“However, it does not matter if 2 percent or 98 percent of your instances will exhibit the characteristic in question; the model must solve the problem in a way that is valid for all of your data. -Karen Coyle, FRBR before and after: a look at our bibliographic models, p. 135.

Usually in my day to day work in graduate school (where I was a musical score and a sound recordings cataloger), I was/still am perturbed by the basic failure of bibliographic description structures to comprehend the ontological1 being of any musically related non-monograph. LCSH confuses the actual content’s form for aboutness, which narrows the music cataloger’s ability to give an actual subject analysis2 Much in the same way the FRBR model can’t encompass the ways of being that music exhibits. Maybe I have the bias of having thought extensively about the ontological being of music in a different context: copyright law.

The best illustration of the points I will come to make through this post can be demonstrated by Joanna Demers:

Consider “The Star Spangled Banner.” You can own a copy of its sheet music, sing it at the opening of a local baseball game, and listen to Jimi Hendrix’s famous Woodstock performance. You can physically hold a recording of Hendrix’s rendition if you find it at a used record shop, and you can also download this same recording as intangible, digital data through a file-sharing service like Kazaa or Morpheus. All these permutations qualify as “The Star Spangled Banner” but possess different commercial and legal statues. There is no single definitive form of a piece of music in the way that there is one first edition of A Tale of Two Cities from which subsequent editions have been copied. Joanna Demers, Steal this music: how intellectual property law affects music creativity, p. 18.

Demers’s point is that all of those modes of musical being are subject to different, overlapping, and sometimes at-odds copyrights under current US law. But it is a point that bears repeating for the sake of the bibliographic universe. The ontological multiplicity of a musical work, combined with the temporal issue of determining ontologically prior expression and manifestation of a work make FRBR an absolutely unfit model for musical data handled by libraries3. I am, of course, not the first one to say that456. However, even the critiques of the FRBR model from the music world still deal primarily with academic forms: classical and jazz. These critiques also do not think over a long term temporality–the critique of jazz only implicates the act of improvisation-as-composition, while the Variations document makes only minor edits to what appears to be a mostly in line system. These critiques do not take into account the progenation of the work, nor do they take into account how interrelated the family of a particular FRBR musical ‘work’ is conceived of by users.78 To provide examples, I will recount to you some anecdotal methods of musical compositions I’ve been a part of or witnessed as a performer with composer friends and colleagues inside and outside of the academy.

Example 1

A composition student is working on a musical piece for an end of term project that will eventually be workshopped into a published musical score. The student writes the work in musical notation while fleshing out the parts using a single hand at the piano. More notation is worked out on the score after a rehearsal with the student orchestra, which is recorded. The composer later calls the debut with the local professional philharmonic the first and definitive recording.

What’s the expression and what’s the manifestation? Is the expression representative of the work and/or the manifestation? Do we care about anything prior to the professional publication of the score? How do we deal with multiple instances in the mind of the library versus the mind of the composer? Why does the library have what should be archival material? How does the archival material differ from the library material in the bibliographic universe - same “thing,” described differently.

Example 2

A jazz composition student writes words to throw over Rhythm changes at the last minute for an end-of-semester performance. The lyrics are later tweaked slightly and released on this composer’s debut album with a significantly different medium of performance (quintet vs. big band).

Where does the new work begin and where does “I’ve got rhythm” end? How does the whole/part relationship function first within the recording of the semester performance and then within the aggregate work that is the album? What about the written chart, which is an entirely different ball game.

Example 3

A local punk band plays a show at your library with a song titled “Up the Stax.” No one recorded it. The title is on the set list on their website and on your program, which is deposited in your institutional archive.

If a work is expressed and no one is there to manifest it, does it really exist? It exists in the bibliographic universe according to the FRBR report, which doesn’t necessarily require all instances of group 1 entities to be a valid instance of the data model. However: this entity also exists in the bibliographic universe as a group 3 entity, “event.” Works can be related to events, but works cannot be events. The Event-as-Work problem shows up a lot in conceptual performance art, which, like music, is mostly video/audio recorded in terms of manifestation. An artist will tell you that your new library manifestation is a literal instance of a new expression (read: E and M are one in the same, every time).

This goes on and on. We can ask multiple questions about every instance of musical composition anywhere ever. I needn’t say more about how unfit FRBR is for music and musical materials. If this is supposed to be the model of the bibliographic universe for the future of cataloging, quite a lot is going to be missing. Quite a lot is going to be undescribable even if collected. It’s another instance of beating square pegs on round holes. Music catalogers can and will find a different way to deal with this issue. I’m not saying we’ve been right in the past (hello Western classical music-as-ideal), but music is infinitely more complex in the linked data world than your average monograph.

Example 3.5

A work takes four minutes and thirty-three seconds of tacet performance. Expressions and manifestations do not functionally differ from one another but each expression is in a different place with different people in the audience at different times and the manifestations only differ in terms of the “background noise,” which is only functionally part of the work because theoretically the composer said so.


  1. I am referring exclusively in this post to the metaphysics definition, not the computer/information science definition.

  2. I often ask catalogers some variation of the following question: “Have you ever met someone that catalogs Beethoven’s 9th Symphony with ‘Deafness – Psychological effects’ as a subject?”

  3. For clarity, since the terms in this sentence come from my previous research in musical copyright law: ontological multiplicity refers to the multiple modes of being that a particular musical work can be expressed and manifested in [^3.5] while ontological temporality and the notion of ontologically prior refer to the timeline through which the work/expression/manifestation comes into being–the ontologically prior, then, is a progenitor of a later derivative work/expression/manifestation.

  4. Coyle, FRBR, 120.

  5. Raymond Schmidt, “Composing in real time: jazz performances as ‘works’ in the FRBR model,” Cataloging and Classification Quarterly 50, no. 5-7 (June 2012), 653-659.

  6. Jenn Riley, Caitlin hunter, Chris Colvard, and Alex Berry. Definition of a FRBR-based metadata model for the Indiana University Variations3 project, 2007. Retrieved from http://www.dlib.indiana.edu/projects/variations3/docs/v3FRBRreport.pdf. I was an expert user of the Variations2 system and was cataloging at the university when the progeny of Variations3, the Avalon project, was undergoing early implementation.

  7. Coyle is right, in my humble opinion, to say that the musical user community most strongly identifies with the idea of the FRBR work. It’s the rest of the model that is wack.

  8. Coyle is also right to identify the aggregate as a massive problem. Turns out the rate of aggregate works in the music community is absurdly higher than anywhere else (think: every sound recording is an aggregate work).