Lian and colleagues
This approach raises several important issues.
I think there are logical reasons why acts and scenes in a play should be considered part of the work, but characters or roles should not.
We should be clear about what "characters and roles" refers to. A part of a work is a work in itself, and must be realized as a part of an expression of the whole work. Any expression may be embodied in a manifestation, and a manifestation that embodies more than one expression is an aggregate. A manifestation may embody all or some of the parts of a whole expression, and it is easy to think of examples such as "10 infamous first acts", "5 famous love scenes", etc. where the acts are embodied completely in a volume as texts or in a videodisc as two-dimensional moving images, etc.
It is also easy to think of a manifestation that embodies an extract from an expression, such as "50 first lines of love poems" or "Quotations from Shakespeare". This was treated as an "expression fragment" in the first version of LRMoo, but the distinction was removed in later versions. A discussion paper on the RDA accommodation of an extract/fragment/pseudopart of an expression was tabled for future consideration when it was discussed by the RSC just after the 3R Project. That time may be now, but the current approach of RDA is to consider one or more expression fragments as a new derived work.
I think that a character/role, in this context, can at best be the dialogue and actions assigned to the character in the context of the work. An expression of the work includes the spoken word or text assigned to the character/actor, and stage/film directions undertaken by the character/actor. I think these must be treated as expression fragments, with a derived work that has a creator independent of the works/expressions that have been fragmented and reassembled.
If the dialogue and actions assigned to a character in a play (excluding single character plays) are treated as a part, then I would expect such a 'part' to be embodied in one or more manifestations. Where are the manifestations that embody only the dialogue and stage directions of the character Hamlet? Of course, there are reading scripts, etc. but I doubt if they do not also embody some indication of where in the overall sequence of dialog does one section start and stop. There are Youtube videos of, say, "Best Indiana Jones close-ups", but they are clearly assemblies of disjointed fragments.
Previous discussions by RSC have considered the more general case of contributions to performances that are embodied on video and audio carriers, including the 'role' of costumes (via costume designer), set designs, camera operation, etc. These do not have whole-part or aggregate structures, and require the concept of 'amalgamation work'. Again, the first version of LRMoo had Performance Work as a subtype of Work, but it has been removed in later versions. There are also issues with 'combination works' such as operas, cartoon strips, etc. which combine contributions by (usually) different agents of different basic content given by the RDA/ONIX Framework Character values (https://www.rdaregistry.info/termList/rofch/). An opera combines language and music, a cartoon strip combines language and image, etc., with different creator agents and potentially treated as distinct expressions (e.g. a translation of language, colouring of image, etc. Again, now may be a good time to revive these discussions.
I also have some technical comments about the example:
1. Usage of RDA namespaces. The 'a' statements are not required; <ex:w1 rdaw:P10223 "Animal farm"@en> implies <ex:w1 a rdac:C10001> in RDF and this can be inferred by any automatic linked data reasoner/processor (because the domain of the element is Work). For this reason, it is better to use elements of the appropriate Agent subclass: <ex:a2 rdaa:P50413 "Alexander Raskatov"> has the broader inference that ex:a2 is an agent; but <ex:a2 rdaa:P50117 "Alexander Raskatov"> has the more precise inference that ex:a2 is a person, as well as the broader inference that ex:a2 is an agent.
2. The use of 'kinds' of entity directly as controlled values of category of entity is dangerous. The Toolkit lists kinds of entity that are referenced in RDA guidance and instructions to indicate a controlled terminology that is local to the Toolkit, and defines the kinds in RDA Terms together with other controlled terms that are local to the Toolkit. The terms are published in an RDF value vocabulary as SKOS concepts so that they can be automatically integrated into the Toolkit Glossary with the element value vocabularies and element sets, and to support external implementations of RDA guidance and training documentation. The RDA Steering Committee or ALA Digital Reference may change RDA Terms at any time to suit the needs of the Toolkit, including versions in other language, refinement of guidance, etc. RSC has discussed the provision of entity category value vocabularies on several occasions before and during the 3R Project, and decided each time that no globally acceptable categorization can be developed for any RDA/LRM entity, so it must remain a community matter.
The best way of avoiding danger is to clone the terms required into a separate local value vocabulary for the appropriate entity category element. it is easy to envisage terms such as 'village', 'town', 'city', 'mega-city' as values for RDA 'has category of place', bur a wee bit more difficult to come up with aggreeable terms for 'has category of person' :-)
A local value vocabulary can be published in the Community Resources section to encourage uptake, and in the case of categories, persuade the RSC to reconsider its previous decisions. I therefore strongly discourage the use of RDA Terms URIs in the context of this example. Instead, I strong encourage the use of Community Resources to share RDA problems and solutions from the bottom up :-)
------------------------------
Gordon Dunsire
------------------------------
Original Message:
Sent: Oct 03, 2025 06:40 AM
From: Lian Wintermans
Subject: Musical roles in RDA
Dear all,
I participated in the same project as Thomas and would like to share another solution that we came up with for structured and actionable data about theatrical roles and musical parts. We are curious to hear what you think about it.
You can consider the role or character of Hamlet as a specific part of Shakespeare's play of the same name, just as you can consider the clarinet solo in Mozart's clarinet concerto as a specific part of that concerto.
We believe there is no logical reason why acts and scenes in a play should be considered part of the work, but characters or roles should not. The same applies to movements vs. parts of a musical work.
Here is a (very brief) example:
@prefix ex: <http://example.com/pknet/> .@prefix rdac: <http://rdaregistry.info/Elements/c/> .@prefix rdaw: <http://rdaregistry.info/Elements/w/> .@prefix rdae: <http://rdaregistry.info/Elements/e/> .@prefix rdaa: <http://rdaregistry.info/Elements/a/> .@prefix rdaterm: <http://rdaregistry.info/termList/RDATerms/> .@prefix som: <https://data.muziekschatten.nl/som/> .ex:w1 # Opera Aminal Farm a rdac:C10001 ; # work rdaw:P10004 rdaterm:1118 ; # has category of work: musical work rdaw:P10223 "Animal farm"@en ; # has preferred title of work rdaw:P10053 ex:a1 . # has composer agentex:e1 # World premiere of opera Animal Farm a rdac:C10006 ; # expression rdae:P20331 rdaterm:1153 ; # has category of expression: performance rdae:P20315 "Animal farm"@en ; # has preferred title of expression rdae:P20231 ex:w1 . # has work expressedex:w2 # Part/role of Mollie in Opera Aminal Farm a rdac:C10001 ; # work rdaw:P10004 rdaterm:1118 ; # has category of work: musical work rdaw:P10019 ex:w1 ; # is part of work rdaw:P10223 "Animal farm: Mollie"@en . # has preferred title of workex:e2 # Performance of Mollie during World premiere of opera Animal Farm a rdac:C10006 ; # expression rdae:P20331 rdaterm:1153 ; # has category of expression: performance rdae:P20315 "Animal farm: Mollie"@en ; # has preferred title of expression rdae:P20074 ex:e1 ; # is part of expression rdae:P20025 ex:a2 ; # has singer agent rdae:P20215 som:um2531 ; # has medium of performance of musical content: soprano rdae:P20231 ex:w2 . # has work expressedex:a1 a rdac:C10004 ; # person rdaa:P50413 "Alexander Raskatov" . # has appellation of agentex:a2 a rdac:C10004 ; # person rdaa:P50413 "Holly Flack" . # has appellation of agent
And here is the corresponding graph visualisation:
The disadvantage of this approach is that it seems somewhat cumbersome.
RE The need for 'agent' and 'collective agent' instrumentalist elements in RDA was discussed by RSC and members of the music cataloguer community during the 3-R project. There are sufficient examples, such as two persons playing a single bagpipe (one on chanter, one on bag inflation :-), to support their inclusion.
I had always assumed that collective agent instrumentalist and collective agent singer were intended for orchestras, bands, choirs, etc. I mean, two people who play one instrument together, but do not perform as a duo (but simply as two individuals) are still two instrumentalist persons, not an intrumentalist collective agent, right?
Thank you in advance for your feedback!
------------------------------
Lian Wintermans
RKD – Netherlands Institute for Art History
Original Message:
Sent: Jul 09, 2025 02:23 AM
From: Thomas Op De Coul
Subject: Musical roles in RDA
Dear Keith and Gordon,
Thank you so much for your generous comments on my question! That certainly is excellent food for thought for our project.
As regards a vocabulary for Medium of Performance, I'm not sure if any of you are at the IAML in Salzburg, but one member of our projectgroup, Eric van Balkum, has given a presentation there on a multilingual thesaurus of vocabulary/thesaurus that was developed by him and myself, initially intended for Podiumkunst.net. Here, the scale is significantly larger, amounting to about 3.000 terms... Time will tell if that is too much of a good thing, but you can find it, provisionally, here:
And it will in due time be included on the Dutch 'network of terms' which contains several vocabularies in LOD format:
Thanks again!
Thomas
Original Message:
Sent: 7/8/2025 8:39:00 AM
From: Gordon Dunsire
Subject: RE: Musical roles in RDA
Colleagues
Some wider context to Keith's good advice may help.
RDA provides the Community Resources section of RDA Toolkit to accommodate the requirements for finer element granularity in RDA communities. If a community develops a set of refinements to existing RDA elements, it can be published in the Toolkit and maintained with the same technical infrastructure.
There are two significant issues that the music community must resolve in developing a set of refinements for the RDA element 'instrumentalist person' ***.
I am assuming that if an instrument is used in a recorded performance, then there must be an instrumentalist who plays it.
The first issue is the scale (pun intended :-) of any set of instruments/instrumentalists that is required for international use across all music genres. The IAML Medium of Performance Vocabulary has nearly 400 entries for instruments arranged in several hierarchies. I don't know how many entries for instruments there are in Library of Congress Medium of Performance Thesaurus for Music, but it is of a similar magnitude, and there are similar hierarchies. 400 instruments corresponds to 400 instrumental roles and 400 custom elements. Each custom element must have a distinct label (for example 'bagpipe player person' , a clear definition, and appropriate scope notes. For integration in RDA Toolkit, the custom element hierarchies (for example 'woodwind player person') must be semantically coherent.
The second issue is that both the IAML and Library of Congress vocabularies are in widespread use, and should be consolidated into a single set of instrumental roles and a single hierarchical structure to support the international RDA music community. I did some work on this several years ago, on behalf of the Permanent UNIMARC Committee, and identified matches, gaps, and mismatches if the vocabularies were to be merged. Unfortunately, this has not yet been followed up by IAML or MLA. A merged list is obviously useful for interoperability of the RDA 'medium of performance' element.
I do not think there are any RDA Toolkit technical barriers to resolving these issues.
*** The primary function of RDA element labels is to provide a unique string that distinguishes one element from another and to indicate appropriate hierarchy structures between elements. This is not always 'user-friendly' for anglophone cataloguer agents, but RIMMF has shown that cataloguing interfaces can easily display different labels for different users. The more difficult process is to develop a set of unambiguous labels that are a one-to-one match with the RDA labels (that is, synonyms).
The need for 'agent' and 'collective agent' instrumentalist elements in RDA was discussed by RSC and members of the music cataloguer community during the 3-R project. There are sufficient examples, such as two persons playing a single bagpipe (one on chanter, one on bag inflation :-), to support their inclusion.
------------------------------
Gordon Dunsire
Original Message:
Sent: Jul 07, 2025 04:11 PM
From: Keith Knop
Subject: Musical roles in RDA
Hello Thomas,
First of all, you are correct that RDA itself does not get more specific than "singer" or "instrumentalist" in terms of agent-to-expression relationships for musical performances. If you consider it vital to include the specific instrument/voice role of that performer, you have a few options, depending on exactly what your final product is intended to look like. The simplest option is to provide that information in a note. In the context of MARC cataloging, that level of specificity is usually given in a participants note, which might look something like this:
Dynamis Ensemble (Birgit Noite, flute ; Rocco Parisi, clarinet/bass clarinet ; Paolo Casiraghi, clarinet ; Sergio Armaroli, percussion ; Candida Felici, piano ; Dominique Chiarappa-Zyrd, violin ; Teresa Felici, violoncello) ; Javier Torres Maldonado, conductor.
However, if you want everything to be structured data, you could consider extending RDA with your own elements (and probably labels for those elements as well, since the RDA element names do not exactly trip off the tongue…) at whatever level of specificity you require. So for example, you could extend, say, the element Expression: instrumentalist person with narrower elements for violinists, guitarists, etc.:
- instrumentalist person [RDA element]
- guitarist person [custom element]
- oboist person [custom element]
- violinist person [custom element]
- …
- singer person [RDA element]
- alto person [custom element]
- soprano person [custom element]
- …
Medium of performance of musical content is not going to be much help in that context-though of course you might wish to include it in its own right-simply because, as an attribute element, it describes an expression without relating it to any other entity. However, I do want to note that there is nothing intrinsically limiting it to notated music expressions, and the Music Library Association Best Practices for RDA recommend routinely recording it for performed expressions as well.
As for the different meanings of "role" in the context of theater, again the simplest option is to record that information in a note. Here practices vary, but it is fairly common for opera recordings to have a participants note that specifies the character instead of (or in addition to) a voice type, e.g. "Joyce DiDonato (Donna Elvira)."
However, if you want that data to be structured and actionable, there is one particular hurdle to overcome, which is that fictional characters and non-human "agents" are both outside the scope of RDA. It is, however, possible to link an RDA entity to a non-RDA entity using the very generic "related entity of…" elements, which, again, you could extend with your own more specific narrower elements. So to use your example of John Williams playing the role of Hamlet in Hamlet, and assuming for the sake of simplicity you use Wikidata for theatrical characters, you might end up with a pair of relationships like this:
<some specific performance of Hamlet> <has actor person> <John Williams>
<John Williams> <has related entity of person> <Q2447542>
------------------------------
Keith Knop
Head, Music Cataloging
University of Georgia
Original Message:
Sent: Jul 07, 2025 05:08 AM
From: Thomas Op De Coul
Subject: Musical roles in RDA
Dear collective wisdom,
I have a question that we are struggling with in a project in the Netherlands. The project is called
Podiumkunst.net (site only in Dutch) and aims to fashion a central website where visitors can search and find resources in the performing arts, from, e.g., manuscript or printed scores to audio recordings - and resources related to specific musical or theatrical performances. It has been decided to adopt RDA for modeling the metadata for this website.
Now, one crucial part in modeling the metadata that is concerned with music and theatre is the role which a given person plays in a performance. Of course, in RDA, there are several roles available, such as composer (of a musical work), conductor (of an expression of that work), dancer (in the expression of a ballet work).
For the performer in an expression of a work there are, it appears, significantly less terms available for roles. The terms 'singer' and 'instrumentalist' are available, but more specific terms like 'pianist', 'soprano', 'violinist' etc. are, it seems, not available.
There is, in RDA, of course, the medium of performance of musical content. But, if we understand correctly, that is especially intented for the musical work and (notated) expression - but not for the (performed) expression. Put differently, an agent can be related to an expression of a work, but that agent cannot be further defined/specified by what role that agent has.
In theatre, the situation is different again, as 'role' has a somewhat different meaning there: the agent/person 'John Williams' might, for example, perform the 'role' of Hamlet in the play.
I hope this makes sense... would anyone have any insights on this topic?
yours,
-- Thomas Op de Coul
Mediabeheerder / muziekweb.nl
M 06 - 424 76 535
Kamer 3.02
Werkdagen: di, wo, vr
Media Parkboulevard 1, 1217 WE Hilversum
Postbus 1060, 1200 BB Hilversum | beeldengeluid.nl