Bailey& Graham (CHArt 1999)

In this paper I want to outline the rationale and methodology of a project which is being carried out at the University of Northumbria. The researchers are myself, and Margaret Graham. Margaret's background is as an information manager and she is currently Research and Development Manager of the Institute of Image Data Research at the University. I am a design historian and my role in this project is to frame the questions about the discipline, which we hope to answer using questionnaires and structured interviews. The initial phase of the project is proceeding with Institute funding although further grants are being applied for.

What is the problem?

Our motivation for the project was set out very cogently for us by Lynne Brindley in 1996

... I would like to make a plea for bringing the user of networked information services to centre stage. As Chairman of JISC's new committee on electronic information I will give priority to the development of this more analytical and holistic approach to the evaluation of use and user behaviour. There is a danger otherwise that we might be opening up an increasing range of services with precious little understanding of the frustrations they will cause users, their true economic costs, and costs to users in lost time, or thwarted ways of working.¹

Brindley goes on to argue that very little is known about the impact of networked resources on user behaviour. She asks,

... have you ever really wondered what academics and students are really doing with this information? What is its real impact?²

Although her arguments are at least partly driven by economics, as academics we also should be interested in the answers to these questions. Perhaps the centre of gravity of the discipline of art history, and its sub-disciplines, is shifting. Surely, it would be surprising if it were not? In particular it would be of interest to us to examine how the use made of digital resources by art historians differs from that made by other academics. With exceptions such as the publication of papers over the last decade and a half in the CHArt Journal, and special issues such as Art Bulletin in June 1997,³ art historians have been slow to tell the story of the impact of new technology on their practice.

This stage of the Compare and Contrast project is concerned with laying the philosophical and methodological foundations for an empirical study of a sample of the community of teachers and researchers in the subject. The paper investigates three propositions. The first is that the approach to art history, as to many disciplines, has been driven by the information providers' need to find uses for new technology, and is based on a limited model of the discipline. The second is that the discipline's core skills differ from those of the other humanities but that they have been only partially elaborated. Finally, the rapid expansion and methodological development of art history in recent decades, have tended to diffuse the earlier consensus about art historical method.

While art historians may have been slow to relate their experience, other professionals have increasingly sought to make sense of patterns of digital image use. This is not surprising when we consider how very widely distributed and accessible all forms of visual imagery now are. Most services offering networked images have a broader user base in mind than just art historians, whose numbers are not sufficient to justify the investment required to establish a networked service. Most publicly funded services would describe their typical user in generalities, and as we shall see, he or she is expected to adopt certain image seeking behaviours, following a stereotype drawn from a limited range of historiographical sources.

The question of the most appropriate methodology for user studies, also discussed by Brindley, is worth lengthier consideration than I can give it here. Perhaps in future we might, as Peter Stone has envisaged, adopt a style of user analysis rather like the JICTAR survey of media readers, listeners and viewers (excuse me, but would you mind telling me which internet subject gateway you accessed last night?), or alternatively, automated logging of the subject specialisation of site visitors might be done by the SBIGS and cross-tabulated with an audit of each user's search tactics. Both would be of immense value both to service providers and to the academic community, but it seems likely that the more traditional quantitative and qualitative surveys will also be part of the "market research" strategy.

We can identify three types of user studies in the literature. The first is the survey of the potential or actual user to find out what the needs are, and what features of the service are found most valuable. The second attempts a more abstract process of modelling user behaviour, usually by looking at the mediation of researchers' queries either by a librarian or image manager, or at records of submitted verbal queries. The third attempts to study directly the behaviour of image users, either the researcher or the information manager, to assess the non-verbal aspects of user behaviour, such as search strategies.

The emergence of networked services such as the CTI for Art and Design, the AHDS 'daughter' services for performing and visual arts, PADS and VADS, and the subject based information gateway ADAM, has prompted each to carry out surveys, in each case by print and web questionnaire, of their actual or potential user base to assess the extent and character of user needs.

The Performing Arts Data Service, which covers the film and television industries, conducted their User Needs Survey in Autumn 1997.⁴ This highlighted a significant level of interest in networked resources, especially databases, but cited a familiar list of deterrent factors, such as inadequate hardware or network connections, lack of knowledge about resources, difficulty in searching for these resources and lack of user experience, especially amongst the student population.

The CTI Art and Design Survey similarly revealed technical and organisational barriers to adoption, and not surprisingly discovered that art and design applications, which tend to require substantial computing power, were the predominant concern alongside training in the effective use of software tools. Although the researchers usefully identified the subject specialism of the respondent the survey was not structured to enable the patterns of use characteristic of each discipline to be deduced.

The ADAM User Needs Survey⁵ was conducted in Spring 1996, and found that over one-third of respondents were using the Internet as a source of information. While many still preferred print media because of its more comprehensive coverage of the subjects, respondents valued those networked resources which were easy to use, and offered some guarantee of academic quality. The ADAM service inferred from this that a more transparent interface (for instance, one permitting browsing of a tree of index terms) and relatively 'fine-grained' cataloguing (such as allowing retrieval of the details of the images held at a museum site) should be traded off against sheer quantity of resource metadata. This was, to some extent an acknowledgement of the needs of the subject groups making up the ADAM 'constituency'.

The VADS Survey⁶, conducted in 1997 showed a strong interest in its planned core activity of access to archived datasets as a research resource but re-iterated the concerns of potential users referred to above. All these investigations, revealing though they are of the potential of the technology, have two weaknesses. Firstly, they do not permit us to determine what might best support the research agenda in art history, as opposed to the professional practice of a design manager, or the learning of a museum studies student, all of which are areas catered for by these services. Secondly, they are of their nature snapshots rather than longitudinal studies. Even though the VADS survey summarises earlier survey results and thus captures evidence of some change over a three year period, the impact of networked resources on academic practice can really only be understood if measured in consistent ways over time.

Another group of user surveys has emerged, not from the service providers, but from social scientists and library schools aiming to model user information seeking behaviour in more general ways. Often they have proceeded by looking either at the kinds of query put, usually to human information managers, by 'customers' of image collections, or at the ways in which information managers themselves seek to provide images which satisfy users' requests. The former sort of project has been reported on by Peter Enser.⁷ The Hulton Deutsch Query Analysis Project, for instance, divided query types into requests for Unique Subjects, examples of which are the Jarrow march, Evelyn Waugh's living room, Peter the Great, or Beethoven's signature, and Non-Unique Subjects, such as smog, physicists, and 'wanted' posters. Either could be a general or refined request. The study concluded that, for this collection at least, the greatest demand is for Unique-Refined subjects, an example of which would be "Edward VIII looking stupid".

Enser's more general concern is with the extent to which the business of retrieval of images can be reduced to a systematic query language. His typical user, like that of Holt and Hartwick, in "quick, who painted fish?", a title which serves to summarise a well researched problem in effective image retrieval by means of image and text query automation, is out to satisfy a particular need for an image, usually to illustrate a verbal text. As Holt and Hartwick state

Because our user population is diverse, many patrons are unfamiliar with art historical periods or terminology and, in many instances, are interested more in what an image depicts than who made it.⁸

Much of the literature on unmediated, and therefore less 'expert' access to iconographical research collections assumes that searchers will have a hierarchy of more or less abstract 'keywords' for what they expect the image to contain. The analysis of the subject of images is, for most authors, based ultimately on Panofsky's pre-iconography/iconography/iconology schema ⁹, or on Karen Markey's work on access to specialist picture collections, which draws upon it. ¹⁰ The pessimistic conclusion of most researchers looking at this problem is that the burden of access will fall more heavily on the users, given the variability of subject cataloguing amongst diverse collections.

For our purposes the principal weakness of these user studies is that they are of mediated searches, that it written or recollected queries given to an information manager. How, we must ask, do the image users themselves, the art historians, typically frame their questions? Although there are user studies, typified by the VISOR Project at the University of Northumbria, which look, not at written or recorded queries, but at the behaviour of image managers and other professionals, close attention to the unvoiced search strategies of image users themselves has not been carried out since before the invention of the World Wide Web.

What is special about art history?

The closer we come to the basic elements of the practice of art history, it seems, the less likely we are to find explicit discussion of them. The acts of discrimination and judgement of visual data are both basic human capabilities, on which survival has always depended, and the core skills of the art historian. The difficulty of verbalising this activity is acknowledged. When a student asks a tutor, faced with the task of saying something non-trivial about an abstract work of art, to "tell me what to say about this painting", the questioner may really mean, "tell me how I can convert what I can see into something I can say". Sometimes this translates into, "tell me how to look at this". In many works of art, the existence of a 'subject' disguises this problem - there is something to attend to. If the work of art is not a Constable but a Pollock, what does the eye attend to?

The most significant study looking at the requirements, research behaviour and aspirations of art historians was the 1988 Getty AHIP project initiated by Director Michael Ester. If the search is not mediated by a human being or equivalent text interface, then the conceptualisation of the search strategy need not be verbalised at all.

Stated simply, professionals use physical arrangements of images as a way to think. Arrangements may reflect stylistic relationships over time, different artistic or historical expressions of a similar theme, separate studies by an artist for an evolving whole, or the visual outline or story board of a scholarly argument.¹¹

Object, Image, Inquiry: the Art Historian at Work was the published report of this research, based on interviews with a sample of eighteen art historians. ¹²At that point, in the mid-1980s, image libraries were at an early stage of development and the internet was a specialist's means of retrieval of image files, likely to be carried out only with technical assistance. Speculation about the impact of IT was possible, however, and the foresight of the interviewees tells us a lot about the expectations of the discipline at that point. But the experience of using some research tools which now strike us as of central importance, such as networked image resources, was at that point insufficient to gauge any impact on working methods.

Ten years on it is appropriate to explore the extent to which those earlier expectations have been achieved. These might be expressed as follows;

Given the substantial impact on the discipline of research methods drawn from literary theory and the social sciences, it might be doubted that the old stand-by of compare and contrasting paired images is still inflicted on art history exam candidates, or used habitually by professionals as a way of thinking or expressing their ideas. In her primer for students of the subject, Pointon¹³ explores the debt of modern art historical methods to those of the early connoisseurs, for whom the comparative method was fundamental. She lays emphasis first on the physical nature of the object, then on what it represents. She stresses the usefulness of intimate knowledge of the object, if necessary by re-constructing it using the same materials. Whether dealing with an analysis of the object in the context of the movement, or as a part of its times, the art historian proceeds, first and foremost, by immersion, either in a series evincing differences and similarities, or by minute and close mapping of the particularity of the context of the object.

The degree of unanimity on this amongst art historians of very different traditions is remarkable. For instance, here is Berenson,

It is enough to know when and where an artist was born and what older artist shaped and inspired him. We must look and look till we live the painting and for a fleeting moment become identified with it.¹⁴

Or Swarzenski, uniting objects of wildly different scale and function in his monograph on Romanesque Minor Arts,

the artistic aims of Romanesque sculpture and painting found their purest expression in these small works ...¹⁵

The art historian Richard Brilliant¹⁶ gives a similar account of the discipline as, "at its most basic, the connection of the art object to information". Through an iterative process, which Brilliant characterises as the "evocation of visual memory", the historian compares object to object, and text to object, and thus constructs a meaningful context for the object. Most importantly, it is the non-verbal nature of the processes of visual memory which carries most implications. This 'immersion' requires the availability, either internally as human memory, or externally as image collections, of great quantities of visual data.

Despite its apparently greater emphasis on verbal exposition, the 'new' art history also seems to rely implicitly on the interiorisation of an immense catalogue of possible variations in ways of making significant marks in order that meaning can be credibly ascribed. This is T J Clark,

A work of art may have ideology as its material, but it works that material; it gives it a new form and at certain moments that new form is itself a subversion of ideology. ¹⁷

The amassing of large amounts of comparative data in human visual memory could be to make a detailed assertion as readily as an attribution, or a broad generalisation. This is Eric Fernie on Norwich Cathedral,

Round double-splayed windows are of course found in buildings of Anglo-Saxon date, but this does not prevent them being used after 1066, and like other features such as long- and short- work, they occur in a number of securely dated post-Conquest contexts. ¹⁸

The architectural historian Richard Morris is currently working with the Institute on a project which aims to use shape based retrieval methods to find and match architectural moulding profiles. ¹⁹ A range of thorny problems arises in what is mainly a practical computing project, such as how to ensure the software knows which side of the drawn line is stone and which is air, and where moulding begins and wall surface ends, the scale of the image, and how to deal with completely undercut foliage. But the clear implication of the existence of such a visual glossary is that issues such as first and last appearance of a given feature could be conclusively answered. Such assumptions lie behind the seductiveness for funding bodies of many of the applications to digitise corpora of visual and material culture.

Art history as a discipline has been affected profoundly by the intellectual currents of the last thirty years. As a very crude generalisation, history of art, architecture and design now lays less stress on the canonical object, and more on the meanings which permeate them. Objects previously seen only in relation to other members of a single linear sequence are regarded as members of a class of objects in material or visual culture, each have a multiplicity of meaning relations. This challenge to the older formal, comparative methods could, in theory, be reversed as large classes of artefacts become more readily available in digital form. Marilyn Lavin, in a survey of the uses of IT by art historians in Art Bulletin, ²⁰ gives an example of the way the discipline could change, showing how a database enables more effective deductions through relationships of flat data tables and contextual information. It is possible that we might observe a return to the collections-based paradigm of research and teaching by which the subject was defined at the turn of the century, when the minute inspection of the original art work was impracticable for more than a few scholars. But does having a corpus (and appropriate retrieval technology) make it easier to ask the right questions? As Gombrich testily commented of the information driven paradigm of humanities research in Ideals and Idols, "Do we need a corpus of all door-knockers?" ²¹

Is Post-panofskyism a new paradigm?

As the earlier discussion of user studies suggests, the central question of 'aboutness' bedevils those who seek to automate retrieval of image collection. Everyone is familiar with the account of art historical method given in Panofsky's famous 1939 essay. ²² Its hegemony has been remarkable yet, as we have seen, it offers little support for information scientists striving to develop a consistent visual thesaurus for image delivery systems.

In another member of the Viennese School, Otto Pächt, we might find a more congenial conception of the way art history constructs or induces meaning relations in objects. In Christopher Wood's happy phrase, Pächt was a "more consistent relativist" than Gombrich, Wind or Panofsky himself. Instead of an iconographic 'circuitboard' in which images are attached to their idea, becoming in the process little more than a pictograph, Pächt substitutes a theoretical system in which the visual logic of a period possesses its own value and meaning. It is self-constructing. In Gombrich a normative logic must always drive representational convention in one direction. Perhaps this "forcing open of the field of the aesthetic" ²³ with its preference in Pächt's work for the early and late medieval rather than the classic, and in Meyer Schapiro's work, on the grotesque and the marginal, has a welcome resonance with the expanded field of scholarship which is promised by the accessibility of networked resources.

It is far too early to limit the range of possible issues of practice and method which might emerge from the research planned as part of this project. Here I can only hint at what seem to us to be some possible synergies between practical options and the working habits of art historians. If we succeed in shedding light on our subject, we hope that the outcomes of the project will be;

How will the project work?

The methodology which we plan to adopt can be summarised as follows. A literature review will be published following the background research phase which is currently underway and should be complete by March 2000.

During the same period we intend to recruit participants for in-depth interviews. In order to identify participants, we propose using a mixture of personal contacts, professional groups (e.g. Association of Art Historians), email discussion lists, etc. We recognise that, in such a small scale study, it will be impossible to collect together a fully representative sample covering all aspects of art history. However, to guide the choice and selection of participants, we propose establishing a set of criteria for each group, such as level of use of research materials in print and digital formats, record of publishing and giving conference papers. We will also attempt to achieve a balance of length of research experience, age and gender in each group.

With some trepidation, we have divided the subject into five sub-disciplinary groups on the basis of some rough and ready assumptions about how academics in each of these sub-groups typically engages with source material.

The expertise of the team in user studies will then assist us in developing and formalising the interview framework. This will be based on the findings of the literature review and the results of the pilot survey. We anticipate that, apart from purely factual and demographic details, the framework will be semi-structured. Questions will be mainly open-ended permitting subjects to reflect on their individual research process, to develop their own ideas and views on the research questions posed. We will then conduct a pilot interview to test the research instrument, and amend it as necessary prior to carrying out the in-depth interviews with the full sample of 25 art historians.

The interviews will aim to investigate the research concerns and practices of the art historians in the sample and will cover methods of research such as their use and handling of field notes and other evidence, and their writing process, in the context of the availability of digital images. The ethnographic nature of the interview process will aim to encourage personal recollection and reflection. Questions may focus on aspects of resource discovery on the Internet, including cataloguing, metadata and search techniques, digital image resources in research and the pedagogic implications of digital image resources and electronic delivery.

We envisage that each interview will be carried out in the place of work of the participant and recorded on audio tape. Where applicable and appropriate, participants may be observed whilst they undertake their research activity, particularly where it concerns access to digital images. Each interview will last about 6 hours (which may be conducted over two days, according to circumstances and at the convenience of the interviewee, and to allow time for reflection, etc). Analysis of the data will be carried out as the data is collected and transcribed, using a software tool such as ATLAS/ti. This software was chosen as it has already proved very effective in the VISOR research project based in the Institute for Image Data Research. The very loose conceptualisation of the "immersion/construction" process outlined above is already a sufficiently organised paradigm with which to embark on analysis using ATLAS/ti. The open-endedness of the coding permitted by the software will allow the paradigm to evolve as data is analysed.

We propose analysing the results of each interview, and then returning the analysis to the interviewee for their comment and correction. In this way, all participants become involved in validating the analysis. Participants' agreements, disagreements, elaborations, and reassessments become in turn a part of the data collection process.

Final report writing and dissemination of results should take place by June 2001. A project web page will be maintained on the IIDR's web site www.unn.ac.uk/iidr/CC/survey.html. A copy of the final report will be deposited with the Visual Arts Data Service.

Conclusion

Finally, I would like to report briefly on the results of a limited pilot of our initial questionnaire. The questionnaire is currently being completed, unprompted, by a small sample of art historians working at the University of Northumbria. The comment of one respondent,

I realise that I am a complete dinosaur. I really must come and talk to you about some of these resources and how I might use them suggests that we may need to alter the questionnaire to minimise the impression that we feel that art historians oughtto be using these tools and resources, and to give more space for respondents to give reasons why they either cannot, or choose not, to use them.

It is a common observation that teaching and learning strategies in Universities tend to be led by what students want, and much as by the staff who teach them. While most staff might have had their approaches to research in art history formed in an era when the key to a successful career was the discovery of a new cache of data (a group of letters, or drawings, whose existence hitherto unexpected, precipitates re-evaluation of the canon), their students are increasingly working on publicly available data, often in multi-disciplinary collaborations, with the aim of proposing new constructions or relationships. This may be a stage in the evolution of any discipline.

Our guess is that the obstacles to adoption will be substantially the same as those identified in past studies; lack of access to a desktop PC, absence of an adequate IT network connection, lack of time to acquire skills to use digital resources, and lack of knowledge about what is available to support the teaching and research of art history. The politics of organisations, of discipline formation, and of working methods, are all relevant factors and could, in themselves, provide a fertile research agenda. But, even assuming permeation has been slow, the percentage of staff with the necessary access to digital resources seems unlikely to have declined since 1988. Given the years which have elapsed, even on a worst case scenario, our respondents will surely enable us to frame some securely founded hypotheses about how art history has responded to the digital age.

Notes

3 (various authors), 'Digital Culture and the Practises of Art and Art History', Art Bulletin, June 1997, pp.187-216.

6 User Survey Report: a report of the results of the VADS User Needs Survey carried out between December 1997 and February 1998 VADS, 1998.

7 Enser, P.G.B. , 'Pictorial Information Retrieval', Journal of Documentation, vol 51, No 2 June 1995, pp126-170.

8 Holt, B. and Hartwick, L.. '"Quick who painted fish?" 2: searching a picture database with the QBIC project at UC Davis', Information Service and Use, 14 (1994) pp.79-90.

9 Panofsky, E. 'Iconography and Iconology: an introduction to the study of renaissance art' in Meaning in the Visual Arts, Peregrine, 1970, p.66.

10 Markey, K. 'Access to iconographical research collections', Library Trends, 37(2), 1988, p.154-47. See also Shatford, S. 'Analysing the Subject of a picture: a Theoretical Approach', Cataloguing and Classification Quarterly, Vol.6 (3), Spring 1986, pp.39-62.

12 Bakewell, E., Beeman, W. O. and Reese, C. M. (1998),Object, Image, Inquiry: The Art Historian at Work, Getty Art History Information Program.

13 Pointon, M., 'History of Art and the Undergraduate Syllabus', in Rees, A. and Borzello, F. (1986), The New Art History, Camden Press.

16 Brilliant, R., 'How an art historian connects art and information', Library Trends, Fall 1988, pp.120-129.

18 Fernie, E. (1993) An Architectural History of Norwich Cathedral, Oxford, Clarendon Press.

19 Morris, R.K., 'An English Glossary of Medieval Mouldings: with an introduction to mouldings c 1040 1240', Journal of the Society of Architectural Historians of Great Britain, Vol 35:1992 pp.2-17.

20 Lavin, M., 'Making Computers Work for the History of Art', Art Bulletin, June 1997, p.198.

23 Wood, C. Introduction, Pächt, O. (1999), The Practice of Art History: Reflections on Method, Harvey Miller Publishers, (trans. Britt, D.) p.18.

Digital Environments: Design, Heritage and Architecture

Chris Bailey and Margaret Graham

Compare and Contrast: the impact of digital image technology on art history