Computer-Assisted Comparative Analysis with
 the MelAnaly Software Package


by J. Marshall Bevil

        MelAnaly is, as its name indicates, a system of melodic analysis. Specifically, it is designed for the comparative analysis of melodies existing either wholly or partially within oral-aural traditions, including folk tunes, popular songs that are recognized and learned mainly via recordings and broadcast, and melodies such as hymn tunes that are originally fashioned as written works but become known mainly through being performed and heard. I developed both the procedure and the supportive computer software in connection with my dissertation research (1980-1983). I have modified some details of the system since that time, but the basic principles remain unchanged.
        Three central premises underlie MelAnaly. The first is that the oral-aural processes governing the emergence, transmission, perception, assimilation, recognition, and recall of melodies are different from the forces that are at work when music is created, passed along, perceived, and performed entirely through the standard literate process.
The second premise is that not all sounding events in a melody are of equal significance and that a hierarchy of importance, based on objective criteria, needs to be established and observed consistently. Moreover, the hierarchy that prevails within the oral-aural continuum is more marked and also in other ways different from that which governs written and visual processes.
        The third premise is that, when one is considering melodies within the oral-aural sphere, both melodic contour, at a multiplicity of levels of structural-perceptual complexity, and short formulaic patterns (called cells in the MelAnaly system) that open and close melodic sections and act as mnemonic anchors are important. Further, they complement each other, and they work together to define a melody. Therefore any analytical procedure connected with the oral-aural process that ignores either overall melodic morphology or the most important stock figures will not produce consistently reliable or defensible results.
        As the premises above imply, there existed a need for an analytical procedure that was independent of the standard academic models that were developed for the study of music from within the Western written tradition. In addition, there was a clear need to reconcile a long-standing conflict between the oral-aural traditions specialists of the holistic school, who argued for consideration of overall melodic contour and relegated stock melodic patterns to a position of little or no importance, and the specialists of the connectionist school, who maintained that formulas and combinations of formulas were the essence of identity and argued that holistic views were too broad and vague. MelAnaly, both as I initially developed it and as I am continuing to refine it, was and is my answer to those needs. 
            MelAnaly, with its companion set of task-specific computer programs, was originally designed to assist in the comparative analysis of British and British-American folktunes, specifically the archetypal dual-strain melodies that are governed by a melodic morphological norm and the tonal constraints of the anhemitonic gamut. While the software package in its entirety remains best suited to melodies of that genre, the main parts of it (i.e., the primary cell and contour comparison and the comparative melodic graphing) have broader applications to other dual-strain melodies falling outside the folk species and have been used successfully outside the area of folksong scholarship since the late 1990s, most notably in forensic musicological investigations and in a study of music played during the Titanic disaster.
         Following melodic transcription, preliminary manual analysis, and preparation and inputting of raw data, the full course of computer-assisted comparisons involves two steps, the first of which is a scan of arrays that contain numeric codes for the pitch, duration, and stress factors of specified control and test variants at multiple structural-perceptual levels. The second step is the application of seventeen sets of programmed parameters defining the various natures and extents of melodic kinships. Results of the array scan are printed in a table that is illustrated by a set of computer-generated melodic contour graphs. Additional programs can be used to process the results of array scans in determining norms of melodic behavior within specified batches of variants, within delimited geographic areas and/or chronological periods, and between and among individual singers.
         The final step in the analysis is the manual review and evaluation of the computer-generated scan results. Concordance tabulations and percentages, melodic contour graphs, and pre-programmed conclusions are compared to both a visual analysis of the melodic transcriptions and an auditory comparison of the sounding musical examples. The computer is used to manage the large amount of data that is involved in each scan, to reduce substantially the likelihood of clerical errors, to expedite the comparative process greatly, and to assure the consistent application of criteria. The ultimate analysis is a human activity, not a mechanical one.
        Since the completion of my dissertation, I have published and presented several studies for which MelAnaly was used as an investigative tool, and I have published and presented information about both the analytical system and its software. In addition, copies of my dissertation are housed in the libraries of several major universities and other research institutions. Information concerning my publications and presentations, as well as the locations of some of the copies of my dissertation, can be found via the online bibliography of my works.   
        In its present form (reflected in Version 6.1 of the software), MelAnaly is of limited applicability to genres not possessing, at a minimum, a basic bipartite song structure and a primarily melodic center of interest. However, the analytical principle of emphasis on both main motifs and overall contour at a multiplicity of strata is potentially adaptable to a broader range of formal structures through alteration of array sizes and other modifications. Repertories within which harmonic support criteria are of critical significance eventually will be addressed through the addition of one or more array dimensions. Other modifications will include the ability to take into account upward and downward dislocations of contour through manual highlighting and shifting of relevant portions of the contour graphs. In the current version, this adjustment has to be made by by preparing separate data files from transcriptions. These and other expansions of the capabilities of the software, including possibly the online availability of the analytical software and data base via subscription, are projected for Version 7.0 and beyond.
         The forthcoming version of the computer software (7.0), whose completion is projected for 2010, will incorporate the most recent updates of the original analytical system and software of the early 1980s. At the time of its initial design and at the various times of its subsequent revisions, both the analytical system and the software paralleled similar efforts in the United States and elsewhere to meld computer technology with  reduction, or layer, analysis. In addition, the underlying analytical principles paralleled other efforts to merge the disciplines of comparative musical analysis and linguistics.


Dr. J. Marshall Bevil is a native of Houston, where he also currently lives. He is both a string music educator and a musicologist (M.Mus., University of North Texas, 1973; Ph.D., UNT, 1984) with specialization in the history of bowed string instruments, oral-aural musical transmission, and British music of the late nineteenth and early twentieth centuries. His doctoral dissertation has been published by University Microfilms, International ( U M I  No. 8423854, "Dissertation Services" ), and he has published post-doctoral studies in professional journals and presented papers in his areas of specialization at regional, national, and international academic convocations in both the United States and Great Britain. He also is the author of encyclopedia articles on John Avery Lomax, Alan Lomax, and Percy Aldridge Grainger; and he has published on the Internet. In addition to his pedagogic and academic pursuits, he is a performer on the Welsh crwth, a composer and arranger, and a forensic musicological consultant and expert witness  ( links:  1    2  ) in copyright and intellectual property misappropriation disputes.  





"PARADIGM"   ABSTRACT  (American Musicological Society, New Orleans, 1987)

TEXT OF RESPONSE (Society for Ethnomusicology, Milwaukee, 1994)

Titanic  Orchestra (music performed by, near climax of disaster):  ABSTRACT           FULL ESSAY

"SCALE" ABSTRACT  (College Music Symposium, 1986)


. - - - - - - - - - - END OF DOCUMENT - - - - - - - - - -