Re: [Ltru] ISO 639-6 (was: Geocoordinates)

John Cowan <cowan@ccil.org> Wed, 11 March 2009 15:25 UTC

Date: Wed, 11 Mar 2009 11:26:05 -0400
To: Doug Ewell <doug@ewellic.org>
Message-ID: <20090311152604.GA15999@mercury.ccil.org>
References: <3FF1C2BC1E164A1D99E5BA5B6CA09C46@DGBP7M81>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <3FF1C2BC1E164A1D99E5BA5B6CA09C46@DGBP7M81>
User-Agent: Mutt/1.5.13 (2006-08-11)
From: John Cowan <cowan@ccil.org>
Cc: LTRU Working Group <ltru@ietf.org>
Subject: Re: [Ltru] ISO 639-6 (was: Geocoordinates)
Precedence: list

Doug Ewell scripsit:

> I continue to be strongly opposed to adding hundreds or even thousands 
> of variant subtags, using the same mechanism we currently use to 
> register variants one by one.  The boonts and fonipas and njivas would 
> be completely overwhelmed by the oceans of 6xxxx subtags.

And a Good Thing Too.  Boontling is a variety of English, and ought
to have a 639-6 code element.  Fonipa is a kludge to compensate for a
limitation in ISO 15924.  With 639-6 we don't need to register variants,
we make people go to 639-6/RA instead.

> If these don't qualify as language subtags, for whatever reason, then 
> they should be encoded using the extension mechanism described in 
> Section 3.7 of RFC 4646, preferably with an appropriate singleton such 
> as '6'.  This would be orders of magnitude more appropriate than 
> variants.

Most of the 639-6 codes correspond semantically to variant subtags.
Consider 'yoli' (or '6yoli' under my proposal): it's the Yo-Li variety of
the Xinan dialect of Mandarin.  We know this because of the 639-6 parent
data, which shows the following chain:  yoli > xuuk (Xinan dialect,
spoken) > xghu (Xinan dialect) > cmn (Mandarin proper) > micr (Mandarin
cluster) > mycx (Mandarin-Yue complex) > zho (Chinese) > ssic (Sinitic)
> snot (Sino-Tibetan).

This is not, of course, the One True Tree; what's more, it's preliminary
data and subject to change.  But just the fact that 'yoli' is subordinate
to 'cmn', a known individual language code (from the 639-3 data), and
that it isn't cross-referenced with a 3166 or 15924 code, tells us this
is a variant.  The code 'biar' (Bai cluster) > ssic > snot, on the other
hand, is not subordinate to any 639-3 individual language and therefore
is a collection.

-- 
"Well, I'm back."  --Sam        John Cowan <cowan@ccil.org>

Re: [Ltru] ISO 639-6 (was: Geocoordinates) Doug Ewell
Re: [Ltru] ISO 639-6 (was: Geocoordinates) Doug Ewell
Re: [Ltru] ISO 639-6 (was: Geocoordinates) Peter Constable
Re: [Ltru] ISO 639-6 (was: Geocoordinates) John Cowan
Re: [Ltru] ISO 639-6 (was: Geocoordinates) Doug Ewell
Re: [Ltru] ISO 639-6 (was: Geocoordinates) Mark Davis
Re: [Ltru] ISO 639-6 (was: Geocoordinates) Peter Constable
Re: [Ltru] ISO 639-6 (was: Geocoordinates) John Cowan
Re: [Ltru] ISO 639-6 Randy Presuhn
Re: [Ltru] ISO 639-6 Peter Constable
Re: [Ltru] ISO 639-6 Gerard Meijssen
Re: [Ltru] ISO 639-6 Peter Constable
Re: [Ltru] ISO 639-6 Randy Presuhn