Problems with Noto CJK for (Traditional) Chinese users (Part 2)
Noto Sans CJK and Noto Serif CJK has been developed with comprehensive coverage of Simplified Chinese, Traditional Chinese, Japanese and Korean into a single font family. However, there are two main problems with the actual implementation which affects the usefulness of Noto CJK for Chinese users. This article mainly highlights the problems faced by Chinese users, with additional information for Traditional Chinese users in Hong Kong.
Part 1: Character Set Coverage
Part 2: Orthography
Orthography
Background
Noto Serif CJK is a font in the Mingti/Songti typeface style. Mingti typeface originates in Ming dynasty when woodblock carvers adjusted the strokes to be more regular and rigid than Kaiti (Regular script), for more efficient carving. The typeface has since evolved to become one of the two main typeface styles used in publishing in CJK area.
As the Mingti typefaces evolved, the strokes used increasingly diverged from their handwritten counterparts in the Kaiti (Regular script, handwriting). Educators in mainland China and Taiwan saw that as an issue affecting mass education and conducted reforms to change the orthography of the Mingti script.
Orthography Reforms
The reforms by mainland China brought the script somewhat closer to the Kaiti forms, while still preserving many features seen in the Mingti script. For example, the initial strokes of characters were often changed to follow the Kaiti style, while the “feet” in character components are kept as is.
Meanwhile, the reforms by Taiwan were much more radical and invented new components to distinguish characters of different etymology. The new orthography is mandated for education use in primary, and the Taiwanese government actively promotes its adoption in primary school education.
However, the new orthography removes much of the inherent differences between the style of Kaiti (handwriting) and Mingti (printed serif). The reforms by Taiwan were greatly resisted and are still resisted by many publishers in Taiwan and Hong Kong.
Actual use and implementations
The traditional orthography, similar to that used by Japan for characters outside of Jōyō kanji, are still readily used by publishers and commercials in Taiwan, Hong Kong and Macao. The traditional orthography is also exceedingly popular such there are derivatives (2) (3) of Noto CJK Serif which replace the traditional Chinese character’s reformed glyphs with the traditional orthography.
A less radical reformed orthography, somewhere between that of mainland China’s and Taiwan’s, is also used by major font vendors in the region, including MonoType and Dynacomware, which have produced such orthography in-house in the 70s-80s. Even so, bolder typefaces produced by these companies tend to use the traditional orthography, because the traditional orthography is superior in clarity and readability.
So-called “Hong Kong” orthography
Hong Kong has recently published new glyphs for all Big-5 characters, based on the Reference Glyphs for Chinese Computer Systems in Hong Kong, published by OGCIO, the agency in charge of information technology policy, in 2001.
That document was based on an older version of 常用字字形表, which contains characters written by pencil in Kaiti, using a strict transliteration approach similar to Taiwan, and suffers from many of the issues that plague the Taiwan reformed orthography. The choice to design reference glyphs by a strict transliteration of the Kaiti glyphs, without any consideration to actual implementations by major vendors such as MonoType and Dynacomware, has been repeated questioned, but struck down by the agency for reasons along the lines “we did not design anything, we simply referenced the books” and “the glyphs are for reference only, the font vendors can make adjustments for aesthetic purposes”.
(The glyphs contained in 常用字字形表, its current version published by the Education Bureau in Kaiti font, is also increasingly being challenged, as the glyphs contained in that book deviate extensively from the actual norms. The Education Bureau defends the glyphs in the book by saying that they referenced Kaishu calligraphy, and the glyphs are also provided for reference only.)
The “Hong Kong” orthography also asks for completely straight strokes in characters which are nearly universally printed with bent strokes to avoid closed apertures. Such forms are absent in all commercially successful fonts sold in Hong Kong.
Noto CJK Serif
The forms used by Noto CJK Serif are often too literal copies of the official forms, including the inconsistencies found in the standards.
For example, the treatment of 艹 is not consistent in the Songti forms promulgated by Taiwan, unlike those in Kaiti. Noto CJK Serif inherits those inconsistencies, as illustrated by tamcy on Github (https://github.com/adobe-fonts/source-han-serif/issues/36#issuecomment-341677511):
The four vertical strokes are slanted in Kaiti, and the sides of 早 is changed to completely vertical in Songti, but the two vertical strokes for 艹 are kept slanted. They are completely vertical in every major commercial font. They are also completely vertical in Noto CJK Sans.
The forms used by CJK Serif also treats minor typographical variations as region specific, when they should be treated as part of the font design. For example, the existence of the serif in the unconnected “N” stroke, as illustrated by Dang Zhihong (https://github.com/adobe-fonts/source-han-serif/issues/36#issuecomment-364917960):
Here are some more differences contained in Noto Serif, which seem completely superfluous to me:
The traditional design has always been identical to the CN version. No commercially successful Mingti/Songti in Taiwan or Hong Kong uses the left two designs.
The initiating point of “N” stroke is purely aesthetic (with the JP/KR/CN forms being superior in typography in terms of open aperture) and shouldn’t be treated as a regional difference.
Is it even necessary to vary the side where two dots touch (U+7136: JP/KR vs CN, U+347E: JP/KR/CN/TW vs HK)?
U+347E also has a TW source, but it is in Plane 4, so it is out of scope for Noto Sans CJK. However, HKSCS doesn’t put rarer characters in a different plane. Chances are, it’s just as rarely used in Hong Kong as for Taiwan. It would be better for Taiwan users if the glyph for U+347E was designed with [falling stroke] + [upwards stroke] like other characters. The new glyph for Hong Kong is practically worthless — pick any random character in Extension A that’s just not in HKSCS, and boom, you get the CN-style joining.
Noto Serif CJK (and to a less extent, Noto Sans CJK) is wasting a lot of precious glyph space for encoding immaterial glyph differences as region-specific glyphs. Such glyphs could be better used for characters that are used in minority dialects, or people’s names.
