On Mon, 10 Feb 2003, Nicholas Bodley wrote:

> Slightly off-topic for this message, but it seems to me that the
> mostly-unconscious design of CJK characters is to make them as easy to
> differentiate as possible, to minimize misreading confusion.
> Nevertheless, I'm aware that there are some detailed counterexamples,
> which people new to them need to be careful of.

One book about that is Yaeko S. Habein and Gerald B. Mathias' _Decoding
Kanji: A Practical Approach to Learning Look-Alike Characters_ (Kodansha
International, 2000). I've only seen the book in stores, but to some
degree it must be biased towards Japanese usage--not only particular forms
of characters used in Japan, but also taking into account characters that
have similar readings in Japanese. (The most difficulty results when
appearance, reading, and meaning are similar, e.g., U+66C6 and U+6B77.)
What's difficult will of course depend on your ability--e.g., beginners
might not pay attention to small details like hooks or dots.

> Btw, just for fun, what's the largest stroke count Qalam people
> know of, for a CJK character? In my infrequent perusals of
> dictionaries, I don't think I recall seeing as many as 45, and
> perhaps not even 40, but am not sure.

Typically the answer is the 64-stroke character (U+2A6A5) that looks like
four (traditional) 'dragon' (U+9F8D) arranged in two rows of two,
'garrulous', but smaller dictionaries may only include up to the 48-stroke
U+9F98 (itself just a pyramid of three 'dragon'), 'appearance of a dragon
flying' (I believe this is the limit of Nelson's).

However, not as widely accepted is the much larger 84-stroke character
shown at the bottom of http://nohara.u-shimane.ac.jp/ekanji/newfonts.html
, which is for a Japanese surname Taito, collected by NTT. But HIDA
Yoshifumi and SUGAWARA Yoshizou's _Kokuji no jiten_ (Tokyo: Toukyoudou,
1990) says that it is a fusion of two characters tai (the top half, the
pyramid of three 'cloud') and tou (the bottom half, U+9F98 mentioned
above) (176), i.e,. tai + tou = taito. (You decide if you want to think
if its really one character or two.)

Realistically, among characters in actual use, it might be the 29-stroke
U+9B31 'melancholy' (which is more complex than the above). Some of the
others in the low-30s stroke count might also be used, but many of them
have variant characters with less strokes that may be substituted instead,
so their actual use and necessity is low.


Thomas Chan
tc31@...