The purpose of this paper is to summarize the varied structural characteristics which may be present in retrieval languages.
The languages serve varied purposes in information systems, and a number of these are identified. The relations between structure and function are discussed and suggestions made as to the most suitable structures needed for various purposes.
A quantitative approach has been developed: a simple measure is the number of separate terms in a retrieval language, but this has to be related to the scope of its subject field. Some ratio of terms to items in the field seems a more suitable measure of the average specificity of the terms. Other aspects can be quantified – for example, the average number of links in hierarchical chains, or the average number of cross‐references in a thesaurus.
All the approaches to the analysis of retrieval language reported in this paper are of continuing value. Some practical studies of computer information systems undertaken by Aslib Research Department have suggested a further approach.
