The Roman typefaces used in Chinese and Japanese text | Hacker News Reader

The Roman typefaces used in Chinese and Japanese text

bellisk.blogspot.com

49 pointsmzehrer33 comments

Spotting the particulars of what people native to a different script do when using another is one of my hobbies.

- East-asian students will have a hard time spotting typos even in important words as their own name

- Even on designs with the most extravagant typography choices for their native fonts, choices for roman fonts are not prioritized

- "Letter people" do the same thing the other way around ("OK, we got a nice asian feel to this poster ... now to add some cool chinese characters... done") It's exactly the same feeling for people used to seeing asian scripts, to see the basic printed version everywhere and the same old standard fonts.

Next time you make something using a foreign script, please keep in mind that how you're looking at it != how natives look at it!

For more on different traditional script styles see http://en.wikipedia.org/wiki/Chinese_script_styles However, the best information on font choices for Ch, Jp and Kor is in the languages themselves, unfortunately.

Can you give some examples of the boring standard font vs. more sophisticated Chinese font faces?

I doubt. In Japan advertisers aren't playing with a ton of different interesting fonts the way we do in the west. Rather they they play with thickness and add outlines and colors, simply because building a font is expensive, so there aren't that many to choose from, so they focus on a different approach to differentiation.

Related -- Due to Han unification[1], many glyphs you see on the screen in different languages share the same Unicode codepoint. The problem with this is that you no longer have an easy one-to-one mapping to have certain languages render in a desired font. It just so happens that Chinese business users prefer a much more traditional font than Japanese users. This doesn't necessarily affect "documents" that can have a defined language, because the language selection will be taken into account when the font is chosen. When you are writing software, however, that can mix any/all available languages on one screen (e.g. displaying a list of news headlines or even tweets), the programmer has to make a hard choice whether to do all the extra work to send down an explicit language tag (if one exists) next to each group of text elements that share the language so the font system can choose the correct font for a codepoint such as U+5168. If that work is not done (or parts of the pipeline don't carry that information along), the best you can have is a global setting that users can pick "I favor Japanese fonts" and then a Chinese headline could wind up rendered in two fonts -- all of the unified Han codepoints rendering in a much more modern looking Japanese font.

edit: [1]: https://en.wikipedia.org/wiki/Han_unification

I couldn't find the exact page the author used, so I took a screenshot of that "ancient virus" story instead. This[1] is how it looks on a standard English Ubuntu 14.04 system (no tweaking and no extra Chinese fonts installed). The English in it doesn't look half bad; there's even kerning!

As an aside, if you look at the top of my screenshot, you'll see a little lyric display on my taskbar. I use it to learn Japanese while I listen to anime songs. I wrote it in Python3 and there was not a single instance where I had to bother with Unicode or encoding issues.

   with open(file) as f:
       lines = f.readlines()

works as is on Japanese files; I don't even have to supply the encoding. PyGTK and Gnome handles the Japanese text perfectly when I pass it to them. I'm throughly impressed by how far we've come in terms of Unicode support in programming languages, libraries, and the OS.

[1]http://i.imgur.com/rLNVchh.png

ｉ　ｗｏｒｋ　ｉｎ　ｃｈｉｎａ，ｆｅｅｌｓ　ｌｉｋｅ　ｈａｌｆ　ｍｙ　ｊｏｂ　ｉｓ　ａｄｄｉｎｇ　＆　ｓｕｂｔｒａｃｔｉｎｇ　ｓｐａｃｅｓ　ｆｒｏｍ　ｔｈｅ　ｓｔｒａｎｇｅ　ｔｙｐｅ　ｆａｃｅ　ｈｅｒｅ．

Ｉ　ｆｅｅｌ　ｙｏｕ．

Unicode-wise, "ｗ" {U+FF57 FULLWIDTH LATIN SMALL LETTER W} is the composition of the compatibility tag "<wide>" and {U+0077 LATIN SMALL LETTER W}. You can get rid of the compatibility tags by using NFK* (normal form compatibility):

    >>> s = "ｉ　ｗｏｒｋ　ｉｎ　ｃｈｉｎａ，ｆｅｅｌｓ　ｌｉｋｅ　ｈａｌｆ　ｍｙ　ｊｏｂ　ｉｓ　ａｄｄｉｎｇ　＆　ｓｕｂｔｒａｃｔｉｎｇ　ｓｐａｃｅｓ　ｆｒｏｍ　ｔｈｅ　ｓｔｒａｎｇｅ　ｔｙｐｅ　ｆａｃｅ　ｈｅｒｅ．" 
    >>> import unicodedata
    >>> print(unicodedata.normalize('NFKC', s))
    i work in china,feels like half my job is adding & subtracting spaces from the strange type face here.

It doesn't actually have to do with typefaces, the typeface is just the system trying to find a font able to display those specific glyphs, which are compatibility forms for dot-matrix printers and fixed-width terminals: http://en.wikipedia.org/wiki/Halfwidth_and_fullwidth_forms

I'm guessing their continued use is because there are also aesthetic considerations at play for native readers, for whom variable-width latin script could look plain weird.

Me too ,I know how you feel。

Note to HN readers: the final display is also relevant to your system's installed fonts and rendering priority settings.

Op's screenshot is also different from what normal Chinese user could see.

I can't find op's screenshot of the exact linked page, but here is my on rMBP+Safari screenshot anyway

http://edu.qq.com/BBC.htm

http://i.imgur.com/s7OCT2S.jpg

The screenshot looks like a mobile version for a feature-ish phone. Which could explain the use of fullwidth characters (expectations of encoding or hardware limitations)

The parallel that springs to mind is that they look very similar to the NLQ fonts ('Near Letter Quality') used by dot matrix printers in the 1980s/1990s - and made, for the most part, by Japanese manufacturers: Epson, Star Micronics, etc.

At the time, NLQ was a passable imitation of real typesetting. But somehow it looks like that style has become fossilised as "how you write Roman characters".

Part of the reason for the “weirdness” is the blatant fact they’re fixed-width, which is hugely important as so are kanji. That way you can intermix the two and the spacing won’t completely fall apart.

Monospace fonts are fixed-width too and they don't look bad. The thing is that they're full-width.

> I assume this is similar for Japanese and Korean

FYI Korean script (Hangul) is actually an alphabet, so relatively easy to type.

Still, the topic of font choices remains interesting. While we have been accustomed to very differently looking typefaces for Roman scripts for a few hundred years, we know relatively little about CJK fonts. Does anyone know a good article describing Asian font choices in English?

It's not just "symbols" - there's different ways to draw these lines, too, taking into account the purpose of the design.

This article draws some parallels between English and Simplified Chinese fonts:

http://www.kendraschaefer.com/2012/06/chinese-standard-web-f...

It's more of a modular syllabary than what most people would think of as an alphabet; while the individual elements are alphabetical in nature, they're grouped into metacharacters that function as syllabics. That difference may not matter much when writing/typing, but it will matter when reading.

Not really. Hangul is perfectly readable linearly letter by letter (where a letter is a sub-element of a block/metacharacter), and the use of the null consonant effectively allows spanning syllables over more than one block, so they need to be read continuously. However, the orthography isn't purely phonemic but rather morpho-phonemic and the morpheme boundaries happen to line up with block boundaries a lot - but in some sense most applications of the Latin alphabet aren't any less abstract; English orthography is anything but consistently phonemic after all. The blocks also aren't purely phonemic, either, since e..g verb conjugation postfixes can often attach a final consonant to a stem's last block.

(My post history has some longer, less dense comments on Korean.)

> Hangul is perfectly readable linearly letter by letter

Here's an example of printed linear Hangul:

http://imgur.com/yasWNfW

Neat :)

I remember there was a "Linear Hangul" movement around the middle of the 20th century or so to do away with the blocks entirely, but it didn't catch on.

I'm kind of glad it didn't because I think the blocks forcing the retainment of more morphological information is actually a really cool trait of the writing system. Basically, you sometimes have multiple options for how to distribute letters over blocks when writing down a word, and the ortography then fairly consistently prefers the variants that keep the same morpheme spelled consistently across different words. It promotes morpheme reuse and makes the writing more LEGO-like.

Purely phonemic spelling is sort of overrated. Hangul lets you write phonetically if you want to (though it's more rigid than Latin because of the block grouping forcing certain consonant/vowel patterns) but for practical use of an established language there's many other concerns.

> FYI Korean script (Hangul) is actually an alphabet, so relatively easy to type.

Japanese is typed using kana, though they're syllabaries and have about twice the number of characters as hangul, the end-result is mostly similar (so is the on-the-fly transformative effect, in japanese kana entry is transformed into kanji OTF, in hangul jamo get merged into natsori).

Japanese can be typed as kana, but most people type rōmaji (Latin characters).

Most people on a computer. Most people type kana on their phone.

Yes, that's true. Japanese is, luckily, a language quite suited to cellphone number pad input.

The biggest problem might be compatibility, old computers (common in China) only contain fonts that have low quality Roman parts like SimSun (http://zh.wikipedia.org/wiki/%E4%B8%AD%E6%98%93%E5%AE%8B%E4%...) but this font is default Chinese typeface in Windows.

The newly designed fonts in recent years for Chinese have much better Roman parts which are already distributed by operating system or could obtained for free:

- Microsoft YaHei in Windows http://www.fonts.com/font/microsoft-corporation#product_top

- Hiragino Sans GB in Mac (http://blog.jjgod.org/2009/06/04/updates-on-font-changes-in-...)

- Source Han Sans free from Adobe (http://blog.typekit.com/2014/07/15/introducing-source-han-sa...)

Many Chinese websites have already using one of them as their default font. There are also a dozen of commercial Chinese fonts which contain higher quality Roman typeface (although could not catch up most classic ones).

And for professionals (like book publishers or designers) in China, they often combine one Chinese font and one Roman font in their product for better result. (e.g. FZNew ShuSong + Sabon for Serif).

It reminds me of how old German books written in Schwabacher used roman typeface for Latin and other foreign words. In this case, it was intentional, but I have always found it a bit irritating from the aesthetic standpoint. For example: http://en.wikipedia.org/wiki/Antiqua%E2%80%93Fraktur_dispute...

Similar to how we Americans put foreign words in italics, no? Does that look aesthetically irritating to others?

Ugh. I've done a lot of graphic design in Chinese, English, and bi-lingual documents and I absolutely hate that Roman typeface.

There are just far better fixed-width serif fonts out there and I always have to replace the English characters within Chinese documents with those fonts (I usually try to find one that matches in weight and blends in). Also not to mention a lot of English text within Chinese documents does not need to be in a fixed-width font to look good. There's a reason why there exist separate Unicode points for the fixed-width English characters when you do need them.

http://hangeul.naver.com/ is a Korean typography and font distribution site by Naver.

So, do speakers of these languages (maybe specifically in countries where this language is natively spoken) prefer these typefaces? As in, should I purposely be using these typefaces for "small snippets of English embedded in a page that is otherwise written in Chinese or Japanese" when I am optimizing content for these audiences?

A/B test it, it's the only way to be sure. I doubt most people can CJ(K) can even tell the difference. When proof reading Germans' English texts, one of the things that was regularly necessary was fixing punctuation to use English commas and apostrophes. I did this for people doing English degrees. Normal people will not notice this on any (conscious) level without training/education in it. It's like how you never noticed bad kerning before you knew what kerning was (typography, the gift that keeps on giving).

I did not specify whether the preference would be conscious or unconscious, and I disagree that me running the A/B test myself instead of asking if anyone else already knows the answer to the question already (maybe by having run experiments themselves) is the only way to be sure (well, OK, it is always the case that people I am targeting might be different than other people already studied, but at that point there is no longer any value in trying to learn things as a species at all, which is a pointless precision).

I like this breed of fonts, they remind me of the LaTeX "Computer modern" and Garamond of old.

Reply on news.ycombinator.com