doug at ewellic.org
Thu Dec 29 23:52:47 CET 2016
John Cowan wrote off-list (moved back to the list with permission):
> [argument in favor of -t-c0 extension instead of one-off variant]
> I agree, but overloading -t- makes it impossible, as I pointed out, to
> properly represent translations like Michael's English-to-Spanglish
> translation of Alice. A proper generative solution should involve a
> new singleton such as -c- for code-switching.
The more I think about this, the more I agree with John.
The -t- mechanism posits one language as having been transformed from
another, and we have argued, without convincing objection, that
Spanglish (et al.) does not really consist of a source and a target
language. There are sampling variations in which English is sometimes
dominant, and some where Spanish is dominant, but we do not necessarily
want to force the user to tag these differently. And we don't a
situation where "en-something" and "es-something" don't match, when they
really mean trivial varieties of the same thing.
A new -c- extension would indicate explicitly that both languages (or
more than 2, if desired) are presented in free combination and that none
is dominant over the other(s). It should also specify an
extension-specific matching rule that any tag involving both (or all) of
the indicated languages should compare as equal, so that "en-c-es" is
explicitly equal to "es-c-en", and so forth.
Other solutions don't solve all of these problems. "Urgency" should not
keep us from finding the best solution. This won't take that long. We
have a hammer in our hand, but let's not allow that to view this new
problem as a nail when it's really a screw.
Doug Ewell | Thornton, CO, US | ewellic.org
More information about the Ietf-languages