Any attributes for XML is optional, it will be only used when
necessary -- that is the advantage of XML over  MARC which you need to
specify if the subfield is required or optional.

The use of language attribute is needed when the element or field is in
different language from the one that metadata uses, or it carry
specific importance to the user or system usage (limited search by
keyword language/text...)

The best way to control what the language of the data is in should be
done by the metadata, not computer software for good reasons.

Foster Zhang

Quoting Karen Coyle <[log in to unmask]>:

> Here's my "usual" reply to the idea of adding a data element to
> identify
> the language of each field.
> There are fields, like the author's name, which do not render
> themselves
> easily to the "language" concept.
>    Carlos Weintraub
>    Marie O'Reilly
>    etc.
> Which is ok because you can just skip the language data element for
> proper
> names. There are also many titles that are not clearly in a single
> language, or that at least would be hard to identify:
> Title:         Ciao, bella, ciao : roman
> Chacon, Jorge.
>        John F. Kennedy.  [Quito, Ecuador, "La Prensa catolica",
> 1964.]
>   Dollen, Charles.
>        John F. Kennedy.  [Boston, St. Paul Editions 1965]
> Which doesn't mean that a field-by-field statement of the language is
> not
> useful, it's just that it may have to be applied selectively. It's
> much
> easier to attribute language to the elements of metadata that are
> provided
> by the cataloger rather than taken from the item itself, and it is
> vital to
> attribute language to elements like subject descriptors.
> So I think the question becomes whether we need a field-level
> language
> indicator, or if Donna Dinberg's "language of cataloging" serves
> enough of
> our needs.
> kc
> *********************************************
> Karen Coyle           [log in to unmask]
> **********************************************