The limit of language codes
Ciarán Ó Duibhín
ciaran at oduibhin.freeserve.co.uk
Tue Feb 20 18:06:39 CET 2007
It may help to think about what a tag like "ga-GB" would mean.
As Doug recently pointed out, such tags should be interpreted as referring
to language varieties and not to the location of the language producers.
The above tag would indicate a linguistic variety of Irish Gaelic,
distinguished by its association with the geographical entity GB (by which
ISO 3166 means what the rest of the world calls UK). It is not a label for
Irish Gaelic originating in the geographical entity, unless this is
consistently distinguishable in nature from Irish Gaelic originating
elsewhere, which it is not.
There are indeed languages which have different spoken varieties, or more
commonly different orthographies, in different countries, and that is why
country subtags can appear in language tags. But, as the Irish Gaelic
spoken in the United Kingdom is not different from that spoken elsewhere,
there is no such language variety as "ga-GB", nor would the label "ga-UK" be
any more applicable.
I think it is true of the other Celtic languages also, that none has a
distinct form in the United Kingdom, as opposed to elsewhere — unless
Scottish Gaelic in the UK is sufficiently different from that in Canada to
make a distinction, but I doubt it. With that possible exception, there is
no need for -GB at all in tagging Celtic languages. Or for -IE or -FR.
On the other hand, there are dialect variations and orthographic variations
within Irish Gaelic which would be candidates for subtags (variant, not
country). One of the dialects is Ulster Irish, for which we could (and
probably should) have a variant subtag. The correspondence between Ulster
Irish and the geographic entity of Northern Ireland is so inexact that one
would not consider a country subtag. Unless... let us imagine that a
standard orthography was developed for Ulster Irish and that it was adopted
as official in Northern Ireland — sadly, two events for which there is
little demand from Ulster Gaelic speakers — how far we lag behind the
Valencians! — then a tag of ga-GB would be conceivable (though a variant
subtag might still be preferable). And only then would we arrive at the
point where this discussion of -GB vs -UK would have substance.
But in any event, ISO 3166 ought to rename GB to UK. It is not a matter of
etymology but of common usage — no one is likely to interpret "computer" to
mean an office clerk, whereas the correct interpretation of GB is that it
excludes NI. ISO 3166 is simply out of step. But the implications of the
change for tagging Celtic languages would be minimal.
Ciarán Ó Duibhín
More information about the Ietf-languages