`:lang(lzh)` elements in non-traditional-Chinese CJK elements should not switch font to Traditional Chinese one

[tats-u]

An old Chinese literal text (known as Kanbun in Japan) is occasionally quoted in non-Chinese region. However, a Taiwanese or Hong Kongese font is always used if you specify lang=lzh:

<div lang=ja>
  <blockquote lang=lzh>子曰、後生可畏、焉知来者之不如今也</blockquote>

  <p lang=ojp-t-lzh>子曰く、<ruby>後生<rp>（</rp><rt>こうせい</rt><rp>）</rp></ruby><ruby>畏<rp>（</rp><rt>おそ</rt><rp>）</rp></ruby>るべし。いずくんぞ<ruby>来者<rp>（</rp><rt>らいしゃ</rt><rp>）</rp></ruby>の今にしかざるを知らんや</p>

  <p lang=ja-t-lzh>年が若いとは、将来に希望があることだ。今後の世代が、現在の世代を乗り越えていかないとはいえないのだ。</p>
  
  <blockquote lang=lzh>子曰、道聴而塗説、徳之棄也</blockquote>

  <p lang=ojp-t-lzh>子曰く、道に聴きて<ruby>塗<rp>（</rp><rt>みち</rt><rp>）</rp></ruby>に説くは、徳これ棄つるなり</p>

  <p lang=ja-t-lzh>
聞きかじったことを右から左へ受け売りして得意がる。これでは徳は身につかない。</p>
</div>

https://codepen.io/tats-u/pen/MYKPoXa

A text marked as lzh is not concerned with modern Chinese font specifications today. It should be respect the language of ancestor elements whose language is CJK.

• Expected: rendered using a Japanese font (if ancestor elements are ja)
• Actual: rendered using a Taiwanese or Hong Kongese (traditional Chinese) font

1. If the element's language is lzh,
2. Find a nearest ancestor whose language is CJK
3. If find one, inherit its font preset of its language
4. If no one (e.g. when the page is written in English), use the font preset of traditional Chinese

[xfq]

We discussed this issue at the 2025-11-20 internationalization WG teleconference, and we felt that rather than adding more heuristics to the browser, we preferred to directly select the Literary Chinese text through CSS selectors and specify the font, as this approach is more reliable.

[tats-u]

In that case how can we inherit the font specified by sans-serif of the parent element?

lzh-Hans uses simplified Chinese fonts but lzh-Jpan does not use Japanese fonts.

[xfq]

If your parent element specifies sans-serif as its font-family, then there may indeed be no way to solve the problem. However, if your parent element specifies a specific font, such as Hiragino Sans, then this problem can be solved.

[tats-u]

if your parent element specifies a specific font, such as Hiragino Sans, then this problem can be solved.

It is not a good practice to specify specific CJK (especially Japanese not for Apple devices) font names in font-family generally speaking. sans-serif should work well here.

[bobbytung]

Discussed in CLREQ meeting 2025/11/25.

Strictly, 「文言文」in Chinese and 「漢文」 in Japanese should use old glyphs（旧字体／舊字體） like Kangxi Dictionary（康熙字典）。

Modern Traditional Chinese font and Japanese font may not suitable for language code 'lzh'.

In PRC, there is a national standard GB/Z 40637-2021 古籍印刷通用字规范字形表(standard glyph list of generally used Chinese characters for ancient books publishing). However, it's now widely adopted.

As result, I think a font contains old glyphs for 'lzh' will be a better approach rather than match to modern Japanese or Traditional Chinese fonts with new glyphs.

[tats-u]

High quality Japanese fonts can switch to the kanxi dictionary's traditional forms by using the CJK Compatibility Ideographs Block, SVS corresponding to it, or Adobe Japan-1 IVS.

https://www.wakufactory.jp/densho/font/ivs_aj.html

The traditional form of U+795E 神 can be notated as U+FA19 神, U+795E U+FE00 神︀, or U+795E U+E0100 神󠄀.
The traditional form of U+8FD4 返 can be notated as U+8FD4 U+E0101 返󠄁.

Korean fonts always use the kanxi form unlike Taiwanese or Hong Kongnese ones.

[tats-u]

High quality Japanese fonts can switch to the kanxi dictionary's traditional forms by using the CJK Compatibility Ideographs Block, SVS corresponding to it, or Adobe Japan-1 IVS.

https://www.wakufactory.jp/densho/font/ivs_aj.html

The traditional form of U+795E 神 can be notated as U+FA19 神, U+795E U+FE00 神︀, or U+795E U+E0100 神󠄀.
The traditional form of U+8FD4 返 can be notated as U+8FD4 U+E0101 返󠄁.

Korean fonts always use the kanxi form unlike Taiwanese or Hong Kongnese ones.

[tats-u]

Allowing Jpan or Kore designation after lzh is the easiest compromise solution. Hans or Hant has already been allowed and effective.

[xfq]

Let me summarize the discussions from 2025-11-25 Chinese Layout Task Force teleconference:

Ideally, we would use the glyphs in Kangxi Dictionary before the differentiation of Traditional Chinese, Simplified Chinese, and Japanese glyphs. No one has yet created such an ideal font, but this is actually a need that many people have.

Moreover, the final fallback being "Traditional Chinese" will most likely adopt the Taiwanese glyphs, which will face a lot of opposition in both Mainland China and Hong Kong (for example, there is a GB/Z 40637-2021 national standard glyph list of generally used Chinese characters for ancient books publishing in Mainland China).

We can consider raising this requirement to the IRG. Then the IRG/UCS can add a column of ancient glyphs to their list. With this list, we can collaborate with font companies to create such fonts.

Currently, IRG uses a complex rule to imagine the ancient glyphs and then count the strokes. Having this column would simplify IRG's future work too.

This approach may take longer, but the results are more desirable.

We believe this default value is more reasonable. If there are special requirements (such as needing to use a Japanese font), a specific Japanese font can be specified.