[vlc-devel] ISO 639-3 (was: vlc player languages)

[John Cowan]

Bruno Haible scripsit:

> This is all correct. Yes, gettext uses the locale names from glibc, and these
> use the abbreviations from ISO 639. So far only ISO 639-1 and ISO 639-2 have
> been required. ISO 639-3 abbreviations can change over time, as can the
> shorthands in www.ethnologue.com (not a standard, but helpful in understanding
> the relations among languages).

In fact not, according to the usual understanding of "change".  In no
case does a code refer to something and then cease to refer to it, or
begin to refer to something else.  Here's the Registration Authority's
policy statement from http://www.sil.org/ISO639-3/changes.asp (I've put
RFC 2119 keywords in CAPS here):

    Care SHALL be taken in making changes to ensure that existing uses
    of the code set are not seriously compromised.  To ensure continuity
    and stability, the identifier for any given language SHALL NOT be
    changed.  On occasion, given compelling reasons, a code element MAY
    be retired from current use.  When a code element is retired, the
    identifier for that code element SHALL NOT be reassigned.  Retired
    code elements remain part of the code set and retain their identifier
    and denotation.  Continued use in existing data or implementations
    remains a valid application of this standard.  On-going generation of
    data using a retired code element is generally discouraged, however.

    The names of languages listed for a given identifier MAY be changed
    given sufficient reasons without implying any change in intended
    denotation.  Other information relating to a code element -- the
    scope of the language, the type of language, its participation
    in a macrolanguage or collective grouping (aspects intended to be
    informative, not normative, with reference to the standard) -- MAY
    also be changed, provided that such change clarifies or corrects
    the intended denotation, and does not confuse or alter the language
    denoted by the code element.

    Given sufficient reasons, new code elements MAY be added, or the
    denotation for a given code element MAY be broadened.  The denotation
    of a code element SHALL NOT be narrowed, however, as this can result
    in an unknown proportion of the existing uses of a code element
    becoming invalid.  If it is determined that the denotation of a code
    element was too broad, such as if a linguistic variety that was
    thought to be a dialect of a given language was later determined to
    be a distinct language, the existing code element would be retired
    and replaced with two new code elements.

    If a language variety thought to be a distinct language is found on
    further investigation to be a dialect variant encompassed by another
    language, the former denotation MAY be merged into the latter.  This
    has the result of broadening the latter denotation. The code element
    for the former variety will be retired, and the relationship to the
    code element for the latter denotation will be documented.

Ethnologue is updated every three years, ISO 639/3 every year, and ISO
639/2 sporadically, but they all share the same set of abbreviations
and the same change-control policies.

-- 
John Cowan    http://ccil.org/~cowan    cowan at ccil.org
Mr. Henry James writes fiction as if it were a painful duty.  --Oscar Wilde