Which method and font for which language

Improve this page! Feel free to draft a pull request on GitHub.

Index Fonts • Loading the language

As explained in the manual, there are currently two methods to load a language. With release 25.14, babel selects the recommended one. However, you can find of interest the basic criteria, especially if you are still using an old version, which is reproduced below .

There are pages on specific locales in the menu above and here (but note they are still using provide=*, which is often unnecessary).

See also the examples here for languages like Arabic, Thai, Hebrew, Polytonic Greek, Japanese, Syriac and Malayalam, among others.

I’d like to thank Loren Davis for his help in preparing this page, including the list of additional fonts.

---

Fonts

With non-Latin scripts, you need to select another font, which can be done with \babelfont.

If compatibility is not a concern, remember currently the most important systems (Windows, Mac, Linux) come with many fonts. Even if they are not installed by default, there are options in the system settings to do it.

With luatex and in versions ≥24.14, the default renderer in \babelfont is the default one in Latin, Cyrillic, Greek and scripts of the same family. In the rest of scripts, like Southeast Asian ones and Arabic, the default renderer is Harfbuzz.

TeXLive

Here are some basic fonts included in TeXLive. This section is not meant as a font reference, but as a quick guide for a language to work with a more or less standard font. I’ve found no fonts for Telugu or Kannada.

If you know the codepoint of a character in the script you need, you can find fonts containing it with albatross (requires Java) or with something like fc-list :charset=1033C family in the commands line (in this case, a Gothic character, the script required by the gothic language).

Font(s)	Languages / Scripts	Notes
IPAexMincho, IPAexGothic	Japanese
Baekmuk Gulim, Baekmuk Batang, Baekmuk Dotum	Korean
FreeSerif	Multilingual, including Greek, Cyrillic, Coptic, Arabic, Hebrew, many (but not all) Brahmi scripts (Thai, Devanagari, Bengali, Tamil, Malayalam, etc.), Amharic, Georgian, Cherokee, Armenian.	A comprehensive Times-like, although with some oddities.
Amiri	Arabic
ALM Fixed	Arabic	A CM-like monospaced font.
CMU Serif, CMU Sans Serif, CMU Typewriter Text	Cyrillic, Greek	A CM-like family.
NewComputerModern	Cyrillic, Greek, Hebrew, Cherokee	Another, more modern, CM-like family. It doesn’t support African orthographies.
Padauk	Myanmar
Shobhika	Devanagari	With an excellent coverage of Vedic accents, Latin transliteration for Sanskrit, mathematical and technical symbols, and Cyrillic.
Rachana	Malayalam	It was the first Unicode font to support traditional orthography. This is the version by smc.org.in. There are several other fonts floating around under the name “Rachana,” and some of them do not work. Manjari is another popular font suitable for these body text, designed on the same theoretical principles as Inconsolata.
FandolSong, FandolHei, FandolKai, FandolFang	Yi, Chinese
David CLM, Miriam CLM, Miriam Mono CLM	Hebrew	Free implementations of the most common Hebrew serif, sans-serif and monospace fonts. If you need cantillation marks for liturgical texts, Culmus also has a Taamey David. There are many different versions of font faces.
DejaVu Sans	Greek, Cyrillic, Armenian, Hebrew, Arabic, Lao, Georgian	Also Canadian Aboriginal, Ogham.
Noto	Greek, Cyrillic	A large family, but TeXLive includes only the fonts for these two scripts (and emojis).
Junicode	Runic	Also Medieval and Ancient Latin script.

Additional fonts

There are also some fine fonts out there, but you must install they because they aren’t bundled with TeXLive. This list, of course, is just a very small selection. As a complementary source, see the Font Lister.

Font(s)	Languages / Scripts	Notes
Noto	Multilingual	This family has a very comprehensive coverage and come in many variations of weight and width to match your other fonts. However, some fonts don’t cover the ASCII range. You can see how to combine Noto fonts at the end of the video LaTeX – Multilingual Unicode strings. There are 4 huge fonts covering many languages: Noto Serif Living, Noto Sans Living, Noto Serif Historical, Noto Sans Historical (but only the Regular variants).
Babelstone Han	Chinese	It supports both Simplified and Traditional Chinese and has an especially comprehensive selection of traditional Chinese characters.
Abyssinica SIL	Abyssinica SIL
Nirmala UI	Indic scripts	A Windows font with 14 scripts: Bangla, Chakma, Devanagari, Gujarati, Gurmukhi, Kannada, Malayalam, Meitei Mayek, Ol Chiki, Odia, Sinhala, Sora Sompeng, Tamil, Telugu.
Segoe UI Historic	Ancient scripts	A Windows font with 25 scripts: Imperial Aramaic, Brahmi, Carian, Cypriot, Egyptian hieroglyphs, Glagolitic, Gothic, Old Italic, Kharoshthi, Lycian, Lydian, Meroitic Cursive, Ogham, Orkhon, Inscriptional Pahlavi, Phoenician, Inscriptional Parthian, Runic, Old South Arabian, Shavian, Syriac, Ugaritic, Old Persian, Sumero- Akkadian Cuneiform.
Lohit	Indic scripts	A set of Linux fonts covering Assamese, Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, and Telugu.

You may also want to have a look at some articles on tex.textackexchange on Arabic.

---

Loading the language

There are two ways to load a language, which use different definition files. At first this duality may seem inconvenient, but this way the transition to a more modern system where locales can be easily defined and interoperate can be done smoothly.

The first one is:

\usepackage[italian]{babel}

It will load the old good Italian language definition file. This procedure is the preferred one in most languages where an ldf exists, but there are exceptions (for example, Arabic and Hebrew). In general, you should do this for European languages written in Latin and Cyrillic scripts, as well as for Vietnamese. For Greek, it may depend on your needs, and very often both methods will be fine. If you want modern and ancient Greek at the same time, the next one is usually better.

The second method is:

\usepackage[thai, provide=*]{babel}

(With ≥25-14, you can omit provide=*.) This will load data from an ini file. This is usually the correct method for languages written in some of the 40-something non-European scripts, currently supported by babel, like Thai, Hindi, Sanskrit, Marathi, Malayalam, Lao, Khmer, Bangla (Bengali), Gujarati, Telugu, Kannada, Sinhala, Tamil, Amharic, or Burmese.

In short simple CJK texts, this is also the preferred method, but in more complex ones you should resort to some CKJ framework like luatexja, xeCJK, CTeX, kotex, and the like.

Some languages may require additional options. Examples are Arabic, Hebrew, and Syriac. So, with LuaTeX:

\usepackage[arabic, bidi=basic, provide=*]{babel}

With XeTeX:

\usepackage[arabic, bidi=bidi-r, provide=*]{babel}

(With ≥25-14, you can omit provide=*.)