Re: [geolang-comment] what is a language

[Lars Marius Garshol <larsga@garshol.priv.no>]

* Bernard Vatant
| 
| The only thing a neutral PSI publisher can do, as Steve proposed as
| far as I can remember, is to give extensive, authority-referring
| definition(s), like : "Language, as instance identified by an ISO
| 639 language code". 

This keeps coming back, but unfortunately it does not work. Not all
codes in ISO 639 identify a single language, or even a language at
all. 

In Barcelona we agreed to define language (for use with ISO 639) as
follows: "The class of all languages; the notion of language as
implicitly defined by ISO 639."

| So many authorities, so many definitions. And let field experts
| agree on the instances, and map those together if they feel like it.

Exactly. What we will do is to provide a published subject for ISO
639's undefined notion of language. Those who want to can use it,
those who (quite reasonably) claim that it's far too vague can make
their own definition and their own published subject.
 
| I think we should in GeoLang set an example of best practice for all
| those more or less ill-defined classes of more or less well-defined
| instances, in the following way:
| 
| -- If class C is used by authority A to publish a list L of
| instances.  Then a PSI for "class C as defined by A" shoud define
| its subject in an extensive and relative way, like: "Class of which
| instances are defined and updated by authority A in the published
| list L" Where A and L are individuals defined by a PSI ...

There are several problems with that. What if the list L contains
items that are not instances of the common-sense notion C? What do we
do when the instance list changes? Formally, the list then defines a
*different* class from the previous list...

I think the only way to handle this is to go the intensional route as
we did above.
 
| That way, all PSI structure is supported by individuals, which is
| the only way to have effective identification.

Well, I am not sure how effective it is in practice.
 
-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >