Proposal: Language code "de-DE-trad"

Torsten Bronger
Mon, 11 Feb 2002 22:15:01 +0100

The advantages are that date tags are more systematic and flexible,
and allows for gentle fallbacks of the software, even though the
RFC3066 wouldn't like that.

However, for symmetry reasons, the significant date is 1901, when the
(now) traditional orthography was accepted.  (Because in 1996, the new
orthography was accepted.)

It was followed by a transitional period, but it's difficult to say
when it was over; maybe around the publication of the first uniform
Duden without ambiguities in 1915.  However, I doubt that one could
say that these 14 years had a distinct orthography variant that
deserved a tag of its own.  [By the way, the new reform, too, includes
an (official) transitional period, that will last until 2005, if I
remember correctly.]

Since then, we've had the normal development of a language.  You could
give every of the 22 editions of the Duden a tag, but actually they
just reflected the usual minor changes, e.g. new words that came with
the digital age.  The development was quite smooth, until 1996, when
the first "unnatural" change since 1901 was enforced.

So, if you really want year tags, I think -1996 and -1901 are the way
to go.  I admit that this would mean that, in order to keep the
semantics, no further dates *between* 1901 and 1996 could be added, so
the year must be carefully chosen.  But I think it is.

There is one problem though: This -1901 thing is not very intuitive.
You must know that many Germans (including myself) decided to stick to
the old variant (partly for silly reasons, I admit).  If I use the
authoring tool for which I needed these tags originally, I'd had to
write "de-DE-1901" at the beginning of e.g. every letter to the family
or of my PhD thesis.  For me this is just odd and funny, for others it
may be very confusing.  -1945 or -1955 sound more normal, however, as
already mentioned, there was no break in the language at all in these
years, so they would be arbitrary and somewhat artificial.

After all, handling every case elegantly is a good thing to do, but we
mustn't forget what will be the by far most frequent cases.

Maybe "-trad" as an alias for -1901?  That would leave the (possibly
language spanning) systematic approach of time tags intact, but would
also allow for more pragmatic use.