ISO 639

ISO 639 is a standard by the International Organization for Standardization (ISO) concerned with representation of languages and language groups. It currently consists of four sets (1-3, 5) of code, named after each part which formerly described respective set (part 4 was guidelines without its own coding system); a part 6 was published but withdrawn. It was first approved in 1967 as a single-part ISO Recommendation, ISO/R 639, superseded in 2002 by part 1 of the new series, ISO 639-1, followed by additional parts. All existing parts of the series were consolidated into a single standard in 2023, largely based on the text of ISO 639-4.

Use of ISO 639 codes
The language codes defined in the several sections of ISO 639 are used for bibliographic purposes and, in computing and internet environments, as a key element of locale data. The codes also find use in various applications, such as Wikipedia URLs for its different language editions.

History
The early form of ISO's language coding system was manifested by ISO/R 639:1967 titled Symbols for Languages, Countries and Authorities, which aimed chiefly to regulate vocabularies signifying languages, countries, and standardization agencies of ISO member bodies. Its "language symbols" consisted of one- or two-letter variable-length identifiers in capitalized Latin alphabets, e.g.  or   for English; ,  , or   for Spanish; and   for Indonesian. It was also allowed to use (the pre-1993 version of) UDC numeral auxiliaries to indicate languages.

After decoupling the country code into ISO 3166 in 1974, the first edition of the standard ISO 639:1988 Code for the representation of names of languages was published with a framework of uniformly two-letter identifiers in lowercase Latin alphabets, mostly identical in format and vocabulary to that of the current ISO 639 Set 1.

Since then, the standard has been adopted as a fundamental technology of the rapidly expanding computer industry (RFC 1766), leading to development of more expressive three-letter framework, published as ISO 639-2:1998, largely based on MARC codes for languages. The original two-letter system was redefined as ISO 639-1 in 2001.

Seeking for more extensive support of languages for widening applications, separate supersets of the ISO 639-2 namespace that cover individual languages and groups were established as ISO 639-3 and ISO 639-5, respectively. There was also an attempt to code more precise language variants using four-letter identifiers as ISO 639-6, which was later withdrawn and to be reorganized under another framework, ISO 21636.

Relatively constant updates in parts of ISO 639 had been handled by each own authority in charge until the publication of ISO 639:2023, which harmonized and reunified the body text of former standards and brought about organizational change with a joint maintenance agency supervising all sets and issuing newsletters.

Current sets and historical parts of the standard
Each set of the standard is maintained by a maintenance agency, which adds codes and changes the status of codes when needed. ISO 639-6 was withdrawn in 2014, and not included in ISO 639:2023.

Characteristics of individual codes
Scopes:
 * Individual languages
 * Macrolanguages (Set 3)
 * Collections of languages (Sets 1, 2, 5). Some collections were already in Set 2, and others were added only in Set 5:
 * Remainder groups: 36 collections in both Set 2 and 5 are of this kind — for compatibility with Set 2 when Set 5 was still not published, the remainder groups do not contain any language and collection that was already coded in Set 2 (however new applications compatible with Set 5 may treat these groups inclusively, as long they respect the containment hierarchy published in Set 5 and they use the most specific collection when grouping languages);
 * The only collection which previously assigned with two-alphabet code is Bihari (bh) during the Part 1 era, which deprecated in June 2021.
 * Regular groups: 29 collections in both Sets 2 and 5 are of this kind — for compatibility with Set 2, they can not contain other groups;
 * Families: 50 new collections coded only in Set 5 (including one containing a regular group already coded in Set 2) — for compatibility with Set 2, they may contain other collections except remainder groups.
 * Dialects: they were intended to be covered by former ISO 639-6 (proposed but now withdrawn).
 * Special situations (Sets 2, 3).
 * Reserved for local use (Sets 2, 3). Also used sometimes in applications needing an two-letter code like standard codes in Sets 1 and 2 (where the special code  is not suitable), or a three-letter code for collections like standard codes in Set 5.

Types (for individual languages):
 * Living languages (Sets 2, 3) (all macrolanguages are living languages)
 * Extinct languages (Sets 2, 3) (599, 5 of them are in Set 2:,  ,  ,  ,  ; none are in Set 1)
 * Ancient languages (Sets 1, 2, 3) (124, 19 of them are in Set 2; and 5 of them, namely,  ,  ,   and  , also have a code in Set 1:  ,  ,  ,  ,  )
 * Historical languages (Sets 2, 3) (89, 16 of them are in Set 2; none are in Set 1)
 * Constructed languages (Sets 1, 2, 3) (23, 9 of them in Set 2:,  ,  ,  ,  ,  ,  ,  ,  ; 5 of them in Set 1:  ,  ,  ,  ,  )

Individual languages and macrolanguages with two distinct three-letter codes in Set 2:
 * Bibliographic (some of them were deprecated, none were defined in Set 3): these are legacy codes (based on language names in English).
 * Terminologic (also defined in Set 3): these are the preferred codes (based on native language names, romanized if needed).
 * All others (including collections of languages and special/reserved codes) only have a single three-letter code for both uses.

Relations between the sets
The different sets of ISO 639 are designed to work together, in such a way that no code means one thing in one set and something else in another. However, not all languages are in all sets, and there is a variety of different ways that specific languages and other elements are treated in the different sets. This depends, for example, whether a language is listed in Sets 1 or 2, whether it has separate B/T codes in Set 2, or is classified as a macrolanguage in Set 3, and so forth.

These various treatments are detailed in the following chart. In each group of rows (one for each scope of Set 3), the last four columns contain codes for a representative language that exemplifies a specific type of relation between the sets of ISO 639, the second column provides an explanation of the relationship, and the first column indicates the number of elements that have that type of relationship. For example, there are four elements that have a code in Set 1, have a B/T code, and are classified as macrolanguages in Set 3. One representative of these four elements is "Persian" / /.

These differences are due to the following factors.

In ISO 639 Set 2, two distinct codes were assigned to 22 individual languages, namely a bibliographic and a terminology code (B/T codes). B codes were included for historical reasons because previous widely used bibliographic systems used language codes based on the English name for the language. In contrast, the Set 1 codes were based on the native name for the language, and there was also a strong desire to have Set 2 codes (T codes) for these languages which were similar to the corresponding 2-character code in Set 1.
 * For instance, the German language (Set 1: ) has two codes in Set 2:   (B code) and   (T code), whereas there is only one code in Set 2, , for the English language.
 * 2 former B codes were withdrawn, leaving today only 20 pairs of B/T codes.

Individual languages in Set 2 always have a code in Set 3 (only the Set 2 terminology code is reused there) but may or may not have a code in Set 1, as illustrated by the following examples:
 * Set 3  corresponds to Set 2   and Set 1
 * Set 3  corresponds to Set 2   but lacks a code in Set 1.

Some codes (62) in Set 3 are macrolanguages. These are groups containing multiple individual languages that have a good mutual understanding and are commonly mixed or confused. Some macrolanguages developed a default standard form on one of their individual languages (e.g. Mandarin is implied by default for the Chinese macrolanguage, other individual languages may be still distinguished if needed but the specific code  for Mandarin is rarely used).
 * 1 macrolanguage has a Set 2 code and a Set 1 code, while its member individual languages also have codes in Set 1 and Set 2: /  contains  /,  / ; or
 * 4 macrolanguages have two Set 2 codes (B/T) and a Set 1 code: / /,  / / ,  / / , and  / / ;
 * 28 macrolanguages have a Set 2 code but no Set 1 code;
 * 29 other macrolanguages only have codes in Set 3.

Collective codes in Set 2 have a code in Set 5: e.g.  in Sets 2 and 5, which stands for Australian languages.
 * Some codes were added in Set 5 but had no code in Set 2: e.g.

Sets 2 and 3 also have a reserved range and four special codes:
 * Codes  through   are reserved for local use.
 * There are four special codes:  for languages that have no code yet assigned,   for "multiple languages",   for "undefined", and   for "no linguistic content, not applicable".

Two-letter code space
Two-letter (formerly "Alpha-2") identifiers (for codes composed of 2 letters of the ISO basic Latin alphabet) are used in Set 1. When codes for a wider range of languages were desired, more than 2 letter combinations could cover (a maximum of 262 = 676), Set 2 was developed using three-letter codes. (However, the latter was formally published first. )

Three-letter code space
Three-letter (formerly "Alpha-3") identifiers (for codes composed of 3 letters of the ISO basic Latin alphabet) are used in Set 2, Set 3, and Set 5. The number of languages and language groups that can be so represented is 263 = 17,576.

The common use of three-letter codes by three sets of ISO 639 requires some coordination within a larger system.

Set 2 defines four special codes,  ,  ,  , a reserved range   (20 × 26 = 520 codes) and has 20 double entries (the B/T codes), plus 2 entries with deprecated B-codes. This sums up to 520 + 22 + 4 = 546 codes that cannot be used in Set 3 to represent languages or in Set 5 to represent language families or groups. The remainder is 17,576 – 546 = 17,030.

There are somewhere around six to seven thousand languages on Earth today. So those 17,030 codes are adequate to assign a unique code to each language, although some languages may end up with arbitrary codes that sound nothing like the traditional name(s) of that language.

Alpha-4 code space (withdrawn)
"Alpha-4" codes (for codes composed of 4 letters of the ISO basic Latin alphabet) were proposed to be used in ISO 639-6, which has been withdrawn. The upper limit for the number of languages and dialects that can be represented is 264 = 456,976.