Typst Docs: You should LOCALLY change the text language just for those parts that contains passages in a different language than the main language.
- Typst is awesome to write English or Latin docs. But when it comes to non-latin, e.g. Greek, Arabic and CJK, things become complex.
- Compare to local, global setting is a bit more convenient.
All World Languages in One Visualization
What languages use the Latin alphabet:
- Germanic and Romance languages: English, German, Spanish, French, Italian, Portuguese, Romanian.
- Slavic languages: Polish, Czech, Slovak, Croatian, Slovene.
- Asian-Pacific languages: Indonesian, Malay, Tagalog, Vietnamese, Hawaiian, Samoan, Fijian.
- American-Australian languages: Quechua, Nahuatl, Guarani, Warlpiri, Arrernte.
- African and other languages: Swahili, Hausa, Yoruba, Igbo, Turkish and Somali.
English and Latin
Greece
Look for a best practice
Arab
Look for a best practice
China
Because pretty long history, there are many kinds of Chinese spoken and written around the world. But most used nowadays are Simplified and Traditional.
Simplified (Typst score: 68.3)
Chinese is a ltr and mono language, you care about punctuation marks only. I’m NOT a China senator or ISO expert, but here are some formal documents I searched out:
- International Standard, reviewed in 2021
- characters
- Pinyin
- …
- punctuation
- marks similar to those in Latin are transcribed as their Latin counterparts
- specific marks are transcribed as follows: 0x3002 → 0x2e, 0x3001 → 0x2c, 0x2022 → 0x20, 0x2026+0x2026 → 0x2026 (.<-。,<-、…<-…… space<-·)
- National Law, updated in 2025.12
- Putonghua (Hebei dialect) as the legal general language
- <General Regulated Chinese Characters Sheet> 2013
- <Chinese Pinyin Scheme> 1958
- National Standard Recommended(non-forced), 2012
- punctuation, 29
- 0.5em, single, 3: -·/
- 1em, single, 21: 。，、：；？！’‘“”（［❲【《<>》】❳］）—～underPoint underLine「『』」
- 1em, double, 4(1): ？？！！？！！？
- 2em, triple, 2(1): ？？？！！！
- 2em, single, 2: ——……
- 4em, double, 1: …………
- order: 一、（一）1.（1）circleOne or a.
- National Press and Publication Standard Recommended(non-forced), 2018
- punctuation
- Chinese sentence: end with Chinese marks, e.g. 。？！
- English sub-sentence: put in Chinese quote marks
- marks inside English sentence: use English marks
- /: add spaces if need
- font size: Chinese small 5 + English 9P, 5 + 10.5P
- Ministry Web page, outdated perhaps, 2009
- punctuation
- similar transcribed as Roman counterparts
- under circle point → italic
- under wavy line → italic
As you can see, no forced law on Simplified Chinese (SC) punctuation marks. And GB or CY doc made some typesetting errors, i.e. they each violated themself.
Merriam-Webster joke: Is themself a word?
What’s the status of Typst SC typesetting?
- *0.99: arbitrary font and glyph fallback if not set
- -2: no under circle point and under wavy line
- -5: wrong marks width: -·/❲？！❳
- -2: smartquote width or height fails
- $“score” = 100*0.99*(29-9)/29 = 68.3$
Best Practice:
#set text(10pt)
Libertinus _Serif_ *En:* a,b.c?d!e'f'g"h"i\`j\`k\~l\@m\#n\$o%p^q&J\*K-L\_M=N+O\\P|Q(R[S{T\<U\>V}W]X)Y/Z;
无豆腐_宋体_*简中：*〇-一·二/三。两，仨、四；五？六！七“八‘九’十”廿（拾［玖❲捌【柒《陆「伍『肆』叁」贰》壹】零❳字］词）句—段～节？？章！！篇！？书？！册？？？典！！！籍——古……诗…………文“出游从容，‘鱼之乐’也。”惠子曰：“How to tell? "After all, you are not 'a fish'".”。```pl print "花径不曾缘客扫" #春风不度 ``` $e^(i pi) + 1 = 0 "玉门关"$
#block(spacing:0.6em, line(length:100%))
#set text(lang:"zh", region:"CN", font:(
(name:"Libertinus Serif", covers:"latin-in-cjk"),
"Noto Serif CJK SC"))
#show smartquote: set text(font:"Libertinus Serif")
#show raw: set text(font:(
(name:"DejaVu Sans Mono", covers:"latin-in-cjk"),
"Noto Sans Mono CJK SC"))
#show math.equation: set text(font:(
"New Computer Modern", "Noto Serif CJK SC"))
#show emph: it => box(skew(ax:-12deg, it))
Libertinus _Serif_ *En:* a,b.c?d!e'f'g"h"i\`j\`k\~l\@m\#n\$o%p^q&J\*K-L\_M=N+O\\P|Q(R[S{T\<U\>V}W]X)Y/Z;
无豆腐_宋体_*简中：*〇-一·二/三。两，仨、四；五？六！七“八‘九’十”廿（拾［玖❲捌【柒《陆「伍『肆』叁」贰》壹】零❳字］词）句—段～节？？章！！篇！？书？！册？？？典！！！籍——古……诗…………文“出游从容，‘鱼之乐’也。”惠子曰：“How to tell? "After all, you are not 'a fish'".”。```pl print "花径不曾缘客扫" #春风不度 ``` $e^(i pi) + 1 = 0 "玉门关"$
Traditional
Look for a best practice
Japan
Look for a best practice
Korea
Look for a best practice