5.4 Representative Expressions

“In a strict formal sense, within the model all the expressions of a work are equal as realizations of the work. However, research with users indicates that they recognize that works have original or “canonical” expressions, those that can be said to best represent the initial intention of the creators of that work. Other expressions can, if the full history of the work is known, be seen as taking shape from a network of derivations or transformations starting from this original expression.” p. 62


4 thoughts on “5.4 Representative Expressions

  1. Pingback: World-wide review of the FRBR-Library Reference Model, a consolidation of the FRBR, FRAD and FRSAD conceptual models | FRBR Open Comments

  2. It seems like a lot of people have doubts about this, but I was really happy to see representative expression included. This contextual information about the original, intended version of a work is important to users, especially for the arts, literature and film. OLAC recommended a version of this in its investigation into what a moving image work is and how it might be used (http://olacinc.org/drupal/?q=node/27) However, I think that it is only practical to model these attributes as associated with the history of the work or else as an abstract expression that is not fully fleshed out or associated with a specific, actual expression. So it wouldn’t be that one particular existing Greek version of the Iliad would be chosen as the representative expression. In fact, perhaps the representative expression for the Iliad might have only the attribute of language = Ancient Greek if that is all that is known and commonly accepted. It seems to me that almost any expression attribute could be part of the representative expression in a particular implementation. In the draft, they seem to consider category, extent and rights not to be appropriate for a representative expression. For moving images, extent should clearly be an attribute of the representative expression. I can see why you wouldn’t want to use content type (listed under expression category) for music, but perhaps it would be useful to consider the original content type of the Iliad to be spoken word. Nothing is coming to mind right now, but someone else could probably think of a use case for the rights for the representative expression. FWIW, I also wrote a longer post on this at http://lists.ala.org/sympa/arc/rda-l/2016-03/msg00178.html


  3. This is part 3 of my personal response to the FRBR Review Group, based in comments to the RDA, FRBR, and PCC lists (revised somewhat from the versions Heidrun posted above)

    3. Representative Expression

    The general feeling that a work has some expression-related attributes that apply to what most people think of when they think of work (e.g., it was originally a text, written in French, for children) is completely understandable. FRBR-LRM proposes the solution of choosing a “representative expression” and defines specific expression attributes to be used in the description of that “representative expression.” I have reservations about solving the problem in this way.

    FRBR-LRM doesn’t explain very clearly how this is supposed to work. It states “The model does not prescribe the criteria that must be applied in making the determination of representativity” and “Whether an expression is the original expression [meaning the representative expression, though the document states earlier that the original expression is not necessarily the representative expression] of the work will often be a component of this decision-making process.” A response might be that FRBR-LRM is a model and it’s up to codes to work out the details. But even a model needs some sort of framework.

    On the apparent presumption that the original expression is most likely the representative expression, for textual and most other kinds of works “the” original expression is almost never the expression a librarian (or user) has in hand—unless the librarian happens to be holding the author’s original manuscript, which is a different expression from that contained in the first published manifestation since there are always differences between the manuscript and publication. In my experience I’ve probably never touched an item containing “the” original expression of any work I have cataloged. So talking in terms of “the original expression” might not too useful.

    To get specific, when describing the work Iliad, which is the representative expression? Beginning with Greek expressions, those in the language Homer is presumed to have composed the work using, there are hundreds if not thousands of published editions of Greek expressions of the Iliad. All have different texts. Is one of them going to be chosen specifically as the representative expression? How? Or what about one of the thousands of manuscripts? These are all different expressions because they all contain different texts. Is one of those the representative expression? Which one? This situation exists with any author who wrote before the modern period, and it exists for many modern writers as well (e.g. James Joyce). The idea of picking one “expression” as the “representative expression” seems fraught with difficulty. Perhaps it is not prudent to introduce this notion into the model until it is clearer how it might work, if at all.

    Also, a practical implementation question. The attribute “representativity” (LRM-A5) is given as a “yes-no” proposition: an expression is either the representative expression or it is not. Presumably there can only be one representative expression of a work. But in a given system, particularly a cooperative system, how would this work out? When a cataloger or other metadata professional is preparing a description of a work and expression and having a look in the ER database is the “representative” expression going to stand out somehow? What would prevent folks from describing more than one expression within the same database as the representative expression? One possible response could be in a linked data environment it doesn’t matter if more than one expression is designated as the representative expression. That seems dubious to me, but if so, what’s the point, then?

    Instead, I recommend that the attributes given in FRBR-LRM as “representative expression” attributes be instead recorded as work attributes. The attributes of expression marked with an asterisk could remain as general expression attributes, but narrower versions could appear as work attributes with names such as “original (or canonical) intended audience”, “original (or canonical) language”, “original (or canonical) key”, “original (or canonical) medium of performance”, “original (or canonical) scale”. It could be reasonably seen as simpler to dispense with the judgment-laden decision about which is “the” representative expression and instead make these “canonical” attributes of the work. I recommend that this might be a better way to go.

    By the way, I agree that (non-representative) intended audience, key and medium of performance should be expression attributes, not work attributes as in current FRBR/RDA.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s