Ok, let me throw out some actual statistics to give some solidity to this rampant speculation:
Number of Joyo kanji with kun-yomi: 1081

Number of kanji needed to read:
95% of a newspaper: 1050
95% of a magazine: 1350
99% of a newspaper: 1700
99% of a magazine: 2200

Number of phonetic compound characters in the toyo-kanji: 1211

Helpfulness of phonetic component in phonetic compound characters (toyo-kanji):
exactly the same: 57.6%
partly reliable: 32.7%
no help: 9.7%

christine Wrote:so that would be consistent with the theory that if you learn 1000 characters with kunyomi readings, then the remaining ones are usually used in compounds only.

Before you were saying that if you learn the 1000 most useful characters, then you know most of the kun-yomi. That's not the same as saying that you only need to know 1000 characters with kun-yomi to know most of the characters with kun-yomi. I hope you realize those are remotely different things.

christine Wrote:Remember: even if the technique is only effective 80% of the time, it's still makes a big difference, because it takes you from a reading ability of 94% (assuming you know 1000 characters) to >99%. And surely the ability to read more than 99% of newspaper content knowing only 1000 kanji is pretty good. By the way, I didn't make up these figures - 94% comes from Kanji in Context, but I did make up the 80% figure. But even if the technique was correct 50% of the time (ie., it's marginal), it would still raise the reading ability from 94% to 97%. So, regardless of whether you believe the technique is effective or not, it should provide a significant boost to reading ability.

You may be able to guess the readings about 35% of the time, but it would be far, far less successful as far as guessing characters' meanings.
And 95% is a misleading number until you think about it. It means that on average every 20th kanji you encounter will be one you don't know. I think that's a lot.

yukamina Wrote:The way I see it, if I see a new kanji word, even if I know the kanji in it, I'd have to look it up to be sure. I would hate to assume and then get the wrong reading stuck in my head. The use I see in semantic and phonetic markers(or whatever), is that once I look it up, it really makes sense and I can remember it easily. But before I look it up, my guess is just a guess.

Absolutely. Being able to guess a character's reading about 35% of the time (and significantly less often for whole compounds), but it doesn't amount to anything if one can't do it with much certainty. And if you do check on the character's reading, it's going to be hard to keep that information in your head if you haven't specifically learned that character, so we're back to the whole "you need more than 1200 kanji" thing.
And someone who knew that 化 is pronounced か might guess that 靴下 is pronounced かか or かげ, but of course they'd be wrong on both counts. And this isn't a rare exception but actually a pretty common case.

Hello Wrightak,
Your question to Cristine might not have been a challenge, but ... My sense is that you have studied Japanese for many years and that these questions are pretty much theoretical for you at this point. Other than that, you seem like a decent sort.

So, De Roo's story for the kanji you mentioned (I haven't really figured out how to get kanji to come out correctly in my posts) would be something like:
First kanji--a bunch of people standing around the corridor waiting to be let in to see the boss. That seemed pretty clear to me.
Second kanji--land around the shrine.

There is a lot of cultural info which can be applied here. I knew that we were not talking fields, or the kanji probably would have included that. It helps in the interpretation if you understand the details of family and institutional shrines etc.

That said I took a guess. It doesn't matter if I'm wrong because if it's an important word for me, it will come up again. The part that people don't seem to get is that it doesn't matter what your initial accuracy rate is. If you keep trying to read things, it will naturally improve. For example, I started seeing the kanji for kouen meaning play or theatrical production; I was hearing the same word watching television interviews. I put it together. This is not rocket science, because if it were I couldn't do it.

Laura Wrote:De Roo's story for the kanji you mentioned (I haven't really figured out how to get kanji to come out correctly in my posts) would be something like:
First kanji--a bunch of people standing around the corridor waiting to be let in to see the boss. That seemed pretty clear to me.
Second kanji--land around the shrine.

And with Heisig you would see the word and think: "honorable" "company"

If one is as versed in the vast complexities and multitude of layers of the Japanese persona, one might be able to guess that this means "your company"...in fact, now that I think about it, an orphan could figure it out, don't even need "institutional shrines". (Sounds like a great band though)

EDIT: Could you DeRoo this word for me I would like to see what it turns up: 梨園 because the Hesig keywords don't work well for it "pear tree" "park". I am curious to see if it comes up with anything better.

But, you get the benefit of De Roo after a few days study.

and it will remain in your memory for a quarter of a century with no review.

You know, I was just thinking, it is possible that De Roo inspired Heisig or perhaps they were good friends sharing ideas. They really are very similar in concept.

dilandau23 Wrote:You know, I was just thinking, it is possible that De Roo inspired Heisig or perhaps they were good friends sharing ideas. They really are very similar in concept.

Maybe, but as you can see from the article, the idea of 禾=rice plant is not just something that those guys cooked up in their heads.
I think Heisig's primitives are a lot more etymologically acurate than most people realize; it's the mnemonics themselves that are mostly works of imagination.

According to my 漢和辞典 (which just happens to be published by 旺文社), 禾 is a pictograph of a 穀 (which of course would include rice) plant drooping over after it has grown fully and become ready for harvest.

wrightak Wrote:I chose 御社 completely at random and I made it clear in my post that any other word could be used to explain the character break down method with more clarity. The word I chose was a bad example because its reading is irregular but I was curious to see if the meaning could be deduced.

Let me see if I get this straight. You chose "御社" at "random" but already knew it was a bad example because it's reading is irregular. Then why did you choose this example?

Forgive me for saying so, but it seems that your example was not a genuine attempt to test whether the technique works, but potentially designed as "entrapment" to make me look silly (given that I've already said I am not experienced enough to apply this technique).

If that is your intention, I'm sorry to say I won't participate any further.

wrightak Wrote:

Christine Wrote:That's how I built up my vocabulary in English

What is your first language? Since your mother is a Chinese teacher, and given the above quote, is it Chinese?

I've mentioned before, I can't speak Mandarin or Cantonese, apart from a few words. And it's pretty obvious that if I already knew the characters, then I wouldn't be studying them, would I?

But I am curious how you have mastered English. How did you build your vocabulary? Was it by reading (like me), or watching TV, or talking, or did you memorized an entire dictionary? It strikes me that any systematic study of kanji is kind of like memorizing a dictionary. Sure, it can be effective, but I wouldn't want to study that way.

JimmySeal Wrote:Ok, let me throw out some actual statistics to give some solidity to this rampant speculation:
Number of Joyo kanji with kun-yomi: 1081

Number of kanji needed to read:
95% of a newspaper: 1050
95% of a magazine: 1350
99% of a newspaper: 1700
99% of a magazine: 2200

Number of phonetic compound characters in the toyo-kanji: 1211

Helpfulness of phonetic component in phonetic compound characters (toyo-kanji):
exactly the same: 57.6%
partly reliable: 32.7%
no help: 9.7%

Can you cite the source of your statistics?

yukamina Wrote:Okay, how do you know what the phonetic marker is? Not every kanji has one(assuming I understand what you mean by phonetic marker). For 務, how would you know the marker is 矛 and not 力(assuming this is right)?

Take this with a grain of salt, but the impression I get is that the phonetic marker is typically (but not always) the part of the character that is NOT the radical. So in the above example, if you can look up the radical for the above character, the phonetic marker will be whatever is not part of the radical.

Of couse this works only when the radical can clearly be distinguished from the rest of the character. In cases where the radical is not obvious, I am not sure how the technique works.

Another hint I received is that the phonetic marker is usually the most complex and elaborate part of the character, and usually written a bit bigger and more pronounced than the rest of the character. That's why good calligraphy is very important, it's necessary to get the proportions between different parts of the character correct.

For example, in "旺" you can see the right hand part is written larger and more pronounced than the left part. So clearly "hi-hen" is the radical and "ou" is the phonetic marker. That's how I was able to guess (more by accident than anything else).

For me, the easiest way to determine which is the phonetic marker is simply look up Henshall. He normally points out the phonetic marker in the text.

JimmySeal Wrote:Before you were saying that if you learn the 1000 most useful characters, then you know most of the kun-yomi. That's not the same as saying that you only need to know 1000 characters with kun-yomi to know most of the characters with kun-yomi. I hope you realize those are remotely different things.

Yes, and this is what I wrote:

Christine_Tham Wrote:I would think that the first 1000 kanji probably need to cover off all the basic characters (that act as building blocks for the important phonetic and semantic markers), plus as much of the commonly used kunyomi as possible. You are right in that this is not necessarily the most popular 1000 characters. And the minimum number is probably closer to 1200 than 1000.

Once again, can I suggest rather than assuming I don't know or haven't realised, perhaps you could actually pay a little bit more attention to what I have said?

christine Wrote:You may be able to guess the readings about 35% of the time, but it would be far, far less successful as far as guessing characters' meanings.

Can you provide some justification as to how you have arrived at this figure?

Quote:And someone who knew that 化 is pronounced か might guess that 靴下 is pronounced かか or かげ, but of course they'd be wrong on both counts.

But if someone really knew how to apply the technique, then they will have realised the left hand side is not a valid Bushu hen radical, so therefore that would be an incorrect application of the technique. Something tells me you are not really applying this technique correctly, hence the reason why it does not work for you and hence you are seeing all these "exceptions".

Christine_Tham Wrote:[But I am curious how you have mastered English. How did you build your vocabulary? Was it by reading (like me), or watching TV, or talking, or did you memorized an entire dictionary? It strikes me that any systematic study of kanji is kind of like memorizing a dictionary. Sure, it can be effective, but I wouldn't want to study that way.

I learned English by growing up in an English-speaking country (tho the question wasn't directed at me, I can't stay silent on this one). Heisig says again and again that the key insight of his system is that there is no reason to believe that the right way to teach kanji to Japanese schoolchildren living in Japan is the right way for Western adults (not necessarily living in Japan) to learn kanji. In fact, he goes on to give many compelling reasons why it isn't (for example, Japanese schoolchildren already know the spoken language). Nowhere does he imply that his approach is a good one for Japanese natives to use to learn to write their own language.

Same thing for vocabulary. One generally does need to "memorize the dictionary" by drilling vocabulary to learn a foreign language in a reasonable amount of time (unless one is lucky enough to immersed 24x7 in the country), although it would not be a very good way to learn ones own (although Malcolm X would disagree, see his autobiography).

Christine_Tham Wrote:

JimmySeal Wrote:And someone who knew that 化 is pronounced か might guess that 靴下 is pronounced かか or かげ, but of course they'd be wrong on both counts.

But if someone really knew how to apply the technique, then they will have realised the left hand side is not a valid Bushu hen radical, so therefore that would be an incorrect application of the technique. Something tells me you are not really applying this technique correctly, hence the reason why it does not work for you and hence you are seeing all these "exceptions".

My understanding is that 革 is indeed a valid bushu （かわへん）. The kanji 靴 does have the onyomi reading か however the kunyomi is used in JimmySeal's example above.

In most(?) cases a kanji's bushu is relatively obvious. However as Heisig's radicals correspond so closely to kanji bushu, if RTK1 highlighted each kanji's bushu (say in red, or whatever) perhaps we could pick up the bushu almost for free.

Christine Wrote:Once again, can I suggest rather than assuming I don't know or haven't realised, perhaps you could actually pay a little bit more attention to what I have said?

I'm trying, but it's hard when most of your posts are the length of encyclopedia articles. You're right, though. You did say it's about 1000-1200.

And as synewave pointed out, 革 is indeed the bushu for 靴 and its on-yomi is entirely regular, yet still, you could guess between two likely readings for the compound (and how do you choose which of the two?) and still be wrong in both cases. This happens all the time.

Christine_Tham Wrote:

jimmyseal Wrote:You may be able to guess the readings about 35% of the time, but it would be far, far less successful as far as guessing characters' meanings.

Can you provide some justification as to how you have arrived at this figure?

Just a rough estimate, but my thinking went along these lines:
The likelihood of guessing a character's on-yomi is about 58% (as shown in my statistics) so right off the bat, that is the maximum success rate for guessing a compound with one unknown character.
Now assuming that the character is actually using its on-yomi in the compound, if the other character in the compound has 2 on-yomi or 1 on- and 1 kun-yomi (that you already know), you have a 50% chance of picking the right one, so your success rate in that situation is .58 * .5 = 29%.
And what the other character has 2 on-yomi and 1 kun-yomi (still a common situation)? Then you've got .58 * .33 = 19%

If both characters are unknown, the maximum chance of success is .58 * .58 = 34%, but with the inclusion of kun-yomi, that again would be significantly lower.
There are a lot of variables and this isn't a bulletproof statistical analysis, but I think 35% is a pretty generous estimate.

And I will cite my statistics from the top of this page later tonight.

[rant deleted]

I wouldn't want any screen clutter in these threads!

Nice use of the rant generator. Now would you mind deleting it? It's cluttering my screen.

how did the rant-generator come up with this line? i'm afraid it perfectly describes the study section of this site...

jonzep's rant Wrote:He then boasts about how he'll project a stream of wishy-washy images of death, sex, disaster, material goods, celebrities, and other fixtures in a mock-Olympian firmament in the near future.

perhaps fabrice should quote this as a sort of warning-label disclaimer.

Woah, did you read the whole thing? :bugeyes:

JimmySeal Wrote:I'm trying, but it's hard when most of your posts are the length of encyclopedia articles.

If you don't want to actually take the trouble to read what I post, then why bother replying? Particularly since this is now the 3rd or 4th time you have done so and what you are posting does not reflect what I have actually said?

JimmySeal Wrote:The likelihood of guessing a character's on-yomi is about 58% (as shown in my statistics) so right off the bat, that is the maximum success rate for guessing a compound with one unknown character.

There's one critical flaw in the above estimation. 58% assumes you are trying to guess all 2000+ characters.

But if you are trying to guess the full Jouyou list that would imply you don't know any kanji at the moment. And if that is the case the success rate is actuallly 0% because you simply do not have the background to do the guessing.

As I've pointed out several times before, this technique works ONLY if you already know at least 1000 charaters. So all the exceptions that you are factoring into your estimations are irrelevant, since you already know those 1000+ characters you don't need to guess their meanings and readings.

the rant insertion probably wasn't such an appropriate move in this thread. but it did have its moments.

especially given the human ability to "read into things" even if they are totally irrelevant and randomly generated by nature, computers, or george bush, it was actually quite funny. and a momentary distraction from the heated discussion at hand. so, yes, i read the whole, rather interminably long, thing.

synewave Wrote:My understanding is that 革 is indeed a valid bushu （かわへん）. The kanji 靴 does have the onyomi reading か however the kunyomi is used in JimmySeal's example above.

Two points:
- the technique cannot be applied to kunyomi or any irregular reading so right off this would be an incorrect use of the technique.
- secondly, wouldn't have used the technique necessarily anyway since (assuming that the character is unknown to the applier, which would certainly be true in my case) in this case it breaks the principle of the phonetic marker generally being more elaborate and complex than the radical.

Furthermore:
- even if the technique was applied incorrectly, it doesn't matter. Remember, the context is being able to read newspapers (by oneself) without necessarily consulting a dictionary. In this case, assuming the wrong reading is acceptable. Just like if I'm reading in English and I come across a word that I have never seen before and pronounces it incorrectly, it doesn't matter. If I encounter it often enough, eventually I'll get it right (or someone will correct me).

Christine_Tham Wrote:ｔhe technique cannot be applied to kunyomi or any irregular reading so right off this would be an incorrect use of the technique.

I think this is what JimmySeal is pointing out.

Christine_Tham Wrote:secondly, wouldn't have used the technique necessarily anyway since (assuming that the character is unknown to the applier, which would certainly be true in my case) in this case it breaks the principle of the phonetic marker generally being more elaborate and complex than the radical.

There are quite a few cases where the bushu is more complex than the phonetic marker. My guestimate is that it runs into the 100s.

I'm still confused as to what your technique actually is. One minute you seem to talk about learning individual characters, the next, learning through context.

Both valid techniques (I try and do both).

If possible can you consisely tell us about how this "techinque" is supposed to work as well as provide some examples to illustrate its application.