Menu
▾
▴
#326 Review of language, script and region code recommendations
closed-fixed
5
2012-07-08
2011-12-08
No
Various points in the Guidelines discuss the use of language, script and region codes, and as a result of ticket #3304622, we cleaned up our usage in @xml:lang values in the Guidelines prior to the Laurentian release. However, I believe that the section in Chapter CH which covers language identification (CH.html#CHSH) should be reviewed and compared against the latest version of BCP 47, which was updated in September 2009, and in particular in view of the fact that RFC 5646 superceded 4646 at the same time; I have updated the CH chapter to show 5646, but we should review the new version and make sure our recommendations still fit with it.
While we're at it, we should address this ludicrous description for @xml:lang values:
"The value must conform to BCP 47. If the value is a private use code (i.e., starts with x- or contains -x-) it should, and if not it may, match the value of an ident attribute of a language element supplied in the TEI Header of the current document."
I can't figure out what was intended by "it should, and if not it may", but it shouldn't and it may not be left like that, surely.
Seems clear enough to me, tho I agree its not very elegantly expressed. It means that if you use a private use language code you really ought to explain it in the <language> element. And if iyou use a non-private use code, you are still graciously permitted to do so. Very decent really.
It may be decent but it completely threw me. I took it to mean:
If you use a private use code, it should, and if not should then may, match...
From your explanation I see that you mean:
If you use a private use code, it should match the value..., and if you don't use a private use code, it may match the value...
Even that seems confusing at best. I contend that if I can't understand it, there will be others in the world who also fail to grasp it. Think of the <choice><sic>children</sic> <corr>non-native-speakers</corr></choice>.
Proposed rewording:
The value must conform to BCP 47. If the value is a private use code (i.e., starts with x- or contains -x-), a language element with a matching value for its ident attribute should be supplied in the TEI Header to document this value. Such documentation may also optionally be supplied for non-private-use codes, though these must remain consistent with their Unicode definitions.
The need to check what we now say against the latest version of BCP47 is probably a separate issue, which needs a separate ticket.
I think the primary purpose of the ticket was to review CH against BCP 47, so we might as well leave this ticket open -- it has a suitable title. The re-wording was a minor issue.
Have made proposed rewording change and changed ticket status to amber, since it needs to be assigned to someone to do the comparison.
I've now considerably simplified our explanation of language subtags, referring the reader to good external authoritative sources (rev 10650). I'll also assign the ticket back to me (the Michigan minutes gave me this task, I think), and I'll close it when I've had one more look through all our uses of @xml:lang.
Final changes to three @xml:lang values committed in rev 10651. Closing this ticket.