The thorny issue of fonts

679 views
Skip to first unread message

Mikhail Skovoronskikh

unread,
Nov 22, 2023, 1:01:05 AM11/22/23
to pm...@googlegroups.com
Dear colleagues,

Recently I have been puzzling over a seemingly marginal issue with the potential to not only cause quite a few pesky problems, but also challenge the ways we think about "language" in our field.

Everything began with my dissatisfaction with the (relatively) new Yu Mincho font, which now seems to have become the default Japanese font in MS Word. To be honest, I still fail to see the need to revamp the good old MS Mincho, but this is really beside the point. However, having to constantly switch between CJK fonts when inputting characters in Word has made me realize that neither Yu Mincho nor MS Mincho is ideal for working with pre-1945 texts. I am not a specialist when it comes to Unicode, but some characters in the Mincho family don't "translate" well (or at all) into other CJK fonts: it will suffice to bring up the example of 説 vs. 說, which must have peeved many of you at least once. Sometimes these seemingly insignificant graphic variations can cause a lot of trouble when it comes to searching databases, changing document formatting, or even printing things out. Mutual CJK intelligibility suffers as well, at least in the digital realm. For example, plugging in the "wrong" Mincho character into Korean IME may not work; conversely, Japanese IME may not be able to process a Batang hanja.

There is an additional factor that makes things even more complicated. As you are well aware, we are often encouraged to use different fonts for "Chinese," "Japanese," and "Korean" when preparing manuscripts for publication. Curiously, I do not recall any injunctions to this effect in the Monumenta style sheet, but the Harvard Journal of Asiatic Studies clearly states, "[t]he preferred Chinese font is PMingLiU, Japanese MS Mincho, Korean Batang." However, rules like this can turn highly problematic when it comes to Chinese characters alone. The three font "families" mentioned above are not fully interchangeable, which can make formatting a nightmare. As someone who works primarily with texts in literary Sinitic, I am faced with a dilemma: Do I need to put my text in MS Mincho if it was written in Japan despite the fact that MS Mincho is incapable of rendering  and ? What about something written by a Japanese author in China? What about a Korean in the Tang or a Parhae envoy in Heian? The "supravernacular" (please pardon this coinage) nature of Sinitic--where in many cases the very question "in what language is this written?" is counterproductive--strongly militates against the need to choose a particular font for "Chinese," "Japanese," or "Korean" (for some reason, Vietnamese is rarely mentioned in this context).

Now, to my question. Given the issue I have outlined above, could anyone suggest a satisfactory solution, at least to the more practical part of my query? What is the best way to ensure CJK mutual intelligibility (as well as esthetic consistency) and maximize "searchability" in digital files? Should we all simply stick to, say, PMingLiu when citing Sinitic texts regardless of their origin? Or should we still use different fonts when quoting Sinitic pieces by someone like the Tang warlord Huang Chao, Silla literatus Ch'oe Ch'iwŏn, and Sugawara no Michizane in a single article? I would very much appreciate your comments and thoughts on this.

Best regards,
Mikhail Skovoronskikh

Howell, David L

unread,
Nov 22, 2023, 7:26:01 PM11/22/23
to pm...@googlegroups.com
Dear Mikhail and all,

The font question is an interesting puzzle, but it may be best not to overthink it. Yes, I know the irony of this advice since I'm the editor of HJAS, home to the most overthought style sheet in Asian studies.

At HJAS we ask contributors to distinguish among fonts used in the rendering of terms in Chinese, Japanese, and Korean (the issue hasn’t come up in Vietnamese during my nine years as editor) out of a desire to remain true to the subtle differences among fonts used in the various languages, even when displaying the same traditional characters. During the editing process we encounter various problems that add time and hassle, but we haven’t found a one-size-fits-all font. In general, we follow the font of the country in which a text was produced—so a Japanese text in kanbun would be rendered in a Japanese font (MS Mincho) rather than a Chinese one (PMingLiU). I should say, however, that the fonts we request for manuscripts aren’t the ones that appear in the published version. Our aim is to maintain the CHJ distinction rather than focus on the particular fonts per se. Incidentally, few contributors bother to honor our style preferences, but if someone felt strongly that everything in Literary Sinitic should be in a Chinese font, I wouldn’t object. It’s never come up in real life.

If you are writing for publication, the publisher will decide on the font. I’ve never been asked about font choice, which is left to the book designer or the journal’s compositor. If you are writing for yourself (e.g., a dissertation, conference paper, etc.), I’d say you should do whatever brings you joy. If Yu Mincho annoys you, set your default to MS Mincho. (I’m a Hiragino Mincho user myself.) So long as you’re consistent, no one should mind, even among the few who notice.

Best,
David Howell

David L. Howell
Robert K. and Dale J. Weary Professor of Japanese History
Chair, Department of East Asian Languages and Civilizations
Professor of History
Editor, Harvard Journal of Asiatic Studies
Harvard University



--
PMJS is a forum dedicated to the study of premodern Japan.
To post to the list, email pm...@googlegroups.com
For the PMJS Terms of Use and more resources, please visit www.pmjs.org.
Contact the moderation team at mod...@pmjs.org
---
You received this message because you are subscribed to the Google Groups "PMJS: Listserv" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pmjs+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pmjs/CAK9hvFsAp2PF145aVyYKeGR78Sk8uu33UgTMP9_%2BK4rxxBCrVw%40mail.gmail.com.

Ash Price

unread,
Nov 23, 2023, 1:27:41 AM11/23/23
to pm...@googlegroups.com
Dear all,

I think it unlikely that a font for everything exists. I would stick to a style guide, until I run into characters that are not supported - then I would override the style guide and tell whoever is relevant why I have done so. It is fine to simply change the characters for some situations where a variant is an acceptable representation of another, but if you're doing any kind of historical language work or treating original sources, I would be relatively insistent. 

If your Japanese or written-in-Japan Chinese text has characters unsupported by the style-guide's font (for that language), I would use a font in the style guide that has those characters for all such text of that language if possible, even if it is not the font they originally specified for that language variety, with a note. Failing that (and there are no fonts in the style guide with adequate coverage), I'd request the usage of a different font altogether that has the relevant characters.

When I am free to choose, my process is: first, I consider what characters I need rendered, and I pick fonts to suit that. If there is a font or group of fonts that maintain a consistent look across CJK and beyond, great. If not, I try for pairings that are as little jarring as possible. If I don't have to worry about pairings, I just go for aesthetic preference, but it's rare that I both get to pick and don't have to think about pairings!

If you are looking for fonts with broad coverage, the Noto CJK (previously Adobe Source Han) set of fonts offers the advantage of wide coverage while allowing for a uniform look - though that look may not be the nicest - with variants for several widths and all of serif, sans, and monospaced.

Missing from Noto are a number of historical CJK characters, for which the go-to I would say are the Hanazono fonts, which have a rather sizeable coverage. See here: http://fonts.jp/hanazono/ . The rendering of these fonts is not as good as some more recently developed fonts, though.

A final thought is that the environment within which you write may increase or decrease your pain, depending on how it works. I haven't used MS Word in a long time, but for example, if a font has two variants for a single unicode point, can it handle picking the correct one without much pain? Ideally, the answer to that question would be yes!

Best regards,
Ash Price.

Alexander Zapryagaev (萌覺・Miǒgacu)

unread,
Nov 23, 2023, 9:57:22 AM11/23/23
to PMJS: Listserv
Everyone,

as somebody who thinks about fonts a lot (and sometimes draws missing glyphs by hand), here is some of my experience.

The only way to solve the problems of "showing all characters in a single style" and "maintaining the CJKV local differences" at the same time is employing a pan-CJKV font family. There is basically only one in existence: namely, Source Han Serif/Sans/Mono (which is also branded as Noto CJK, but I prefer the Source originals, as they are developed first and the Noto copies may be late in updates). They are lacking in trickier characters but perfectly sufficient for the core usage in the regions of Japan, Korea, Mainland China, Taiwan, Hong Kong (no Vietnamese version or a Kāngxī-esque Classical version is sight). Note that most of the versions of these fonts actually expect you to mark the text with a correct language in your text editor to show the correct glyphs (to have guaranteed regionally-correct forms in the most basic of situations, only the "Subset OTF" forms are fit).

Hanazono (I strongly, strongly recommend the Hanazono Mincho AFDKO Ex A1, Ex A2, Ex B, and Ex C fonts, available at https://github.com/cjkvi/HanaMinAFDKO) are basically the merge of GlyphWiki glyphs, made in 2018. They do not cover Extensions G, H, I, or any glyphs added after 2018 to earlier extensions, but for all earlier, they are likely the only option to get all the Unicode characters, even though not really appealing in their form. A very winning point for them is being the only font series supporting all the Ideographic variation descriptions; say, these are probably the only way to get all the 21 versions of 邊 encoded in Unicode, from U+908A U+E0100 down to U+908A E+0114!

For a more specifically Japanese usage, it is highly improbable to ever need anything beyond the font called IPAmjMincho (https://moji.or.jp/mojikiban/font/). It is the font connected with the Moji-Joho database, so it is basically guaranteed to contain all characters with a registered Japanese usage in any databases. Surely there are characters connected with Japan that are not included (say, I requested nine characters appearing mostly in the 西本願寺本 edition of the Man’yōshū, two of which are now pipelined to the Extension J, the rest to the Extension K - obviously, none of those are in the font, as they are not in Unicode yet), but most are - and it also has hentaigana and allows attaching dakuten to it, which makes it a very comfortable choice to pick as THE font for anything Japan related.

As a postscript, some non-Japan related fonts I think worthy of attention:
- Andrew West's BabelStone Han, an extremely large font containing characters in proper PRC forms, constantly extended and improved;
also has a large private-use font supplement BabelStone Han PUA of characters not yet in Unicode;
- I.Ming (https://github.com/ichitenfont/I.Ming), the best font for Classical Chinese, fulfilling the Kāngxī-esque form demands in the most meticulous details possible (only one character currently missing to have the complete coverage of Kāngxī itself);
- The CNS fonts (https://www.moedict.tw/fonts/truetype/cns11643/), TW-Kai, TW-Kai-ExtB, and TW-Plus. They cover only up to Extension D included (though almost exhaustively), but also provide (in Plus) a staggering 22000+ characters not encoded in Unicode in the shared regular style, probably the best option if kaishu forms are needed for rarer characters;
- Nom Na Tong (https://github.com/nomfoundation/font), the absolutely definitive font for Vietnamese Chữ Nôm.

Best wishes,

Alexander Zapryagaev (萌覺)
четверг, 23 ноября 2023 г. в 09:27:41 UTC+3, Ash Price:

Hanna McGaughey

unread,
Nov 23, 2023, 7:25:34 PM11/23/23
to pm...@googlegroups.com
Dear Mikhail, dear all,

While I doubt my suggestion will solve the issues raised, I would love to mention a font developed by a group that is headed by Kobayashi Tatsuo, a fonts specialist who started with Shōgakukan and included a directorship at Unicode. The IMAmjMincho font covers the relatively recently (2019) approved Unicode code points for hentaigana, a selection which we might also argue over, but before we do, there exists a review article by Okada Kazuhiro that points out a few of its weaknesses:

Okada, Kazuhiro. “Unicode 10.0 ni okeru hentaigana shuroku shūroku.” Kanji bunken jōhō shori kenkyū 17 (January 25, 2018): 174–79.

In any case, it is possible to download the IMAmjMincho font through a third party site linked from the creators' website:


While I will not vouch for the reliability of the third party site, I have yet to experience problems with the font or the download two years in. Rather, I have been quite satisfied with the integration of the hentaigana in the Mincho font family, although inputting them (on Mac) still requires either copy and pasting or specifying Text Replacements in the Keyboard settings, which has been a real time saver.

Warmly,

Hanna

Mikhail Skovoronskikh

unread,
Nov 27, 2023, 1:26:06 AM11/27/23
to pm...@googlegroups.com
Dear all,

Thank you very much for your wonderful comments and suggestions.

I especially appreciate all the links to alternative fonts: I never realized how many additional CJK fonts there are!

Best regards,
Mikhail  


Siva Kalyan

unread,
Nov 27, 2023, 5:44:02 AM11/27/23
to "'ノット ジェフリー' via PMJS: Listserv"
Yes, thanks to Alexander for the incredibly useful font links! It’s good to know that there are better options than Source Han/Noto.

Siva

Ash Price

unread,
Nov 30, 2023, 2:42:10 PM11/30/23
to pm...@googlegroups.com
Dear all,

I just wanted to write a late reply to say thanks to Alexander, much of this information is new to me, and was far more helpful than my initial reply by far.

It's good to learn as well that Adobe Source Han is still under development and is actually farther ahead, as well. But I wonder, is there somewhere other than the github repositories to follow development?

Wishing you all the very best,
Ash.

Reply all
Reply to author
Forward
0 new messages