when I add cards, that first review to actually put them into the decks is easy enough. however, when it comes to the 2nd review, I had a real wake up call today - I got only 55% correct. Is there a fundamental flaw in my stories, or is it normal to have such a low percentage of remembered kanji on the 2nd review?

It sounds like you are just starting. This is a long term process where you will likely forget many kanji many times before you get them. It definitely helps to get strong stories. But more importantly, don't expect to memorize them perfectly at first. When you fail one or many, just go back later and review them, retest and pass or fail again, then keep moving. Don't bite off more than you can chew, but also don't try to get 100% before you add more cards. I hope this makes sense.

That's perfectly normal, we've all been there. Keep adding and reviewing, being honest with yourself when passing/failing cards, and you'll be fine.

The pitfall of digital SRS flash cards is not spending enough time on learning. I don't know if you've fallen into that pitfall, but if you find you're just adding cards to your deck with not much more than a glance at them, then it'd be worth fixing your study habits before you get too far.

I find myself spending at least five minutes per kanji, either writing and revising my own story or trying to set a concrete image into my brain with someone else's story. It's not enough to just go to the kanji section on this site, read the top story, realize that it kinda makes sense (lol, you advance your turkeys on the road, huh... brilliant), and then move on. You've really got to make sure you are using a story that you can visualize all the primitives acting as characters and the keyword being obvious from the connotations of the plot because when you encounter these kanji in the wild, all you'll have is what you remember about the primitives.

For what percentage I remember on the second review, that can vary. Generally I get about 70-80% correct on the first review (day after I learn them), but sometimes it's as low as 60%, and about 90-95% on the second review (usually the following day), but there are cards in my active deck that I have been screwing with for weeks because I can't pass them with confidence. Of course, I've gotten the hang of this by now--that took me a long time to do. If you're just starting out, 55% on the second review is not something to be ashamed of. Just keep reviewing and you'll get the hang of it too. All it takes is patience and persistence.

I'm similar, probably something like 50-80%, sometimes worse sometimes better. It varies a lot. Remember you're aiming for long term retention, so you should try and keep the spacing reasonably long, even if it does result in failures at the beginning. Be honest with your SRS. If you keep at it this way you'll find it getting easier as the intervals get longer.

Personally, now that I've got the hang of it I spend about 1-2 minutes per card for the prewritten ones. I find this more time-efficient than spending 5 minutes per card, when you consider how much time each extra review takes due to a failed card (about 30 seconds).

jajaaan Wrote:The pitfall of digital SRS flash cards is not spending enough time on learning. I don't know if you've fallen into that pitfall, but if you find you're just adding cards to your deck with not much more than a glance at them, then it'd be worth fixing your study habits before you get too far.

...

For what percentage I remember on the second review, that can vary. Generally I get about 70-80% correct on the first review (day after I learn them), but sometimes it's as low as 60%, and about 90-95% on the second review (usually the following day), but there are cards in my active deck that I have been screwing with for weeks because I can't pass them with confidence. Of course, I've gotten the hang of this by now--that took me a long time to do. If you're just starting out, 55% on the second review is not something to be ashamed of. Just keep reviewing and you'll get the hang of it too. All it takes is patience and persistence.

This is probably one of the most useful pieces of advice I've seen on learning Japanese in while.

I fallen into this pitfall many times, and keep on getting myself out by either deleting cards (not a strong enough memory), or reviewing ineffectively. When I took the JLPT today, the cards that I spent time learning properly stuck like glue, and the others just barely stuck.

That said, I'm going to heed your advice jajaaan and work more on learning. Any advice on when you think you've "learned" something?

It can be hard to tell, which is why I usually try to err on the side of spending too much time on each kanji. One test is if, after you've finished with a set of new cards, you can go back through them and be able to recognize them right away, but I've found that even when I don't recognize half the cards at this stage, that doesn't mean I won't remember it just fine even as early as the next review.

Another piece of advice is that if you're using this website as your primary resource and have never read the physical RTK book, I'd recommend doing that. True... the SRS section on this website has a lot of good stories, but the book is loaded with advice on how to set these stories into your memory and there's lots of examples of stories that show you how to do things like order the primitives, remember superfluous dots. Eventually it shows you what not to do when writing your own stories.

I remember Heisig's advice for the kanji for "pond," one of the first ones encountered in the section where you start writing your own stories. The elements are "water" and "scorpion." Don't, he says, just picture a scorpion sitting contentedly beside a pond. Instead, try to picture the animal letting its venom until it's filled up an entire pond full of venom. To me, this means that the difference between a good story and a bad story isn't simply one of merit, it's whether you can take yourself, as it were, on the journey required to set that story into your memory and have a meaningful experience there.

If you're failing more than you'd like, you might want to try using Anki instead of this site. With Anki you can learn something from nothing via brute force by repeatedly failing and re-reviewing it over a short space of time (not how you're supposed to use it, but it works), whereas RevtK is more suited to people who follow Heisig's advice to the letter (spending five minutes with each kanji, visualising all the elements in your mind, etc).

I think what jajaaan wrote on this is spot on. I average 10 minutes on each new kanji - time includes the initial review when entering the cards for the 1st time. I get similar review percentages to jajaaan - normally 70~90%. Sometimes I get more but there are occasions when I get below 60%, especially if I'm tired.

In the absence of a good Heisig story I spend a few minutes trying to think of one of my own. If that fails there is almost always a good story in the srs section, or at least some ideas to inspire.

Bear in mind that, if you are inputting new kanji daily, many of your review kanji will be recently learned ones which you are more likely to fail. On occasions, when I've been too busy to learn new kanji for a few weeks, and only had time to review the expired ones, my pass rate has gradually risen.

I find it useful to scrutinise the review summary as it highlights problem kanji. Ones that have been failed many times indicate a problem. I note these down and then review those stories and either tweak or change them. I like to give a story a fair "crack of the whip" especially at the beginning. It can take a few reviews for a story to settle in the mind. I sometimes experience an odd phenomenon when I try a completely new story: Having learnt a new story, I discover I can suddenly now remember the old story which I couldn't previously remember - but then cannot recall the new story! It's as if the old story, afraid of being discarded, suddenly starts to perform on cue. So, it can be difficult to forget an old story which isn't quite working. For this reason, as javaaan said, I think it's wise to spend time on your story initially otherwise you may have the problem of interference between stories if you have to change them.

Finally, I think it's better not to try and overdo it. I've reached about 1500 kanji and input about 20 new ones per day at present. This represents a few hours work for me but, on top of that, there are daily reviews of maybe 90 kanji which eats into my time. On odd days when I learn a lot of new ones (max 50) I find it better to "learn > input >initial review" in small batches of say 15, rather than all 50 in one go. I think this is due to the "primacy and recency effect" - if you try to learn too much in one go, you find you can recall the first and last things but forget stuff in the middle.

Some people seem to be able to spend just a few minutes per kanji. Perhaps I'm at the other extreme. Each of us is different but I guess somewhere in the middle is about right.

Quote:I sometimes experience an odd phenomenon when I try a completely new story: Having learnt a new story, I discover I can suddenly now remember the old story which I couldn't previously remember - but then cannot recall the new story!

Happens to me all the time. Frustrating and rewarding at the same time. I've never thought about it before, but I wonder if a trick like that could be used under controlled circumstances. Like say there's some kanji you've been having trouble remembering... to deliberately change the story in order to somehow refresh the old story. Maybe I'm thinking too far into this, though...

jajaaan Wrote:The pitfall of digital SRS flash cards is not spending enough time on learning. I don't know if you've fallen into that pitfall, but if you find you're just adding cards to your deck with not much more than a glance at them, then it'd be worth fixing your study habits before you get too far.

Personally, I've found this to be the advantage of SRS, not a pitfall. I spend about a minute per kanji learning the first time, and my retention rate is around 50% when they come up for review 1-3 days later. I spend an extra minute with the half I get wrong, and then get half of those right on the second try. I generally spend some time on the remaining 25% thinking up a good story or finding one on this site, which takes a few minutes. On average that works out to be a little over 2 minutes per kanji. No reason to spend 5 minutes on a kanji that's going to stick with me easily, but of course I don't know which ones those are and which ones I'm going to have a hard time remembering in advance.

I only add about 10 per day so this may be disastrous for people that add 20+, but it's been working for me so far. I'm closing in on 500 kanji, and looking at a quick sort of my cards, I have 0 fails on about 45% of them, 1 fail on 25%, and 2+ fails on 30%. My goal has been to keep my "one review" pile at 60. If I do poorly reviewing on a given day, I add few kanji; if I do well, I may be able to add 15-20.

Maybe this is just me, but i keep a log of all my kanji with a brief note that is enough to recall the story, and before I enter anything onto the website I cover up the kanji with my rubber and review everything recent, not in order obviosly, my page layout allows me to review everything in a different order. This is where I identify the good and bad stories. I often fail kanji due to keywords ambiguity, I click fail but it's not entirely true. I got greasemonkey and tweaked many keywords, like adding a second meaning/kunyomi. I think the best thing you can do is find what works for you, so spend time trying to find out why stories didn't stick and why others did, I think giving each primitive a single solid reliable meaning is very important, sometimes I would review just primitives, but I suspect that's a consequence of my rushing.

jajaaan Wrote:I remember Heisig's advice for the kanji for "pond," one of the first ones encountered in the section where you start writing your own stories. The elements are "water" and "scorpion." Don't, he says, just picture a scorpion sitting contentedly beside a pond. Instead, try to picture the animal letting its venom until it's filled up an entire pond full of venom.

So Soooooo true, don't just make a 2D picture, put the elements together that lead directly to the meaning. Try looking at your failed kanji and see if you feel this is the case.

Otherwise if the story is good, and lets you write the kanji and see its meaning clearly, then the problem might just be associating the keyword, which again to me doesn't feel like a true fail, just a tad of focusing should fix that, close your eyes and let your brain do it.

Monkeys_Spiders Wrote:I often fail kanji due to keywords ambiguity, I click fail but it's not entirely true.

I agree. I typically fail a small percentage due to keyword overlap. For example "condolences" and "consolation" or "rot" and "decay". There are many like this and I seemed to encounter them with increasing frequency one I got over the 1000 frame mark. Many of these kanji have similar meanings so Heisig has to inevitably choose English keywords with similar meanings. When I have a problem like this I often add a note to my story e.g. With 'Consolation story' I will put "Avoid confusion with condolences {1233}". But I don't display the character 弔. This allows me to test myself on the spot. As you learn more and more kanji you are likely to encounter this interference between similar characters more and more. Another example (this one is with sound rather than meaning but principle is the same) would be:
I learn 海 うみ sea - easy, no problem.
Then I learn 馬 うま horse... okay
and later I learn 梅 うめ Japanese apricot, umm...

The more I learn, the more confused? Well no, with practice (or reviews) these ambiguities fortunately fall away.

I sometimes have a quick check in my Kodansha Kanji Learner's Dictionary (KLD) for the range of meanings when I'm learning a new keyword. On occasions, when the keyword has more than one English meaning, Heisig doesn't always make it clear which one to use. A recent example for me was "defer" where no further explanation was given. Well, in English defer can mean (1) postpone or (2) submit or yield (to someone else's opinion). The actual meaning of this kanji is close to (2), not (1). Having a deeper understanding of the meaning helps me tailor my story (or choose some else's) which is closer to the real meaning. I have started to put the KLD ref. number alongside the Heisig number in my RTK1 book to allow me to quickly cross reference.

I imagine Heisig took great care selecting appropriate keywords and I'm fine with most of them but I can see the benefit of having the option of using custom keywords.

gavmck Wrote:I sometimes have a quick check in my Kodansha Kanji Learner's Dictionary (KLD) for the range of meanings when I'm learning a new keyword. On occasions, when the keyword has more than one English meaning, Heisig doesn't always make it clear which one to use. A recent example for me was "defer" where no further explanation was given. Well, in English defer can mean (1) postpone or (2) submit or yield (to someone else's opinion). The actual meaning of this kanji is close to (2), not (1). Having a deeper understanding of the meaning helps me tailor my story (or choose some else's) which is closer to the real meaning. I have started to put the KLD ref. number alongside the Heisig number in my RTK1 book to allow me to quickly cross reference.

I imagine Heisig took great care selecting appropriate keywords and I'm fine with most of them but I can see the benefit of having the option of using custom keywords.

I used to do this too, but then I realized that it doesn't matter if I get the connotations right or not. If I can think of 屈 as yield, whether it's yielding to traffic, to a higher authority, or as in the yield of crops, all that matters is I remember the keyword "yield" correctly and don't think of a synonym when I see the kanji. Ultimately, I'm going to be changing over to a Japanese keyword anyway and "re-learning," as it were, the meaning of the kanji as I learn various compounds. So it seems to me that there's no way around this initial, albeit inefficient, step unless you use a Japanese keyword from the get-go. The only reason I can think of to remember the English keyword in the proper sense is if you don't plan on going on to learn the readings and only plan to decipher Japanese writing by looking at the meanings of the kanji.

That said, I'm not knocking your method. I also use a kanji dictionary to occasionally change a keyword when I come upon a Heisig keyword that, for one reason or another, sucks. I just don't worry about getting the proper sense of an English keyword all the time, especially when a good story using an improper sense of a keyword suggests itself to me.