J. Trant, Which Image? Which Text?
"Imaging in Museums: Issues of Relating Text to Imagery"
CIDOC, Washington, D.C.
29 August 1994
... a talk about the relationship of text and image given without showing either text or image - no bias in presentation ;)
When I began thinking about the issue of relating text to imagery within the museum context, I was overwhelmed by both the possibilities and the problems presented by the topic. Museums, especially art museums, consider images their stock-in-trade. We study them, exhibit them, analyze them, and preserve them as both aesthetic objects and carriers of cultural meaning. We describe them, index them, catalogue them and photograph them; and we do these things in order to make our collections available to both the visitor and the researcher, both on site, and now increasingly, on the networks.
As Helene Roberts has so aptly pointed out, much of what we know about an image is information not necessarily inherent in its visual content. In order to identify and understand images, we associate text with them. But what kind of text is it? What does it tell us about the image? What information should a textual description of an image contain? And what standards are necessary to make it possible for us to share information - both textual and visual?
When we associate text with an image, we attempt a number of things. We are trying to identify that image, distinguishing it from others. We are trying to describe the image, conveying some sense of its appearance and content. We also attempt to group images together, by classifying them. And we try to analyze the image - recording its meaning. Each of these approaches requires a different type of text, and each of these textual descriptions will have a different nature.
An identification of an image could be as simple as an accession number. This serves to distinguish one image from another, but is not very helpful in terms of communicating meaning or significance. Descriptions of images have traditionally taken the form of structured texts, originally found on catalogue cards, and now stored in databases. Classification of images takes the form of controlled terms, drawn from structured vocabulary resources; such terms are now a standard part of any database. Analysis of images is most often found in free text form in the art historical literature. As Helene has mentioned, it is seldom systematically recorded or even referenced.
These are some of the types of texts recorded. What about the images? “Image” is an all inclusive word, subsuming various kinds of visual materials, each different from the next. I have already spoken about the work of art itself. At its simplest, this is The Image. But it is not the only image. We regularly take documentary photographs of works of art - which we like to call surrogate images. These may take any one of many forms, from a Polaroid shot taken when a work arrives at the loading dock, to a 35mm slide or black and white photograph used for reference or teaching purposes, to a detail of a particularly interesting aspect of the work, through a 4x5 or even 8x10 color transparency made when a work is to be reproduced in a high-quality publication. Increasingly these photographs are being converted into digital images of works of art as well. Digital images are now forming part of our databases, figuring in our public information systems and are regularly the basis for images in print publications, whether we realize it or not!
So far, the distinctions between different types of images seem pretty clear. We have a nice duality – the work of art, and the reproduction of it. Each of these could be seen to have their own descriptive needs. But these categories are not absolute. The reproduction may be considered a work of art in itself, especially when we are considering a print by Hendrik Goltzius after Raphael. (As Helene mentioned, Vasari could have used reproductive engraving to illustrate his Lives.) The historic photograph may record a work of art or show it in situ, and may be considered a document, or a work of art in itself. The work of the Alinari Studio comes to mind here Recent scholarship by Anthony Hamber and others has begun to study this class of work, more closely, analyzing how our sense of a work of art is tempered by the reproductions we see of it.
It thus becomes clear that physical form alone cannot be relied on to define the image at hand. A video could be a work of art created by a video artist. It could be part of a work of art, having figured in an installation. It could be a record of a work of art, such as a performance piece, or it could be documentary - having been broadcast on the evening news and having nothing to do with 'art' at all. Digital images are equally problematic. They often reproduce works of art; but their source is not always apparent. A reproductive digital image can be derived from a photograph of the work, from the work itself, or from another digital image. Or more and more frequently a digital image can be the work of art itself.
How an image is described, and therefore the kind of text associated with it, must depend on the kind of image it is. Why an image was created, and how it has been used are critical clues in understanding both the visual information it contains and its significance. Just as an art historian looks critically at an original work to understand its significance, so should a critical eye be cast over the reproductive image. No image truly reproduces. Some things are included, others are left out. Knowing what isn't there is often as important as seeing what is. Good text again becomes key.
Structuring that text in a standard way will be critical for the long-term viability of descriptive information, and for its interchangability. Within the art-world, it is apparent that we require standards for describing both works of art and the images which depict them. Determining how an image is to be described is the first step towards linking appropriate text to it.
There has been a long history of standards development in the museum community, which I am sure is familiar to many here today. Recent achievements that come to mind are the publication of MDA's Spectrum and the completion of the CIDOC Minimum Data Standard which is being discussed elsewhere on the conference program. Within North America, the Art Information Task Force, or AITF, has been working since 1990 to define the content of a scholarly description of a work of art.
A collaborative project of the College Art Association and the Getty Art History Information Program, partially funded by a grant from the National Endowment for the Humanities, the AITF brings together representatives from broad range of communities. Art historians, museum curators, registrars, visual resource curators, information managers and systems developers collaborated to articulate the information needs of those who use information about works of art in their professional activities.
The information recorded about a work of art is complex. We document a work's physical nature, we give it names to identify and describe it, we record the circumstances around its creation and we study and analyze it, recording what it is about. The AITF has organized this information into 25 Categories for the Description of Works of Art. They begin with the Object or Work itself, and progress through its Title or Name, recording the Materials and Techniques used to create it, and providing details of its Creation. They also provide the facility to record the history of its Ownership and Exhibition, an analysis of its Style or Period, historical and cultural Context, and the critical responses to it. As well, categories are provided to identify the other works of art, visual and textual references related to it. For those of you who would like further information about the AITF, Fact Sheets are available.
In my time remaining today, I would like to focus specifically on two of the AITF categories which draw out difficulties in relating text and image: Subject Matter, and Related Visual Documentation.
Understanding a work's subject matter is key to its appreciation. what a work depicts – whether an heroic battle scene, or a red stripe on a blue ground – and how the artist depicted this subject, together inform our interpretation and assessment of the work. Helene has already pointed out that much of the meaning carried by a work of art is extrinsic to its physical manifestation. Associated texts are often key for the identification and understanding of its iconography – often we cannot tell what a work is about without textual clues. [how many of you could have placed that bundle of grapes carried on a stick to Numbers 13:23?] As our culture moves furher and further away from the religious and mythological traditions of the past, text becomes essential if the general public is to have any hope of understanding historical works.
The AITF Subject Matter Category takes an incremental approach to the description, identification, classification and analysis of the content of a work of art. It follows the approach outlined by Panofsky, which Helene outlined briefly..
The first of the Subject Matter subcategories is Description. This is a generic description of what is in the image. It identifies what is seen in a scene – a Man and a woman stand under a tree with a snake in it. The woman holds an apple. Provision is made for recording terms which classify this description, from vocabulary sources such as the Art and Architecture Thesaurus: man, woman, apple, snake, tree (you get the idea).
Once an image has been described, it is identified, or named, in Subject Matter - Identification. Here the man and woman are called Adam and Eve in the Garden of Eden. Controlled vocabulary sources appropriate here include ICONCLASS and the Library of Congress subject headings.
The meaning of the work is recorded in Subject Matter - Interpretation. Varying interpretations of a work can exist, and these should be recorded. We are now able to introduce concepts such as the Fall from Grace, or Original sin. Sources for controlled vocabulary would include
The final AITF subcategory pushes the boundaries of traditional iconographic studies: Subject Matter - Interpretive History records the iconology of a work of art - or the development or changes in its meaning over time. Here we would record changing theological interpretations of Genesis, as well as the feminist reading of the story as the source of the Roman Catholic justification of the continued subjugation of women.
This points to one of the greatest challenges in relating text to images – the meaning of visual works is not fixed. Their nature insists that they are open to interpretation. Different analyses of the same work co-exist, for it is in the nature of art historical discourse to challenge and interpret visual evidence. The systems we develop for recording these meanings must be flexible enough to accommodate multiple opinions.
The varying kinds of information that the Subject Matter category records, illustrates the different types of texts we associate with an image. Subject matter is often best described as a free-text narrative, but it must be indexed in controlled vocabulary. The ability for various types of texts to co-exist is one of the reasons why the AITF is exploring SGML (Standard Generalized Markup Language) as a technical means of expressing the categories. I am sure that John will speak more fully about this project, which the AITF is now pursuing in collaboration with CIMI.
The second AITF Category I would like to discuss is Related Visual Documentation. There is an irony in the fact that we must use words to record the relationships between images. Related Visual Documentation begins to systematize the information we need to record about an image of a work of art. This is, however, still a work in progress. Related Visual Documentation was defined in conjunction with the Visual Resources Subcommittee of the AITF. This work has formed the basis for the Data Standards Committee of the Visual Resources Association, chaired by Linda McRae of the University of South Florida.
Our task in defining Related Visual Documentation was to identify the pieces of information necessary to reference a "surrogate image". This prompted an examination of the descriptive needs of the surrogate itself, a task which has been taken over by the VRA Data Standards Committee. In many ways, a surrogate image is about the work of art that it represents. Often, such images are catalogued as if they were the work themselves; within teaching collections, for all intents and purposes, they are. However, there are additional pieces of information recorded about a surrogate, which document the process of its creation, and its place within a particular collection, which the VRA Committee is identifying.
What was of interest to the AITF was the intersection between the text associated with the surrogate image, and the description of the work of art itself. A description of the work shown was key to describing the surrogate. An identification of available surrogate images was seen as an important element in the description of the work of art. More interestingly, we found that an additional piece of information, now titled "view" was required to adequately document the relationship between an image and a work of art. Not all images show a work in the same way, or even show it in its entirety. “View” records the aspect of work shown in the image, and the way in which it is depicted.
Much more work is necessary to identify the standard description of an image of a work of art. This is particularly true when we add the layer of complexity that digital images bring to the picture. As a resource, digital images are dependent upon a reference to an external source for their identification. As files, alone they do not carry enough information for us to evaluate their contents, even simply in terms of the image data they contain.
As a community we are only beginning to think about what we need to record about an image file to ensure its utility across systems. What information is required in a description of a digital image file? How do we create bodies of digital visual reference that will function as a coherent whole? How do we ensure we have the information necessary to correct for biases introduced at image capture? When will we move past the 'one-off ' approach, and begin to treat images as we are starting to treat text, as a resource that can be reused in many ways, for purposes not necessarily envisioned at their creation.?
These are issues which I am sure Michael Ester and John Perkins are going to address, as they talk about standards for texts and images, and image management information. They are also concerns that are at the heart of AHIP's new Imaging Initiative.
Digital images offer the potential to make our cultural heritage available to a wider audience in ways never before envisioned. The Distribution of images and associated texts over communications networks is changing the nature of teaching and research. For this transformation to take place, however, [a critical mass of] digital information about works of art must exist, and it must be available in standard forms.
The Getty AHIP Imaging Initiative sees the development of common standards as essential for the sharing and preserving of digital image and textual information. For the network to fulfill its promise, we have to work together to define it, and to provide its content. Documentation, capture, storage and transmission standards for images and art information must be the foundation of a critical mass of high-quality images and information capable of supporting innovative research and teaching. Without them, we risk being adrift in a sea of references and cross-references, with no point of reference at all.
Trant - Introduction
Jennifer Trant is the recently appointed Manager of the Getty AHIP Imaging Initative. A specialist in artsinformation management, she has worked for several major Canadian institutions, including the National Gallery of Canada and the Canadian Centre for Architecture. She authored the report of the Art Information Task Force,Categories for the Description of Works of Art and participated in the development of a forthcoming CD-Rom publication of Frank Lloyd Wright's architectural drawings. Throughout her career, Jennifer has been actively involved in the establishment museum data standards, regularly publishing articles and presenting papers about issues of access and intellectual integration.
[text last updated August 29, 1994.]