German as used in Liechtenstein

Doug Ewell dewell at
Sun Dec 26 03:48:54 CET 2004

JFC (Jefsey) Morfin <jefsey at jefsey dot com> wrote:

>>> I have 42624 locations in the UN/LOCODE. I have 33000 French cities
>>> in my Web of France reference data base and more than 500 listed
>>> local language associations I should attach to them. The algorithm I
>>> use permits to roughly relate them to one of the 266 "business
>>> areas" considering that such areas are also historical and cultural
>>> areas. (For the time being I am focusing on European France only).
>> My Thomas Guide has hundreds of thousands of public and private
>> streets in Los Angeles and Orange counties.
> Street listing is an obvious must, but today no many countries have
> standardized their description (may be for radio frenquency maps?).
> However there is a centuries old algorithm which permits mail to be
> distributed. I am documenting the need of such a canvass algorithm.
> And I have no problem in accepting RFC 3066bis tags as language
> "zipcodes".

Well, that wasn't quite what I meant.  You mentioned a database that
contains 33,000 French cities and implied that these would be suitable
for qualifying language tags.  My point was that narrowing the
geographical scope down to every little hamlet in France, or for that
matter every residential street in Fullerton, is probably too precise
for MOST language tagging purposes.  I can't say ALL, because I don't
know your specific needs.  But in this age of easy travel and frequent
relocation, at least in North America and Western Europe, it doesn't
seem likely that the inhabitants of town A can be characterized as
speaking a recognizably different dialect from those of town B, 10 km

>> None of these is usable under RFC 3066bis or 3066 or 1766.  There was
>> a proposal many, many months ago to allow UN/LOCODE identifiers, but
>> there was apparently no consensus for that level of precision in
>> language tagging.
> Sorry to tell you you are wrong. You currently propose UN/LOCODE with
> a country neuronal level which is ISO 3166. What you want and only cas
> say is "there is no consensus right now for that level of precision in
> RFC 3066bis language tagging", what I support.

When I say there was no consensus, I mean most people *on this list at
the time* did not think it was needed.

I don't propose UN/LOCODE for anything related to language tagging.
UN/LOCODE contains over 600 codes for locations in California alone,
including separate codes for La Habra, Brea, Fullerton, Anaheim, and
Yorba Linda.  This is much too precise for language tagging for me,
although it might be great for other purposes.

I have no idea what you mean by "neuronal" in this context.  UN/LOCODE
starts with ISO 3166-1 country codes, and appends codes for "trade and
transport locations" to them.  The only references I can find for
"neuronal" have to do with the brain.

But you are right, I am thinking in terms of RFC 3066bis language
tagging, not in terms of a dramatically different scope of effort.  This
does not bother me.

> But what we need is a more flexible and well devised canvassing
> algorithm permitting to consistently address all the not fully RFC
> 3066bis supported needs, through solutions fully RFC 3066bis
> interoperable. They will most probably be based upon and adapt the
> smallest granularity which is the human being. Shakeaspeare English is
> not exactly Whodehouse English which is not exactly 5th Avenue English
> which is not exactly my Franglish (hopefully :-).

I have no desire to narrow down language tags to the individual human
being.  Maybe you do.  But in any case, RFC 3066bis will provide two
ways to allow you to define language tags as narrowly as you like:

(1)  Create an extension subtagging standard.  To do this, you would
write an RFC detailing the extension mechanism, in accordance with the
process described in Section 3.4, and get it approved.  You would
probably want to start an e-mail list to discuss this, or see if you
could use this one, since Section 3.4 includes "discussion list e-mail"
among the information to be provided in the RFC.

For example, the extension singleton 'l' (lowercase L) could
theoretically be defined as starting an extension subtag that specified
the exact latitude and longitude where the given language variety is
used.  So you could write "fr-l-4318N-00525E" to refer to a variety of
French spoken in a part of Marseille.  Or, the singleton 'u' could
indicate a UN/LOCODE identifier, so the tag "fr-FR-u-MRS" would
accomplish pretty much the same thing.

(2)  Use private-use subtags.  Despite popular belief, these are not
only appropriate for use between individuals whispering passwords to
each other.  They can be used by anyone or any company, in any protocol,
as long as it is understood that not everyone or every protocol will
recognize them, and that others may use the same codes for different
purposes.  The benefit is that no prior registration is necessary.

With private-use subtags, one could write "fr-x-4318N-00525E" or
"fr-FR-x-MRS" to mean French as spoken in Marseille.

This illistrates how RFC 3066bis can be used to meet the needs of
different types of users, without requiring a whole new thought process.

> Also, an important point: languages are an important part, but for a
> computer scripting is more. In the case of multi-scripted languages it
> makes a real difference.

I don't know what you mean by "computer scripting."  Computer languages
like C and Pascal and JavaScript are not what we mean by "language
tagging."  If you mean "scripts" like Latin and Cyrillic and Katakana
and Hangul, that is one of the major advances of RFC 3066bis.

> Another point important to consider is the country related synonyms. I
> am a cofounder of Eurolinc for European languages to be supported on
> the Internet and the web. One of our main interest is European
> multilingual e-governement issues. When you create an automated
> registration form local legally used synonyms are to be used, because
> a word can have locally added meanings that is ported by another word
> in the same language but in another country. A simple example is the
> name of Ministries, University Grades, personnal situation, health
> status ... were people may not understand, be mislead or hurt by a
> correct word used in a wrong way in their national legal language
> flavor. You refer to language tags. Another form of tags are icons.
> Which icon to use for a language version of a website - this is as
> much important than a code : an icon is a tag for the brain.

Correct language tagging, using existing methods, can help with this
type of localization problem.

I strongly believe that RFC 3066bis (or 3066 or 1766) tags make better
visual identifiers for languages -- on Web sites, signs, brochures, and
so forth -- than little pictures of national flags.

-Doug Ewell
 Fullerton, California

More information about the Ietf-languages mailing list