I think you're right that there's an optimal rate to work at (or at least, each individual will have an optimal rate).  An 80% pass rate when you're doing 25 kanji/day is probably better than a 95% pass rate at 5 kanji/day.  Of course, I don't really know what that rate is.  :-) 

I'm generally trying to stay above 75%, at least.  I don't mind going a little lower for kanji in the 2nd stack, since that's generally my first or second review of that kanji.  But I shoot for above 90% on kanji in the 3rd or 4th stack.  When my pass rate dropped to ~50% the other day, I figured that was a sign to slow down a little bit.  (I added 145 kanji over the holiday weekend, which was almost certainly too much for me.)

For kanji in the 4th pile ahead, it is surely above 90%. If is not close to that, there is something wrong with your method.

As rich said, try to do the way you think it is the best. I think you shouldnt worry about retention rates before the 3-4th stacks. But that is the way I do. I worry about amount of reviews. I allways try to go the fastest as I can while coping with all my reviews. I firmly believe in clearing the failed pile every day. For as long as I can manage that, crappy recall rates are ok.

Notice that is not an excuse for not putting effort in the stories, as they save precious review time.
I'm aware there are people who can do it with 90% recall rate from the beggining. But mine is way lower. Around 50-70% for the first review.

Last edited by mentat_kgs (2008 July 10, 7:16 pm)

An ideal retention rate for the long-term is probably around 90%. More than that means the scheduling is probably too aggressive and you're doing more work than you really need to. Obviously, your retention will be lower than that for cards you've recently added and 60-70% will be more typical.

When I started using this site, I had 1827 kanji (I used to do Heisig with SuperMemo and I was switching) and I hadn't reviewed kanji in a month, and I drilled myself on all 1827 over a period of a few days. My retention was 85%.

- Kef

Fast and sloppy for the win alyks

I think it's best to make a nice, solid image and move on. Screw how many per day; screw retention rate. It'll sort itself out as long as you made a nice image, story, and mnemonic.

Yeah, but sometimes a keyword kanji relationship doesn't stick, but I already know the kanji in Japanese. I don't bother trying to fix those, because there's no point to it. And then there are keywords like "row" and "rowing" (and many more similar words), and I'm not going to waste brain cells differentiating between the two. The English keywords are important, but only temporarily so.

It should be noted that this implies that many people probably shouldn't be using this site in its current form, as the repetition timing isn't adjustable.

(No, I'm not really encouraging people to leave, and certainly not going to myself, but that seems to be the logical conclusion from CharleyGarrett's post, and I can't come up with a solid counterargument to the base (his post))

~J

woodwojr wrote:

It should be noted that this implies that many people probably shouldn't be using this site in its current form, as the repetition timing isn't adjustable.

I think the convenience of using this site outweighs any such disadvantages, even if it's ultimately less efficient. It's true, you can't control your retention rate (other than by actually paying attention when you study and review), but I think the retention rates we have could be useful for the site administrators, who do have control over it.

pitwo wrote:

After all, what you're trying to associate to the kanji is the story, not the keyword.

I'm not sure I agree. I often end up forgetting the stories while still remembering the kanji themselves. When I do, I still look up the story (unless it's a character like 先 which I already know by heart), but I'm not particularly bothered by it. What really matters is if I'm going to remember the kanji or not. I find that I do always know the names of the elements, though... if I forget one of the elements, I won't be able to write that element at all.

- Kef

Last edited by furrykef (2008 July 11, 1:33 pm)

Other threads have delved into retention rates as well, and I think 90% is a good target, with 85 being satisfactory. Anything less and some aspect of your technique - either story creation, or studying, etc - needs improvement. To get all the kanji (or nearly all) into the final box, you'll need to maintain good retention - which is the long-term goal anyway.

Alyks wrote:

I opened a Japanese manga once and saw 国 like three times on a page. After that I had no trouble remembering it from heart, and can easily sidestep the mnemonic.

This is the great thing about the Heisig system. Once you start seeing and using the kanji is books, manga, etc, they stick really well. This is when the payoff comes.

Does everyone else review them box by box? I just review the whole bunch at once.

Timcampbell, I'm sure most do it as a batch. They're just talking about not worrying about cards that might be in box 2 or 3 being a higher fail rate.  Kind of difficult with RevTK to know what box the card you're looking at is in, but you probably know if it was a recent card or one you have not seen in awhile.

It may be hindsight talking, but always seeing your 4th stack getting bigger is a great sign of progress. Yeah, you'll be missing cards and have a growing fail stack, but knowing that other kanji are sticking in your head for the long haul should be motivating. Still, some see 100 cards due to review and get depressed. Others see 200 red cards and get depressed. The advice given above addresses that problem in a good way.

My retention rates for kanji i recently learned is like 50% lol, any OLDER lesson i get 98+ percent.
Seems kinda strange that the kanji 'fresh' in my head have the worst recall rates, but i suppose thats because the story isn't perfected yet

Basically, what furrykef said. Well, rather, they haven't gotten firmly implanted in long-term memory.

See, the thing is, the terms "short-term" and "long-term" memory don't get used
entirely appropriately in casual speech. If you remember it for more than a handful of minutes, it's in long-term memory. For the rest of this post, I'll use scare quotes when I refer to the idea meant in casual speech, while the plain term refers to the meaning in memory research.

Without pretty regular rehearsal, in twenty minutes you'll have lost pretty much everything you were holding in short-term memory. Anything you still have is in long-term memory. However, without some kind of special salience or attention, recent additions to long-term memory will degrade quickly?matter of hours or days.

I haven't actually read any of the research on SRSs (I've read summaries of it, but that's not the same thing at all), but assuming they work as indicated, if you've been doing your reviews the kanji you learned earlier should be in long-term memory still with very slow degradation ("long-term memory"), while for new kanji the degradation time is so quick that even if the kanji does in fact make it out of short-term memory it may well be gone by the next review.

~J

Last edited by woodwojr (2008 July 12, 12:36 am)