Hey everyone,

At the moment, I'm learning via the dodgey old rote methods, however, I'm using SRS and learning the kanji in the order Heisig has them in the book. What I do, is I pick out 5 kanji in a row, and then do the RevTK flashcards for them repeatedly till I have good recall on them, wash, rince, repeat. So far, I've been doing about ~80 a day (I have lots of time on my hand) and I just hit 200. Will I run into problems if I don't create/use the sentences to remember the Kanji, and instead, just remember what they mean/concepts and ideas implied, and how to write them?

I would think so, but since this is working out quite well for you, perhaps you could give it a try for a week or two and see how it works out?

In my experience, and in the experience of everyone I've helped start learning kanji, "over-learning" the kanji before putting them into SRS wastes time that could be better used learning new kanji, and actually hinders the building of long term memory. The method I use is just to look at the kanji & the primitive list, make up a sentence or mental image, and then move on to the next. I add them to RevTK and review them the next day. No writing, no drilling. It works beautifully.

Although your case is kind of strange in that despite using an SRS (RevTK) you're not using it like an SRS since you repeatedly re-review the cards when you add them, putting them into higher stacks. If you really want to keep at it this way you'd probably be better off with a traditional flashcard program (or real flashcards) since the SRS stuff would only get in your way. I really do recommend using an SRS properly though.

Also, while you don't really need to make well fleshed out stories (at least for easier characters) to remember the kanji, you should at least have a basic framework. You might be fine with the early characters, but once you add more and more characters that get more and more complex, you WILL run into issues. How will you avoid confusing something like:
(wisteria) 藤 flower flesh quarter rice grains
with
(victory) 勝 flesh quarter power
? Many characters resemble each other, differing by only one or two primitives.

You have to link the keyword in with the primitives in the form of stories, sentences, or mental images, otherwise your memory will just reach critical mass and everything will become a jumble. The failure to do this is probably why most people end up getting stuck about halfway through RTK (600-800字 area seems most common).

Ahh, I'm starting to see what you mean.

Now that I'm at the 200 mark, it takes forever to do my daily reviews. However, the SRS seems to be working well, my recall rate is about 85%.

**edit** I just finished reviewing all 200 that I've done, and my recall rate is actually 94% :O!

I do have a bit of a basic framework when I learn them, for example, when I learn bribe, I remember it as Fish possesion and anxiety as fire page etc; so I do have a bit of framework, but I think I'm just using the primitives as a mnemonic.

I find that as I create the stories, I really only actually go through the story maybe the first time or two that I review the cards. After that I can kinda just remember the primitives without using the story, and eventually I just kinda "know" the kanji.

I'm not really sure that a story can particularly help you remember the specific primitives a whole lot better than just, well, remembering the primitives. The reason for this is because even though you may know the general idea of your story, unless you know the specific words and phrases that you used to make it up, the story doesn't help a lot, particularly when you are dealing with keywords that can have similar meanings.

There are some cases where the story does help to initially cement the kanji into my head though. I've got a lot of rather risque or vulgar stories, which seem to be good for remembering some kanji :p

Zarxrax Wrote:I find that as I create the stories, I really only actually go through the story maybe the first time or two that I review the cards. After that I can kinda just remember the primitives without using the story, and eventually I just kinda "know" the kanji.

That's kind of the point of the Heisig system. I believe he talks about it in one of the chapters infact. You use the stories at first but they eventually fall to the waysides and you just know the kanji. I also just use the story the first time, later reviews I might just recall the primitives, and then finally I just write the kanji without thinking anything or even remembering what my story was. Sometimes I do have to pull a story out of the cobwebs though to remember one of the parts of a kanji, or to differentiate some kanji that are similar in meaning or in appearance. THATS why you should be making stories. Forgetting one of the primitives in a kanji is something that many Japanese people have a problem with because they don't have stories for their kanji. They just learned them as a pile of primitives. There is a DS software just to help people with that (DS 丸書いて覚える I think it's called).

Duranix: That can work some of the time, especially early on; but when you get into kanji with more and more primitives, or that only differ from other kanji by one or two primitives, you will likely run into trouble. My example from my earlier post stands. What is stopping you from confusing flowers flesh quarter rice grains from flesh quarter power, or what the keywords for them are?

Jarvik7 Wrote:Although your case is kind of strange in that despite using an SRS (RevTK) you're not using it like an SRS since you repeatedly re-review the cards when you add them, putting them into higher stacks. If you really want to keep at it this way you'd probably be better off with a traditional flashcard program (or real flashcards) since the SRS stuff would only get in your way. I really do recommend using an SRS properly though.

Actually, I think Duranix may be using an SRS properly.

Unless my understanding of an SRS is wrong, there is nothing that says someone cannot repeatedly review new info the first day they learn it.

They can enter the new info (in our case, kanji) into an SRS and repeatedly review it that first day, they simply don't mark it as passed. Or they can mark this new info as passed (and move it up the stacks) as long as at the end of the day they use the fabulous "Manage Cards" on RevTK to delete that day's learned cards and then add those cards back into the system fresh for the first "proper" review the following day. (IMHO, the first proper SRS review of newly learned material should always be the day after you learned them.)

When I was going through RTK I found that it really helped to go through the new kanji several times the day I learned them. I would study them in the morning. And then I would review the material several times during the day. (Either using an SRS that I would reset to zero for those cards at the end of the day or a paper list that I could carry around with me if I wasn't going to be near my computer.) At first I would just go through my stories, really trying to let the story play out in my mind and write the kanji 1x every time I went through the story. Then I would test myself on these kanji. Again, seeing if I could recall the story as well as the kanji. Then I would do my first "proper" SRS review the following morning. And I would stick with the SRS schedule for those cards from that time onward.

Doing this may have slowed me down. I usually only leaned 20-25 kanji per day. (Although for the last 600 kanji I did 50-90 kanji per day.) But it certainly did not hinder my long-term retention.

After RTK 1 I started in on RTK 3 and did the first 300 or so of these but I eventually burned out and did nothing -- no reviews at all -- for over six months. When I went back to reviewing I still had an exceptionally high pass rate. (98% for the first 500 kanji. And between 80-95% for the remaining kanji.) In fact, the later kanji that I did the best with were the ones that I had taken the time to review through several times the day I learned them.

IMHO, an SRS is all about helping you to move newly learned info into long-term memory the most efficient way possible. It does not dictate how you should go about the initial learning.

Some folks find they like to take the extra time at the initial learning stage to go though things carefully (and repeatedly) often with the result that they have high retention rates throughout the SRS stacks, other folks like to speed through using the quick and dirty method of letting the SRS take care of learning the material usually with the result of initial low retention rates that gradually buildup on repeated exposure. However, I believe these people may have to go back and review/revise stories more later on. It is all about what learning style suits you and how quickly you want to go through RTK.

I am very interested to find out how Duranix's method works for him/her. It will be an interesting test to see if someone can successfully go through the entire 2042 kanji of RTK using its ordering principle but without using full stories.

I, too, feel that the sheer number and close resemblance factor will cause a large burden from about kanji #600-800 or so. Even using the Heisig method I hit a wall around kanji #1200, where it seemed difficult to make stories for those kanji. I think it will be even more difficult trying to brute-force primitives as mnemonics. But who knows? Perhaps the order, a mnemonic use of Heisig's primitives, and an SRS will be sufficient.

Duranix, please let us know how it goes for you.

The stories and keywords for the signal primitives are what make the system work so well for me. Yeah, there are some patches that are rough and will slow you down, but for the most part, I find it much easier to remember them all that way.

The whole idea of Heisig's method is to create as many ways to remember how to write and recognize them as possible. Just trying to memorize them as pictographic patterns puts you at a disadvantage.

Being able to assign any kind of meaning to primitives also makes it much easier to dissect new kanji in the wild. If I already know how to label the constituent bits, I can recreate the new kanji pretty easily.

Well, a quick update, I find that adding new primitives is the really hard bit. Once I learn a primitive and I can use it in a few words, it becomes really easy for me to remember new words using that primitive.

I'm currently pushing my way forward at #230, progress is slower than when I first started, however, its still not THAT slow, I'm managing around 50-60 Kanji a day as opposed to the 80 when I started. I suppose that number will slow down, but I can't seem to learn using Heisig's sentence method. His order of Kanji, however, is pure genius! It's so much better than the text book way of "Learn the most common ones first", especially when the most common ones have nothing in common (Pun not intended).

So I'm going to go do my 230 Kanji review now, and them I'm going to move on and see how many I can do. I'll keep you guys updated

Do you mean his story method? Don't confuse Heisig with AJATT. 2 different critters there.

Also, do you have the actual book? If you don't, then that would explain why you're having so much trouble with it. You need the book, and you need to read all of his out-of-order chapter introductions to understand how to make it work.

Nope, I mean the order in which he arranges the Kanji for you to learn. For example, In the book that I have for school, the Kanji that you learn are in the order of that they get used the most. But with the order of Kanji in Heisig's book, you learn a kanji, and then all related ones, for example, the first 50 or so are variations using the primitives moon and sun etc.

And yes, I have the book but I'm learning the Kanji using rote memorization with SRS in the order that Heisig has set out in his book I might try AJATT some time soon too. I want to achieve a high level of proficiency before I graduate from school

And I'm not having trouble remembering anything I just wanted to know if there was something incredibly wrong with what I'm doing (rote memorization using SRS)

Also, if you have only been doing this for 4 days (50-60 a day, at #230) you should still remember everything. However, if you were using the SRS, you would now be able to not review the first 50 for almost a week and still remember them...

I didn't HAVE to do a 230 card review, I just left them all on the 4 and above reviews column on revTK, so that whenever I wanted to review a few, I had to review all of them -_-. I Just discovered Anki, and now I only have to do about 10 or so.

Duranix Wrote:I didn't HAVE to do a 230 card review, I just left them all on the 4 and above reviews column on revTK, so that whenever I wanted to review a few, I had to review all of them -_-. I Just discovered Anki, and now I only have to do about 10 or so.

Yikes, it does sound like you aren't using an SRS properly.

As I said in my earlier post, it is OK to "review" newly learned material multiple times the day you first learn it as a part of your learning process, but at the end of the day you need to reset all of those new cards to zero and then test them properly the next day (in order to do your first check of retention). You then need to follow the SRS's review schedule for the cards from that time onward in order to maximize your retention.

This means that you will always have your new batch of cards to test (i.e., the ones you learned the previous day) and then various amounts of other cards as they become due for review at the various stack levels -- thus testing your retention 3, 7, 14, 30, 60 days out, etc.

Perhaps you should read the "Learn More" section of this website as well as Anki's documentation.

I'm still curious to see how using just the Heisig order and an SRS (but no stories) will work out, but you've really got to use the SRS properly -- you need to let it space your reviews out to help you most efficiently retain what you have learned.

Good luck!

Just to emphasize what Chadokoro_K is saying, I think you need to learn why and how a SRS works.

If you are just going to completely review all items, you might as well use a wordlist. It would be easier and you could do it anywhere, anytime.

heh, I wasn't using it properly before

as I said, my review deck is only about ten cards now