ARLIS/NA 30th / VRA 20th Joint Conference, Hyatt Regency, Union Station, St. Louis, Missouri - March 20-26, 2002

Seminar 6

Common Ground: Standards for Cataloging Images and Objects

Moderators:

Linda McRae, University of South Florida
Lynda White, University of Virginia

Speakers:

Gregory Most, National Gallery of Art
Ben Kessler, University of Chicago
Sara Shatford Layne, UCLA
Margaret N. Webster, Cornell University
Dustin Wees, Columbia University
 

Recorder:

Trish Rose, Indiana University

McRae opened by stating this is the third part in the “Common Ground” mini-institute. This seminar looked at issues that cannot be “crammed into a kind of bibliographic framework.” The issues that were addressed are hard to quantify, rely on subjective decision-making, and can be resolved in a variety of ways.

Gregory Most opened the seminar with his talk, “The New Crusades: Image Curators and the Quest for a Uniform Title,” in which he explained the futility of searching for one absolute and true title for each work of art because it does not exist. Unlike guides such as AACR2 for bibliographic resources, image catalogers do not have a readily available “chief source of information” to which they can turn for constructing titles. Our most reliable sources would be the museum’s collection catalogs, online websites, or catalogues raisonnés. If the information from these various sources does not agree, then the catalogue raisonne is your most reliable source because its information does not usually change over time.

Most described five ways that titles evolve. Actual titles are attached to artworks by modern artists and often change. To track these changes, he suggested adding an alternate title to the record or placing the alternate in parenthesis after the primary title. Traditional titles are the commonly associated titles, such as Whistler’s Mother. Untitled works are often given descriptive titles based on what is depicted in them. Location/ownership titles get their name from an artwork’s location, place of origin, or owner. For architectural works, the most consistent identifier is the address. Groups/series titles refer to the larger work to which an artwork belongs.

He concluded by stating while it is not possible to establish one rule for dealing with the wide variety of titles, it is still important to begin making rules and setting standards for image cataloging. The best access we can provide is through multiple forms and hierarchies of titles.

In Ben Kessler’s talk, “Parts-to-Whole: What makes a Work a Work?,” he pointed out that visual resources catalogers have to deal with the challenge of cataloging not only the “object in hand” (e.g., slide, photograph, digital image) but the intellectual content of what that object depicts (the work itself). That information often depends on external sources, separate from the container. VRA Core 3.0 has made it fairly easy to differentiate between these separate layers of records as long as the work is a discrete entity or self-contained unit.

What has been more difficult to address is how to deal with the large proportion of works that are part of a larger whole or contain many parts (e.g., decorative cycle within a building, leaf from a manuscript). This begs the question of whether there should be a given rule to guide catalogers. Is it better to collocate different components together into a hierarchical structure under the broadest possible context (lumping) or to define a work as the smallest indivisible component (splitting)? Splitting would seem to be the better approach, but some indivisible components would not be recognizable enough as works of art in their own right.

While this is a matter best left to individual catalogers to decide, they do need to determine whether the parts-to-whole relationships are essential (when a work cannot be fully identified except within the larger context of the other) or informative (related works that are not dependent on the other for identification). If essential, the Title.Larger Entity element can be used to express this. If informative, the Relation element can be used. These are important theoretical distinctions that will only become clearer as the VRA Core 3.0 has been in use for some time and patterns of commonly accepted usage emerge.

While Sara Shatford Layne’s topic may ring familiar to librarians who have read her articles on a subject’s “ofness” and “aboutness”, this talk, entitled “What is a Subject? And, Why Does it Matter?” expanded on those ideas by addressing how to differentiate between these two aspects of a subject using coding or formatting rather than vocabulary. Benefits of coding include: identifying the entity (e.g., person); identifying its role (e.g., creator); and helping the end user understand the options and content of the database they are searching.

MARC21 is a good model for this type of coding. In MARC21, if a person is an author of a book their name is entered in the 100 field; if they are the subject of the book, it is entered in the 600 field. The “00” portion of the field represents a person, while the “1” or “6” portion of the field represents either an author or subject respectively.

This approach reduces the burden on the user. For example, users can search for Goya without specifying Goya’s entity type or Goya’s role. The results can then be presented to users more meaningfully by informing them that Goya is the creator of ninety works of art and is depicted in ten works of art. They can then choose which is more relevant for their purposes.

Margaret Webster’s talk, entitled “The Built Environment: What is the Work? What is the Related Work? What is the Subject?,” explained how images of architecture and their supporting materials present challenges to the image cataloger. The container of the image, such as the slide or digital scan, needs to be cataloged along with the intellectual content of the images. The content of architectural images often consists of parts related to larger works. The cataloger must determine whether the image portrays an integral, and therefore dependent part of another work (essential relationship) or whether it is an independent but related work (informative relationship).

Three elements of the VRA Core 3.0 (Title, Subject, Relation) were designed to assist catalogers with both describing a work of architecture and its related collateral material. A qualifier can be added to Title so that Title.Larger Entity can be used to express essential relationships in which “the described work includes the referenced work.” The Subject element includes the “terms or phrases that describe, identify, and interpret the Work or Image and what it depicts or expresses.” This allows the cataloger to identify an object depicted in or on a work and relate it to “the record of the quoted work.” The Relation element allows the cataloger to express an informative relationship between works. The link both identifies the related work and the nature of the relationship.

Associated drawings of architecture can present another challenge to image catalogers, but if they are viewed as “visual documents” for the building rather than works themselves, they can be easily cataloged with VRA Core 3.0 in the same manner as digital files or slides. Whether a document is cataloged as a work or a visual document of the work is not as important as consistency in applying metadata and linking the records in meaningful ways.

In the final talk by Dustin Wees, entitled “Hallowed Ground to Common Ground,” the audience was informed that it is not as difficult as we think to bring together records from museum collections and visual resources collections. There are interesting parallels between the growth of slide collections and data collecting on works of art in museums. Descriptions of these collections developed idiosyncratically based on the needs of the faculty or curators. Access by a broader public was typically never considered. Automation has forced museums and slide collections to develop standards as demonstrated by the creation of descriptive standards. AMICO is also having a positive impact on visual resources metadata standards because museums are being forced to think about their information in a broader context.

Despite the late hour on the last day of the conference, the audience seemed eager to discuss the issues brought up by the speakers.

One unidentified speaker wanted to know, now that we understand how to link records through the VRA Core Elements, how do we do it structurally in our respective databases? Kessler responds that the VRA Core’s intent was to standardize the information but leave the structural implementation up to the individual institutions. David Bearman added that we need to focus less on physical implementations and more on named relations, which allow data to exist in a variety of physical models. Eileen Fry commented that it is often easier to express these relations spatially than it is verbally. Jennifer Trant added that relationships have to be explicitly expressed.

Fry then posed a different question to Webster and Shatford Layne about whether buildings have ofness and aboutness. Webster believed they do. When, for example, an architectural drawing depicts something, it is about that depiction. Shatford Layne questioned whether everything has to have a subject. Often, the subject field becomes a catchall field for information that does not seem to fit anywhere else.

Christine Sundt then directed a request to museums that they retain information on changes of titles for historical purposes. Webster added that this is also needed for buildings. A final question was asked by Fry of Most about the definitiveness of the information about an artwork found on museum webpages. Most felt it is a good bet, if a museum has focused its efforts on presenting its collection over the Web, that the information has been verified before it goes public.