Dear colleagues,

Admittedly, I have been rather much of a spectator in the discussion lately
over variant forms, and so also in the most recent matter touching on
transliteration/transcription. However, I feel I have to make something so
that you know I am still around.

1. Variant names etc. 

Both the parts of 639 define codes for languages, not names of languages.
The names are there to help users to identify the language meant by a code
or the code defined for a language. The search process for the former use is
uncomplicated, while the search for the code has to use a common name of the
language as the search argument. There, I feel we have to look to user
friendliness as the first priority and be ready to accept into the database
any name that is known to be used, with the English, French and vernacular
language names as an initial minimum. (This is where the matter of
transcription comes in; I realize that.) 

Furthermore, if these names consist of two words, the first of which is a
chronological or spatial definition of the second one (the real language
name), also an inverted search argument must be supported
        Old English - English, Old

My idea of including vernacular names is to some extent already followed but
not very often for Western European languages. We see both Myanmar and
Burmese, one vernacular and one English, but we can not look up German under
Deutsch or Swedish under Svenska. On the other hand, when vernacular names
have been accepted into English they occur alone: Inupiaq, Gujarati.

2. "Part 3"

I agree with Håvard, and I have felt all the time, that the matter of
additional coding for geographical and chronological variants is not
successfully handled by the present standards and the resulting lists. Other
additions are managed OK, at least as long as the suggested additions are
not regarded to be geographical variants. We may find already that difficult
to decide.

It would seem a practical way of handling them by using something similar to
RFC1766, i.e. adding something to the 639 code. However, neither the
geographical nor the chronological units needed are covered by relevant
international standards, but we must go to national bodies and their coding.
So, "fre-ca" works for Canadian French, but for the dialects of the Swedish
province Dalarna, used in locally published documents every now and then,
you do not find a code to add to "swe" outside the Swedish administrative
coding that uses a format quite unlike ISO3166.

So, the best we could do is to apply the given standards the way they are,
with their pros and cons, but also to keep track of the situations where we
feel that the standard is not sufficient, evaluate that material in good
order and, in due time, report the experience to our TC's. This may lead to
the revision of the standard, and in that work the documented weaknesses are
important input, or to the 
formulating of tasks for new projects. I feel this is about what Håvard is
after in the exercices described.

So, we must not try to introduce novelties "on the fly" but stick to the
approved text AND prepare for a controlled revision of that text in due
course. Of course, this may lead to situations where we feel that we can not
meet all requests and suggestions in a positive way, but there is no real
great harm in letting the languages concerned continue under the code used
up til now some time more.

---

As for the nine distinctive languages distributed as some sort of enquiry by
Håvard, I have to thank Håvard for his clarifications and corrections of my
cases of misunderstanding. I do not add anything now but look forward to
instructions from Rebecca as for the continuation of the matter.

Regards,
Sten 
*********************************
Sten Hedberg
Expert on cataloguing and standardisation
Uppsala University Library
POBox 510, S-751 20 UPPSALA Sweden
Voice +46-(0)18 471 3970
Fax  +46-(0)18 471 3975