proposed ISO 639 change for "arn"

Doug Ewell doug at
Mon Dec 10 23:57:05 CET 2012

Peter Constable <petercon at microsoft dot com> wrote:

> A proposal has been made to the ISO 639-RA/JAC to make changes for
> Mapudungun. This is currently coded as "arn". The problem is that that
> ID was derived from an exonym that is considered offensive by some/all
> of the language community.

In a sense, the precedent for changing short alphabetical code elements
to achieve political correctness may have been set in 2002, when ISO
3166/MA changed the 3-letter code element for Romania from 'ROM' to
'ROU'. The code element was printed on passports held by Romanians, who
didn't want to be identified as Rom (i.e. Gypsies).

Registration and maintenance agencies for ISO standards sometimes remind
users that with a limited number of available short alphabetical code
elements, not all of them can be natural abbreviations, or intuitive.
They can't always be flattering, either. But requests like this come in

> "arn" has been coded in 639-2 since its inception in 1998, and was
> used in MARC before that. What has brought this issue up at this time
> is, IIUC, that members of the language community want to start
> creating Wikipedia content and, since Wikipedia uses ISO 639 IDs in
> domain names, that would entail using "arn" in the domain name for all
> such content.

There shouldn't be anything special about Wikipedia here, or about
domain names. Existing applications of ISO 639 (or BCP 47) and data have
already been using 'arn', both before and after the name of the language
was changed from Araucanian, and will have to make adjustments if this
is changed.

In any case, we know Wikipedia uses domain names with non-standard
identifiers when it suits them; see "simple" and "roa-rup" and "bat-smg"
and "map-bms" for examples. At least the present request is less silly
than the one to change Wawa to something other than 'www' because that
interfered with Wikipedia's URI naming conventions.

> Two proposals have been floated. Both involve creating a new ID for
> Mapudungun (let's call it "mxx"); they differ in their treatment of
> "arn".
> - Change the scope of "arn" to macrolanguage, encompassing Mapudungun
>   (mxx) and Huilliche (huh)
> - Change the scope of "arn" to language group (it would get added to
>   ISO 639-5 as a result, and technically should be removed from ISO
>   639-3 as out of scope)

Both macrolanguages and language groups are poorly understood as it is;
many people think they are the same, judging by the requests that have
come in to ISO 639-3/RA over the last few years. Unless there really is
a macrolanguage or a language group to be created, one that would still
have been created if not for this vanity request, these approaches would
only expand the confusion.

I'm kind of siding with either Philip (leave the identifier alone, and
explain to people that it's just a bunch of letters and not a value
judgment) or Gordon (withdraw and reassign). The latter option would be
consistent with the Romania change, and is at least honest about what is
being done. On the other hand, as Mark observed, the deprecated 'arn'
would continue to live forever in the Language Subtag Registry, and in
existing data.

Of course, as Gordon pointed out, these are options for the RA/JAC, not
for the Registry, which will follow whatever the RA/JAC decides to do.

> There's no formal action being taken yet in the JAC; this has just
> come up. But one voting member and one observer are suggesting swift
> action. (I'm definitely not in favour of any swift action in this
> case.)

I agree; changes of short, cryptic code elements for political
correctness, even if made, should not be viewed as urgent.

Doug Ewell | Thornton, Colorado, USA | @DougEwell ­

More information about the Ietf-languages mailing list