I understand the Leitner system as reviewing things at increasing intervals. 1, 3, 7, 14, days, etc. Am I subverting the implementation and effect of this method by studying kanji more frequently than those intervals? In other words, am I defeating the purpose when I don't wait the allotted period of time? I like to be productive and study my flashcards on the train, walking in town, on the pot, etc. But should I be waiting to study those cards until the "correct" day?

Arguments I envision:

By studying kanji more frequently than planned in the Leitner system:

Con:
I solidify the kanji in my short term visual memory but do not allow them to form in my long term memory thus ensuring I will forget them if i DO wait a while to review them.

Pro:
Reviewing the kanji more frequently can only HELP remember them. Practice makes perfect.

Any official word from Leitner? Your own thoughts?

I don't think reviewing the same cards every day helps very much. It's the painful task of pulling it back up after 3 or 7 or 14 days of having the discipline *not* to look at it, that really trains the brain, in my opinion.

If I were you, I'd spend my time on the pot learning memorizing new kanji rather than reviewing, if you've run out of cards to review.

Try it and see how it works -- there are many systems out there, and you can design your own, even within constraints of this site, by choosing how many cards to add, and when and which boxes to review. One thing that I can say for sure is that for different people, different things work better.

Reviewing more often does not make anything worse, it just takes more time. The whole point of spaced repetition is not to take so much time while still producing good results. Reviewing too often just gets you inefficient items memorized / time used ratio. It's better to study more new items and let the flashcard system ensure you remember them. If you want additional practice, I suggest reading something in Japanese. (with furigana, if you need it)

[ファブリス]

Quote:1. Intervals should be as long as possible to obtain the minimum frequency of repetitions, and to make the best use of the so-called spacing effect3, which says that longer inter-repetition intervals, up to a certain limit, produce stronger memories
2. Intervals should be short enough to ensure that the knowledge is still remembered

source

I don't know if Mr. Piotr Wozniak, author of SuperMemo based his research on that of Leitner. I haven't read Sebastian Leitner's book on "How to learn to learn" either, but I think they both make use of the same principle of spaced repetition. Programs like SuperMemo, Mnemosyne or this website's flashcard system are the modern, computer assisted equivalents of Leitner's box. Admittedly, software like SuperMemo have refined and optimized the spaced repetition algorithms much further than what this website's flashcard system does; but I believe this website's Leitner box implementation is well enough efficient for the purpose of learning RTK1, paired with its simplicity of use.

I think you shouldn't review ahead of time. Especially if you doubt your own recall ability. Also if you wait long enough, you will be able to separate the good stories/mnemonics from the less effective ones earlier. Review less but take your time to review, try to recall as much as you can, even if it takes several minutes to do it. It's happened to me to take several minutes to trace my way back to a complete character. That does not necessarily mean the story was bad, if I could retrace the character completely, then the story worked.

From my experience, it took at least 4 or 5 reviews until the characters came to mind easily anyway. The last ones studied always take a while to recall so even today those in the range #1500-#2042 takes longer to recall, as are those that are seldom seen in books and other reading material.

For the past 5 months or so I have been reviewing very seldom, you may have noticed that "fuaburisu" is not a regular on the "members reviewing today" list on the homepage. And yet since completing RTK1 around January 2006 I still have "only" 276 cards in my failed stack, 200 of which were already there when completing RTK1, which I have yet to clear up. Currently I have 1372 expired cards in the last stack

So I think that approach worked well. Review less often but apply the same effort and concentration to reviews as you did when studying the characters.

I'm not sure. It might be true, but it is also a matter of where that 'certain limit' is, and it will vary from person to person. I think the review system should produce the intervals short enough so that one does not have to think much before answering -- I generally noticed that it never pays off, and if I need to think about the answer for more than a few seconds it's best to just fail the card, as I would have to think more still next time if I don't. Twinkle has a nice 'drill' feature to help with such cards (as well as new cards), probably supermemo/mnemosyne/etc do too.
I guess if someone finds himself in that position with many cards then the default intervals might be just too long for him.

If you're going to review more, then I think it's better to make new flashcards that are related but not the same and review those. E.g. from kanji to reading and from full words in hiragana to full words in kanji (e.g. 〈会〉釈 --> え and えしゃく --> 会釈). Memory works better the more associations you have, so adding more of them instead of just strengthening one is more fruitful.

Laxxy wrote:
if I need to think about the answer for more than a few seconds it's best to just fail the card, as I would have to think more still next time if I don't.

I've been working on the opposite theory -- if it's relatively new, and I think of it immediately, it means I'm probably using visual memory and I expect I'll probably have trouble with it next time around. But if it takes a long time and I finally manage to pull it up through Herculean effort, then I feel like I've done some real learning, and it should go much quicker the next time. I can't say as I've checked to see if that's really true, though; it just my intuition about it.

Maybe this is an individual difference in learning styles? I don't know. But I don't insist on a quick recall to pass a card, and I'm making steady progress despite that, I think.

-chris

I have the opposite problem from Fuaburisu....I finished a while ago but I'm still pretty active because I let over a thousand kanji accumulate in the last pile while I was learning the last few hundred....so some of the cards I feel like I'm learning for the first time....though not completely....I'm using flashcardexchange.com to practice the onyumi readings of my flashcards from the the "blue box" of Heisig cards I got from Amazon....with their membership option ($20 for a lifetime) you can set up cardfiles....it works differently from here....you set up the flashcards in sets (usually 20 is a good enough pile) ....when you have successfully reviewed every card in your pack 5 times in a row...you are sent an email to review again in 4 days....then 7....etc...up to 11 years...I was using it to review just regular kanji but I realized that this website was enough....and I really need to get going on studying Japanese vocabulary and actually learning to communicate...look forward to hearing more of what people have discovered and used.

cbogart Wrote:Laxxy wrote:
if I need to think about the answer for more than a few seconds it's best to just fail the card, as I would have to think more still next time if I don't.

I've been working on the opposite theory -- if it's relatively new, and I think of it immediately, it means I'm probably using visual memory and I expect I'll probably have trouble with it next time around.

I think for me it works the way I described, or at least that was quite firmly my impression so far. I think this is because those stories were not good (but when I failed a card, I'd review it some more and story would get polished in the process).

As about the visual memory -- ex post i think I didn't use it enough. It is a great and powerful tool, it's just that its' capacity is limited in the sense that you can only use it to memorize several items per day, and some of those have to be primitives. But the speed of both memorization and recollection is much higher than with the associative story based memory. I don't think I've ever made any mistakes recalling any of Heisig's named primitives, ever, for example. I now think I should have just used it to memorize a few of the harder kanji, and I definitely wish Heisig used a few more primitives.

laxxy wrote:
As about the visual memory -- ex post i think I didn't use it enough. It is a great
and powerful tool, it's just that its' capacity is limited in the sense that you can only use it to memorize several items per day, and some of those have to be primitives.


Interesting; I've been kind of dogmatic about going for the story-memory and not paying much attention to the visual memory, figuring that would come later. But you're right that visual is faster -- for a few it comes fast and unbidden and my stories feel like a pointless recital after the fact. I also think the visual memory works better for reverse recall from kanji to keyword.

The nice thing about the stories has been that when I have a vague visual memory, I can fill in the details. Like "line" -- it's an old friend from trying to navigate through train stations in Japan, but I always saw it as an impenetrable thicket of strokes supporting "white" in the upper right. But the story lets me write it.

Anyway, I'm going slowly enough that I may be able to use visual memory more than I have been -- I'll give it a try this week and see if my recall is better.

[ファブリス]

I think somewhere in RTK1 Heisig recommends to recall the story and images as you write down the kanji during reviews, even if the kanji comes quickly to mind. Especially, if it's an early review. That way you know you're not relying on short term/visual memory alone (not to confuse with imaginative memory).

My feeling has been that when I spend time retracing a kanji' story, primitives and finally the character, however long it took, it "refreshes" the path in memory to that character, so it is fruitful.

Admittedly, I've found that after 3-4 reviews (one month delay and over), if the characters still take a while to recall, I'll be better off making a better story. I have the problem with "funnel" for example. For some primitives like that my stories are just too convoluted and I hit the same memory "bottleneck" each time I come accross a character including this primitive. So even though I could recall them after 1 month or more, they ended up in the failed stack.

ファブリス Wrote:For the past 5 months or so I have been reviewing very seldom, you may have noticed that "fuaburisu" is not a regular on the "members reviewing today" list on the homepage. And yet since completing RTK1 around January 2006 I still have "only" 276 cards in my failed stack, 200 of which were already there when completing RTK1, which I have yet to clear up. Currently I have 1372 expired cards in the last stack

So I think that approach worked well. Review less often but apply the same effort and concentration to reviews as you did when studying the characters.

Having 1372 expired cards seems like a sign that the system isn't working optimally for you. A while back you mentioned that you were going to increase the maximum review spacing beyond 60 days. That would help the problem somewhat because it would keep your backlog from filling up as rapidly with well-known kanji, so you could devote more of your limited review time to less established kanji. Is this change still in "the cards"?

Thanks for the responses! Here's some of my thoughts in response:

TIME isn't an issue. I have plenty of it to fill which is why I'm reviewing my cards then. 15 mins on the train; 5 mins walking somewhere; 5 mins waiting for a friend, etc. I review quite quickly also. ~7 kanji per minute, so a 20min review will cover ~130 kanji.

THIS SITE is the only place I learn new kanji (in conjunction with the book). This, for me, is the most productive method. I have a vivid and active imagination. But I can't think of anything better than picking and choosing from everyone's brains. Improvement is the goal of cooperation. (this site embodies that whole 'Web 2.0' concept). I take the concepts that best trigger my imagination, embellish them in my own way, and make them my own. So, since I only use this site, carrying the book with me is pointless since I won't be learning new kanji away from the internet.

VISUAL MEMORY is an underutilized tool, as laxxy wrote. Why NOT use it? My understanding of Heisig's theory is that 2000 random lines is too much to stuff in visual memory, so we need a systematic new method to recall kanji. But that does not it visual memory isn't a splendid tool when utilized sparingly! I too find myself using visual memory in situations where stories are just too convoluted. And as another thread is currently discussing, I wish I would have known this before. Heisig could easily have added a few more primitives. And once we know that certain primitive combinations will occur frequently, it's too late to go back and assign new meanings to them.

I still rely on mental imagery as I write every kanji though. This is key! I often make simple mistakes if I rely only on visual memory, such as switching "muzzle" 咅 and "sound" 音.

I'M FINDING that my best path to learning is to review a new kanji several times over the first few days to solidy a clear story. After those first few quick reps, I don't have trouble recalling it later. I find the enforced delay (1 then 3 days) invites me to forget. I will review a new kanji, along with ~50 others I learn each day, 5 times within the first 2 days. Those I haven't done this for are extremely difficult for me to retrieve.

dingomick Wrote:Thanks for the responses! Here's some of my thoughts in response:

TIME isn't an issue. I have plenty of it to fill which is why I'm reviewing my cards then. 15 mins on the train; 5 mins walking somewhere; 5 mins waiting for a friend, etc. I review quite quickly also. ~7 kanji per minute, so a 20min review will cover ~130 kanji.

THIS SITE is the only place I learn new kanji (in conjunction with the book). This, for me, is the most productive method. I have a vivid and active imagination. But I can't think of anything better than picking and choosing from everyone's brains. Improvement is the goal of cooperation. (this site embodies that whole 'Web 2.0' concept). I take the concepts that best trigger my imagination, embellish them in my own way, and make them my own. So, since I only use this site, carrying the book with me is pointless since I won't be learning new kanji away from the internet.

VISUAL MEMORY is an underutilized tool, as laxxy wrote. Why NOT use it? My understanding of Heisig's theory is that 2000 random lines is too much to stuff in visual memory, so we need a systematic new method to recall kanji. But that does not it visual memory isn't a splendid tool when utilized sparingly! I too find myself using visual memory in situations where stories are just too convoluted. And as another thread is currently discussing, I wish I would have known this before. Heisig could easily have added a few more primitives. And once we know that certain primitive combinations will occur frequently, it's too late to go back and assign new meanings to them.

I still rely on mental imagery as I write every kanji though. This is key! I often make simple mistakes if I rely only on visual memory, such as switching "muzzle" 咅 and "sound" 音.

I'M FINDING that my best path to learning is to review a new kanji several times over the first few days to solidy a clear story. After those first few quick reps, I don't have trouble recalling it later. I find the enforced delay (1 then 3 days) invites me to forget. I will review a new kanji, along with ~50 others I learn each day, 5 times within the first 2 days. Those I haven't done this for are extremely difficult for me to retrieve.

I absolutely agree on all counts. My experience and my impressions were quite similar indeed. Particularly about the points about a dense initial review schedule and a desire for more primitives. Maybe this site could help other people with that a bit, it did help me when people described new primitives they invented when they first appear, and I think I mentioned a couple of my own too.