尻 can be pretty common. 伯父さん is one of the possible ways of writing おじさん, but I think I only know it because I was reading a manga with a Count Something-Or-Other (伯爵）.

凶、庶 I know only vaguely.

顕、糾、酬 I cannot recall having ever seen before.

My point is that, indeed, as you get to the less common kanji (1500 and up, maybe?) you start running into more and more kanji -- like 伯 and 尻 -- that are way more common in certain contexts than in others, and kanji that you can get by very well without knowing depending on what it is exactly that you need to read.

I would say I pretty much agree with your last post Yudan. Until someone finds a study that tells us how many kanji Japanese speakers know on average, it's impossible to come up with a hard number. My feeling though is that 2500-3000 is not exaggerated.

Looking at the kanji that Fillanzea has brought up, I think this tells us that kanji in the 1600-1615 range are still pretty common and do not qualify as rare or exotic by any means. Out of those, I don't remember having seen 糾 and 凶. 顕, I have seen has part of けんびきょう, microscope. 酬 is part of ほうしゅう　and 庶 is part of しょみん.

Fillanzea Wrote:酬

報酬？

顕、鈍、隅、凶、糾、塾
誓、酬、冗、紺、漂、桃
伯、庶、晶、尻

I think that list is a good example of kanji that you don't necessarily need to specifically study in isolation because most of them only show up in a small number of words, sometimes only one word (e.g. 桃色, 誓う, 報酬, 尻, 塾, 庶民, 結晶). 糾 is the only one there that is totally unknown to me; 凶悪 shows up sometimes in video games.

In addition, there are several examples of kanji there that you might see a lot if you're reading the right things, but you might not see at all in other types of reading -- I would say that 凶、酬、伯、and 庶 may qualify there.

But now look at the higher numbers -- here's 10 kanji around the 2500 mark for frequency:
彗 牌 醍 塑 漸 穣 砦 醐 沓 迪

I recognize the first one from 彗星, the second one from Mah Jong terms (聴牌 means you only need one more tile to win), 砦 from video games, and 醍醐 because Emperor Daigo shows up a lot in writings about Genji. But obviously we're getting into heavily specialized terms and I think you would find a lot of variation even among native speakers as to which of those they would know.

Now move up to 2900:
靡 顯 躬 贅 蜃 糺 秉 猥 熾 杣
顯 is kind of a gimme because it's just the old form of 顕. 贅沢 is probably fairly recognizable, and for some reason I feel like I've seen 猥褻 before, but the other 7 are totally opaque.

(怯 is almost at 3000, but 怯える and 卑怯 are pretty common, maybe just not in newspapers.)