Same Kanji is being rendered differently

[Kurious]

Hi, I have observed that the character 社 is being rendered differently on my computer, depending on the font. Is it possible for any of you to check if you have the same behavior? If so, are these differences normal and acceptable; or is this some sort of computer bug? Not only are there thousands of Kanji to memorize, it now turns out that there is more than one way to write the same!

Details: I open OpenOffice Writer (equivalent to Microsoft Word, which you can use), and copy and paste that Kanji in the document. The following fonts render differently than the rest (but I only checked A through M):
Batang
Dotum
Gulim
Gungsuh

Of 礼, 社, and 神; only 礼 renders the same on any font; whereas 社 and 神 render differently depending on the font. (I chose these characters because they share the same radical on the left.) I noticed this because I use Opera web browser, which uses different fonts than Internet Explorer. (If you have both browsers, you can try opening this web page in each.)

Any feedback is appreciated.

EDIT: Add text "I chose these characters because they share the same radical on the left."

[richvh]

At a guess, the fonts you have issues with are Chinese fonts. There are differences between Japanese kanji, Traditional Chinese hanzi, and Simplified Chinese hanzi.

[clay]

Can you include a screenshot to show us what you mean? My guess is you are seeing 'cursive' (stylized) Japanese. If that is the case, you don't have to worry too much about it. Books, newspapers, online, etc will use easy to read fonts.

See here for examples

EDIT: Changed link from a tattoo site to a non-tattoo site

[richvh]

The left hand radical is being rendered in its traditional form: 示. I suspect that 礼 is unchanged because it isn't even used in that font, and it's being substituted in from some other system font (MS Arial?), note that the Batang 礼 is in a completely different style than the other Batang characters. I would suspect that this is a Traditional Chinese rendering.

[JaySee]

Those four fonts are Korean though, I'm not sure how many Chinese characters are included in those - not so many by the looks of it. But seeing that in Korea no character simplification took place, if there are any included they would (more or less) be the same as the traditional Chinese variants.

I think MingLiu and SimSun are the standard Chinese fonts that come with windows XP, so if you want to compare then it might be better to use those.

[richvh]

Simsun and MingLiu added to the comparison.

[Kurious]

OK, it makes sense that the versions using 示 are Traditional Chinese. The question now is, do they mean the same? If they had different meanings I would guess that computer systems would have separate characters instead of making them font-dependent. And if I used the 示 versions for Japanese writing, would the Japanese understand it?

Wow, I didn't know about gyosho and sosho. I can see that learning Japanese will take longer than I thought... bring it on! ^.^

[furrykef]

What we have here is something called "Han Unification". This is a little complicated to explain in a non-technical way, but I'll give it a try. I suppose I'll start at the very basics just to be sure.

All text in a computer is really a bunch of numbers, because numbers are the only thing that computers really understand. For example, in the vast majority of character sets ever used, the capital letter A is really the number 65. Every symbol is assigned to a number so that it can be stored in the computer. Somewhere, sometime, some guy had to decide what number went with which symbol. The first popular system of doing this, called a "character set", is called ASCII, and it's what all modern character sets are based on. Unfortunately, though, ASCII only defined 127 symbols, which wasn't nearly enough, so everybody modified ASCII in their own way so that people can type letters like é, ö, and of course あ, etc.

This whole situation became a mess, because you can't use two character sets at once, so making two programs or computers using different character sets talk to each other was a big pain -- the number 128 could (hypothetically) mean á on one computer and あ on another, and there might be no way to reconcile them. So some guys got together and created this big character set called Unicode, which essentially defined a number for every single character that is possible to type.

This is where Han Unification comes in. "Han character" is just a language-neutral term for kanji/hanzi/hanja. Now, as you're aware, different ways of writing Han characters appeared: Traditional Chinese, Simplified Chinese, Korean, and Japanese. They realized that if they created a different character for every single possible form, there would just be way too many of them, so they decided to "unify" some of them: the same number will print a different version of the character depending on the font you use, because they're essentially the same character, just written differently depending on the language. And, naturally, that's what causes the situation you see there.

Note that not all variants of a character have been unified. For instance, 竜 and 龍 are still different characters, even though they're identical in pronunciation and meaning, as are 学 and 學, although, in both cases, generally only one of the two will be correct for a personal or place name. Unification has affected mostly radicals, like the 礻 radical in question here, not simplifications of individual kanji.

Oh, by the way, you'll often hear the term "encoding" used. Encodings are a similar idea to the "character sets" I was talking about earlier, although sometimes one character set has more than one way it can be encoded. For the purposes of this discussion, you can consider them synonymous, since it involves the same idea: it has to do with how numbers are translated into symbols.

- Kef

[Kurious]

furrykef, thank you so much for the explanation. Now, I don't know how you will react to this, but I'm a programmer myself. O.o /me hides/ So I do know about Unicode and character sets in general. I didn't know about Han unification though; nor had I heard about hanzi or hanja. It's good to know all of this. I still wonder, if I used the 示 versions for Japanese writing, would the Japanese understand it? I wonder if they use these variants (示) themselves.

I now noticed that if I select "View -> Encoding -> Japanese -> Automatic Detection" then I get the appropriate Japanese rendering, however, if I select "View -> Encoding -> Automatic Selection" then it doesn't. I suppose that because of Han unification, Opera (nor any other program) could never perfectly guess the right encoding. Maybe they should adopt the same defaults as Internet Explorer.

[Yudan Taiteki]

Because of context, small variations (or even big ones) in characters will generally not cause comprehension problems for native Japanese, but I don't think anyone except maybe old people write the left hand side as 示.

[furrykef]

furrykef, thank you so much for the explanation. Now, I don't know how you will react to this, but I'm a programmer myself. O.o /me hides/ So I do know about Unicode and character sets in general.

Hehe. Well, hopefully it's proved enlightening to someone out there.

Oh yeah, I might as well say that richvh's guess is correct: if a character does not exist in the font you've selected, the system will fall back on another font to try to render the character, which is why his images display inconsistent fonts. I'm not sure how the fallback mechanism works, though.

- Kef