Hey all, wanted to share what’s been working for me, in case it helps anyone else. There are similar methods, but I haven’t seen anyone doing exactly what I’m doing.
**Background**: I want to be better at reading: it’s something I can do on my own, for as many hours a week as I want, it isn’t embarrasing, and it’s actually really fun.
**Goal**: I want to read ‘extensively’, so I can pick up new words in-context as I go, rather than having to use a dictionary all the time. For that, you need:
* A lot of vocab. It seems like 3000 words is a good starting target if you want to read stuff for teens.
* A bunch of kanji. With the most common 500 or 1000 kanji, you can guess pretty well on words you don’t know, especially in-context.
**My Method**: Focus on the most useful words first, that you’re likely to see a lot, and learn the kanji for those words along with them. In brief, here’s what I’m doing:
* Get a list of the most common 3000-5000 words, in order.
* I used Kanshudo’s ordered list of vocab. I copied down the most common 3000 words.
* Find the kanji frequency *only among those words*, and then order all the kanji by that frequency. Ignore any kanji that aren’t in those words.
* In other words, I ignored any online kanji frequency lists, and basically did something like this:
* Say 同 appears in words #7, #68, #235, and #1004 (ordered by frequency)
Therefore, 同 gets a score of 1/7+1/68+1/235+1/1004 = 0.16. Order the kanji by this number.
* Then split these kanji into lists of high and medium frequency kanji. High frequency are the first (say)100 kanji, and medium are the second (say) 200 kanji. The lists might look like this:
* High: 同通代教同料当体作集…
* Medium: 流感図予察番風説使院頼新類約落性着題知起…
* Each day, learn (say) 3 new kanji (I use Anki flashcards)
* I use the RtK mnemonics, and I use the community stories on Kanji Koohii. It’s a little more work, since I’m not doing things in the RtK order, but the system still works great. I find the radicals for my word, create flashcards any new radical(s) as needed, then learn the word + the radical(s). I typically only add 1 new radical per day.
* Find all vocab that use the new kanji + kanji you already know. I have a little spreadsheet & program to do this.
* I use Kanshudo flashcards to study vocab, because they have lots of extra info, like multiple examples for every word. I have both Anki and Kanshudo set up on my phone, so I can practice before bed.
**Example:**
* Let’s say I already know \~350 kanji
* On Tuesday, my 3 new kanji are: 同 (high), 仕 建 (med)
* These give 10 new vocab to learn: 同じ , 同時 , 同様 , 共同 , 同一 , 仕事 , 仕方 , 仕組み , 建物 , 建つ
So I’m learning 3 new kanji and \~10 new vocab per day in this example. These vocab focus on the kanji I’m learning *and* are all in the 3000 most common words, so I’ll see them in my reading before long. Yay!
**Benefit**: Learning this way immediately reinforces the kanji you’re learning, which is fantastic. And you’re always focusing on the next-most-useful vocab *and* kanji at the same time.
One nice thing is, as you go on, you end up getting around the same number of new vocab per day even when you get to less common kanji. The reason is, as you go down the list, each new kanji will show up in fewer words, but there will be more words that use *all* the kanji you know, so it just seems to work out.
(In other words: 物 shows up in a lot of words, but you won’t get to 建物 until you learn the lower frequency 建. So when you learn 物, you aren’t immediately overwhelmed.)
Hope this method helps someone else!
5 comments
How can you find the kanji frequency among those words?
It’s quite interesting approach and looks quite balanced. I especially liked your idea of using accumulative frequency value.
This method is very interesting!!
Best of luck with studies… 毎日頑張って下さい！
It’s an interesting approach, and it focuses on associative memory, which is an important aspect of long-term language memory. I’d love to see your spreadsheet, as this sounds promising, but I understand just how much work this might be.
It’s somewhat expensive, but Routledge published a Frequency Dictionary that operates on a similar approach, but focuses more on vocabulary than Kanji acquisition as a target. Also, I seem to recall a similar concept as a spreadsheet floating around, but it was more focused around the 5-10 most common combinations per Kanji, rather than your more comprehensive methodology.
Nonetheless, sounds exciting!
That sounds like a very cool learning method.
It also sounded fun to implement, so I wrote a short python program, that rates the kanji. In case anyone wants to come up with their own learning method based on this, the code can be found here [https://github.com/RRasche/KanjiSort](https://github.com/RRasche/KanjiSort)
I used a different rating function, since I was not sure, if it’d be okay to use yours.