A DESCRIPTIVE STUDY OF SUBJECT INDEXING AND ABSTRACTING IN INTERNATIONAL INDEX TO MUSIC PERIODICALS, RILM ABSTRACTS OF MUSIC LITERATURE, AND THE MUSIC INDEX ONLINE.(Muic Index, RILM Abstracts of Music Literature, International Index to Music Periodicals)

There are now three major online databases dedicated to current indexing of the literature of music: The Music Index (hereafter MI); RILM Abstracts of Music Literature (RILM); and International Index to Music Periodicals (IIMP). [1] The online version of MI provides citations, but not abstracts, for periodical literature in music going back to 1979, and is available both on CD-ROM and via the Internet. It is produced by Harmonie Park Press, which has published the print version of MI since 1949. The Web version was used in the present study. Chadwyck-Healey, after briefly marketing an earlier CD-ROM version of MI, launched its own music database, IIMP, in 1996. Now owned by Bell & Howell Information and Learning, IIMP, provides indexing and abstracts for music periodical literature since 1996 and is steadily adding citations (without abstracts) for pre-1996 literature. IIMP is available on CD-ROM and in two Web versions: a basic product with traditional citations and abstracts, and IIMP Full Text, which add s access to the full text of articles from over forty journals. Trial access to IIMP, Full Text was used for the present study, though no use was made of the full-text options.

RILM is a joint project of the International Musicological Society and the International Association of Music Libraries, Archives, and Documentation Centers. Since 1967, it has published abstracts of the whole range of scholarly literature in music, going beyond journal articles to include books, dissertations, catalogs, Festschriften, conference proceedings, and other formats. RILM exists in two electronic versions: it is available on CD-ROM and via the Web from National Information Services Corporation (NISC USA), and Web access is also provided through OCLC FirstSearch. The OCLC product was the primary means of access for the present study, though trial access to the NISC product was arranged for comparison of a few points.

These three databases differ greatly in the ways in which they index the literature. The present study identifies a group of articles that have been treated by all three databases and compares those treatments in both quantitative and qualitative terms. [2]

The quantitative section involved randomly selecting a large group of articles and comparing the number of subjects assigned, total words in the subjects, and unique words in the subjects in each of the three databases, and the number of words in the abstracts in RILM and IIMP. The qualitative aspect required looking more closely at a smaller number of items to compare the appropriateness of subjects chosen for indexing, the content and style of abstracts (for IIMP and RILM), and the accessibility of vocabulary in both.

A search of the library literature found no studies comparing both indexing and abstracting of the same articles by different databases. Though many articles can be found comparing various utilities' coverage of particular topics or journals, very few actually examine subject access at the individual record level. MaryEllen C. Sievert and Alison F. Verbeck [3] come closest with a document-level comparison of subject indexing of the literature of online searching in Library and Information Science Abstracts and the education database ERIC, but their study did not include comparison of abstracts.

QUANTITATIVE ANALYSIS

In order to carry out the comparisons envisioned for this study, it was necessary to generate a group of articles for which full treatment could be found in all three databases. This study was meant to be descriptive rather than predictive, and care was taken to make the sample generation process as objective as possible. The sample was limited to articles from 1996 in order to eliminate pre-1996 citation-only records in IIMP and post-1996 brief records added to RILM as part of its current citations project. A list was created of ninety-eight journals which each of the three databases claimed to index comprehensively. [4] Twenty-nine titles for which 1996 coverage could not be found in at least one of the three databases were dropped from this list, and the remaining sixty-nine journals were then divided into four classes: musicology; performance; theory/composition; and music education. In order to examine specific trends in the treatment of non-English material, non-English journals were isolated into a se parate class. [5]