斈倉頡 Learning Cangjie

[amhoanna]

1. Firstly, getting it out of my head that the 倉頡 Chong Kiat components are not necessarily (and in most cases, are not) the 康熙 traditional radicals themselves.

2. Applying the rule of maximum inclusion, i.e. identifying as large a chunk within the character as possible.

These two are related. It kind of surprised me at the start too, that Chongkiat didn't break things down according to the principles of calligraphy. Pretty counterintuitive.

3. When the character starts getting complicated, when I can apply the skipping rule, i.e. identify the key first couple of blocks, and then jump to the last one (in order to stay within the constraint of maximum 5 alphabets per character code).

I think I know what U mean. Some characters break down right-left, top-bottom, or both. Other hanji are treated as one unit, where the Chongkiat code is usually the first three elements plus that last. It can be hard to tell beforehand which it's gonna be. Sometimes U just have to try both and see what works!

Another thing to keep in mind is that different editions of Chongkiat have different codes for a few handfuls of hanji. The most expanded version of Chongkiat that I've seen takes in all versions.

4. In general, being creative with breaking the characters down and 'seeing' the blocks.

U'll probably do great once U get issues #1 and #2 out of the way. Learning Chongkiat is kind of like learning how to write hanji. Everything adds up.

U might want to try typing up part of the 三字経 or some other writing and see which characters have U stumped.

[Yeleixingfeng]

Not to change the topic, neither to diversify Mark Yong's option in stroke-based input methods - people often get distracted when alternatives are presented. It makes you doubt your choice. >.<

Is there a learn-Cangjie pdf online? I have been using Zhengma for so many years, and since the only other alternative input method I have is pinyin, I find Zhengma rather efficient. Nonetheless, I would want learn more stroke-based input methods, because I am trying to compare between them and conclude which would be the most suitable for new-learners, for example, to be taught in school.

This is ultimately because, if everyone types in stroke-based inputs, no one would ever say typing in Chinese would be a chore!!

[Mark Yong]

amhoanna wrote:
I think I know what U mean. Some characters break down right-left, top-bottom, or both. Other hanji are treated as one unit, where the Chongkiat code is usually the first three elements plus that last.

Actually, what I was referring to were those characters with a large number of strokes (by that, I mean in excess of ~15-16) where the first three (3) blocks of the code are already consumed by just the left-hand side of the character alone (or, in some extreme cases, just the top-left corner), leaving you with only 1-2 alphabets left to complete the residual 50-75% of the character. In those cases, I recall that you are allowed to 'skip' the top-right corner (and even the bottom-left corner, if it has not already been constructed) and jump straight to the last block (which, in 99% of the cases, would be the bottom-right corner). You know what I am referring to?

Yeleixingfeng wrote:
Is there a learn-Cangjie pdf online?

This is the only website I am aware of that has some PDF's on 倉頡:
http://cocoanutstech.com/cang_jie/www.c ... method.htm

[amhoanna]

Well, not sure what U mean. If a hanji breaks down into left and right sections up front, then at most two strokes will be allotted to the left side, and at most three to the right. 囍 = GRGRR.

The first-three-and-last characters can be a tricky set. B/c of the nature of hanji, these are usually "tall" characters. Sometimes it's hard to tell whether a character should be top-and-bottom or first-three-and-last. 哥, for example, is MRNR (first-three-and-last), not MRMNR (top-and-bottom).

What kind of hanji is 剪 TBNH? Damned if I know. If it were top-and-bottom, with the split btw 前 and 刀, it should be TNSH. If it were first-three-and-last, it should be TBLH...

Wait, no. It is first-three-and-last as TBNH. Why? B/c in CJ, the 亅 in 前 actually comes "before" the 丨. Why? B/c it's higher up.

I may've just demonstrated that it's easier to be fluent in CJ than it is to understand it. CJ is actually not as romantic as it may seem. It's at odds with the art of calligraphy. I'm sure the inventor of CJ thought all of this out... This was probably just a more economical way to get to every kanji in five strokes.

Do U have any examples of the problem U described?

[Mark Yong]

Yes, I do. Here are a few examples:

1. 廣. The code is ITMC, the respective blocks being 土廿一ハ. So, in this case, the 由 component in the centre was ‘skipped’.

2. 將. The code is VMBDI (sorry, can't generate the blocks here). But the centre vertical stroke was ‘skipped’.

3. 制. The code is HBLN, the respective blocks being ノ冂|亅. One ー and one | were ‘skipped’.

I have a feeling I am not getting some fundamentals down pat in the first place.

[amhoanna]

廣

At the character-level, this follows the outside-inside pattern. The outside gets a max of two keys. The inside gets up to three.

The rest is just familiarity with the forms. 广 is YH, but in every other hanji, the form 广 is called up with just one key: 戈, or I. So it's yet another manifestation of 戈. So here we take care of the outside part, 广, with just one key, I.

The inside, 黃, is TMWC: 廿一田ハ。CJ doesn't analyze it as 廿一由ハ, which would be 廿一中田金. This is just something that CJ users have to know. It's possible that the latest versions of CJ allow for both analyses.

Since the inside is capped at three keys, we take the first two and the last, TMC.


將

This is a right-left character: up to two keys for the left side, up to three for the right.

爿 is VLM. We take the first and last, VM. The vertical stroke was skipped b/c the left side only gets two keys.

If I'm not mistaken, in calligraphy, the stroke-order is actually more like VML. But CJ doesn't go strictly by stroke order. Also, I think, but I'm not sure, that CJ could've -- but doesn't -- analyze 爿 as LVM. Why? B/c in many fonts, the vertical stroke on the right "sticks up higher" than anything else in the character. But CJ analyzes it as "same peak height", so we go from left to right. Don't worry too much about this...

寽 is exactly three keys. No need to cut anything out here.


制

Another right-left character.

CJ analyzes the left side as either HGLB or HQB. I don't know which b/c it's not relevant to anyone that's still working through their first 20,000 kanji. Since the left side only gets two keys anyway, we take the first and last: HB. The middle key or keys are skipped b/c there's only room for two, in the CJ scheme of things.

On the right side, 刂 is simply LN, with left-right taking precedence over high-low (w/i 刂 itself).

[amhoanna]

Hope this helped. It's late, and my jaw is killing me.

[Mark Yong]

Most certainly helped, thank you. I guess I just need to remember the ‘max 2 left, max 3 right’ rule now.

With all the skips and subjectivity, it amazes me how 倉頡 Chong Kiat can still maintain a strict ‘one-character, one-code’ feature. Or unless the reverse is not true, i.e. one code could map to several possible characters?

I am going to assume that you work off a Chinese keyboard, or at least one that has a 倉頡 Chong Kiat overlay on it. Would be a b*tch to memorise which block corresponds to which key (I only remember A to G, for obvious reasons).

[amhoanna]

Computers sold in TW and probably HK too come with the Chongkiat elements printed on the keys, but I don't rely on that. Unless U can count on always using your own computers, U might not want to either.

U coulddo one of two things. U could forget about the Chongkiat elements and just remember everything by the letters, by the ABCs. Or U could associate each Chongkiat element to its key. This is what I did from the start. Here are some mnemonics that "hardwire" the Chongkiat elements to their letters in my mind:

火 F; Canto fó; English fire; Siamese fai (I didn't know any Siamese when I learned CK, but I think CK helped me to remember the Siamese word)

土 G; Eng ground; ground floor;

竹 H; H for Hsinchu 新竹, city in Taiwan;

十 J; something to do with the word for TEN in Japanese;

大 K; really no mnemonic;

中 L; no real mnemonic, but lowercase l corresponds to one of the forms of 中;

月 B; bulan

人 O; orang

廿 T; twenty

手 Q; "qiu", using Pinyin to write the Hoklo word

心 P; no real mnemonic; it's just kind of hardwired...

弓 N; the word 弩

女 V; sexiest letter in the Latin alphabet; va-va-voom vixen; Malagasy word vehivavy, prob. cognate to Tagalog babae, meaning WOMAN

U'll have to come up with your own...

For me, 卜 is Y, period. There's no two ways about it.

As for 1-to-1 mapping, there are hanji, but very few, that have alternate codings in different editions of CK. Meanwhile, most codes pull up multiple characters, maybe as many as 8. If U restrict your CK to just Big-5 characters, though, and keep the expanded character sets turned off, which is probably what most users do, there's not that many repeats. 么 and 鬼 are one, 死 and 恐 another. 新倉頡 works in real-time to detect the context of each hanji. For example, just now I was trying to type 艸源. It came out as 出鴻. As soon as I adjusted the second char to 源, CK automatically corrected the first to 艸 -- not sure why.

FYI just now I went in my CK options and checked some boxes. Now once again I have the huge character sets with chars like 𤆬, 𢶀, etc.