Specifying available languages and scripts leiden

The TEI provides a langUsage element in the teiHeader whereby the languages and scripts used in a given document or project must be declared before they may be used in the document itself. The langUsage contains a list of language-and-script codes, drawn from international standards (see: Language tags in HTML and XML; the standards document itself is RFC 4646: Tags for Identifying Languages (September 2006)). Each such code may be glossed with an appropriate description (at the discretion of the project or editor). Examples follow:

language id="la"Latin/language language id="grc"Ancient Greek/language language id="en"English/language language id="el"Modern Greek/language language id="grc-Latn"Ancient Greek in Latin characters/language language id="la-Grek"Latin in Ancient Greek characters/language

The id of each language should contain the 2-letter ISO language code, where one is available, or the 3-digit code where one is not (as in the case of ancient Greek). The four-letter, case-sensitive script code should not be added as a subtag unless it is not the usual (default) script for the language in question. These languages are then available to be referred to by the lang or xml:lang attribute on any element in the file. See tei:foreign and languages.xml.

Note that established practice of looking up language codes in various ISO-sponsored lists is now deprecated. These lists function as the source for only some of the full set; The Internet Assigned Numbers Authority (IANA) is the final aggregator and arbiter for language codes. IANA's lists of language/script codes and related information are available at:

IANA Language Subtag Registry IANA Language Tag Extensions Registry

Note that, by definition, language subtags are case-neutral and all processing environments are required to treat them in a case-insensitive manner. There is, however, a recommended captialization scheme — intended to help human readers — which is reflected in the IANA registry. Note, however, that the TEI P4 DTD defines the standard lang attribute as an IDREF; this means that — in order for the file to be valid — there must be one, and only one, exact copy of each language code value in an id attribute somewhere in the file (i.e., on a language element in the header, as described above). EpiDoc users therefore must ensure that they make consistent use of capitalization in language tags across their project, and preferably across the EpiDoc world.

One possible safeguard might be to produce a modification of the EpiDoc DTD to constrain the legal values for the lang attribute so that they match the codes used in the id attributes of the language tags in the header. So, for example, the default EpiDoc DTD defines lang as follows:

lang IDREF #IMPLIED

To match the example language code declarations above, this definition could be altered as follows:

lang (la | grc | en | el | grc-Latn | la-Grek) "en"