[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Ltru] going back to the roots to find a solution to "zh"



> From: ltru-bounces at ietf.org [mailto:ltru-bounces at ietf.org] On Behalf Of
> 'Stephane Bortzmeyer'


> > I missed the whole extlang debate so I don't really know all the
> > reasons why it was dismissed.
>
> [My personal summary]
>
> * stability issues (what if a "standalone" language becomes
> encompassed or the opposite?).

Wasn't an issue if we used extlang only for pre-existing cases, not new cases.


> * "dummy" right truncation, as allowed by extlangs, is not a good idea
> for macrolanguages, anyway (two encompassed languages of the same
> macrolanguage may not be mutually intelligible).

An issue only for the cases in which the macro-language ID has been used for some dominant, encompassed language: content would be tagged with (and hence matchable against) the macro-language ID only if there were some conventional form of linguistic expression it was associated with. That would happen only if it were associated with a dominant encompassed language, or if there were some variety of common communication (à la Chinese or African cases Don has alluded to). But in both those scenarios, the fallback probably *would* provide intelligible results for both.


> * politically, it may send the message that the encompassed language
> is inferior in some way.

It seems to me that would only arise if the macro-language ID is particularly associated with one dominant variety -- and that's true whether extlang is used or not. Thus, if we ended up using zh for Mandarin and either yue or zh-yue for Cantonese, that attitude might be conveyed.


I think there were a few other reasons cited:

* matching processes need to be made more complex / right-truncation doesn't work as expected in some scenarios -- an issue for tags with extlang and script or region subtags

The counter-argument made was that matching processes need to be made more complex without extlang in order to capture useful fallback associations between encompassed and macro-language IDs.

The rebuttal to that was that the whole point of extlang was to facilitate right-truncation matching involving existing usage of macro-language IDs, yet extlang breaks some desired right-truncation behaviours.

I'd note that limiting extlang to certain, pre-existing cases such as zh would cover the bulk of cases of concern, though as we know zh is a case for which script or region subtags may have particular significance.


* Not all macro-languages in ISO 639-3 are useful for matching and fallback.

Not an issue if not all macro-language IDs are used with extlangs.

* Fallback requirements are more involved than what extlang can support.

True; but extlang wasn't proposed as a general fallback solution, and the issues exist with or without extlang.


> As Randy said here, you can also make a similar list of the good
> reasons to keep extlangs. The case is really a 50-50 but we have to
> decide on one solution.

That's mostly what it amounted to: a compelling case couldn't be made either way, and those arguing for extlang gave up sooner than those arguing against extlang.



Peter
_______________________________________________
Ltru mailing list
Ltru at ietf.org
https://www.ietf.org/mailman/listinfo/ltru



Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.