In many ways, this paper is an extension of my presentation at ARLIS last year in the panel session, "Paving the Cow Path" - this time the subject path. The idea, again, is to put things into some historical perspective - in fact, to learn where the problems and inconsistencies come from. To that end, I will be discussing the development of Library of Congress Subject Headings. I will do this generally and then in terms of art specifically. And to close I will say a few words about the Art and Architecture Thesaurus.
At the 1984 International Congress on Automatic Processing of Art History Data and Documents, Elaine Svenonius speaks the truism that "language is the embodiment of information." The language of interest to us here is that of controlled subject vocabularies - not for the purpose of describing images, but for providing access to materials which discuss images (although some of these vocabularies can do both), in other words, subject cataloging. It has been noted that the practice of subject cataloging lacks a code, an AACR of some sort. This lack of a code has a long complex history.
It began, as all good American library stories begin, with Charles Cutter. His Rules for a Dictionary Catalogue, beginning with the first edition in 1876, include guidelines for assigning subject entries. It is important to note that he struggles with many of the same issues that continue to baffle us: inversion and compound subjects and the formulation of geographic headings. It is the case that Cutter's rules can be said to constitute a code of a kind, written in his unmistakable chatty style, interspersing instructions with rambling expressions of doubt and the intricacies of his thought process. Our current body of rules is quite dry by comparison.
Cutter's "objects" which apply to the function of a dictionary catalog, can still be applied as the purpose of subject access for any type of library catalog:
1. To enable a person to find a book of which the subject is known, and
2. To show what the library [database] has on a given subject and in a given kind of literature.
There is no question that Cutter greatly influenced those who followed him. The principles which drive his rules are behind current cataloging practice. They include the notion of specificity, the consideration of the user as the principal basis for subject heading decisions, the practice of standardizing terminology, the use of cross-references to show preferred terms and hierarchical relationships, and the complex attempts to solve the problem of order of elements. However, an important difference between Cutter and those who followed (especially LC) is that the basic philosophy under which Cutter operated, inherited ultimately from the Enlightenment, consists of the notion of a harmonious and rational universe created by a benevolent deity and capable of being understood by analytical procedures. This is heady stuff to be considered in the context of subject cataloging, but the point is that Cutter's approach to the work is in terms of creating topics which fit into a universal scheme.
Nothing else resembling a set of rules, or code, was to appear for nearly 50 years after Cutter. In 1897, James C.M. Hanson became chief of the Catalog Department of the Library of Congress, and began the process of developing its subject heading system. His was the pragmatic approach, rather than the philosophical - the subject terms came from the books themselves in the Library's collection. This is not news but it is an important point. Certainly, Hanson made use of the work of others, including the A.L.A. List of Subject Headings for Use in Dictionary Catalogs, first published in 1895. But for the most part, LC's list grew by accretion and was based on decisions of individual catalogers, many of whom were hired as specialists in appropriate fields. This system produced a variety of sequences and patterns the logic of which was not always clear to anyone who had not been privy to the rationale behind it.
The Library began to publish its list of subject headings in 1908, and any principles attached to the list (and its subsequent 22 editions) evolved organically from the exercise of creating headings for Lc's collections. The next code-like publication from the Library of Congress is the 1951 Subject Headings: a Practical Guide, by David J. Haykin, first chief of the Subject Cataloging Division, created as such in 1941. Again, this is a set of principles derived from established practice. As we know, LC was to produce many other useful guidelines later in the century - particularly the Subject Cataloging Manual: Subject Headings (first edition published in 1984).
To return briefly to the notion of language embodying information, if we want to access the information, we want to ask what sort of language is needed. It is the case that some systems make use of natural language for subject access, but most of us are accustomed to controlled vocabularies. The controlled vocabulary does not foster a sense of immediacy - it places a kind of intermediary between the searcher and the database, requiring expertise on the part of the user as well as the cataloger. The choice of terms and their structure are important - each heading must have a specific meaning; headings must be current, even politically correct, but, we hope, not trendy.
In LCSH there is a great deal of variety in heading structure, including nouns alone, nouns with adjectives, compound phrases, and headings with subdivisions. As a practice inherited from Cutter, many of the headings with adjectives were established in the inverted form in order to bring the significant word into prominent position. As filing order becomes meaningless in an online environment, LC made the decision in 1983 to establish new headings in direct form. Because of the very large number of headings already established as inverted, many categories retain this pattern, including many art related headings.
Another significant phenomenon affecting the development of subject vocabularies, particularly the LCSH, was the growth in the early part of the 20th century of specialized fields of knowledge, including art, and also special libraries to accommodate scholars in these new fields. The Library of Congress was part of this trend early on, Hanson hiring subject specialists to develop unique terminology and special patterns of collocation. This practice continued with Haykin later in the century, the result of which has been to perpetuate "specialty" even in the nation's foremost general library.
Until the publication of the Subject Cataloging Manual in 1984, any documentation from LC pertaining to subject headings after 1951 was in the introduction to LCSH, which was reprinted as a separate work in 1975. In 1990 Lois Chan prepared for the Library: Library of Congress Subject Headings: Principles of Structure and Policies for Application, as they themselves were feeling that the documentation was becoming unwieldy. For us art catalogers, two sets of guidelines were generously provided by Anna Smislova, former art subject specialist at LC, in Art Documentation in May, 1983 and Summer, 1983. Currently, guidelines for the application of subject headings in art are found in instruction sheet H1250 in the SCM, which first appeared there in 1988. Aside from cosmetic adjustments (the addition of MARC tagging and later $v for form subdivisions), H1250 has remained basically the same. However, big changes are afoot, and we can expect a new version of H1250 within the year.
There have been changes in art cataloging practice over the last 25 years - for example the elimination of plural forms for individual art genres (paintings) in 1977, the abandonment of the distinction between movable and permanently located art in 1981, and the allowing of the subdivision "Exhibitions" under artists' names; these in addition to the general updating of terminology mentioned before.
In an effort to correct inconsistencies which have accrued in the list since the beginning and to bring subject specialties in line with each other, the Library of Congress held the infamous conference on The Future of Subdivisions in the Library of Congress Subject Heading System, at Airlie House (VA) in May 1991. Recommendations from that conference were widely distributed in the form of a report written in 1992. There were five recommendations summarized as follows:
1. That the order of subdivisions be standardized in the pattern topic - place - time - form except when the first element is geographical
2. That there should be authority records created for headings with subdivisions and for subdivisions alone. Cooperation is encouraged in the creation of authority records.
3. That chronology expressed in subject headings should reflect the content and not the publication date of an item, and that it should be expressed in numerical dates rather that words which describe a period - this latter is a machine sorting issue. In conjunction with this is a recommendation in favor of free-floating chronological subdivisions.
4. That the Library of Congress develop a subfield specifically for form subdivisions.
5. That LC continue wholeheartedly the practice of indirect subdivision of geographic names and even add the general indirect form to authority records for individual place names.
6. That there be a general trend toward simplification of subject headings and their subdivisions, reducing a tendency toward fine distinctions.
Following this pivotal event were a number of implementation studies and further reports and reviews, of course. Of interest here is the Nov. 1993 skillful, mind-bending report by Bethany Mendenhall on "Implications of Art Subject Headings." She was among those assigned to the task by a subcommittee of A.L.A.
Other general and art-related position papers were written by LC and others and distributed throughout the art cataloging community. Partly as a result of this effort, in the fall of 1999, a proposal came straight from the CPSO of substantial revisions to art subject headings. Responses were requested to be submitted by Dec. 31, 1999. There is no telling how much our comments will be taken into consideration. I could not get a sneak preview from CPSO although I tried to, in light of the conference and this particular session. I was told that the proposal went to A.L.A. Midwinter, that responses were still being received, and that CPSO was just beginning to sort it all out.
As with many subject areas in LC practice, art headings are fraught with vagaries and inconsistencies. A long-standing aberration is the distinction made between the "fine arts" and the "useful" or "decorative" arts, the latter being assigned with general patterns prescribed by LC. To understand the origin of this divided thinking, I direct you to an interesting article written by Steven Blake Shubert of the U. of Toronto, which appeared in Art Documentation, Summer, 1993. This distinction is made quite clearly in H1250 and the instruction sheet offers guidelines only for "fine arts."
With this background in mind, I will touch on a few problems and inconsistencies specific to art headings - the troublesome offspring, so to speak, of past practice - a combination of specialized decision-making and codeless accumulation. The first involves the practice of inversion. Cutter's Rules for a Dictionary Catalog include the instruction: "Enter a compound subject-name by its first word, inverting the phrase only when some other word is decidedly more significant..." This practice continues, quite frequently in the case of art subject headings, although even Cutter admits, "It must be confessed that this rule is somewhat vague and that it would be often of doubtful application...."
Modifying topical headings by inversion is a collocation devise useful in manual systems, but is unnecessary in an online environment. Shifting policies have created inconsistent application. As I said before, in 1983 LC made the decision to create any new headings in direct form except where a pattern of inversion already existed. The list of exceptions includes: those qualified by time periods; those qualified by languages, nationalities, or ethnic groups; and those with modifiers for artistic or musical style (Baroque, Renaissance, Rococo).
In spite of its slippery nature, the distinction, just referred to, between stylistic and national modifiers is significant because you cannot apply national modifiers to certain "real" art headings (not just decorative arts), such as Architecture. You can do "Architecture, Baroque -- France" and "Architecture -- France," but you cannot do "Architecture, French -- 18th century."
It is also problematic trying to make a distinction between period (chronological) and style modifiers, such as "Ancient," "Renaissance," "Victorian." In her guidelines for "Art Subject Headings: General principles," (Art doc, May 1983), Anna Smislova states that they are to be considered as both. But H1250 seems to separate them into two camps.
The next problematic area is that of subdivisions and the current attempt to normalize their order. In most cases, subdivisions serve to bring out various aspects of the main heading. In earlier editions of LCSH, options were comparatively simple, but patterns have been introduced in all subject areas which are quite complex, creating very long subject strings. Instructions for the organization of elements take up a good portion of SCM. There is a logic to the order in which subdivisions are organized - usually one goes from the broad to the narrow - but there are many other factors. Sherman will have a bit more to say on this topic shortly.
Probably the quirkiest use of subdivisions for art headings is the two different concepts of place. When expressing origin, that is where a body of art was produced, a national qualifier is used - as in Painting, Italian. If we need to say that the paintings were produced in Siena, we say "Painting, Italian -- Italy -- Siena." If these very same paintings are located at the Museum of Fine Art, Boston, we have to say "Painting -- Massachusetts -- Boston;" in other words, the unmodified genre signals location. Now it is debatable whether or not this long-standing distinction is understood by art researchers. It may actually be clearer that we are describing location if we were to say "Painting, Italian -- Massachusetts -- Boston," or it may seem as if we were describing the paintings of immigrants. The truth is that there is no fool-proof way to handle this problem.
William Studwell, a supportive critic of LCSH, offers his solution in his 1990 (pre-Airlie) book, Library of Congress Subject Headings: Philosophy, Practice, and Prospects. He would eliminate all geographic modifiers for art headings, using geographic subdivisions regardless of meaning. In other words, "Painting -- Italy -- Siena" could mean paintings from Siena or located in Siena. This solution would make fine arts consistent with decorative arts headings and with "Architecture," neither of which can be modified geographically. To express both origin and location on the same record, you would continue to have a pair of headings. For Milanese furniture located at the Philadelphia Museum, you would have both "Furniture -- Italy -- Milan" and "Furniture -- Pennsylvania -- Philadelphia." For British style architecture located in India, you would have both "Architecture -- Great Britain" and "Architecture -- India."
The LC 1999 revisions in cataloging of art materials includes a proposal to combine the idea of origin and chronology. Instead of the two headings "Art, French -- France -- Paris" and "Art, Modern -- 20th century -- France -- Paris," headings with not entirely overlapping meanings, LC is proposing use of a single heading: "Art, French -- France -- Paris -- 20th century." This is in line with the Airlie recommendation of topic - place - chronology.
Well, of course, there's lots more. But before I get to the point when you might be worried about a pop quiz, I'll quit. My intent was to give you a small sampling of thorny art subject heading issues. But before I do sit down I want to touch on the fact that there are other vocabularies besides LCSH, indeed other whole systems of providing subject access to the contents of books and other materials. I will mention one: the Art and Architecture Thesaurus, which I bring up here to make two points. The first is to highlight some ofthe differences between the LCSH and the AAT (the list and the thesaurus). In some ways, the developers of AAT returned to Cutter's approach to organizing the universe of knowledge - the terms were chosen and set down in hierarchical relation to one another in an overall faceted structure. Combinations of ideas take place when the vocabulary is applied (to objects or to works about objects) - the terms are not precoordinated as they are in LCSH. A basic principle of the AAT is that of "image warrant, rather than "literary warrant," one of the principles guiding LCSH. Image warrant dictates that "a term is warranted if there is an object of interest to art scholars that can be described by it."
That aside, my other point is that, when all is said and done, the task of constructing or even applying any sort of controlled vocabulary (list or thesaurus) for material culture, for works of art and works on art, is a frustrating, limited, subjective effort. Anywhere language is used creatively and symbolically there will always be a lack of precision. Art is not science.
Which is maybe the main reason that, in spite of efforts made and desires expressed for a century, we still do not have a code.