Every known language gets a three-letter code, and animal calls get their own
ISO 639-3 tries to give every human language, living or dead, spoken or written, invented or ancient, a unique three-letter tag. Published in 2007, it held 7,916 entries by January 2023. It even reserves zxx for material with no language at all, such as recordings of animals.
The International Organization for Standardization issued the standard on 1 February 2007, building on the shorter lists in earlier parts of the series, which had focused on major languages with large literatures. Much of the expanded coverage came from volumes of the Ethnologue catalogue produced by SIL International, which now acts as the registration authority. The codes serve as metadata: websites and software use them to declare which language a text is in, libraries use them for cataloguing, and linguists cite them because language names can be obscure or ambiguous. In 2023 this edition was formally withdrawn and folded into ISO 639:2023.
Simple arithmetic sets the ceiling. With 26 letters in three positions there are 17,576 possible codes, but 546 are already taken by special codes, a reserved block and older bibliographic codes, leaving at most 17,030. A handful of catch-alls handle awkward cases: mis for languages not yet coded, mul for data mixing several languages, und when nobody has identified the language, and zxx when there is none. The range qaa to qtz, 520 codes, is left for local use.
The standard admits its own limits. Deciding where one language ends and another begins is partly subjective, especially for varieties without schools, media or a written tradition, so it presents itself as a precise labelling tool, not a verdict on how many languages exist. It groups 58 entries as macrolanguages, so Standard Arabic can sit as an individual language under a broader Arabic code, a way of handling varieties whose speakers regard them as forms of one tongue. Dialects are not coded, and neither are reconstructed ancestors such as Proto-Indo-European. Constructed languages qualify only if built for human communication and backed by a body of literature.
Anyone may request a change. Proposals are posted publicly, announced on the Linguist List discussion forum and given at least three months of review, with decisions issued at the end of each yearly cycle, usually in January, and an archive explaining every ruling.
Source: ISO 639-3
