> In the past, I believe we used xml:lang for the literals
> that represent a natural language like English or Spanish.
> I'm curious, is it generally considered good practice to use
> xml:lang for literals in a broader sense?

When doing so is meaningful it is good practice. "Squeezing" literals into a
language isn't.

> > > is
> > >       defined simple as: "Identifies the class of a resource."
> > >
> > >    Perhaps we should eventually create a dcu:type -- a Usage
> > >    Board element specifically defined as the "DCMI Type of
> > >    Term".  For now, rdf:type seems like perhaps the better fit?
> >
> > I think this is a mistake.  dc:Type is a perfect fit.  I don't
> see a reason
> > to create another term, in a new schema, to describe something that can
> > already be described using the DCES.
> Or using RDF itself
> (

It seems to me that dc:type should be a subPropertyOf rdf:type. I can't
think of any counter-examples. Whether the inverse is true (which would
effectively make them the same) I'm less sure of, as convenient as it would
be to be able to ignore it when composing RDF/XML.

Question, all instances of dcterms:TypeScheme and dcterms:DCMIType seem to
be also of type rdfs:Class. Is there a hypothetical counter-example, or
should they be subClassOf rdfs:Class?