Using Language Identifiers (RFC 3066)

Language identifiers as specified by RFC 3066, can have the form language, language-country, language-country-variant and some other specialized forms. The guidelines for choosing between language and language-country are ambiguous. 

To clarify which form should be used, John Cowan and I have posted this list for review. This is currently a draft document. It will be continually revised as we get feedback from linguists and internationalization experts. (Although, I have a day job, so allow some time for me turn it around.) The topic is being discussed on the W3C www-international mail list and the IETF ietf-lang mail list. (The feedback thread starts www-international 2004OctDec-0131 and ietf-lang 2004-12-002431.)

This page is intended to list combinations of language code and country code that have been found useful in distinguishing language variants from each other. Please write the authors if you think either that there is no variation between two code pairs listed, or that there are additional code pairs that are useful in distinguishing written variants.

Send comments to Tex Texin and John Cowan. To improve turn-around of edits, please make sure the mail uses the subject: "Language Identifier List Comments". This will help us easily identify the mails and process them. Please use a different subject heading for other discussions.

There are a number of suggestions for deciding whether to use a one-level (language only) or two-level (language-region) tag. They require some discussion and will be added here shortly.

Language codes used as primary subtags are from ISO 639. Country codes used as secondary subtags (or tertiary if script tags are secondary) are from ISO 3166. Note also that several unique language tags are also defined in the IANA Language Tag Registry. Script codes, generally 4 letters in length, such as "Hans" for Simplified Chinese, "Hant" for Traditonal Chinese, "Cyrl" for Cyrillic, etc. are defined by ISO 15924 Codes For The Representation Of Names Of Scripts.

Proposed List of 1-level Language Identifiers

Some languages have no significant variations, and therefore can be adequately represented by a language subtag alone, as opposed to a language subtag and country subtag. We would like to clarify which languages are in this category.

Proposed List of 2-level Language Identifiers

Some languages differ by region and therefore a language subtag and country subtag may be used to identify the language more precisely. Note that depending on the purpose or intended audience of the content, a very precise label may not be needed and a one-level language tag may be adequate. For example, using "en" for this page may be sufficient instead of en-US. If the tag is being used with an application (as opposed to labeling content) the requirement for precision may also vary.

John Cowan provided the original data for this list. The list has since been amended based on suggestions from a number of folks and considerable research.

Table Legend

In the following table, the language codes are followed by a parenthesized expression which contains the language name and if the region code is different from the region represented by the row, a region name. For example, en-US (English, U.S.).

Tags will look different depending on their source and whether they have been deprecated. Their appearance is as follows:

ISO 639 tags IANA registry tags deprecated tags