Documenting Performance and Contemporary Data Models: Positioning Performance within FRBR and LRM

One important aspect of performance documentation is the structure and models of data relating to performance. Documenting performance is a rapidly developing and changing field, as attested to by the work of various performance scholars across the world, the Documenting Performance project at City, University of London (Documenting Performance, 2017), among others. However, in a related universe, the bibliographic world has seen great changes in how they model bibliographic data over the last twenty or so years, through the model called Functional Requirements for Bibliographic Records (FRBR), which has recently been superseded by the IFLA Library Reference Model (LRM). So, this article is going to put these developments together by considering performance documentation through the lens of the FRBR and LRM models. There are a number of reasons why exploring FRBR and LRM in relation to performance documentation is worthwhile. As libraries hold some materials relating to performance (for example, collections of theatre programmes), understanding how these materials fit into the dominant library models is useful. For those working specifically with performance and its documentation, the FRBR and LRM models provide an alternative way of modelling the performance world, and ask interesting questions about the nature of performance documentation. 
This article starts with a brief precis of FRBR and LRM, followed by a summary of existing literature which discusses the treatment of performance within FRBR/LRM. A significant article by Miller and Le Boeuf (2005) is introduced: this paper models performance in FRBR, and forms the starting point for discussion. Then, three specific areas of performance documentation are discussed, describing and questioning the consequences of Miller and Le Boeuf’s (2005) realisation: performance ephemera such as programmes, and the interplay between performance document and performance-as-document; the relationships between performance and recordings, in particular how the Miller and Le Boeuf model fits (or not) within current realisations of FRBR; the issue of whether all performance (for example, dance, theatre, music) can be treated as one within a FRBR or LRM universe. The article shows how there is a tension between taking a pure FRBR approach which only places traditional ideas of performance documentation within its structure, and the Miller and Le Boeuf (2005) position of performances being the central unit, which could be seen as a proto-realisation of performance-as-document.


Introducing FRBR and LRM
FRBR is "a conceptual model of the bibliographic universe" (Tillett, 2003).The FRBR model was developed by IFLA and first published in 1998, after development during the 1990s (for a brief history of the initial development of FRBR, see IFLA, 2009, pp. 2-3).FRBR is structured as an entity-relationship model (IFLA, 2009) meaning that at its essence, there are (bibliographic) things and relationships between those (bibliographic) things.As a bibliographic model, FRBR does not give rules or guidelines about cataloguing; instead, FRBR is a structure and a way of breaking down and visualizing the bibliographic world.Its purpose for modelling the bibliographic world, rather than a broader concept of information, is important when we consider its application to performance materials.
Post-1998, two important developments happened relating to FRBR.First, FRBR became the fundamental structure and conceptual engine room of the new cataloguing guidelines, Resource Description and Access (RDA); these guidelines were first disseminated in draft form in 2005, and first published through the RDA Toolkit in 2010.So, although FRBR is a model, its structure is very much enshrined in a real-world cataloguing; therefore, while this paper focusses on FRBR, we cannot entirely ignore the treatment of performance materials in RDA, as sometimes this represents the practical realization of performance documentation within FRBR.
Second, FRBR expanded and evolved after its initial development.New models were published which enhanced the initial FRBR model, such as Functional Requirements for Authority Data, known by its initialism FRAD (IFLA, 2013).In addition, there were developments to harmonize FRBR with models from other information environments; for example, an IFLA working group created an extension to the museum conceptual model of CIDOC CRM, which brought together FRBR and CIDOC CRM into the formal ontology known as FRBRoo (IFLA, 2016;Le Boeuf, 2012).In 2016, the first draft of a new model was published, which brought together various FRBR family models and developments.IFLA LRM (initially given a working title of FRBR Library Reference Model) superseded FRBR and FRBRoo, and a final version was approved and produced in August 2017 (IFLA, 2017).So, this paper will discuss both FRBR and LRM: while LRM is the more recent model and so provides the most contemporary thinking about bibliographic structures, its newness means that most of the literature about bibliographic modelling of performance and other event-like ideas are discussed in terms of FRBR rather than LRM.
FRBR and LRM contain a number of entities, but this paper will focus on four of these: work, expression, manifestation and item.In simple terms, the work is the creative act.This work is realized in communicative form in the expression.This expression is disseminated or published through the manifestation.Any individual realization of this manifestation is an item, which may or may not be corporeal, and indeed items are frequently electronic.In FRBR, entities are divided into groups, and this set forms the Group 1 entities (IFLA, 2009); conversely, in LRM, these groupings have been disbanded, so the entities are just four amongst a general group of entities.The entities of LRM are visualized in Figure 1 in their hierarchical format.As Res is the overarching entity of LRM, this is shown as the top of the hierarchy, with entities such as work, expression, manifestation and item shown as being parts of Res.Similarly, person and collective agent are types of agent, so again are shown in a hierarchical pattern as part of the agent entity.Finally, the entity of nomen is the naming aspect which is related to the other entities, so Figure 1 shows this in a separate space.1

A Brief Literature Analysis of FRBR, LRM and Performance
Part of the discourse about FRBR and LRM discusses non-textual materials, which includes discussions about performance.An extremely significant paper by Miller and Le Boeuf (2005) asks how performing arts can fit into a FRBR environment, as well as giving historical context to how previous cataloguing rules and models conceived performance.Their paper culminates in a model of the work, expression, manifestation and item entities for live performancesalthough, primarily covering dance and theater-in what they term "extended FRBR" (Miller and Le Boeuf, 2005, p. 168).Other authors also discuss performance and FRBR models: a conference paper by Doerr, Le Boeuf and Bekiari (2008) explores how FRBRoo can be used for performing arts, through the entities Performance Work, Performance Plan and Performance.Unlike the earlier paper by Miller and Le Boeuf, these entities are in the "official" text of FRBRoo (IFLA, 2016) rather than an individual author's "unofficial" extension.There are also papers about FRBR which indirectly relate to performance.For example, Taniguchi (2013) discusses expanding FRBR and FRAD to include events; so, this could have some interest to those considering performance.
The way that performing arts such as music fit into FRBR has received significant interest.It is noteworthy that some of the issues in music also apply to other performance types, whereas other issues only concern sonic communication.The expression entity receives a lot of interest in discussions about FRBR and music.For instance, Vellucci (2007) argues that music used FRBR-like ideas of splitting resources into works and items, long before FRBR and its Group 1 entities came along; Holden (2013) suggests that one of the issues with music and FRBR is the number of different types of expressions and posits a typology of types of expression relating to music.Of course, one type of expression associated with a musical work is the musical performance; Le Boeuf (2005, p. 117) theorizes that the expression of musical performance is the transformation of the musical work into "sonic signs."Meanwhile, FRBR is found to be unsatisfactory when it comes to the issue of defining the musical work, especially for jazz and popular songs.Schmidt (2012) argues that jazz improvisations are new works, rather than the unsatisfactory idea within FRBR that all performances are expressions born from a singular musical "work."Clearly, jazz does not fit into a world where composed, notated text is considered the supreme creative act.This is a particularly useful parallel to other performing acts, and the interrelationships between authors, texts and performance-as-creation.
Finally, discussions about FRBR (and friends) are not limited just to those within the library and information science communities.A recent book chapter by Pendón Martínez and Bueno de la Fuente (2017) discusses FRBRoo and how it could be applied to performing arts, using a specific example of a collection at the Teatro Muncipal Miguel de Cervantes to illustrate their ideas.However, as the particular entities in FRBRoo focused on performance do not seem to appear in FRBR and LRM, this article is not going to discuss this book chapter or indeed FRBRoo in detail, focusing instead on FRBR and LRM.

Analyzing Miller and Le Boeuf
We now turn to the seminal article about FRBR and performing arts by Miller and  Le Boeuf, published in 2005. 2 We are interested primarily in the model proposed at the end of the paper, along with its commentary.These will be used as a base point to discuss particular issues in modelling performance.It is important to note that the proposed model does not strictly follow FRBR, but instead uses a version of it.In the words of Miller and Le Boeuf, they are not "FRBR fundamentalists" (2005, p. 168).So, while Miller and Le Boeuf (2005) follows the spirit of FRBR, it cannot directly be placed back within FRBR as is and the authors state that to be adopted for use within FRBR itself, a series of new entities would have to be developed (Miller and Le Boeuf, 2005, p. 171).
In simple terms, pure FRBR treats a play, symphony or opera as a work; whereas, any performance of that opera, play or symphony is treated as an expression.3Dance is more complicated, according to Miller and Le Boeuf (2005): historically, the Anglo-American tradition has treated the choreography of dance as works in their own right.It is important to note that pure FRBR models the bibliographic universe, and performances are only visible by the evidence they leave behind through capture, recording and documentation, and so on.Conversely, Miller and Le Boeuf's (2005) model has a different conceptual basis and designates performance as a set of entities in their own right.Matching pure FRBR, realization of a work in a spatio-temporal realm is treated as an expression in Miller and Le Boeuf (2005).However, unlike pure FRBR, the work which this expression realizes is not the text (for example, a play, an opera, and so on) but Miller and Le Boeuf (2005) calls "mise-en-scène/choreography"-in other words, the creative act of making a performance.Two types of expression descend from "mise-en-scène/choreography": what the authors (Miller & Le Boeuf, 2005) call "semiotic system: spatiotemporal process," which is the communication of that performance, and from which we get a run of performances (manifestation) and individual performances (item); and, "semiotic system: notation," which is the creative performance in notated form, and leads to choreographic notation, director's notes, and so on.
The model and discussion in Miller and Le Boeuf (2005) are part of an extremely detailed and analytical account of FRBR and performing arts; however, the early date of 2005 means that it is valuable to rethink FRBR and performance documentation using the vantage point of 2017.There have been a number of changes since 2005 which have had an impact on how Miller and Le Boeuf's (2005) model could be viewed today, aside from the advent of LRM which has had little practical effect so far.First, the interpretation of FRBR through cataloguing performance-related materials in RDA during the period 2005 to the present day, conflicts with the core ideas contained with Miller and Le Boeuf's (2005) extensions to "pure" FRBR.
Second, ideas about documentation have changed over time.For instance, Buckland's (1997) seminal paper explores thinkers from earlier in the twentieth century such as Briet, who propose that a "document" can exist outside of just a textual environment.This means that many things which document a performance-such as costumes, set designs, recordings-fall into the document category once a document is not limited to text or two-dimensional objects; over the 2000s and 2010s, the definition and meaning of document and documentation are debated, for instance, discussions take place about whether intentionality is an important part of being a document (Buckland, 2014, p. 179).As documentation discourse advances in the 2000s and 2010s, a particularly relevant question emerges: can performances be documents?Buckland (2015) appears to suggest they can, as he gives performance as something which could be considered a document once the definition of a document is broadened to include any object which can we can learn from.This question of performance-as-document has interested other researchers and students: for example, the idea of performanceas-document is delineated and discussed in detail as part of a master's dissertation in documentation studies which looks at distributed performance (Sømhovd, 2011).Therefore, not only are many performance-related objects discussed in Miller and Le Boeuf (2005) considered as documents in their own rights in contemporary thinking, but there is also a good argument that the performances themselves, which are the central unit of Miller and Le Boeuf, are also documents from the perspective of contemporary documentation thinking.So, from the perspective of the late 2010s, discussions about FRBR/LRM and performance involve contemplating how FRBR/LRM can be used for both performance documentation and performance-as-documents.
Three specific areas from Miller and Le Boeuf's (2005) model will now be discussed.For each area, we will look at how these aspects fit into contemporary ideas of performance, how they relate to practices within the modern cataloguing world, and we will consider any potential complications wrought by juxtaposing pure FRBR and the Miller and Le Boeuf (2005) conception of FRBR and performance.

Area 1. Programs
One of the most interesting areas concerns the treatment of documents such as theater programs.In pure FRBR, a theater program would be considered a work: the programs produced for a particular production run would be a manifestation, and the individual program that I buy at the theater would be an item.However, there is no link from the program to the performance itself (as work, expression, manifestation or item) because, in pure FRBR, the performance does not exist.How can the program relate to the performance, if the performance itself does not appear in the bibliographic universe?Miller and Le Boeuf's model is based on performances having their own entities; so the set of works, expressions, manifestations and items entities for the theater program is linked to the run of performances (the manifestation). 4rom a conceptual level, the program example is particularly interesting as it considers the relationship between one type of performance documentation (the theater program) and a performance (which could also be considered a document in its own right from a documentation studies viewpoint).Thinking about these differing viewpoints is not new.For example, in the 2000s a number of performance-related projects had to make a fundamental decision about whether they would use performances or objects as their central unit-see, for instance, the Royal Opera House's (2017) performance database, as an example of the former.The next question to ask is what happens in a performance-centric model, such as Miller and Le Boeuf (2005), if the performance is taken as a document itself meaning both performance and program are now documents.Documentation theorists discuss the idea of documents based on other documents (Briet, 2006;Roux, 2015), seeing a division between the initial, primary document (so in our example, the performance itself) and the secondary document derived from that initial document (so in our example, the theater program).So, one possibility is that the idea of derived documents from documentation theory could be seen as an extension of the existing idea within the FRBR/LRM universe (and RDA) of derived works; this more general relationship of "derivation" between all documents would enfold documentation theory ideas of the relationship between an initial and derived document into FRBR/LRM, while helpfully formalizing the relationship between performance and program within FRBR/LRM.Thinking about what it means for a program to be derived from a performance, especially considering that performances follow programs in terms of their temporal creation, is an area worthy of further contemplation in the future.

Area 2: Relationship Between Performances and Recordings
The second area to consider is the relationship between a performance and a recording of that performance.In pure FRBR, a recording of a live performance is treated as an expression of the textual or choreographic work from which the performance is based; for example, an audio-visual recording of an opera is an expression of the musical work, an audio-visual recording of a ballet is an expression of the choreographic work, and an audio-visual recording of a play is an expression of the textual work.A worked through example using a performance of Shakespeare's Romeo and Juliet is given in the OLAC guidelines for DVDs and Blu-Ray discs (Online Audio-visual Cataloguers, 2015, p. 20), which implicitly gives the recorded performance as an expression of the work-asplay.Conversely, Miller and Le Boeuf (2005) model the recording as a separate family of entities.The recording work family can be related to performance at expression (performance as process), manifestation (run of performances) or item (individual performance) level, but it is always a separate family of works.See Figure 2 for a comparison between pure FRBR and Miller and Le Boeuf, which shows how the relationship between performance and recording is hierarchical in pure FRBR but more equivalent in Miller and Le Boeuf (2005).Note that only the part of Miller and Le Boeuf (2005) relevant to recordings has been shown, and both sides of the diagram have been adapted from the originals in terms of terminology and visualization in order to aid comprehension and comparison.Conceptually, the Miller and Le Boeuf relationship between performance and recording is attractive.They are stating that the act of recording the performance alters the essential creative work, creating a new work (the recording).This solution assumes that the recording has altered the creation of that performance and the recording is a separate work from the performance.5However, in reality, a wide variety of circumstances and creative processes can lead to a recording, with varying levels of creative input from the recording team; for example, one fixed video camera may not make any impact on the live performance, while making decisions about set design, movement and costume for the benefit of a live broadcast would certainly alter the performance.So, I would argue that there is a question about whether all recordings should be considered as separate works, and if not, where the line between recordings that just record, and recordings which create, should be.

Area 3: Across the Performing Arts
The third area to consider is how FRBR and LRM can be applied not just to theater and dance, but across all the performing arts.To start, the model in Miller and Le Boeuf (2005) is specifically designed for theater and dance.However, there are other performing arts which are not covered, such as music without a staged aspect-for instance, a performance or audio recording of a symphony, rock song or folk song.So, we need to consider what a general performing arts perspective on FRBR and LRM would look like, and whether the model by Miller and Le Boeuf could also be applied to music.
Like theater performances, musical performances are considered to be at the expression level in pure FRBR-see the brief literature analysis above.Also like staged performances, the musical performance itself is not contained within FRBR; instead, it appears in FRBR only through its trail, such as CDs of recordings.Miller and Le Boeuf (2005) give the performance as an expression which is attached to a mise-en-scène or choreography, in other words, the performance-as-work; it is this mise-en-scène or choreography which has a relationship with the play, libretto or musical work, according to Miller and Le Boeuf.So, is there a similar "work" for music, which matches the mise-en-scène or choreography seen in theatrical and dance works?In Western art music, this could be considered to be the creative interpretation of a musical work, for instance, a pianist's interpretation of a particular piano sonata or a conductor's realization of a particular symphony.The problem is that this sort of creation has little which is fixed in the same way as a choreography or a director's vision of a piece (although such a creative act is arguably replicable, at least by the antagonist, through notated means).Therefore, while performance-as-creation can be enveloped into conceptions of music, fitting this into Miller and Le Boeuf's (2005) model is more problematic, due to contemplating and solving how musical interpretation can become the equivalent to a choreography or mise-en-scène.
When dealing with non-art music, the idea of a performance rather than notated creation offers a different sort of advantage.As discussed in the brief literature review above, the composer-centric and musical-work-centric naturethe term "musical work" here being used in a musicological rather than FRBR context-of FRBR and RDA have long been identified as problems by those considering FRBR for music such as jazz or popular music.Kishimoto and Snyder (2015) discuss some of the issues with assuming composer-led rather than performer-led works.In fact, their solution (Kishimoto & Snyder, 2015) to the practical problems caused by giving primacy to composition and text over performance, makes the performed song a work in its own right, which is then a related work to the composed song (with its associated song-writers).The performer-led song would be a companion to mise-en-scène/choreography in Miller and Le Boeuf's (2005) model. So, even if Miller andLe Boeuf's (2005) model could be adopted as is (which it could not), it would lead to issues concerning inconsistency among the performing arts, as well-established practices of music cataloguing are structured around performance being an expression of composed musical works.However, on a conceptual level, if issues over how interpretation could be embodied for Western art music could be explored and resolved, it is clear that the performed work as a central tenet of FRBR/LRM in the manner of Miller and Le Boeuf (2005), could be a pan-performing arts solution to the issues raised when using pure FRBR for organizing and describing performance documentation.

Conclusion
This article has shown how FRBR and LRM are interesting lenses through which to observe performance documentation.While they may contradict current ideas about performance, these models are the present and future of bibliographic description and access, so it is important to understand how these models treat performance documentation.In pure FRBR and LRM, the performance itself is a shadow, only captured by what that performance leaves behind through objects, recordings, or similar.Miller and Le Boeuf (2005) have created a detailed, FRBResque structure, which shows how performances can be written into FRBR.However, this article has shown that despite the conceptual advantages, there are issues with the Miller and Le Boeuf approach which need discussion.
First, on a practical level, cataloguing guidelines and practices developed from 2005 to the present day through the conduit of RDA, have been designed with certain relationships in place, such as expressions connected to textual or musical works, and recordings of live performances linked to textual or musical works.This means there are contradictions between pure FRBR/LRM and Miller and Le Boeuf (2005), complicating any future integration.Second, there are questions about whether the Miller and Le Boeuf (2005) model would work across all types of performance arts; moreover, the desire (or not) for an integrative conception of performance across all arts for the purposes of data modelling, is fascinating in its own right, especially considering the historic variation in the treatment of choreography and dance compared to other performance arts.Third, and perhaps most excitingly, from a contemporary documentation viewpoint, there is an argument that the performance itself could be considered a document.This opens up some intriguing possibilities.If performance is a document, then this could bolster the position of those wishing to draw performance out from the shadows (while also asking questions about the exact boundaries of FRBR's bibliographic world).Furthermore, this sets up interesting connections between documents which are performances and documents which document performances.FRBR and LRM, with their focus on deconstructing types of information, are apposite lenses for reconsidering our conception of performance documentation.
However, this is only a brief foray into FRBR, LRM and performance, and much more work needs to be done.One area of potential future research would be to consider whether the move to LRM has any impact on how performance materials are discussed within data models discourse; in addition, it would be useful to consider whether subsequent changes to LRM-in-practice through the conduit of RDA have an effect on any of the issues discussed in this article.Furthermore, there is more work to be done investigating how pure FRBR/LRM and Miller and Le Boeuf (2005) could be applied across all the performing arts, and even extended to other event-based arts; for instance, would the performancebased approach be adaptable to perennial problems in art cataloguing relating to exhibitions and curators?So, while FRBR and LRM are primarily bibliographic models, they do offer interesting conceptions of performance documentation and performance itself.FRBR-esque models, such as Miller and Le Boeuf (2005), demonstrate how FRBR concepts can be successfully utilized in a world where even a performance can be a document.

Figure 2 .
Figure 2. Recordings and performances: pure FRBR in comparison to Miller and Le Boeuf (2005)