In the light of international co-operation, subject indexing remains the weakest point in information exchange. Besides terminological and language barriers, an essential need exists for better und erstanding of the different existing systems and their conceptual foundations. This concern was addressed in 1990, with the establishment of the IFLA Section on Classification and Indexing's Working Group on Principles Underlying Subject Heading Languages (SHLs). The Working Group's activities and achievements are described, as well as explanations of its major findings.
Commonly accepted principles, standardized practices, harmonized tools and information sources are far from being a reality in subject retrieval between different bibliographic systems. Exceptions, to a certain point, may be considered in the use of classification schemes, namely, those universal in scope or of international use.
On the contrary, alphabetical subject indexing systems suffer from different types of barriers and constraints. These are much more difficult to overcome, particularly if we think about cooperation and transfer of information. These barriers and constraints are of two main kinds: some external, arising from the nature of the information world itself, and others, internal to the systems, deri ving from the nature of the intellectual and technical work involved.
Firstly, is the terminological aspect, expressing the complexity and dynamics of the natural language information. This is not easy to manage, even within a single language environment. Beyond the terminological aspect, are the linguistic barriers, a question that nowadays information systems must address in order to survive. The lack of multilingual facilities in the majority of bibliographi c systems has been partially alleviated by the international use of certain languages. Nevertheless, while this may be true for science and technology, it is hardly so for the social sciences and hu manities, where different languages are by themselves repositories of specific knowledge and information pertaining to local and regional culture.
Secondly, are aspects internal to the information systems that make cooperation and transfer of information difficult: aspects that derive from the subjectiveness of the content analysis process, as well as, from the options and devices of the controlled language used in each system. In addition to all these aspects, is the concept of the subject, and other conceptual foundations of a subject catalogue.
Because of different historical, linguistic and technical reasons, existing subject alphabetical systems have different backgrounds and use different forms to express knowledge through their control led retrieval tools. Different practices also have generated differences in concepts/technical terminology, sometimes affecting professional communication and understanding.
While it may be difficult to settle general and internationally acceptable solutions for terminological questions and linguistic issues, tasks that particular systems must face, it should be possibl e to reach some international understanding about conceptual foundations of subject alphabetical retrieval systems. These should become a focus point for further developments in national and interna tional cooperation.
At this point, the Working Group on Principles Underlying Subject Heading Languages (SHLs) began its task: first to devise the most general principles on which any subject retrieval system should b e based; then, to review existing real systems in the light of such principles and compare them in order to evaluate the extent of their coverage and their application in current practices.
Background and history of the Working Group
An investigation of the feasibility of developing an internationally acceptable statement of principles for the construction of subject headings was part of the IFLA's Section on Classificati on and Indexing's Medium-Term Programme (MTP) 1981-1985. This goal was temporarily removed from the subsequent MTP, when the Section's attention was directed to other issues, such as the content of subject authority files, because of the Section on Cataloguing's work on the Guidelines for Authority and Reference Entries (GARE).
At the 1990 Stockholm Conference, when the work on the Guidelines for Subject Authority and Reference Entries (GSARE) (1) was in progress, the Section decided to investigate principles for th e construction of subject headings, as it was felt that it was an important issue for the library profession (2). In order to proceed, a Working Group of members from several countries represented o n the Section's Standing Committee was established. Since then, the Working Group has met during each IFLA conference and corresponded throughout the year, adding more and more international represe ntatives.
The work was undertaken in two main phases. During the first three years, under the chair of Elaine Svenonius (University of California, Los Angeles, USA), the Working Group concentrated on identif ying and stating general principles of underlying subject heading languages, as well as clarifying the reasons for such a task.
During this phase, the most difficult and important part of the work was accomplished. Several drafts of the Principles were issued and discussed. The definitions and statements of princip les now seem clear and concise. This task of identifying, organizing and expressing the definitions and statements in a way that met all the members conceptual points of view required a great effort . I must stress that this was in fact the most interesting part of the work, where much was learned .
Before 1993, three levels were considered in the conceptual framework for the identification of principles underlying subject heading languages: a Content Level, a Linguistic or Expressive Level and a Pragmatic or Application Level. These covered nine identified principles. In 1993, the work evolved to a more simple presentation of such principles with only two major levels: Construction P rinciples and Application Principles. The already identified principles remained basically the same but the corresponding statements were improved in order to reflect the Working Group's consens us on the matter and, therefore, show the widest possible applicability in real subject systems.
The same occurred with the definitions included in the document, a very important issue for the Working Group. The discussion on what concepts to consider, what terms to use and how to formulate t he definitions, gave the Working Group the opportunity to clarify the various options and points of view that are always context dependent, and relate to the terminology used in different systems, languages and/or countries.
The Section's IFLA Satellite Meeting organized in 1993 in Lisbon (3), coincided in general with the end of the first phase of the Working Group's activities, and provided an excellent opportunity fo r an open discussion about Subject Indexing: Principles and Practices in the 90's. After the Lisbon meeting, a document was prepared giving Background, Purposes, nine Construction P rinciples and two Application Principles.
Also at the Lisbon meeting, the second phase of the work began. International participation on the Working Group was expanded in order to include statements from materials pertaining to different l anguage and country based systems, with examples from corresponding subject heading lists.
A first draft document, with Principles illustrated with information from five systems/countries (Canada, Germany, Iran, Portugal, USA), was issued in 1994 for discussion by the Working Group . After its review at the Havana Conference, a second draft was prepared, enhanced with illustrations from additional countries/systems (France, Norway, Poland, Spain). It should be distributed for worldwide review at the end of this year.
Principles underlying subject heading systems: what are they for?
As defined by the Working Group, its first aim is to facilitate subject access to information on an international level. This statement recognizes that subject information is difficult to pr ovide and to access beyond linguistic and other frontiers of the systems where it is generated.
In order to improve communication in subject retrieval information, in addition to principles, practical devices at the level of real systems are needed. But since we start from an existing situati on where different systems appear with different backgrounds and features, we need to first establish the basis of a common understanding at the conceptual level.
The expression of such conceptual common understanding may assist in developing SHLs by stating what is meant by a good SHL and what desirable construction and application principles are for such languages. This second purpose is a prospective one, assuming that existing standards and guidelines do not meet all the basic aspects that should be considered in the design of new subject ret rieval systems.
Although standardization activities (4) have produced some international guidelines in the field, they have mainly focused on partial aspects rather than on a unified conceptual framework for the wh ole process of subject indexing, in its objectives/principles. Such a conceptual framework is probably out of the scope of standards, because they are, by nature, rule and procedure oriented. But, agreement on principles is a pre-condition for further developments in standardization of specific issues.
On the other hand, standards and guidelines are usually derived from the experience and understanding of diverse realities in the same field. Understanding is therefore a keyword for the third defi ned purpose: to promote understanding of different SHLs by identifying commonalities underlying them and providing a structure for their comparative study. Knowledge about different systems c an be useful in making decisions, especially when the knowledge is provided under the framework of general, common and pertinent points of view. This is why the last defined purpose is to provide a theoretical rationale for particular standards or guidelines for SHL construction and application.
State of the work and major findings
The results of the work to date are presented in a draft document (5) arranged in two parts. Part I introduces the Working Group and provides Background, Definitions and Principles.
Definitions included in the document relate to the concepts of Subject Heading Language, Subject Headings and Principles. The intention is to provide the largest possible scope for the Princ iples, namely encompassing pre- and post-coordinated systems.
Construction Principles cover general issues pertaining to the following aspects:
These principles were selected as the most important ones for any type of alphabetical subject retrieval system, having been derived not only from the knowledge and experience of the Working Group members but also developed independently of any particular system and affirmed by a large and representative sample of SHLs.
Part II includes the Principles reviewed in real systems, from a survey of each principle, with illustrating statements and examples taken from various systems. Reviewed systems appear with different levels of detail, derived from the type and extension of the information available about them (manuals or other introductory or guidance tools and subject heading lists). In the absence o f statements of principles, rules and procedures are given from which principles may be inferred. Texts from sources published in languages other than English were translated, only the examples of s ubject headings remain in the original languages. The idea is to provide basic but sufficiently comprehensive information about each system and at the same time, investigate and demonstrate the exte nt to which the defined Principles may be considered internationally accepted.
Nine countries are represented in the survey, with subject indexing tools relating to eight different languages. All refer to existing particular bibliographic systems, except for Norway's contribu ted standard. In addition, the UK's contribution refers to the British Library's use of the Library of Congress Subject Headings (LCSH) and its principles.
All reviewed systems are universal in scope (6) and provide for some kind of pre-coordination. Almost all of them are managed by or with the collaboration of the national library and are used curre ntly in the national bibliography of the country. Another common aspect is that in the majority of cases, these systems are also used by other types of libraries in their countries. Most of them al so provide manuals or compilation of rules intended for subject cataloguers, as well as subject heading lists, all but one in print.
Major findings that are verified in Part II - The Principles reviewed in real systems are the following: all the systems comply in general with the Construction Principles while Ap plication Principles are less explicit or evident.
Concerning the Construction Principles, there is evidence in all or in most cases, more by rules and procedures than by theoretical statements. Uniform Heading, Synonymy, Homonymy and Sem antic Principles are the most explicit ones, showing good general compliance with the international guidelines of ISO 2788. The Syntax Principle is assumed, but with less detailed informa tion, but underlies all the systems. The Naming Principle is generally observed. Finally, the Consistency, Literary Warrant and User Principles are the least visible ones, although th ey are observed in most cases.
In respect to the Application Principles, there is less coverage and conformity, especially when dealing with the Subject Indexing Policy Principle. While the Specific Heading Prin ciple appears clearly in about 70% of the cases, general statements of Subject Indexing Policy only appear in about 60%, mostly with information regarding rules and procedures which are le ss detailed than that provided for the Construction Principles. It should be noted that only a few systems refer to specific criteria and rules for special bibliographic materials.
Some conclusions, based on the analysis of the survey, can now be projected. However, such conclusions, as well as the analysis provided above, should not be taken as emanating from the Working Gro up because it has not yet discussed them. They are my personal views.
The first and most important objective is that the development of an internationally acceptable statement of principles underlying subject heading languages is feasible and supported in theory and p ractice by the existing subject retrieval systems.
Secondly, it appears that subject systems are currently more advanced and concordant to each other in technical devices for terminological control and semantic (paradigmatic) structure, but less str uctured and similar in respect to syntagmatic aspects of knowledge representation. In this regard, it seems that this needs to be explored to acquire common understanding and compatibility, if libra ries want to pursue pre-coordination (7).
A third aspect, concerns a less developed practice in defining systematic and comprehensive subject indexing policies, showing that the probably managerial aspects of subject retrieval systems shoul d be emphasized and supported by sharing experiences and producing recommendations.
Organizing knowledge to retrieve information in a controlled way is a multidisciplinary task requiring much more than just building and applying a controlled vocabulary, even when this does not appe ar explicitly. Theoretical foundations of subject retrieval systems are therefore of major importance, as well as the skills of its managers and the ability to assume the role played by different di sciplines in the process.
No matter what future directions the Working Group's efforts may take, it seems evident that something of use has been achieved at the international level. The task of the Working Group has been pr imarily an investigative one and extends beyond the Principles. The accompanying survey should be used as a learning tool in the conception or redesign of subject retrieval systems in the educationa l environment and in library practice.
(1) International Federation of Library Associations. Working Group on "Guidelines for Subject Authority Files". Guidelines for Subject Authority and Reference Entries i>. UBCIM Publications - New Series, v.12. München: K.G. Saur, 1993.
(2) The investigation was undertaken following a request by the Croatian Library Association.
(3) Subject Indexing: Principles and Practices in the 90's, ed. by Robert P. Holley, et al. UBCIM Publications - New Series, v.15. München: K.G. Saur, 1995.
(4) See Williamson, Nancy J. "Standards and Standardization in Subject Analysis Systems: Current Status and Future Directions." In Subject Indexing: Principles and Practices in the 90's, e d. by Robert P. Holley, et al. UBCIM Publications - New Series, v.15. München: K.G. Saur, 1995, pp.278-291.
(5) "Principles Underlying Subject Heading Languages: Draft Version" /
Working Group on Principles Underlying Subject Heading Languages of IFLA's Section on Classification and Indexing. May 1995.
(6) There is no intention to restrict the survey to general systems, although the collection of information has been mostly addressed to national libraries and national bibliographic systems. It is the intention to include some specialized systems such as MeSH.
(7) For some time, this has been an important professional issue for discussion. See Svenonius, Elaine. "Precoordination or Not?." In Subject Indexing: Principles and Practices in the 90's, ed. by Robert P. Holley, et al. UBCIM Publications - New Series, v.15. München: K.G. Saur, 1995, pp.231-255.