Kevin Ford said:

>We are looking at defining the relationships that we absolutely need,
>because they are already supported in our data (7XXs) or because they
>are new/different ...

Library and Archives Canada (LAC) has said that if they use
relationship indicators, in RDA/MARC21, they intend to use $4 codes as
opposed to $e terms, because of LAC's bilingual situation.

SLC would follow LAC for three reason: most of our derived records for
Canadian material come from LAC; we have a multilingual client base,
and we would like to avoid having several lists in different
languages; the RDA terms are split up among different lists, as
well as being in some cases too verbose and redundant (e.g., "composer
of music for a sound film" as opposed to just "composer" in a record
for a sound film).  

The MARC code list is shorter and simpler, is in one alphabetical
order, and has a better list of terms following the codes than the RDA
terms, available in more than one language, which we could use for
export to clients who prefer terms to codes.

Our questions are, will Bibframe have coding for codes as well as
terms, and will the present MARC codes carry over?

How we use MARC21 for RDA would be influenced by Bibframe
expectations.  We would want to crosswall all RDA/MARC21 records to
Bibframe, to avoid having three flavours.  We would want the
AAACR2/RDA and MARC21/Bibframe breaks to be the same, and not have a
midgroup of RDA/MARC21.

Speaking of crosswalks, it would be strange to crosswalk MARC21 245 $b
*and* $c to <title remainder>; we do hope you decide to code statement
of responsibility.  That granularity is needed, particlarly if we are
asked to crosswalk from Bibframe to MARC for a client.  With many
libraries' investment in MARC, we anticipate being asked to do that.

   __       __   J. McRee (Mac) Elrod ([log in to unmask])
  {__  |   /     Special Libraries Cataloguing   HTTP://
  ___} |__ \__________________________________________________________