I am trying to study kanji, but find that I can retain only about 30% of what I learn. So I might be able to (on a good day) study 30 kanji, but only remember 10. My question is this, should I keep on reviewing until I lean all thirty, or should I just move on, and come back to the unlearned kanji later on.

I find the method of making stories for the kanji great, but I have never really been great at studying, I prefer a hands on approach, and not have my head in a book, but there really isn't any other way is there?

Thanks!

you should work on your stories, they probably aren't as memorable as they should be.

For each Kanji you learn, you might try something like:
Come up with a story
visualize the story
write the kanji once while saying the story to yourself

And maybe instead of adding 30 at a time, try only 5 or 10 at first until you get a feel for it.

It's probably best to clear your failed kanji stack and then add new cards

I think spaced repetition is great, but not everybody learns the same way. Maybe you're more visual/kinesthetic? Maybe if you draw pictures and color them, of the stories that link the keyword to the primatives? Auditory? Maybe if you say them out loud? What if you made tapes/mp3 of the stories, and then used shuffle on your mp3 player?

For me right now, I think the limiting factor is time.....how many reviews can you do every day? So, let's say it takes you 30 minutes to review 30 kanji. You have 30 minutes to study. You get 10 right, so the next day, you've only got 20 failed kanji to review. You could add enough so that you'd have 30 the next day (10). You'd get another different 10 right. But then the next day, 10 are going to expire and need to be reviewed again. Plus the 20 failed kanji. So, that day, you might not add any new ones.

See? Think about how you learn best, and try to play to your strengths. And then also, keep an eye on how much time you can give to it, and don't overbook yourself.

[ファブリス]

Hi kiwizz, this may sound a silly question ... do you have the RtK book or did you read everything in the RtK sample PDF file? (it should be linked somewhere on the Learn More page).

Stories isn't the bulk of the RtK method. THis is secondary. The real "secret" if you will is mnemonic associations to the components, and learning characters as simple combinations of those components. Stories is one way to bind those components. You'd be surprised at how many kanji I currently remember the writing of, with purely images and no stories. It works because the image is the "key" to the components of the characters.

You can do both ways. At one time, I was going for an empty missed stack (study all missed for the day) then adding in new cards depending on how missed I studied. This meant about 10 new cards a day if at all sometimes.

Later, I began always adding new kanji, and incorporating a number of missed items too for studying. Basicly, I figured if I just sweated the missed stack, I was adding kanji too slowly. If you only added kanji, your missed stack grew too much. So it's a balance of the two especially in the last half of the book. Plus, as you add Kanji, you're gonna get easy to remember kanji.

Here's my exact method: I'd study a chapter a day (add stories, look at story, write kanji) upto 30 cards, which ever was less. Then I'd study 20 to 30 missed cards (to get the missed stack to a round number). By the time I "finished" RTK (about 2 weeks ago?), my missed stack was over 300. Now it's down to 50 with 100 reviews a day for now.

For me, it was important to constantly add Kanji, otherwise it felt I was not making progress. Yeah, I had to go and relearn missed kanji later, but as it kept me motivated to that point now I'm doing clean up. YMMV.

I don't want to discourage you from getting doing better, but I think 30% isn't bad for first time reviews. At risk of putting my foot in my mouth, I think the people getting 85-90% right all the time, are doing so well because they use this site in tandem with something like Anki. Are you only using this site and the book? If so, 30% sounds like what I'd expect for your first review here. If you want to see your numbers shoot up, find an SRS you want to work with.

About reviewing early...on the first stack, if I just know I forgot something then reviewing early helps, for sure! I'm talking about the youngest cards here. If you're scared of moving the ones you do remember ahead too early, you can just click no.

If my image is really fuzzy, I sometimes ask myself. What? How big? What Color? How many? Where? Feeling? Sound? etc. I'm not great at imaging and this helps me get something in my head that sticks.

[ファブリス]

ruisu Wrote:At risk of putting my foot in my mouth, I think the people getting 85-90% right all the time, are doing so well because they use this site in tandem with something like Anki.

Leitner's scheduling works very well, saying that using Leitner instead of SuperMemo's algorithm accounts for such a drastic difference is complete nonsense!

I'm sorry but 30% remembered kanji is bad, no matter what method you use. kiwizz simply needs experience with the method, keep at it, or choose something other that works better for him.

As of today, a simple query on the database shows that the SUCCESS RATE ACCROSS ALL USERS, ALL CARDS is 82% (out of 6.034.801 total reviews).

Everyone's learning style is different and you need to tailor your study to match what works for you. CharleyGarrett makes great suggestions.

My style:
I work out or pick stories in the evening and write them on paper flashcards printed from Polarcloud's PDF file. I usually do them in blocks of 24 (2 pages of the pdf). Manually writing the stories on the cards puts them in my mind in a vague way that'll be refined later. I think that by writing them down before I go to bed, they're in my head being worked on while I sleep.

The next day I study the cards in earnest, focusing on the story and trying to build a clear image. I do most of this on a walk at lunch time, and get a lot of weird looks as I sketch the characters in the air while recalling the image.

If I feel good about the stack (most days I do), I'll test that evening before working on fresh cards. If I feel shaky about this set, I'll just hang onto the stack and study them one more day.

On that first review, fresh off studying, I almost always get 100%. On subsequent reviews that drops off a fair bit, typically to 80-85%, but sometimes lower.

Every once and a while I check the detailed list here on kanji.koohii, sorted in descending fail order looking for kanji that just aren't sticking. I try to rework the story on these to make them more memorable, which usually fixes the problem.

ruisu Wrote:I don't want to discourage you from getting doing better, but I think 30% isn't bad for first time reviews. At risk of putting my foot in my mouth, I think the people getting 85-90% right all the time, are doing so well because they use this site in tandem with something like Anki. Are you only using this site and the book? If so, 30% sounds like what I'd expect for your first review here. If you want to see your numbers shoot up, find an SRS you want to work with.

Don't get me wrong here, I am a huge fan of SRS, and all the benefits it gives, but using an SRS at the same time as this site for reviewing RTK cards doesn't have any benefits as far as I can see. The cards would be out of sync, no? so with that in mind, wouldn't you be seeing cards again before they were scheduled review??

Doesn't that defeat the object a little?

If you're only remembering 30% of the kanji, you may want to consider spending more time on each frame when learning. This may involve you coming back to a frame a couple of hours after you've first seen it. If you don't remember it after a couple of hours, then I'd say you haven't worked hard enough on it. If you can remember it, it's safe to add it to your `learned' stack.

Good luck.
Misha

Christoph Wrote:Doesn't that defeat the object a little?

I think most people dumped this site in favor of Anki or an SRS, neh? At least I did simply because I require the ability to work online and offline. The story reservoir is really the best part about this site, and why I still come here.

You have to remember with any SRS programme/site, that they only help you remember information, not learn it in the first place. Yes you can use something like Anki's 10 minute fail box to grind things into memory, but I would worry with the kanji that you're just going for repetition learning and not creating vivid enough stories.

Personally, when I was studying I created stories in blocks of 10. After thinking about a kanji I would write it down in my book while thinking of the story/image. When I hit my 10 I would then close Heisig and try and write the keywords next to the kanji. This is normally pretty easy as they're still in short term memory, but it's also easier to check the stories going kanji to keyword.
Next I would study my next block of kanji and do the same thing. Now I have a list of 20 kanji written down my page with keywords next to them. So, I cover the kanji with something and try and write them based on just the keyword. If I can't do it it's no stress, just look, think of the story and write it down. I found the key in the early stages was to keep going and not focus/stress about one kanji.
With my page now looking like kanji - keyword - kanji, I study the next block and then write it's keywords. Now turn back to the previous page, cover all but the second kanji column and try and write the keywords. On the second page now cover the kanji and write the kanji from the keywords. Now do another block.
In this way I would do 40-60 kanji in a sitting, each time after learning a new block, going back to the start and adding another column until each page has

kanji - keyword - kanji - keyword - kanji

Finally I'd add them to this site for review the next day. Working in little blocks and reviewing those while learning new blocks also keeps things interesting as your brain is jumping all over the place

[ファブリス]

drwhitman Wrote:On that first review, fresh off studying, I almost always get 100%. On subsequent reviews that drops off a fair bit, typically to 80-85%, but sometimes lower.

It worked the same for me. Now it's almost two years since I finished RtK, I have over 550 expired cards on the site, and I still remember over 70%. At this point it's more like story refining.

ファブリス Wrote:I'm sorry but 30% remembered kanji is bad, no matter what method you use. kiwizz simply needs experience with the method, keep at it, or choose something other that works better for him.

I guess I would need to know more about what he's done with the kanji prior to adding them as new cards, before I can say if 30% is good or not. Reading what some other members do, 80% recall sounds just about right. Like Mighty_Matt's and dwhitman's process sound like they are going to be those 80+%'ers. But compared to me, they have put a lot more effort into each kanji before ever adding them as new cards here. I'm also not so good at coming up with stories off the bat. It could take three or four reviews before a kanji develops a "personality" that I can work with.

Christoph Wrote:Don't get me wrong here, I am a huge fan of SRS, and all the benefits it gives, but using an SRS at the same time as this site for reviewing RTK cards doesn't have any benefits as far as I can see. The cards would be out of sync, no? so with that in mind, wouldn't you be seeing cards again before they were scheduled review??

Doesn't that defeat the object a little?

I agree. I wrongly assumed this is what people were doing since my recall rate into the 80's until the fourth stack. Yet, in other threads it's quite common to see people saying that 80% is their recall rate, and Fabrisu confirms this in his query.

I shouldn't have said 30% is good for a first review, or assumed that people were using Anki & this site, before considering that people just have much better methods for learning new cards than I do.

However! For a literal first review, I mean from closing RTK to the first time test your knowledge, be it on index card or koohii.com, I think that 30% is not something to worry about (in chunks of at least 20 new cards).

edit:

ファブリス Wrote:Leitner's scheduling works very well, saying that using Leitner instead of SuperMemo's algorithm accounts for such a drastic difference is complete nonsense!

I think you misunderstood me, I wasn't really saying that. I just had it in my mind that people were putting new RTK kanji into an SRS, and reviewing them until the SRS scheduled them. Then I assumed they would add them as new cards here, and that's what accounted for the high recalls. I am probably wrong about this. I now think it's just that people have developed very good methods.

But I'm a confused now, doesn't this site use Leitner?

[ファブリス]

ruisu Wrote:But compared to me, they have put a lot more effort into each kanji before ever adding them as new cards here. I'm also not so good at coming up with stories off the bat. It could take three or four reviews before a kanji develops a "personality" that I can work with.

Perfectly normal and I reckon there's a need for more early reviews, it's on my to-do list. It's about time I add a way to review any lesson for example. Other members have posted about your approach before, sometimes during review you get new ideas, new memory associations with the character come effortlessly.

Quote:Then I assumed they would add them as new cards here, and that's what accounted for the high recalls.

You are right, actually. I remember some members explaining just that. Many users of Anki also choose to continue to review kanji on Anki since they've become comfortable with it, others I guess have a preference and may find better results with SuperMemo's algorithm.

I just thought I'd point out, that before looking at the SRS you need to really look at how you study, because the SRS is not a magic pill and won't do the study for you.

Yes this site uses the Leitner "cardbox" system.

SuperMemo's algorithm ("SM8", "SM11", etc.) is the basis for the opensource MnemoSyne project and I thought it was the basis for Anki as well, but I could be wrong.

I've started using Might Matts learning method...so far its working really well. Im getting 100% recall rates in the next morning.

Not so sure how well itll work for long-term reviews though. But so far Ive learned 80 Kanji in about 3 hours with his method and so far so good. I had to learn a bunch of keywords that I barely can remember the meaning of in English though. I tend to do really poorly with keywords that I have to look up in a dictionary before I learn it.

[ファブリス]

Mighty_Matt's approach looks very interesting, I never thought of reviewing both ways while learning the characters. But I also recommend doing the study in chunks throughout one session. Every 10 or so, go back and do quick test.

I wonder if Mighty_Matt's approach could be an improvement over Heisig's basic technique, in that you work links between kanji and keyword both ways, during the learning stage, and this may improve retention (I say that from my experience with kanji chains, as those require going back and forth, doing "chunks", and testing chains in one direction and the other, and this works very well for recall).

Additional benefit is at this stage you may still adapt your story to reinforce the recall on both ways.

I am one of the slowpokes of this site. I resist adding more than 10 new cards and before I do, I complete all my cards due for review. Usually, I only add 1-5 new cards.

This is a slow way of doing it but my recall rate is good. This success in turn keeps me enthusiastic. If my recall rate were only 30 percent, I don't know if I would be able to maintain enthusiasm as easily.

A user in kiwizz' position should reduce the number of new kanji that he or she adds and put more effort into the stories. Johnzep's method is the same as mine and, if I remember aright, is the procedure James Heisig recommends in his introduction.

This is not to detract from the achievements of those users who complete all 2000+ kanji in a few months or a year, but speed can be overemphasized. Regardless of pace, a user will completely learn the kanji and keywords much faster using the component analysis approach and an SRS than by any other method available.

For me I just re-read the stories. Say I have a new stack of 100 characters and I have all the stories in place, I will go through and read them from 0-100 one more time and really let the imagery set in my mind. Then I wait about 2-3 days after re-reading the stories before I do the initial test to make sure I really retained it. I usually score between 50-80% on the first run.

If you're stalling at around 30% ... well, three possibile culprits come immediately to mind. It's possible that A) You're not using Heisig's method correctly -- B) You're trying to go much too quickly, or C) You're not truly locking the kanji into even your short-term memory before hurling the flashcards into the SRS's vault.

I know we all learn "differently," and everything ... but there are elements to the way that the human brain works that are absolutely universal to all of us, and Heisig taps into that (we're using a triple-pronged attack of IMAGERY, VERBAL mnemonics, and LOGICAL arrangement ... that covers all bases, and it was designed that way). Unless you're literally autistic or dyslexic (which it doesn't sound like you are), 80% should be something that's very easy to maintain.

So, why are you getting 30%? As I stated above:


1) You may not really be using Heisig's method, and only THINK you are. Are you just going through the order of the kanji as laid out in the book, without creating your own imagery/stories/mnemonics for them? Are you actually WRITING the kanji that you learn? It's just that ... a 30% recall rate is the sort of thing that might happen if you were just reading the kanji in the order he presented in RTK1, but relying on plain old drill-and-repeat memorization for your recall.

Remember that Heisig's method is to apply MENTAL IMAGERY and repeat a VERBAL STORY to yourself, AS YOU WRITE. If you're not doing all three thrings, you're not using the method. If you're not literally saying to yourself "a small Sun and a huge Moon make a very BRIGHT sky" as you write the "bright" kanji, and literally picturing it in your mind's eye as you scratch it down ... you're not actually using the method. You're using some hybrid of it you came up with, and it sounds like it's not working so well.


2) You're trying to go at the speed of light. Heisig is fast, yes, but if you're only spending 10 seconds per kanji, I doubt anyone but Lt. Commander Data could keep up. Spend a minimum of 1-2 minutes per kanji. It's about what I do, and it's plenty for me. You may need more, and of course ... some abstract kanji will take longer.


3) You're not performing any reviews at all before adding flashcards to the site. Your very first review should NOT be the one scheduled after three days in the first box. You should do at least one review of all the kanji that you learned at the end of a study session ... if you can't remember one of them, then you haven't actually finished yet. The day ends when you recall all the kanji that you allegedly "memorized" within that particular session.

That way, you're assured that you've got it in short term memory. Only then (in my opinion) should you add them to the site's SRS at all. Come back the next day to take them out of the "never reviewed" column, and the spaced repitition truly begins.


I may be somewhat new to Heisig, and granted, I'm still short of the half way mark ... I've got to say, though, that a 30% recall would indicate to me that I did something wrong with initial memorization ... I would be suspicious that I "missed" a step somewhere.

If you're thorough in the beginning, when you're just barely adding the new kanji ... the foundation for long-term recall becomes that much more solid. I've never failed to get above 80%, even on the longest-term reviews. Not once. In fact, the first time I got below 90% (just a few days ago) on some cards in the second box, I was seriously bummed out. I didn't understand how typical of a recall rate that was for the first big "hump" in an SRS program, but now that I see how common it is ... I feel better.

I think the key is to make sure you've got them seriously drilled into SHORT-TERM memory, before adding the flashcards at all. You review for the first time the next day (from stack 0 to stack 1), then three days out, then seven, and so forth ... the site does the scheduling for you.



It's a beautifully effective procedure, and I love it dearly ... you just have to know how to apply it correclty. Don't worry ... you'll get there. It's also entirely possible that your actual problem isn't related to any one of the three issues I've listed above .. I'd bet a lot of money that you're at least dabbling in one of them, however.