Renato, thanks for your comments.
- Make dc:language an optional element.
- Change the cardinality of dc:language to "unbounded".
- Change the recommendation about the content of dc:language by including
ethnologue codes as another option (probably the main option). Note that it will still be just a recommendation, not a normative statement.
Ok. Perhaps we should add an optional attribute also, for specifying the used code standard, if any? That should not affect current implementations I think. Problem is that you cannot do anything with an abbreviation if you do not know what it means. Making assumptions can be dangerous. For instance you could asume that "SW" means Swedish, or that it means Swahili. If you know that it is an IANA subtag, you can use it and you can also raise an error if there is an abbreviation which is not present in the used standard.
Another comment about the Tapir metadata: when giving courses in installing Tapirlink, I noticed that none of the about 10 (Dutch) students could figure out themselves what 'relatedEntity' means. They all needed help on that. Perhaps the documentation of that element should be expanded?
Cheers, Wouter