By the time you get to the end of RTK1 (I haven't added them all to this website, but I finished my initial study through number 2042 yesterday, hooray!) you're probably pretty good at this game, but at least for me it took a while to understand how to make memorable stories that would lead to easy memorization of the kanji.

There are also some choices involving organization of primitives that you need to make early on which are not that easy to change later, such as whether to lump several small primitives into a single bigger one, and conversely, whether to split a given primitive into two based on either location or how it's drawn.

I think an interesting and potentially useful topic would be what people think would have helped them to know when they were starting out.

Memorization is obviously quite individual, so I imagine that different people may have different views, but the following are some of things that I think would have helped me...

To remember from keyword to Kanji, the most important thing is that the story is memorable, not that the keywords necessarily fit that well into the story. For example, a kanji which has been difficult for me is "wash", which uses "water" and "before". It would seem that there's a natural story here, something along the lines of "wash with water before doing something". The problem is that this story isn't memorable, you'd be better off having a more exciting story even if it didn't connect as well to the primitives.

An example of a story that worked well for me because it was memorable was erikkusan's story for "detach". The primitives are used in a completely arbitrary fashion, and don't fit that well into a coherent story, but because the story itself is fixed in my mind well I have no trouble remembering it.

Often I make up a story by thinking of the keyword independent of the primitives and seeing what images come to mind. I then pick the most vivid image and somehow work the needed primitives into it, even if they aren't that great of a fit.

I initially resisted combining primitives and "personifying" some of the abstract primitives (so "finger" becoming "Shrek"). I felt that this would just add more clutter to my brain, and it also felt a bit phony; if a primitive means "finger" why think of it as something else. But now I think that these kind of mnemonics are useful, the point of RTK1 is just to hang a label on each kanji, how you get there and what the exact label is doesn't matter so much.

For combining primitives, if a given pattern occurs more than about 3 times, I generally tried to combine it. So, for example, the "broom crown crotch" pattern that occurs in "lie down", among other places, I made into "Cinderella".

By the way, "crotch" I found impossible to work into stories, I think viewing it as "buttocks" is much more productive.

Sometimes to help in remembering placement or the way a given primitive is drawn it's useful to split a given primitive into two. Heisig does this to a certain extent, but I think it can help to split more of them. Since a primitive that is repeated a lot is easy to remember, this doesn't add much of a burden as far as memorization.

For "wheat" I use the image of a "wheat stalk" when it's to the left, and "cream of wheat" or "cooked cereal" when it's above. For the moon/flesh primitive I try to use "moon" or "month" when the primitive is curved, and "flesh" or "body part" when it's written with straight lines (unfortunately I started this part way through, so I'm not completely consistent with this). "Heaven" is written two different ways (with the top stroke being written in different directions), for one of these I use the imagine of a peaceful heaven, and for the other a stormy heaven.

There are other primitives which I didn't split but probably would if I were starting over. For example "soil" depending on whether it's to the left or below, and "correct" which is written two different ways.

I could probably go on, but I'll stop here... I'd be interested in what other people have to say...

I like the idea of assigning slightly different meanings to primitives based on where they're located in the kanji. I hadn't thought of doing that before, but it's a good idea! I've had trouble at times remembering whether to place them side-by-side or top-to-bottom.

The following is not a tip for people to follow. I just want to discuss your method...
It sounds like an interesting idea, but how do you know that "wheat stalk" is left or top? That seems completely arbitrary. Another idea for linking in the position would be to define certain actions. For instance, soil on the top would always be burying something with soil. Soil on the left could be some other action... You need not limit yourself to two actions though, otherwise your stories would be so similar.

Sorry, I'm not quite sure of your question. What I actually do is to view the primitive in a top position as "hot cereal", so 香 for example (did you know that if you leave Cream of Wheat out in the Sun for a day it will turn into a fine Incense?). Anywhere else, so 利 or 穀, I take it to be a wheat stalk or grain or plant.

Most primitives I don't do this with, if I have trouble with placement I just tweak the story to try to get it right, but for common primitives that often go in different places I think it helps.

I think splitting primitives is even more useful for primitives which are drawn slightly differently, since it's harder to tweak a story to imply a different drawing. Splitting only works with common primitives, however, since otherwise the burden of remembering an extra primitive is more than the effort saved in having a simpler story.

You could try using different actions, I suppose, but as you say that starts to limit the stories. My guess would be that using different nouns would work better. So for soil you could imagine dirt for one, and soot for the other. As I said, tho, I didn't split that one myself.

I guess that I wasn't clear enough. What I meant to ask is, say you're reviewing, and you remember your story deals with soot. Now, the problem that I imagine is that you forget whether soot is top or left position. After all, nothing about soot implies a certain position, whereas a certain action does: burying from the top, for example.

I suppose for the most common primitives it's not a problem because you review them and naturally remember the link of soot to left, or top. Which makes me think it would be useful to have the primitives data so that one could order based on number of times used. In that case one would know ahead of time to spend a lot of time on a good image/story for this primitive, and also whether it is important to split it as you describe.

Ah, I see... Yes, whether soot went left or on the bottom would be completely arbitrary. This probably only works if both positions happen often enough that you'll easily remember which goes where.

You're right, knowing which primitives and combinations of primitives are going to come up a lot would be useful, the problem is that by the time you notice something has come up a lot it's too late to do anything about it.

For example, a combination that could be possibly combined into something is "flowers over water to the left", it comes up at least 4 times in RTK1 (薄, 落, 藩, 藻). Four is about the point where it might be worth thinking about combining this into some combined primitive, but by the time you know it's going to come up that often there's no point.

I think it's only worth special attention when a primitive's place can vary a lot. With the example give, water and flowers, I would never do anything with it, since these primitives are very fond of their respective positions (left and top). I have more trouble with earth, since that can take a variety of position (I notably mess up left and under, like grave where it goes under, and bury where it goes left). In that case, I change the story to incorporate the position of the radical as best as I can. But only if I get it wrong.

I was thinking about water and flowers because I'd just messed up "full" 満 where for a change flowers didn't get the full top position. The other advantage of combining in this case would be it could simplify your stories a bit, having one less primitive, but yes, I agree that the utility of combining flowers/water is a bit marginal.

[ファブリス]

Usually James Heisig's "primitives" are close to the chinese radicals in their meanings. However for the flattened eye, I think it is the chinese radical for "net", and a simplification of this kanji : 网

You can get a feel for the placement of the primitives when you look at all of them. Some primitives are nearly always stronger and get in the leftmost position. Also often the main primitive (leftmost) is the one that is related to the meaning of the character. The other ones are there to indicate the pronunciation. Other times, the secondary primitives are simplifications of the more ancient characters, and have no relation at all to the original meaning.

For the flattened eye, I used the image of a cyclops' eye (and by extension the cyclops itself). I could have used the image of a net (such as fishnet) to keep a meaning closer to that of the chinese radical.

I'm finding that when I assign a person to a radical, I rarely forget it. Maybe it's just me, but so often when I can't remember a kanji, I at least know it's got Mr. T (person) in it, but I won't remember sun or mouth or heart unless I remember the whole story. Knowing that I would have assigned people to more of my radicals.

radical_tyro Wrote:Which makes me think it would be useful to have the primitives data so that one could order based on number of times used. In that case one would know ahead of time to spend a lot of time on a good image/story for this primitive, and also whether it is important to split it as you describe.

Yes, having primitive data would be very useful. See the thread at http://forum.koohii.com/showthread.php?pid=2941#pid2941 if you are interested in this.

There's probably loads more of us "done" with RTK1 since this thread was around first time...

If I was doing RTK1 again (knowing what I know now) I would have worked "dirtier" as astridtops has talked about in another thread I can't track down right now.

Basically not worring about high kanji failure rates from the first box. Not getting stressed seeing my red stack grow. Rather just keep adding new kanji, not necessarily at a huge rate but just keep adding say 5-10 if not more per day.

Obviously you don't want to ignore your red stack. However rather than spend a long time thinking of an image first time round I would have had a quick scan through the stories on the site and just gone for something that caught my fancy.

Not sure if what I'm advising can be backed up by decent evidence however. I've not yet managed to get over 1900 RTK1 kanji into the final stack...

Push through and finish the first time through.
If you stop, you'll have to start at the beginning and repeat work.

Move at a fast enough pace to finish quickly.
Finish within a few months. Fewer calendar days means fewer reviews, fewer opportunities to grow demotivated.

Learning kanji is challenging enough, don't make the task harder by creating more work for yourself.

I second Ziggr above. Here I am, back yesterday after another long break from studying Heisig (even though I did, sort of, finish. I had studied the first half of the book before discovering the site and the second half using this site). I'd been gone so long I missed Fabrice's Japan tour! Sorry Fabrice, it would have been nice to meet you.

Anyway, I tried to clear my expired cards but failed most except half of the ones from the four or more stack. This is my woeful stack situation now:

Failed or never tested: 1928 (757 failed, 1171 never tested)
One review: 28
Two reviews: 11
Three reviews: 0
Four or more: 75

Ouch! With a bit of work, hopefully I can get back to where I was and gradually add in kanji from the first half. But this would be a lot easier if I worked more consistently.

Consequently my Japanese hasn't been going anywhere recently either. Moral of the story: use it or loose it! And keep learning new kanji/vocab/grammar and not just practising what you know, if you want to improve.

I think consistency of primative "value" helps. Particularly when seeing a kanji and trying to recall the meaning. I've sometimes sat there trying to recall stories that simply aren't there. Only to realise later that I didn't use my standard image for a primative.

I don't really find this to be a problem going from keyword to kanji however. So it wasn't something I really noticed until finishing RTK1.

I'm not done yet with RTK I (but I'm close, at 1978).

If I would start over again, I'd like to change a couple of things:

Use kanji readings in the stories: of course, not all readings, maybe just the reading used in the most "common" (or simpler?) compound. Something like this:

Kanji: 坊 (492)
Keyword: boy
Primitives: dirt/soil/earth + direction/compass.
Reading: BOU.
Story: the boys go for a "foot orienteering" (where you hit the dirt with a compass). One got lost and the others where shouting "BOOOOy" to locate him! When he was found and they finished the race, the boys went BOWling.

I think that if you have to learn the readings one way or another, this CAN save you a lot of time and effort.

Use other reviewing system: I love Fabrice's site, but the lack of support for reviewing "on-the-go" on portable platforms is a big inconvenience for me. I even wrote a "proxy" for it, but I can't always use it since there's no good & free/cheap wireless coverage everywhere I go. When I was around kanji #1300, I entered all kanjis into Twinkle (a flashcard program for Palm Pilot handhelds), and it really improved things for me, since I could review in my spare time anywhere.

Use/write more software, more often: I just can't live without some of the Greasemonkey (GM) scripts I wrote (and the ones that Woelpad wrote/extended/enhanced) for RevTK! The "Adopt a Story" is one that saved me a lot of time. It took me some time (though not much) to decide to write it, but other GM scripts and offline tools I use came later.

If I was to start over, I'd probably write a couple of other GM scripts (e.g., I'd try to implement a rating system for stories; highlight words on my story that are keywords for other kanji as I write/type the story; etc.), and/or a (Palm) software to manage the stories and reviews offline (with some dictionary support).

As you can see, there's not much I'd change in the way of studying (stories, primitives, mnemonics, etc.), but just a few additions and changes here and there. OTOH, I think that most changes could be used by any user to improve his/her studying time, so that's why I'm sharing it here.