Locale

Specification of language variants

---

A locale is an identifier of a language and region, plus an optional writing script. The locale is used in machine translation APIs to specify the language of the source and target text. Locales are used to indicate the language of documents in web crawling to build training data.

Example: frCA means French (fr) as spoken in Canada (CA)

The formatting varies from one system to another. frCA, fr-ca, and fr_CA are all common formats.

Language codes are typically specified in two or three characters according to ISO 639. Regions are typically specified in two characters according to ISO 3166. Scripts are optionally specified according to ISO 15924, such as sr-Cyrl_RS for Serbian written in Cyrillic script in Serbia.

API support

These language variations are supported by many API vendors:

• Chinese (zh):
  • Chinese, Simplified (zh-cn, also zh-Hans)
  • Chinese, Traditional (zh-tw, also zh-Hant)
• Portuguese (pr):
  • Portugal (pr-pr)
  • Brazil (pr-br)
• French (fr):
  • France (fr-fr)
  • Canada (fr-ca)
• Spanish (es):
  • Spain (es-es)
  • Mexico (es-mx)
  • Latin America and Caribbean region (es-419)
• English (en):
  • United States (en-us)
  • Great Britain (en-gb)
• Serbian (sr):
  • Serbia, Cyrillic script (sr-Cyrl-rs)
  • Serbia, Latin script (sr-Latn-rs)
• Norwegian (no):
  • Norwegian Bokmål (nb, nob)
  • Norwegian Nynorsk (nn, nno)

Challenges

• When a translation API uses only a language code without a region code or script, it can be unclear what locale is being translated.
• Not all languages or variants have standardised locale codes, leading to differences between different APIs.
• In some cases, the locale codes have changed over time. For example, old systems may represent Cantonese as zhHK while newer systems use the newer language code yue.

See also

• Languages