Hi Joe!

Joseph Kiegel kirjoitti 19.12.2017 klo 19:54:
> I think, broadly, there are two separate cases:  cataloging data created natively in BIBFRAME, and cataloging data converted from MARC records.  I would argue that standards and practices for language tags should be developed based on the needs and possibilities of native BIBFRAME.  Once these are established, they can be simplified to what is possible for converted MARC records, since MARC provides relatively little explicit information about the language content of records.

Excellent point. I agree completely that the native BIBFRAME cataloging 
use case should drive the specification. With conversion there may be 
limitations to what sort of language information can be inferred from 
MARC records, but that shouldn't hold back BIBFRAME itself.

In my experience, problematic situations arise when a MARC record 
describes a multilingual publication, with several main languages and 
titles in different languages. Then it can be impossible to infer which 
title is in which language.

> In native BIBFRAME, why wouldn't we want a language tag for every language string?  If that seems too much, we certainly want language tags when the same string appears in multiple manifestations, e.g. original script and romanization.  Tagging both clearly will be essential for accurate searching, display, etc.

Agreed. Best practice in the RDF world is to use language tags for all 
strings that are in some respect natural language. So this includes 
things like titles, descriptions, names of people (with some caveats), 
organizations such as publishers etc. In contrast, language-agnostic 
data such as plain numbers which represent just that number (not a title 
like "1984") should not be tagged with language tags, but may be tagged 
with a more specific data type such as xsd:integer or xsd:year.


Osma Suominen
D.Sc. (Tech), Information Systems Specialist
National Library of Finland
P.O. Box 26 (Kaikukatu 4)
Tel. +358 50 3199529
[log in to unmask]