Criteria for languages?

Michael Everson everson at evertype.com
Mon Nov 23 07:43:52 CET 2009


On 23 Nov 2009, at 02:08, Peter Constable wrote:

> If you choose option 2 and there are Latgalian documents out there  
> tagged as lav, then by narrowing the scope of lav to exclude  
> Latgalian those documents immediately become _incorrectly tagged_  
> without warning to anyone that might be adversely impacted, and no  
> particular way to discover the problem other than to inspect all  
> documents tagged lav all over again to discover the minority that  
> are incorrectly tagged.

This begs the question as to whether there *are* any Latgalian  
documents "out there" tagged as lav. Electronic documents? Find 'em,  
please.

The history suggests to me that it is similar to Bavarian or Swiss.  
Doubtless there are some books in those  dialects/languages/speech- 
varieties which have been tagged as "German".

> It is a collective business decision to choose #2 over #3: that the  
> narrowing of scope of lav has no measurable negative consequences,  
> effectively assuming that nothing tagged lav is actually Latgalian,  
> and so effectively saying it is safe to assume that lav excluded  
> Latgalian all along. That is somewhat risky, though in a very few  
> instances since 639-3 was published the JAC has gone in this  
> direction. If there's reason to  say that users may have reasonably  
> assumed Latgalian (or whatever variety is in question) was  
> encompassed by the existing entry, then we should act conservatively  
> and with caution. This is why the text of 639-3 states that the  
> denotations of existing entries are not to be narrowed.

I think it's "reasonable" to assume that until really rather recently  
there was little written in Latgalian at all.

> Now, in the case of lav, the MARC Language Code List (http://www.loc.gov/marc/languages/language_name.html 
> ) explicitly states that Latgalian is encompassed under lav. So,  
> option 2 would break compatibility with MARC. We should not consider  
> option 2 unless the MARC community explicitly indicates that this is  
> acceptable.

To when does that date? The document seems to be dated 15 October 2009.

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



More information about the Ietf-languages mailing list