As we all know, Heisig is all about order.
His order is optimized for learning rather than effeciency in gaining literacy. When learning kanji by frequency, more words can be spelled with a fixed repertoire of characters. I was interested in how grave this difference actually is, and did some plotting:
I took JMDict, which is quite comprehensive and contains about 130000 entries - most of them being kanji words. I then checked how many frames a student must have learned to spell the entire word.
This is matched against the "common" order of learning Kanji. For this i used the frequency index found in KANJIDIC - which is far from accurate, but should suffice the job:


Nothing surprising there. The Heisig graph is exponential, which is to be expected when using any random order.
The frequency is somewhat logarithmic. This is also obvious, as the additional kanji become less useful.

Another statistic arising from this is the number of *new words per frame*. It's the number of new words you can spell when learning an index/frame (given you only know kanji of previous frames). For frequency order it obviously goes downhill (which is the whole meaning of it). Heisig is more evenly distributed, and even increases towards the end.


And this is my motivation to all of you learning RTK! Imagine learning in frequency order: You're at your 2000th-and-something kanji, learning it only to be able to spell a handful minor words. I'd not be very motivated to continue.
Heisig however keeps rewarding you for your studies, even later frames are as useful as any other.
The "best" frame in this context is actually 1885 (気) as it adds over 1000 new words!
So for learning ALL kanji, it might not be the best choice to chose that speedy path
And at the end of the day, common characters aren't necessarily useful ones.

Is 1885 "Spirit" (気) ?
And "Person" (人) is frame 951.

Chaze Wrote:And this is my motivation to all of you learning RTK! Imagine learning in frequency order: You're at your 2000th-and-something kanji, learning it only to be able to spell a handful minor words. I'd not be very motivated to continue.

The counter argument to that is that with frequency order, you don't have to learn up to 2000 kanji from a list, you can start learning kanji from readings you are interested in long before that point.

Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: "There are three kinds of lies: lies, damned lies and statistics."

Mark Twain

Just because you know a kanji, does not mean you know a word that uses that kanji. The reverse is also true.

I've found using frequency lists are useful for getting a large group of items you want to study (say 1000 kanji or 2000 words). One then groups that small sub-group into a more intuitive order such as themes or in my opinion 2001KO kanji order.

Same can work with kanji. Learn the JLPT 2, half of the Joyo, or 2001KO books 1 and 2 Kanji in RTK order (ie RTK Lite). Later own after you've learned enough vocabulary that uses those, you can learn JLPT 1 or 2001KO Book 3 , or the rest of the Joyo. Again, learn about 4000 words that uses those and you'll likely want to learn the rest of RTK3 or Kanken upto level 1.5 or 2.

Its the usefulness of frequency combined with the benefit of efficient learning order.

PS: If you are comfortable with radicals/primitives, then you can learn kanji in any order at anytime. In other words, you don't have to learn kanji in groups, just learn them after they pop up in new words you're learning.

yudantaiteki Wrote:The counter argument to that is that with frequency order, you don't have to learn up to 2000 kanji from a list, you can start learning kanji from readings you are interested in long before that point.

Yes, but before switching to Heisig I've been using a workbook - and these obviously start with frequent characters - ease of writing being another factor.
The first kanji I ever learned was 日, followed by 本 and 人. That's 1, 10, and 5 on my frequency list

jcdietz03 Wrote:Is 1885 "Spirit" (気) ?
And "Person" (人) is frame 951.

Meh, mixed them up. 人 is the best frame under 1000, but it's really 気 I meant.
I must admit I'm merely at frame 300 myself

A side product of this stat fiddling, I have "new words in this frame" lists.
So, for an example, my list for frame 36 reads:

Code:

自  (じ) - self-
自ずから  (おのずから) - naturally
自ずから明らか  (おのずからあきらか) - self-evident
自ずと  (おのずと) - naturally
自と  (おのずと) - naturally
自ら  (みずから) - for one's self
自明  (じめい) - obvious

(Note that words like 自分 are not included, as 分 is still missing at that point)

I have no clue how RTK2 works, but could this be a usable asset when going over to readings? I imagine it to be useful when already knowing Japanese, but not having learned any kanji..

Chaze, I wish I had your skills. 2 comments:

If your new words are coming from the 130,000 JMDict list, then I think you'll be learning too many rare words without the ability to distinguish them. There are other ways to obtain more common vocab containing a particular kanji. For example, cangy made a list of words by kanji (taken from JLPT, 2001KO and Lastfm's core lists). 'Kanji in Context' is also a good source. IIRC, JMDict also marks common words - perhaps you could limit the search to those?

I don't think it's ideal to limit the vocab to only those kanji you've learned up to that point. In many cases, it's better to include the most common words but provide furigana for the unknown kanji.

Speaking of statistics.....

Right now, I'm working on a java program to extract all unique kanji that appear in a file (i.e. each kanji only appears once in the output file).

I ran it on the japanese script for episode 4 of Ohitorisama (Japanese TV drama) and it turns out there are 444 different kanji that appear in episode 4 (lots of 4's :-).

Anyway, with knowledge of 444 kanji, you can read the script for one episode of drama. It's not too bad when you consider that you'll only need to concentrate on the kanji that you fail..... :-)

For the curious, here are all the kanji that appear in episode 4 of Ohitorisama:

一万丈三上下不世両中主之乗了事二井人今介仕付代以件任休会体何使例価係保信候借値健偵備傾働像僕僚償充先入全内円冊出分切刑別刻削前力加助労動勘勝十博原友取受古可合同向君呂呈告味呼員問器回団困在坂基報場増売変夕外多夢大太夫失契女好妙妹始婚嫌子学守安定実室家容宿寄密察対封少局居届屋展山巻布師席帯帰干年幸広店度庫康弁式引張当待後徒従得微心必忘忙応思急性息恵悩悪情想意感慌慣憶成戻手払扱持指挟捜掃探提携支放教文料断新方日早明昨時普暮曜替最月服朝本材村束来板染査校格案梨業極構様機欲止正残段母毎比気水池決沢泊注洗活派浄消淑清済減渡温湯源為然燃父物特犯理生用田由留番疑疲癒発的盗盛監目直相真着知短研確礼社祝神私秋秘移稼究空立第筒節紀約紋紙紛細紹終組経結絞絡給絵絶緒線緩繊罪置美習考者聞職能自臭苦茶荒菊菌落蝶行術衣要見視覚親解触言計記訳証試話詳読課調論謝警議護象財販責貯買貸費購贈起越足跡身車輩込近返迫迷追送逃途通速連週進運道達違遣適部配里重野量金長門開間関限除陳際隠隣集難電音預頑頭頼題額願風食香高麻黒

For comparison, I ran my program on a J-r&b song (散歩道 by Keyco).
51 Kanji in total.

Lyrics:
http://www.utamap.com/showkasi.php?surl=B02000

So one J-pop song a day is very doable.

Kanji used in the song:

人今仕他傷共冗出別午単告呼咲夏女好居届川強彼後恋愛探方日昨時水流海涙然特真眠眺笑終自落行言談軽迷遊隣風

JWPce does that automatically. You get info like this (this list is from Akutagawa's Rashomon):

人　71　ジン／ニン　ひと／―り／―と　person
下　54　カ／ゲ　した／しも／もと／さ(げる)／さ(がる)／くだ(る)／くだ(り)／くだ(す)／―くだ(す)／くだ(さる)／お(ろす)／お(りる)　below, down, descend, give, low, inferior
上　35　ジョウ／ショウ／シャン　うえ／―うえ／うわ―／かみ／あ( げる)／―あ(げる)／あ(がる)／―あ( がる)／あ(がり)／―あ(がり)／のぼ( る)／のぼ(り)／のぼ(せる)／のぼ(す)／よ(す)　above, up
死　32　シ　し(ぬ)／し(に―)　death, die
事　31　ジ／ズ　こと／つか(う)／つか(える)　matter, thing, fact, business, reason, possibly
老　28　ロウ　お(いる)／ふ(ける)　old man, old age, grow old
婆　28　バ　ばば／ばあ　old woman, grandma, wet nurse

(skipping down to the bottom)

学　1　ガク　まな(ぶ)　study, learning, science
新　1　シン　あたら(しい)／あら( た)／あら―／にい―　new
潮　1　チョウ　しお／うしお　tide, salt water, opportunity
社　1　シャ　やしろ　company, firm, office, association, shrine
昭　1　ショウ　shining, bright
和　1　ワ／オ／カ　やわ(らぐ)／やわ(らげる)／なご(む)／なご(やか)　harmony, Japanese style, peace, soften, Japan

You can even make a list so that the program will highlight the kanji you haven't studied yet (or highlight non-Jouyou kanji, or whatever you want).