Some Song-Style Clusters--A Preliminary Study

Abstract

ince February, 1963, a team of investigators, including myself, and supervised by Alan Lomax and Dr. Conrad Arensberg, have been working at Columbia University under a grant from the National Institute of Mental Health. Using, as a point of departure, Cantometrics,2 a rating system for song style developed by Lomax and myself in 1961, the project has incorporated data from the social sciences, linguistics and musicology in an attempt to examine the musical behavior of man from the broadest point of view. The results of this study will be published next year. This paper deals with one method used in this project. In order to illustrate the method, let us draw upon an hypothesis from Alan Lomax' article of 1959 entitled Folksong Style (1959:937). One of the ten style areas which he outlines is called Modern European. Illustrating some of the salient characteristics of this style, Lomax proposed a distribution through most of Europe, listing several areas specifically. Early in the Cantometrics Project, statistics were made available which indicated that this hypothesis was indeed accurate. A statistical summary of the manner in which our Western European sample behaved in relation to each Cantometric parameter indicated that the song style qualities which Lomax had named were characteristic of Western Europe as a whole. This summary was, however, capable only of telling us the degree to which any given song style trait occurred in any given area. It did not tell us whether or not a certain cluster of traits was actually present in any individual song. We might learn, for example, that sixty percent of the songs from Western Europe were solo songs and that sixty percent were unaccompanied. This did not necessarily mean that sixty percent of the songs were both solo and unaccompanied. This summary had the disadvantage, moreover, of failing to indicate whether or not songs of this type occurred outside of Western Europe, nor did it tell us exactly where in Western Europe they could be found. With the aid of Norman Berkowitz, our computor programmer, I was able to work out a very simple approach to this problem. The first steps are informal. I leafed through individual coding sheets from Western Europe, looking for significant patterns. When I found several traits that seemed to go together regularly I searched for those which linked most consistently with one another. This search was simplified enormously by the area summary which we already had. The criteria finally chosen were: the use of many words with little repetition and few nonsense syllables; little to no embellishment; little to no rhythmic freedom; and simple strophic form.3 These criteria were fed into a program which is designed to print the culture name and index number of each song which has all four of these characteristics. The computor searched a sample of 1700 songs from all the principle areas in the world. The area distribution of the 161 positive cases is as follows: