Tagging transliterations from a specific script

[Kent Karlsson]

Den 2011-03-17 11:52, skrev "Michael Everson" <everson at evertype.com>:

> On 17 Mar 2011, at 05:15, Peter Constable wrote:
> 
>>> tt-Arab-alalc92 is Tatar language written in Arabic script and then
>>> transliterated 
>>> into Latin using that specification
>> 
>> Nope! The "Arab" subtag is declaring specifically that 'this content is
>> comprised of Arabic-script text', not that it has some indirect relationship
>> to _other_ content that is in Arabic script.
> 
> Yup! Because the alalc97 [sic] subtag is specifically defined as a
> Romanization, so this would imply (to a human at least) that an Arab-to-Latin
> transformation had been applied. (My point being that this subtag has just as
> strong a definition as the ISO 15924 one.)


But BCP 47 does not specify language tags to work that way. According to BCP
47, leading subtags always trump following subtags, no matter what. I'd say
that that has been the case ever since IANA language tagging was first
devised (even if slightly different terminology was used initially).

Ok, there is some reordering issues with extentions, but there is as yet
only one extenstion (denoted by the 'singleton' u)...

    /Kent K

 
> Michael Everson * http://www.evertype.com/