Thesauri

The ISO has issued guidelines for the establishment of monolingual (ISO 2788) and multi-lingual (ISO 5964) thesauri, but these appear to have been little used by archivists.

Further attempts have been made at national level to control terminology, in thesauri promulgated by national Standards institutions, by ministerial directions, or by the national archives or learned bodies.

In the USA, for example, the national Standard [Z.39.19-1980] on subject thesauri is informed by ISO 2788. A national thesaurus for information retrieval was issued as a national Standard by the USSR in 1980. The Direction des Archives de France has devised a thesaurus (W) governing the description and indexing of contemporary local administrative archives, for use by the archives of départements, whilst other localities have devised their own thesauri.

Work on thesaurus construction and subject indexing is in progress at national level in a number of other countries including Canada, China, Sweden, and under the impact of computerisation in many individual archives elsewhere.

Debates about the technicalities of subject indexing and the compilation of thesauri had been current throughout the archival world long before the advent of computers, and the present survey cannot enter into all the issues involved. As with archival description more generally, however, computerisation has served to sharpen the debate, and has opened up new practical problems and difficulties, as well as presenting opportunities.

Once again it is useful to remember that the needs of different countries, and of different kinds of archives within one country, may be quite distinct. It can be altogether more feasible for a local or specialist archives or even a national library to create a central subject index from its finding aids for archives and manuscripts, even down to item level, than for a large national archives faced with a vastly greater bulk of material to do the same.

Commonly in the latter kind of institution users still have to be instructed in the principles of provenance and the organisational framework of the record-creating bodies if they are to derive the greatest benefit from their searches. This helps to explain the recent crop of guides to national archives based on organisational histories (see above, p. ll) and also the urgency of the need felt in some countries to establish authority controls relating to the organisation and functions of government as well as to the more obvious fields of personal, place and corporate names.

But it is perhaps over subject headings that most interest is now being expressed. In the USA the Library of Congress Subject Headings file has become a de facto national Standard. This can have drawbacks as well as advantages. The assignment of appropriate subject headings to the reports received for the National Union Catalog of Manuscript Collections, for example, is a factor in NUCMC's considerable backlog of entries for publication. (In the United Kingdom an earlier initiative, in a pre-computer age, to introduce subject cataloguing of a comparable kind into the National Register of Archives was abandoned because of the staff costs involved and the unwieldy growth of the index).

Subject access to data can be greatly facilitated by computerisation and is virtually expected by many software packages, some of which come complete with their own built-in thesauri. With the computer, for relatively little extra effort (keying in a few control words at the moment of description, for example) whole new tracts of information can be opened up. But at the same time some discipline needs to be introduced into the process, otherwise terms may be used inconsistently and not be recognised. by the machine, or for that matter by the user, conducting the search.

Some archives and archivists have traditionally been opposed to the creation of general thesauri, for a variety of reasons including the feeling that every distinct fonds has its own character which should be allowed to dictate the terminology used (ADPA 1/1, 1972). Others argue that concepts and terminology in natural language vary so much even from region to region (Playoust 1988, pp. 115-117), let alone from nation to nation, that observance of any central rules would be misleading.

Partly for this reason the French application for the control of modern records, PRIAM 3, has been built up using natural language, with controls inserted only at a later stage.

By contrast, with some modest variations, archives as far afield as Hungary and Zimbabwe appear to be able to make use of the Library of Congress Subject Headings.

Other countries have their own national authorities, like the Repertoire des vedettes-matieres, Canadian Subject Headings and Canadian Authorities used by the National Archives of Canada.

Authorities used with reference to the names of persons, places and companies are almost always of local or national scope, as in the specific instances reported to the survey from Sweden and the United Kingdom.

There are further complications for highly specialised research centres where the archives may need a detailed command over concepts, or even place-names, required almost nowhere else and which are not covered in sufficient depth in any common authority.

The British Antarctic Survey, to take just one example, needs to identify very specific sites within a defined geographical territory as well as highly specialised scientific terms.

In the United States, the further research has progressed on the fundamentals of archival description, the more options it has been found necessary to create for the description and control of specific media or concepts.

A recent review of the position (Zinkham 1989) described the thesauri and controls already evolved or in progress: for art and architectural, cartographic, graphic and moving image materials, for form terms in general and for genre terms, binding terms, printing and publishing evidence and provenance, emphasising the need for careful analysis by form of material.

It is apparent to the outside observer that the main archival institutions, however much they wish to abide by standards or prescribed authorities, in practice tend to diverge in at least some respects to meet their own needs. One possible solution seems to be a loose federation of authorities where the minimum, and perhaps the only, requirement for the compiler of an archival description will be to give a clear indication which particular conventions etc are being followed.

With the free-text searches now possible on many databases, which are capable of producing word-by-word concordances of their data content, some argue that controlled subject indexing is no longer necessary. This depends largely on the scale of the individual application, but the use of some form of thesaurus-like control can be recommended as a means of taming the material for the searcher and reducing the time taken in searches. The many other forms of controlling searches, including the use of 'stop' and 'pass' or 'go' lists, word linkage, right and left truncation of key-words, and Boolean operators, and specific problems associated with marking up existing finding aids to assist such data control have been described elsewhere (Arad 1981 (2), 1987; Cloulas 1985).

In the particular fields of subject indexing and thesaurus control the current priorities would still seem to be at local and national rather than international level, although there is scope for the wider exchange of ideas and approaches to problems such as authority control. The professional associations of archivists probably have the most important role to play in this respect nationally, but some of the problems are far-reaching and require detailed consideration by specialists, for which public sponsorship and grant aid seem likely to be necessary.