Latvian extlang subtags

Randy Presuhn randy_presuhn at
Mon Jan 25 02:25:50 CET 2010

Hi -

> From: "John Cowan" <cowan at>
> To: "Randy Presuhn" <randy_presuhn at>
> Cc: <ietf-languages at>
> Sent: Saturday, January 23, 2010 5:42 PM
> Subject: Re: Latvian extlang subtags
> Randy Presuhn scripsit:
> > (a) was certainly a consideration.
> > As far as I can recall, (b) was not.  I do not know why (b) would have been
> > a valid criterion.  We would (I believe) have come to the same conclusion
> > regarding 'zh' regardless of the Cantonese:Mandarin ratio.
> If so, why did we exclude sq-als and sq-aln?  Precisely because
> neither variety is dominant.  See, again, Peter Constable's list
> at .
> With the exception of sgn, the languages with extlang tags are precisely
> those listed in his category 2, which he defines as "Macrolanguage ID used
> in ICT, and there is a single, dominant variety".  That is no coincidence.

It's important to me anyway) that the language Doug cited earlier in
this thread  (RFC 5646 4.1.2) does not "and" those criteria together.
Indeed, I had not understood (b) to even be a criterion, but rather an
observation about the data.  As I see it, the overwhelming factor was (a),
and it just happens that cases that satisfy (a) typically (and not surprisingly,
considering the history) satisfy (b) as well.

I think such a reading is further supported by the first sentend of 4.1.2:
"To accommodate language tag forms used prior to the adoption of this
document, language tags provide a special compatibility mechanism:
the extended language subtag."  The whole thing was about accomodating
tags forms already in use, and that's precisely the kind of situation we're
faced with here.  Conceptually, it doesn't seem any different to me from
what happens when variant subtags come into use and the need arises
to explicitly identify the "non-variant" (e.g. "mainstream") variety.  This is
inherent in the way tags are constructed.


More information about the Ietf-languages mailing list