Locale naming

The babel multilingual framework for LaTeX employs specific conventions for naming and referencing locale files (both the traditional ldf mechanism and the modern ini one), prioritizing compatibility with established standards (Unicode CLDR, IANA/BCP 47 registry) across the web and international data repositories. Babel adheres to those standards for locale classification, which are based (according to ISO 639) not only on linguistic similarity, but also on the existence of a common literature (which sometimes also means an specific typographical tradition) or of a common ethnolinguistic identity.

To ensure consistency, clarity, and compatibility across systems, the following conventions are used for naming locales at the user level. These conventions apply to both language names and the corresponding file names.

Basic rules

Base language names are written in lowercase English, with no spaces. Diacritics and non-letter characters (ampersands, hyphens, apostrophes) are removed to simplify parsing and avoid encoding issues:

• northernkurdish (not “Northern Kurdish”)
• kinaraya (from “Kinaray-a”)
• lu (from “Lü”)
• bosniaherzegovina (from “Bosnia & Herzegovina”)

Scripts and regions are appended using hyphens:

• serbian-latin (Serbian written in Latin script)
• spanish-mexico (Spanish as spoken in Mexico)

Region codes may be used, for convenience, as a shorthand for long region names:

• ba is accepted as a shorthand for bosniaherzegovina

Locale names are primarily derived from the CLDR (Common Locale Data Repository). When CLDR does not provide a name (e.g., for historical or constructed languages), the following patterns, followed in practice for other names, apply:

• Use the natural English form, lowercased and concatenated:
  • medievallatin (for “Medieval Latin”)
• Prefer names based on the description field in the IANA Language Subtag Registry:
  • polytonicgreek (for Greek written in polytonic orthography)

• When necessary, simplify overly descriptive names to maintain usability and brevity.

• Variants in the tags are uppercased in the CLDR, but in the BCP 47 registry they are lowercases (eg, el-polyton). We stick to the BCP 47 format.

• As secondary references:
  • Glottolog.
  • Wikipedia articles can be taken as a complementary but unreliable source, and its information must be verified; on the other hand, internal data, like this one, is useful for both names and tags.

Special cases

Some locales in CLDR include region-specific names that differ from the base language, with a more precise name:

• ro-MD is labeled “Moldavian” in CLDR.

Compound names like “American English” or “Norwegian Nynorsk” may be shortened by removing the language when unambiguous:

• american instead of americanenglish
• nynorsk instead of norwegiannynorsk
• brazilian instead of brazilianportuguese
• scottish instead of scottishgaelic

When choosing short demonyms, prefer ‘plain’ forms (even if vernacular) over composed or acronym-based names:

• british is preferred over UKenglish
• american is preferred over USenglish

This reflects the evolution of the english style, because the names american and british predate USenglish and UKenglish. Further, the main names in the CLDR for en-US and en-GB are American English and British English. Note the names ukenglish and usenglish (all lowercase) are not supported by babel as ldf files, even if they work in some operating systems.

Deprecated Names

The following locale names are deprecated and should be replaced with their standardized equivalents:

Deprecated	Use instead
brazil	brazilian
bahasai	indonesian
bahasam, meyalu	malay
classiclatin	classicallatin
ecclesiasticlatin	ecclesiasticallatin
frenchb, francais, canadien	french
germanb	german, ngerman
lsorbian	lowersorbian
portuges	portuguese
samin	northernsami
ukraineb	ukrainian
usorbian	uppersorbian
vietnam	vietnamese
russianb	russian

These names are not included in name.babel and should be avoided in new documents. Some legacy or anomalous names (especially related to German and Serbian) are under review and may be revised in future updates.