Optimal Leitner arrangement
I'm using a Leitner Box system to review kanji, vocabulary, and grammar. At the moment I have sets of cards set at one day, two day, four days, one week, two weeks, four weeks, two months, and so on. I have schedule like this:
2 day Sunday, Tuesday, Thursday 4 day Monday, Friday 1 week Wednesday 2 weeks First and third Saturday ( Read more... )
Comments 5
Paper and pencil early version of it is described here http://www.supermemo.com/english/ol/beginning.htm#Algorithm
There may be better descriptions elsewhere.
Reply
Aha! Thanks, that's useful. I found a paper on the site which describes how to run the system manually, and it's much like mine. It suggests varying the intervals but doesn't give any mechanism for doing so.
Reply
recalling something after an interval is better for your memory than looking at it every day.
I expect that testing every day gives better recall than testing after an interval, but it is a lot more work. (I'm not counting 80 hour days and dying of fatigue as being non-better). Surely the staged rehearsals are designed to save time? As suggested on the page. Still this is not relevant to the below.
Perhaps the optimum pass rate is 100%? For example, if you were a robot and didn't know it, and your memory was a dictionary, and key value pairs were annotated with (say) a first learned date and a lifetime, and every lookup that was within the last half of the lifetime caused the lifetime to be doubled, that Lietner Box system is exactly what you would use to make sure something learned stayed learned. If the pass rate was below 100%, that would mean your intervals were too wide and you had to learn something again from square one. This model of memory function is consistent with the system described on the ( ... )
Reply
Yes, it does a little, especially since words and kanji are made up of other words and kanji, most of which have been learned earlier, and are therefore deeper in the stack.
I've modified my system. I now have one "1 day", three "three day" (A, B, and C), and two "1 week" (A and B) rings for each of kanji, vocab, and grammar. My schedule goes like this:
Monday 1d 3dA 1wAg
Tuesday 1d 3dB 1wAv
Wednesday 1d 3dC 1wAk
Thursday 1d 3dA 1wBg
Friday 1d 3dB 1wBv
Saturday 1d 3dC 1wBk
Sunday rest
So I'm processing seven regular rings each day (the one day kanji, vocab, and grammar, then a set of three day kanji, vocab, and grammar, then one 1 week ring from among kanji, vocab, and grammar). Each ring promotes to the other rings on the same day. The week rings promote to the "old generation" which is a big row of rings with individually marked dates. This means I'll often be processing an eighth "old" ring with a period greater than two weeks. However, I'm not ( ... )
Reply
Reply
Leave a comment