Herbert Wender wrote:
> Could anyone explain me the subtle difference
> between the 4-digit-string "1970" as value of attribute
> VALUE in the DATE tag and the same 4-digit-string as
> content of this DATE element?
The difference may not be apparent in the case of
<p> ... He graduated in <date value="1970">1970</date> ...</p>,
but even a simple expansion to
<p> ... He graduated in <date value="1970">1970</date> and married in <date
value="1970">the same year</date> ...</p>
should illustrate that the difference is real and important.
The value attribute on <date> is specified in the Guidelines as holding a
standardised or normalised form of the date which is found in the text
content. A simple four-digit year number, as above, could perhaps be
justified as an adequately normalised form for some purposes, but it might
be preferable to follow the Guidelines more closely and use ISO 8601or any
other public standard more appropriate to the material in hand.
The text content of the date element is whatever appears as the textual
representation of the date concerned in the source: hence it lies outside
the encoder's control. The corresponding value attribute is there so that
the encoder can translate that contingent textual representation into a
normalised representation, specified in the teiHeader, which allows all
dates within the document, or the corpus to which it may belong, to be
treated as commensurate.
Commensurability is of course a distinctive feature of dates, but in essence
a normalised form in the markup alongside the actual form in the text plays
the same role for dates as it does for names and places: it allows reliable
automated recognition of identity and difference, another example of how TEI
markup faithfully records what is randomly implicit, while associating it
with a consistently explicit form of the same information, thus making it
accessible to automated processing.