[ファブリス]

An interesting read to fuel the discussions about applying kanji chains for RTK 2.

http://www.susi.ru/kanji/ChMethod.html

As far as I understand, the document describes a method that could very well be applied to learn both RTK I and II at the same time. I think the method is sound, however for most people I think this is too steep to learn. It is already quite a commitment to finish RTK I, on your own, and to also learn by oneself how to use the imagination to produce vivid images for remembering.

This kind of techniques doesn't seem to make any sense when you read the explanation. Only by doing it, through RTK I we get a feel for how associative memory works, and how vivid images and stories help us remember the composition of the japanese characters.

Of particular interest is the section Memorizing onyomi :

Quote:Remembering Chinese readings is not an end in itself. One should remember them to be able to read compounds. The traditional way is from compounds to the readings of their components; the chain method suggests an alternative. (...) to memorize a jukugo, you have only to memorize the characters it consists from. (...) Thus, it is not an exaggeration to say that the easy and full remembering of onyomi is the most important advantage of "Kanjichain."

Some data :

Number of distinct ON groups covered by RTK I (using one "main/most common" reading for each kanji) : 274.

Number of distinct ON groups covered by RTK I + III : 287.

Top ten larger groups :

キョウ 29
セイ 31
カ 32
ソウ 38
トウ 39
キ 39
カン 47
シ 52
ショウ 54
コウ 66

Based on my experience the largest group of 66 kanji would take 1h30 or 2h to build.

Large groups are not a problem.

But I don't know how to handle the small groups yet :

48 groups of 1 kanji / reading
50 groups of 2 kanji / reading
27 groups of 3 kanji / reading
22 groups of 4 kanji / reading
19 groups of 5 kanji / reading
108 remaining groups.

The first one could be made into a location such as a "single's club", a kanji chain comprising the 48 "single" readings, each with a powerful image to remember the reading.

However this may not be necessary, I have yet to experiment with very small ON groups and "single" kanji readings.

Very interesting stuff. I'll be even more interested to hear about the results people have when applying this method.

A couple of queries. How flexible is the system to handling new kanji and new readings when they appear? Also, through learning the readings like this, it seems obvious that the goal is to be able to know the yomikata of a kanji compound when it appears. How effective do you think this method will be in increasing the vocabulary of words that you can produce when speaking or writing? I think it might help but it would depend on how well you'd researched the meaning of the kanji in RTK 1.

I just hope that everyone realises that once you've learnt how to write the character (RTK 1) and how to read it (kanji chains, RTK 2 or otherwise), I reckon you've only got about 30~40% of the job done. The rest involves learning what the word means and, much more tricky, how to use it.

wrightak Wrote:I reckon you've only got about 30~40% of the job done. The rest involves learning what the word means and, much more tricky, how to use it.

seems like quite a low estimate. the reading of kanji is a massive jump from just looking at compounds and having a "inkling" as to what the word means, surely? i think it has been mentioned in another topic that being able to understand a compounds basic meaning upon seeing it for the first time and being able to pronounce it correctly seems like a great leap forwards for japanese learners.

IMO of course

That 40% is a massive achievement and I certainly wouldn't take anything away from it. I'm just talking about my experience in looking at the amount of time I put into each aspect of learning vocab. Thanks to RTK 1, I spend relatively little time on the writing and when I encounter a word the first aspect of it that I learn is its pronunciation. These things come fairly quickly for me but I spend much more time learning what situations the word is used in, all of the nuances of its meaning and how to use it in a sentence. In order to check understanding, I would encourage anyone to make 3 varied non-trivial example sentences with a word they learn and show them to a teacher or a Japanese person and observe the results. If you get a nod of approval then the word's probably in the bag. If not, more study.

I've also had many occassions where I've made intelligent guesses of meanings of compounds using Heisig's keywords but when I confirm with teachers it turns out that I didn't understand properly. Other people may be different but for me, learning the writing and pronunciation of words forms a small part of the overall task.

Now that I have made my way through RTK1 and am doing RTK2 in what I see is a series of chains, rather than a town, I must say that I believe the work with RTK1 is perhaps more valuable for chaining with RTK2, since you work with discrete elements that you taught yourself in RTK1.

However, working on a few elements in RTK1 I did make chains, I realize that now. Those are also the kanjis that I remember the best. The problem is that heisig has his set way of doing the kanjis, building on elements. If you want to make chains here after any kind of system, its going to be really tricky. I'd say chain where you can, but don't expect it to hold up throgh RTK1.

[ファブリス]

Yes the beauty would have been if RTK I had the kanji grouped by ON reading.

I ran some scripts once and found that the ON readings are completely shuffled in the RTK I frame order. If you built ON chains in the RTK I frame order, you'd have to swap between ON groups every 2-3 kanji, way too often. Swapping between groups and adding to each group is not difficult, but swapping between chains too often makes studying too slow.

So the only small "chains" that exists in RTK I happen because of the "signal" primitives (part of the written character that indicates the sound). A good example is frames 1527-1530 which is a "pure group" in RTK Volume 2. Those can be chained into one common imagined location as you learn them in RTK I. They will stick very well, and you already know they have the same reading (JOU).

Immacolata Wrote:Those are also the kanjis that I remember the best.

Same here. I think that using the kanji chain approach for RTK II is a great way to consolidate the kanji from RTK I. Once they are all neatly tied up in kanji chains, they are much less likely to fade out of memory.

Another way would be chains through elements. I chained rice and altar, because so many came after another. But then a long time passes, and whoops, a little rice or altar element appears, then is gone again for another 200 kanjis. Making chains hard. If you would do the joyou kanji in kanjichains, you would need to sit down first and make a completely different index than the one RTK1 uses.

However, I fear you would lose one major pedagogical advantage of Heisig: his way of building elements on elements. That is also a kind of chain, don't forget that, a chain that doesn't work per kanji as such, but on element basis. Once you learn the various shapes for elements like red/row, you never ever forget them again. So let us give Heisig some credit for that

[ファブリス]

Yes I didin't mean to argumenting the RTK I method. Rather, saying that it would have been great if it had been possible to get the best of both worlds : smooth introduction of new primitives (1 or more chinese radicals), and grouping of kanji by reading.

It's best to keep RTK 1 & 2 separate anyway. Without enough experience the learner may not make a clear distinction between the kanji-story, and the chain-stories. The chain-story is one level higher. If there is no clear demarcation betwen each kanji-story in a chain, then the individual kanji would not be remembered correctly.

That's why I prefer to build chains visually, by creating imaginary locations where each kanji sits in a distinct place, rather than rely on mnemonics and words alone. For example you could have "the brew man deferred his lot to the lass" (4 kanji pronunced ジョウ). How do you remember that "defer" ( 譲 ) is one kanji ? I think it's important to have "defer", a kanji from RTK I, also appear as a separate entity, in a clear location within the kanji chain, and using the elements of the "defer" story to give it an image.

Immacolata Wrote:Another way would be chains through elements.

Yes that works. I did quite a few chains like this in RTK I. Most of my "axe man" stories ("axe" primitive) happened around some location. All the kanji using the "altar" primitive I placed into a fictional church. Those were remembered easier and recalled faster than other non-chained kanji.

I don't think such groups prevent re-linking the kanji into different groups later (based on chinese readings), but the groups themselves have no real use. And THEY STICK. So I'm not sure I would recommend to make such groups through all of RTK I.

[ファブリス]

Something that needs to be taken care of, is a system for making a clear distinction between long a short vowels in the ON readings.

For example I have a "CHUpa CHUps" Factory for the reading CHUU. I didnt have to worry because there is no ON reading with just "CHU".

However we have to be careful with DO and DOU, GYO and GYOU, GO and GOU, etc.

Say that you think of a "Go Club" (where people play GO) for the reading GO. Then you need another location (kanji chain) for the reading GOU, and not mix up the two.

One idea would be to replace the U with another sound in the mnemonics, perhaps V since V is not part of the syllabary. So we would have DO and DOVE instead of DO and DOu.

Simplest would be to make sound associations : MAKU -> makudonarudo, GO -> "go club", DAI -> "die / dining room". But it's not always easy to find english words with long vowels (GOU, JOU, ...).

For the adepts of kanji chain here, did you make yourself a system for this ?

ファブリス Wrote:I have a "CHUpa CHUps" Factory for the reading CHUU...
Say that you think of a "Go Club" (where people play GO) for the reading GO...
MAKU -> makudonarudo, GO -> "go club", DAI -> "die / dining room".

I have no experience with kanji chain so feel free to dismiss what I say but it seems to me that there are far too many English words representing Japanese sounds. Why not use Japanese words instead? マク -> マクドナルド is good. How about ダイ -> 台所 or 大学 instead of dining room?

I fear that using English words to represent Japanese sounds has the potential to have a horrendous effect on your pronunciation when you speak.

[ファブリス]

I wouldn't worry about pronunciation too much, the word associations are only meant as mnemonics to remember which chinese reading the kanji chain corresponds to. I don't think it has any impact on pronunciation (I assume most ppl finishing RTK I have already spent many hours listening to radio, movies, doing a bit of reading..).

Using Japanese words for the mnemonic association is a good idea !

This is probably a better idea than using english keywords, I will just have to learn some Japanese words to use in the mnemonics.

In that case I think it's best to pick a japanese word that also comprise one of the kanji from the group, otherwise it could be misleading.

So for example my "JOU island" (I just remembered it like that) can become "JOUZU island" where everyone is skilled! (cant type kana at the moment). JOUZU contains the kanji "above" which is part of the kanji chain.

That actually makes a lot of sense! Why didn't I think of that?

this is mainly aimed at immacolata as i know that he is well into studying the kanj- chains now.

how do you study them? are you taking them from ア to ゾウ "alphabetically" or what? any tips you can give me as i start out on this voyage?

I go like that yes. REad my blog.

Well, as an alternative, since I still have some kanji from RTK1 to go, I start with all the lists that contain primitives that I can put a kanji to immediately (actually, it helps that I keep the list at my work where I have no Japanese language support, so I can't see the kanji anyway). That way, I started out with little chains from 1 to 4 readings, and my largest one is now 9 kanji. That way I can practice the methods with easier chains before tackling the really large ones. Every week, I go over the list to see if have progressed enough to add a few readings to my list.

[ファブリス]

Hi astridtops,

I would recommend you to learn a ~30 kanji chain as soon as possible, it will give you confidence with the method.

One of the first chains I did was for [kana]SEI[/kana]. Which was 31 characters which I can still retrace completely out of my head. The ability to retrace the whole chain from memory is not directly needed for reading. I think it comes as a byproduct of using the "memory palace" method, rather than a chain based purely on logic. In the memory palace method, you really visualize a location, and each kanji must have its very distinct place. In a classic square room for example, you can place 5 kanji very easily remembered, one at the center and one in each corner. You can stack and connect those rooms, create alleyways, etc etc. This in itself is quite fun and interesting to experiment with.

The other method is more based on logic, there is less emphasis on location, but rather on a story. A memory palace is a little bit like a story as well, but instead of telling a story you just retell a "journey" as you walk from one end to the other, and recall the kanji as you come accross.

Now recently I realised that the memory place might be slightly more difficult for some persons, who may be less proficient with "imaginative memory" however I think it could be well worht the effort because I';ve just realised that the kunyomi could later be added into these "memory palaces". The information stored in these places can be extended later to include the kunyomi!

Maybe we could post our progress here.

I wish I could focus on this atm, but I am only at about 10% with the chain method.
24/275 groups
266 kanji readings covered (main readings)
And the larger chains I did were [kana]KEN (29) SHIN (29) JOU (21) SEI (31) SEN (27) [/kana]

It's funny how many compounds you can look up in a dictionary when you start combining just a few of these chinese readings!

When I got back from work, it was much easier for me to do a chain of 27 kanji beacuse I only have to think of 1 memory association for the chain "tag". Once you start the location is the same, so doing a long chain is fairly easy. If I wanted to do 27 groups of 1 kanji however, I'd have to try to come up with 27 efficient memory associations for the readings, which takes a lot more time.

OK, I've just started chaining. I'd like some tips from the more experienced of you.

I opened that kanjitown template excel sheet and just started with the アs. Since there were so few kanji per yomi, I lumped them into one story.

There is an African/Asian fusion restaurant. Outside of the restaurant a man weeps pathetically, "ai, ai!" He has lost his love. She ate Asian, he ate African. He is an actor (on yomi アク), and has a bad grip on reality. The spices are acting up on him, putting pressure on his ass (アツ). Meanwhile, his love, Ann, goes off into the darkness. She needs to relax and come up with a plan to get another boyfriend.

It holds together in my head pretty well. Have you guys found this kind of thing to be a good way to keep the lonely kanjis together?

Having done this, I'm now thinking that it might be best to hit the signal primitve kanjis first. These seem easy, and you get about 25% of kanjis out of the deal. True that it's tough to rework the stories, but the signal primitive kanjis kind of form a mini chain within a chain, anyway. The new kanjis would stick to the outside.

Also, it seems like not a bad plan to do the longer chains first. Or does that then make the lonely kanjis more difficult.

I've been drilling work-related vocabulary words with Stackz, thinking that this would help with the yomis, however my retention if very poor. But these are particularly difficult compounds. Using Stackz with the the JPLT4 vocabulary words has been more successful, since that is usually just matching a word I already know with a kanji I already know.

The nice thing about RTK1 was that the system was done for me and was a proven method. I would rather spend more time getting a solid methodology down before jumping into RTK2 than waste time with an inferior method.

What wisdom do you have for me?

Wakela Wrote:There is an African/Asian fusion restaurant. Outside of the restaurant a man weeps pathetically, "ai, ai!" He has lost his love. She ate Asian, he ate African. He is an actor (on yomi アク), and has a bad grip on reality. The spices are acting up on him, putting pressure on his ass (アツ). Meanwhile, his love, Ann, goes off into the darkness. She needs to relax and come up with a plan to get another boyfriend.

Nice story Wakela; very memorable. I'm curious - is anybody still using this method? I abandoned something similar over a year ago, and now use this method:
http://forum.koohii.com/showthread.php?tid=109

OK, I've done two days now with making chains for the RTK2 Chapter 2 "Pure Groups." According to Heisig, any time I see these primitives in a kanji I know what the on yomi is (at least one on yomi, anyway). For example, whenever you see 中, you know the on yomi is ちゅう. Is this true? If so maybe chaining is not necessary for the Pure Groups. I don't need to place every 中 kanji in a chain. I only need to associate this primitive with the pronounciation. In other words, I don't need to place Mr. T, the go between (仲), inside the giant head of Chewbacca. I only need to imagine the Chewbacca head as something you put things in. Is this nuts?

I come here and am greeted with the sound of crickets.

Has anyone come up with a list of keywords for the on-yomis? Each keyword would represent a sound, and should start with that sound. ちゅ->Chewbacca, かん->Candyland, etc. I've started, but it's tricky. If such a list does not exist then it seems like a good use of this community to make one.

Wakela Wrote:I come here and am greeted with the sound of crickets.

Has anyone come up with a list of keywords for the on-yomis? Each keyword would represent a sound, and should start with that sound. ちゅ->Chewbacca, かん->Candyland, etc. I've started, but it's tricky. If such a list does not exist then it seems like a good use of this community to make one.

here's a good starting point:

http://www.ravenbrook.com/project/etp24/...mnemonics/

I've tried chains for KANJI with the ON-Yomi of TSUU and TSUI (can't remember third one of this). The keyword is in capitals

Toucan was in PAIN from the TRAFFIC since it somehow got ran over by trucks and other vehicles...

CHASE after REGIMENT of gold diggers with WOODEN HAMMER at Sovereign Hill. Seized the gold diggers and bang them with HAMMER.

On Yomi SHI for characters YELLOW TAIL, LION, EXPERT

LION is an EXPERT at angling for YELLOWTAIL.

I can't remember where I got this, but there is a trick to keep yomis like コ　and コウ straight. Your コ keyword should of course start with the "ko" sound, but the following sound should be a consonant that makes a hard stop like k, p, d, t. For コウ the word should also start with "ko" but the next sound should be a soft consonant like l, m, n, f. For example keywords fo コ could coke, cope, coat, and keyword for コウ could coal or comb.

[ファブリス]

Been looking at this again recently. I'm tired of trying to come up with "linkword" type mnemonic associations, this really slows me down. Creating stories takes some effort but is a lot easier. I'm going to experiment with wrightak's suggestion to use a Japanese word for the sound association. So in theory I have an exemplary compound as the mnemonic association, so I have to memorize ~300 words, the rest of the kanji in each onyomi group, are associated to the first compound I chose, within the story in one way or another.

[ファブリス]

Did anyone maintain a list of onyomi mnemonics ? I'd like to provide one mnemonic for each onyomi group for the "Trinity" update. Good ones are difficult to come by.

I could setup a Google spreadsheet but I don't think there's enough interest yet, in the meantime I prefer the speed of editing the file locally.

If you've got something we can exchange lists by email, thanks.