The mapping of the 382 to a note field is indeed unfortunate, since it 
basically removes all the benefits of the field in the first place--the 
ability to control terms (through the LC Medium of Performance thesaurus 
for music (LCMPT) or other set of terms), and the ability to search for 
a very specific instrumentation--and passes over a rich source of 
curated linked data, by reducing and combining the terms into a string.

The string, as currently converted, also leaves out all the subfields of 
the 382 beyond $a, and thus omits details about the number of performers 
on each instrument ($n), whether an instrument is acting as a soloist 
($b), or an instrument is played by the same performer as another 
(called "doubling" in music) ($d), any note (such as chorus voicings) 
($v) or the total number of performers ($s). The resulting converted 
string may actually mislead a  person searching for a particular 
grouping of instruments since it may be incomplete or simply incorrect. 
As an example, this is what a 382 for Villa-Lobos' Bachianas brasileiras 
(a work for originally for soprano and 8 cellos) would look like:

382 01 $b soprano voice $n 1 $a cello $n 8 $s 9 $2 lcmpt
The conversion would come out like this:
<bf:musicMediumNote>soprano voice cello</bf:musicMediumNote>
This becomes in fact a completely different medium of performance--a 
duet for soprano and cello--and even that is a little confusing to read, 
given the lack of a comma between the two media.

As Kirk-Evan says, this aspect is crucial for those who search for 
musical resources and the conversion of the 382 as it stands does not 
serve them well. The 382 is fast becoming the primary field for 
recording medium of performance in RDA records and will also eventually 
replace medium of performance headings in LCSH. It does need to be 
converted in a more meaningful way and there will need to a way 
developed in Bibframe profiles (at least) to input its equivalent there.

Nancy Lorimer
Music SACO Coordinator

On 9/3/2014 7:52 PM, Kirk-Evan Billet wrote:
> Thank you all for the helpful responses to this question.
> But to continue, I am rather perplexed at learning that MARC bib 382 is being mapped to a note field. Since the LCMPT (Library of Congress Medium of Performance Thesaurus) has recently been released as linked data, why would we want to be so “string-y” about medium of performance in Bibframe? Also, I have noticed that, in current transformations, MARC bib 048 (the “traditional” place for coding instrumentation before upstart 382 came along) is not being mapped at all. (Of course, some kind of programmatic conversion from 048 to 382 may be in our future.) Moreover, I want to be thinking also about new data we may soon begin to generate, outside of transforming what we already have. This aspect of music resources is crucial.
> Best day,
> Kirk-Evan Billet

Nancy Lorimer
Head, Music Technical Services
Stanford Music Library
[log in to unmask]