As of 22 April 2009 this website is 'frozen' in time — see the current IFLA websites

This old website and all of its content will stay on as archive – http://archive.ifla.org

IFLANET home - International Federation of Library 
Associations and InstitutionsAnnual 

64th IFLA Conference Logo

   64th IFLA General Conference
   August 16 - August 21, 1998


Code Number: 033-99-E
Division Number: IV
Professional Group: Classification and Indexing
Joint Meeting with: -
Meeting Number: 99.
Simultaneous Interpretation:   No

Working with LCSH: the cost of cooperation and the achievement of access.
A perspective from the British Library

Andrew MacEwan
British Library,
National Bibliographic Service


In the past few years, following the reinstatement of the Library of Congress Subject Headings to the British National Bibliography in 1995, the British Library has confirmed its long-term commitment to applying LCSH in its catalogues. This has caused the BL to focus more attention on the long-term development of the headings. Although the Library of Congress has editorial responsibility for the system, the cooperative environment creates a new forum in which all users of LCSH can contribute to its development. Improvements are required in vocabulary control and in the clarification of principles and policies for the application of the headings, especially where these can improve standardization for cooperative cataloguing. The British Library has already been involved in some development work on these issues for the indexing of fiction. The LCSH system as a whole has to be improved, but development of the headings must be linked to the development of search systems which fully exploit the thesaural structures already in place. Fully automated authority control is also an essential first step towards large scale improvements to the headings themselves. If resources are given to its development there is huge potential for improving subject access to catalogues through LCSH.


1. Introduction

In 1995 the British Library reinstated the Library of Congress Subject Headings system to the records for the British National Bibliography in direct response to demands from customers and cooperative partners. At the end of 1996 the BL's own COMPASS system of indexing was dropped from the BNB with little or no protest from anybody. Through a process of economic retrenchment and a developing strategy for cooperation the British Library has reached a point of commitment to a single subject indexing standard that is controlled, edited and maintained by another library. It has been suggested that following this path effectively means the end of our involvement in development work on the principles and practice of subject indexing. (1) We have, after all, abandoned the systems which we ourselves developed. In this paper I want to argue that the BL's adoption of LCSH as its sole subject access system merely changes the context in which we contribute to such work. The co-operative context brings with it new costs and challenges, and if we are to achieve good subject access to our catalogues and collections we have to work to address the problems that are shared by all of us who use this standard.

That LCSH is a system which is in need of a lot of development work is a well-attested fact. In thinking about a British Library perspective on the development issues I find that it is to a large extent economics which sets the agenda. As a cataloguing manager I am permanently faced with two questions. What are the costs of using LCSH for us? What are the benefits for our users? It is by answering these questions that we define the critical development issues for LCSH and help to inform the contributions that we all can make to that development. In this paper I want to review some of the familiar issues for LCSH development from this British Library perspective. I will also describe briefly some subject indexing development work in which the BL has been engaged: the indexing of fiction.

2. Using LCSH at the British Library

First a bit more clarification about the context for the application of LCSH at the British Library. The home of LCSH has always been on the records which we create for the national bibliography, where it has been applied as value-added data. This context has been completely changed in the past two years by the decision to go forward with both LCSH and Dewey classification as core elements of records created for the British Library's own catalogues. (2) The significance of this change is that it radically shifts the emphasis in how we think about LCSH: now less in terms of the utility of our records and more in terms of the utility of our catalogue. In other words we are now thinking much more about how LCSH serves to provide access to our collections, with the BNB becoming something of a by-product in relation to this. For instance, a key factor in the decision to abandon our own system is the fact that LCSH provides huge potential for deriving subject headings retrospectively onto records on our catalogues which have never been indexed, and which we cannot afford to go back and index now. Our experiences are therefore informed by our coming to grips with this changed outlook in considering LCSH as the cornerstone for subject access to the whole catalogue.

So what have we made of our experiences with LCSH? Three issues come out uppermost as impinging on how we best achieve good subject access to our collections: vocabulary control, policies for the application of LCSH and the shared cataloguing environment itself.

Good subject access might well be defined as the provision of access in the language and terminology that the end user is likely to search on. Foskett focuses on a central problem in achieving this with LCSH :
"Librarians and other intermediaries realize that LCSH is an artificial indexing language, and use it as such, but users probably do not, thus placing yet another barrier between them and the information they are seeking." (3)

The real issue here is that users should not have to realize that LCSH is a controlled vocabulary in order to conduct successful searches. Access by LCSH in most catalogues is access by the preferred terms established on the authority file, but without the benefit of the entry vocabulary of cross-references behind these headings. Although a lot has been written about the limitations and inadequacies of the thesaural relations on the LCSH file the fact remains that there are reference structures which offer the potential to greatly enrich access for the end user. Making decisions to change headings or references must be linked to the development of systems which will properly exploit these thesaural structures. This line of thinking has been well-developed in the work of Karen Drabenstott and others. (4) I mention it in order to emphasise that we take such systems developments as an underlying assumption in our thinking about the inadequacies or otherwise of headings and thesaural links in the LCSH file.

Inevitably we had an initial concern about Americanisms. The idea of an American subject indexing language forming the basis for access to the collections of the British national library certainly seems to be a triumph for American cultural imperialism. And perhaps it is. But in terms of giving us problems with access I think it is a relatively trivial issue, and certainly one which can largely be overcome through developing thesaural control on the authority file. Many cross-references from Americanisms already exist to provide access in language more natural to a British user, although sometimes with slightly less natural qualifiers in order to avoid conflicts on the file, e.g.

Some references would be more natural than the main heading to an American user too, e.g.

Although not all variations can be dealt with by simple one-to-one cross-references, many can. Participation in the Subject Authorities Cooperative Program (SACO) allows us to propose such references to LC as we identify the need for them in cataloguing, e.g.

More serious concerns for us are common to any user of LCSH and are certainly not unique to the non-American context. For instance the need to modernize awkward, archaic or unsatisfactory subject headings. It is not especially a British cultural perspective which finds fault with expressing the history of England as the history of Great Britain, e.g.
Great Britain--History--Edward 1, 1272-1307,
it is simply a failure in specificity which we would all like to change. Changes on this kind of scale carry with them major implications for authority control of the headings assigned in bibliographic files and this has been a major in-house stumbling block for LC. Lois Chan identifies the tension between the stability of the system and its responsiveness to change as a key issue for LCSH (5): the benefits of change must be weighed against the cost of making those changes. The fact that it is a stable system is of great value in terms of the past investment of indexing in many catalogues across the world, but clearly there are some headings or constructions that fail to express the concepts accurately or in a way that an end user is likely to search for. The trick is to bring in the changes without compromising the coherence of the catalogues where LCSH is deeply rooted over time: a problem for authority control which I will come back to later.

However good an indexing language LCSH might be or become it will only provide good access to a catalogue if it has been assigned consistently according to clear principles for subject analysis. After a gap of seven years the re-introduction of LCSH at the British Library has effectively involved starting from scratch in learning how to apply the headings. Because we have pursued a goal of integrating subject work with descriptive cataloguing in the design of the standard cataloguer job this has meant investing a large amount of time and effort on training cataloguers in the application of LCSH. This experience has served to concentrate our minds on the complexity and difficulty of getting to grips with LCSH as a systematic standard. The system of free-floating subdivisions which contribute so much to the flexibility of the indexing language also bring with them different patterns of indexing in different subject areas and for different kinds of material. The size of LC's Subject Cataloging Manual: Subject Headings (SCM), at four volumes, is itself an indication of the learning curve which cataloguers have to traverse.

Adherence to the principles and policies outlined in the Subject Cataloging Manual is at the heart of the cost-benefit equation that we have to balance in using LCSH. We place prime importance on adherence to the manual in order to achieve consistency in our own indexing and consistency with the indexing practice of the Library of Congress and our cooperative partners in the UK. Following the SCM makes indexing more time consuming and therefore costly, but the quality of our own indexing is improved and we can derive copy from other libraries without losing the benefit of a consistently applied indexing system. Well, that is the theoretical ideal. In reality the complexity of the system means that consistent practice is not always achieved. The SCM is a difficult tool to use and there are certainly arguments for changing, developing or clarifying some of the policies for application. The cooperative context means that achieving indexing consistency has two dimensions: internal, oriented towards training, and external, oriented towards policy development.

Internally our primary concern is to ensure that we are able to maintain and develop our expertise in LCSH. One of the difficulties with our multi-skilled cataloguer job design is that the task of subject indexing comprises only a part of a wide range of specialist cataloguing skills, which include descriptive cataloguing, application of Dewey classification and authority controlling name access points. Consequently gaining basic proficiency in the complexities of LCSH takes a long time and developing deeper expertise has to be balanced with the other demands on a cataloguer's skills. The particular difficulties experienced in applying LCSH are leading us to re-examine the way in which we design cataloguing jobs and the cataloguing process in order to maintain and improve both quality and efficiency. In the past we have maintained a tradition of generalist cataloguing, and this may continue, but we are now considering whether it would improve both job design and the effectiveness of our cataloguing to develop cataloguer expertise by allowing them to concentrate on material in particular subject areas. Something which would allow cataloguers to focus development of their LCSH and Dewey expertise, and which will in turn better inform our contribution of new headings through SACO and our input on policy development issues for the SCM.

Externally our chief concern is to ensure that what we are doing in our indexing is consistent with the application of LCSH on records which we derive from others. We have to face differences in application based on differing interpretations of the SCM and deliberate local deviations based on the different needs or perceived needs of different libraries. Perhaps one of the most difficult principles on which to achieve consistency in LCSH application is that of specificity. Chan identifies a whole raft of problems associated with this central concept: Is specificity defined by the place of a particular heading in the hierarchical structure? Or is specificity purely a matter of co-extensivity with the item catalogued? If so to what depth should an item be indexed to assign specific access points which represent significant parts of the work? Are terms too specific if they are never likely to be searched for by an end user? Is specificity best-achieved by assigning pre-coordinated, highly specific headings, or by assigning several broader headings which post-coordinately cover the specific content of the work? - LCSH allows for both approaches. I do not intend to provide answers to these questions here, but I do want to draw attention to them as an ever-present factor affecting consistency in indexing. Perhaps the most common type of variance is in the use of generic postings: i.e. the practice of indexing a book on Cataloguing with both the heading Cataloging and the heading Library science. The strict SCM policy is to assign only the former, but many consider the policy limits access on their catalogues in some circumstances, e.g. a book about the philosophy of Wittgenstein would only merit the specific access point:-
Wittgenstein, Ludwig, 18...
but a cataloguer may feel that the generic posting Philosophy is also merited in order to provide complete, post-coordinate subject access. The lack of a resolution to such variations in policy between libraries is a real problem for the cost-effectiveness of copy cataloguing. If half of the records in our catalogue have been indexed with generic postings and half without then the coherence of the catalogue as a whole is compromised, carrying with it implications for how we design search systems to best exploit the indexing language. I do not mean to suggest that variations in policy between libraries are either unacceptable or unjustifiable in terms of cost. But I do suggest that variations should be understood, known, justified and dealt with according to the needs of each individual library's catalogue.

This leads us back to my starting point which was the contention that adopting LCSH in preference to our own system merely changes the context in which we have to be involved in the development of subject indexing. Drabenstott argues that since LCSH has by default become the subject indexing standard for many libraries the whole of the cooperative cataloguing community has a responsibility to make the best of it. That is: a responsibility to make it into something better. There is a great deal to gain from sharing a standard system for subject access, both in cost-efficiency and in the benefits to end users of our catalogues, but it is unreasonable to expect that LC can or will carry all the development costs alone. During the past year we have had the opportunity to gain some first hand experience of how development and cooperation can work together.

3. Developing access to fiction

When the system of COMPASS headings was abandoned at the end of 1996 we were faced with the need to make a decision over the future indexing of fiction. It had been a long-standing tradition of access in the BNB to assign general form headings to individual works of fiction: e.g. English fiction. To apply the same level of access using LC subject headings would involve establishing a separate BL policy at variance with the Subject Cataloging Manual. Since this carried with it implications for copy cataloguing we chose instead to adopt a standard closely linked with LCSH which had good credentials for copy cataloguing purposes through its adoption by the OCLC/LC Fiction Project: namely the Guidelines on Subject Access to Individual Works of Fiction, Drama, etc. (GSAFD) (6). LC records for individual works of fiction are indexed by a consortium of public and academic libraries in the US coordinated by OCLC.

The GSAFD provides for 4 kinds of subject access: access by specific form, by setting, by character and by subject. Access by form is provided by a range of GSAFD headings, listed in the standard, most of which conform to equivalent headings on the LCSH file, but which are assigned in the 655 MARC field, rather than the 650, e.g. 655 Science fiction. The other 3 kinds of access are all supported by standard LCSH headings and constructions. It is the headings assigned in the 655 field which open up a range of issues for the future development of LCSH.

The Library of Congress has already announced its long term intention to develop the structured voacbulary of the LCSH file by establishing a separate file of form headings which can be assigned in the 655 field rather than the 650 field. The intention is to distinguish access to works that belong to a form or genre, from works which are about that form or genre:
650 Detective and mystery stories
655 Detective and mystery stories
will retrieve separate search sets. The Library of Congress plans do not impinge on the level of access that they will continue to assign to individual works of fiction in house. Access by LCSH 655 will sit side by side with access by GSAFD 655 on both the LC and the BL catalogues. This opens up questions concerning consistency and collocation on the catalogues where there are variations in the form of headings assigned.

In order to get to grips with the implications of the LCSH 655 for our own fiction indexing the British Library set up a working group to explore the issues surrounding access by form and genre for all works of fiction, both individual and collected works, widening the issues to include consideration of the use of the proposed form subdivision, $v. Contacts with LC and with relevant ALA subcommittees ensured that we grounded our work in debates that were already advanced in the wider cooperative community. The result of our work was the submission of a discussion paper to LC, to the ALA SAC Subcommittee on Form Headings/Subdivisions implementation, and to the SAC Subcommittee to Revise the Guidelines on Subject Access to Fiction. (7)

I will not attempt to rehearse the arguments of our paper in detail. The purpose of the paper was to present a coherent, and complete set of recommendations for standardising the application of form data to works of fiction, in order to provide a focus for discussion and debate which had hitherto been piecemeal. We have taken it as an underlying asumption that the future development of LCSH provides an opportunity to subsume the GSAFD headings wholly into the larger file in order to maximise the benefits of thesaurally controlled access to individual and collected works of fiction. Arguing the case in detail led us into addressing a variety of issues: What does the principle of specificity mean in the context of form data, as opposed to topical data? Does disentangling form data from topical data by means of the new MARC tags present an opportunity for re-visiting certain standard practices for using pre-coordinate strings or post-coordinate headings? How are genre headings defined?

How our contribution of some answers to these questions will serve in the eventual development of LCSH in this area remains to be seen. Working in the cooperative environment means that progress is made through dialogue and it is encouraging to find that both the Library of Congress and the wider community of LCSH users interested in this area have responded extremely positively to the paper. A dialogue has been opened.

4. Some priorities for the future

I would like to finish by mentioning what I consider to be two key priorities for the future development of LCSH. I have concentrated in this paper on the costs and challenges that we have to face in applying LCSH to our records. Subject indexing is an expensive occupation for libraries and the LCSH system needs many improvements to make it an easier tool to use. There is also much that can be done to make it a more effective system for providing subject access on our catalogues. In order to continue to justify the costs of applying LCSH it is vital that where the costs of applying it can be reduced, they are; and where the benefits of applying it can be increased, they are. Since much of the improvement will be gradual and require commitment and resources over time I would suggest that there are two urgent priorities for development which can help to bolster the case for this continuing investment by delivering some immediate benefits:-

  1. Improvements in OPAC software to exploit LCSH

    The reference structures already present on the LCSH file provide a significant starting point for the development of software to enhance subject searching on catalogues. As an example, I was recently treated to a demonstration by Steve Pollitt of his view-based searching system which he has developed with the EMBASE thesaurus, and is now developing for use with Dewey. (8) Potential future developments can be envisaged which would exploit links between Dewey and LCSH.and the reference structures on the LCSH file. The system might also exploit the post-coordinate associations of headings on bibliographic records. Other research, such as Karen Drabenstott's proposal for system generated search trees, offer alternative strategies for developing the user friendliness of subject access through LCSH. By designing OPACs to exploit the structures which are there in LCSH a benefit would immediately be provided for end users. Further development work on reference structures would also become better informed by the demonstration of how LCSH performs in this enhanced environment.

  2. Improvements in authority control

    Major improvements in the authority control of LCSH are already planned and scheduled by the Library of Congress. A full authority file of free-floating subdivisions will finally base the application of LCSH entirely in the online environment. This will provide a platform for resolving the problem I touched on earlier: how to bring in changes without compromising the coherence of the catalogue. Systems with linked authority files have the potential to allow for the retrospective upgrade of the LCSH assigned on catalogues. Savings will be made where it is no longer necessary to maintain consistency by amending headings record by record. Benefits to the vocabulary will also accrue where there is no longer an obstacle to change in the scale of the manual amendment of backfiles.

5. Conclusion

Like any other market product in the modern world if LCSH does not change it will sooner or later be abandoned. Its deficiencies will make it vulnerable to the sideways competition of automated text-search packages. All of us who use LCSH have a vested interest in seeing it improve, which means that it is a cost-effective goal for us to seek to contribute to its development. We can propose new and changed headings, we can contribute to the development of subject indexing policies through open dialogue, and we can develop our systems to take full advantage of LCSH in automated form. In the English-speaking library community the level of past-investment in LCSH makes it the only runner for the general library catalogue. Improving LCSH is a choice between having controlled subject access or having none. Without LCSH, despite all its current faults, the future for subject access to our catalogues and collections would be greatly impoverished.



  1. McIlwaine, I.C. "Subject control: the British viewpoint", Subject indexing: principles and practices in the 90's, Holley, R.P.(ed.) et al, Munich, K.G.Sauer, 1995, 166-180 (UBCIM Publications - new series vol 15)

  2. Oddy, P.M. The case for international co-operation in cataloguing: from copy cataloguing to multilingual subject access. (paper for a seminar on "The Future of Cataloguing", held at the Royal Library, Stockholm, January 1998) Foskett, A.C. The subject approach to information, 5th ed. London, Library Association Publishing, 1996, 346.

  3. Drabenstott, Karen Markey & Vizine-Goetz, Diane. Using subject headings for online retrieval: theory, practice and potential, New York, Academic Press, 1994.

  4. Chan, Lois Mai, Library of Congress subject headings: principles and application, Englewood, Co. Libraries Unlimited, 1995

  5. Guidelines on Subject Access to Individual Works of Fiction, Drama, etc. Chicago, ALA, 1990

  6. The application of form data to works of fiction: discussion paper. British Library Fiction Indexing Group, August 1997: posted at http://www.pitt.edu/~agtaylor/ala/papers/blfictio.html

  7. Pollitt, A Steven, The key role of classification and indexing in view-based searching IFLA '97 Copenhagen Aug 31 - Sept 3 1997 63rd IFLA General Conference Booklet 4, Section on Classification and Indexing Session 95 Paper 009-CLASS-1-E