Generic and non-generic variants (was: Re: Early Modern English)

Doug Ewell doug at
Fri Jan 13 19:45:40 CET 2012

Sean B. Palmer sean at

> On picking a subtag or subtags that sound too generic and may be
> incorrectly used elsewhere, I would say that doing such a thing would
> be the fault of the user who does so against what is said in the RFCs.

RFC 5646 does not quite prohibit the use of variant subtags with tags
that don't match the Prefix field. It uses words like "RECOMMENDED" and
"meaningful" and "suitable" and "makes sense". This was wordsmithed
rather carefully, as I recall.

Most variants are either explicitly generic, like 'fonipa', or
explicitly appropriate for only one or two languages, such as 'tarask'.
But consider a variant like 'baku1926'. The Registry lists 10 specific
languages that might be written in the Unified Turkic alphabet, but
realistically, we're in no position to say there isn't an 11th and a
12th somewhere. So we don't forbid users from creating, say,
"kaa-baku1926", but we do encourage them to make sure they know what
they're doing.

If we have a variant that isn't intended to be used productively (that
is, with multiple languages), but whose subtag value hints to the
careless reader (too careless to read the Description?) that it is in
fact suitable with multiple languages, then there could be a legitimate
problem with {mis|re}use. This isn't the same as a user tagging
"en-tarask", where it's obvious that the user is being either whimsical,
obstinate, or outright stupid.

In August 2006, Mark Davis proposed 'western' and 'eastern' as subtags
that initially would refer to variants of Armenian, but *by design*
could also be used for other languages that have "Western" and "Eastern"
varieties. Michael opposed this type of reuse and ruled that the
relationships between Western/Eastern Armenian and Western/Eastern Hindi
and Western/Eastern Simlish require separate subtags for all. There was
a lot of discussion and a lot of opinions (including mine, now recanted,
that no variant should be registered without a Prefix), but in the end,
we registered 'arevmda' and 'arevela' to get around this problem.

I don't intend my proposal to register "Early Modern English" to serve
as a test bed for reopening this old debate. I would just like to see a
subtag value that isn't obscure in the extreme. 'arevmda' and 'arevela'
worked because these are roots of the actual words used in Armenian to
describe those two variants. They are not named after people who wrote
books in or about those variants, nor are they derived from the year
such a book was written. That is the kind of clarity I would like for
Early Modern English.

Doug Ewell | Thornton, Colorado, USA | RFC 5645, 4645, UTN #14 | | @DougEwell ­

More information about the Ietf-languages mailing list