Marcello Sorce Keller

THE PROBLEM OF CLASSIFICATION
IN FOLKSONG RESEARCH:
A SHORT HISTORY


Folklore, XCV (1984)
no. 1, 100- 104



The problem of folksong classification has been extensively dealt with in the ethnomusicological literature since the beginnings of the discipline in the late nineteenth century. [1] That no ideal solution has been found so far is shown by the constant appearance of new endeavours throughout our century. Today, even though no major breakthrough is foreseen for the immediate future, the use of the computer promises at least the possibility of more efficient methods of handling large amounts of information.
Obviously, one of the requirements that a classificatory system should meet is that of making it easy to locate a given song within a large collection. A second, and equally important requirement, is that of underscoring similarities and relationships among tunes, thereby helping in bringing together groups of variants or, to put it otherwise, of what folksong scholars define as tune families. [2] For the purpose of comparative analysis, the investigator should have at his disposal, in easily accessible form, the entire complex of a given melody-group, including every known variant and derived melodic form. These variants should be arranged so as to permit study that will explain how each one of them is related to all the others. The question, therefore, is to discover a system whereby a large number of melodies are indexed in such a fashion that they can be readily compared and, at the same time, distinguished from each other on the basis of what individual features they may show. Such goals are not easy to achieve. It is for this reason that over the years many different methodologies have been devised, adopted, modified and improved. No single one is satisfying in all respects, since both advantages and disadvantages are inevitably intrinsic to them all.
The first attempt to create a melodic index was made, at the best of our present knowledge, by the Benedictine monk Johannes Werlin in 1646. He catalogued several hundred tunes on the basis of their initial three tones, according to the principles of solmization. His was, however, an isolated enterprise which had no follow-up for roughly two hundred years. In more recent times, another attempt to index melodies was made by Johannes Zahn in his «Die Melodien des Evangelischen Kirchenliedes» (1889-1893). Here the metric features of the tunes are the basis for classification (duple metre, triple metre, either simple or compound). However, in the study of musical folklore, that is, in the study of music living in oral tradition, Zahn's approach is of limited utility. Tunes that are similar in their melodic contour may appear unrelated to each other if classified only according to their metrical and rhythmical pattern.
Looking back at the history of folksong scholarship one can see that the criteria used for classification throughout the whole nineteenth century were solely based on the literary content of the text. [3] This is the case with Child's famous collection [4] in which the same principles are used (form of the text and its narrative content) that had been used by the Danish folklorist Svend Grundtvig, who preceded Child in publishing a collection of folk ballads. It was customary for folklorists of the time to study folksong as poetry, overlooking its musical component. Such an approach, today generally dismissed as inadequate, gave way to others in which music entered the picture and dictated the taxonomies and tabulations based on its sound-durational components.
Toward the end of the nineteenth century, however, with the gradual emerging of musicology as a discipline and musicians taking over from folklorists the task of studying folksong, the necessity of thematic indexing became clearly apparent. The question of how such a problem should be solved was publicly raised by D. F. Schenerleer in the «Zeitschrift der Internationalen Musikgesellschaft» (1901). An answer to this question appeared in two articles by Oswald Koller and Ilmari Krohn, published in that same journal. [5] The latter has been throughout his life the most authoritative advocate of the so-called lexicographic indexing system, although several others such as the above-mentioned Koller and also G. Brandsch, A. Launis, W. Heinitz, L. R. Lewis, H. Mersmann and finally Bartók and Koddly have used and improved it. Ilmari Krohn (1867-1960), however, a Finnish musicologist belonging to a family of scholars in the fields of music and folklore, was the first to show the practical use of lexicographic indexing, applying it to the body of Finnish material he was most familiar with. Many other scholars who used his approach to indexing in later years had to adapt it to the material they were examining.

Some of the adaptations of the lexicographic system propounded by these scholars take into almost exclusive consideration the musical aspect of folksong. In this respect their analysis may be considered just as one-sided as those that had in earlier times, focused on the literary content alone. We might view this as a kind of rebound effect.
Most of these systems were tried and developed for those musical areas in which more research had been done and in which, consequently, the need for ordering that larger repertoire and for locating specific items within it, was more-deeply felt. These areas have traditionally been the music of the Northern Europe (Finland and Scandinavia), Germany and the German-speaking countries, Hungary and Eastern Europe and, last but not least, Britain and the U.S.A.
As scholars began to realise that European folk music, especially the Anglo-American, showed modal characteristics, [6] modality itself came to serve as a criterion for classification. Therefore, an attempt was made to use the nomenclature of medieval musical theory for classificatory purposes. This was the case with Cecil J. Sharp, [7] and, in more recent times, with Bertrand H. Bronson. Later, recognising the fundamentally pentatonic nature of the material he collected in the United States, Sharp based his ordering of songs on a system of five pentatonic modes whose gapped scales could in theory be filled in by one or two notes, making them in turn either hexatonic or heptatonic. [8] Bertrand H. Bronson was one of the last scholars to use a modal classification system, when he published the volumes of «The Traditional Tunes of the Child Ballads» between 1959 and 1972. [9] The limitations of his approach, however, had become clearly apparent; thus it dropped out of fashion in the scholarly community. Modal classification is in fact misleading in suggesting, implicitly, some sort of genetic relationship between the musical tradition of the Church and folksong, and at the same time failing to describe properly a number of tunes that, either because of their unstable intervals or because of the gaps in their scale structure, could not be clearly ascribed to any mode at all. It was probably because of such complex patterns of modal relationships among English folksongs that Percy A. Grainger [10] suggested that all the songs of this tradition could be thought of as deriving from one single, loosely knit, folksong scale in which the third, seventh, and more rarely, the sixth step are unstable and appear now flattened, now sharpened. This was an implicit invitation to do away with the modal terminology, which, having developed within a literate musical tradition, cannot be properly used to describe a musical repertoire which is handed down by word of mouth and which, therefore, is not the expression of any explicit, verbalised musical theory. To Bertrand H. Bronson, however, we owe the establishment of a computerised approach to sorting out tunes and their variants. His approach was the father, in a way, of more recent and therefore sophisticated computer-based svstems like, for instance, Benjamin Suchoff's approach to the analysis of the Hungarian folksongs collected by Bartók. [11] Bronson's approach emphasises the importance of stressed notes, an importance recognised and used for classificatory purposes by other scholars such as Oswald Koller, Sirvart Poladian, and in more recent times by Alan Keagan.
The lexicographic method, conceived by Koller and Krohn, was later taken over by Bela Bartók and to a greater extent by Zoltan Kodàly, after being adapted to a repertoire of a profoundly different nature such as the Hungarian one. The lexicographic classification, to put it briefly, consists in ordering a body of tunes on the basis of the musical intervals they are made of, as words in a dictionary or lexicon are ordered on the basis of the alphabet letters of which they are constituted. Moreover, since in Anglo-Saxon countries musical notes are named after letters of the alphabet, the lexicographic order is arrived at quite naturally (for more details of the method, which can be at times quite complex, see for instance Stephen Erdély, «Methods and Principles of Hungarian Ethnomusicology», Bloomington, Indiana University Press, 1965).
Systems of classification were also developed by archives according to the nature an scope of the collections they were applied to. The best known example in this respect is the Deutsches Volksliedarchiv in Freiburg in Breisgau, [12] whose growth and development is to a large degree a result of the activity of John Meier, In this archive songs are classified by genre (e.g. ballad, song, functional song, etc.) and by melodic type . Thus melodic contour is used in grouping together on cards songs that have identical or similar beginnings. Other systems of grouping variants have been formulated for other repertoires than the German. The system of Sigurd B. Hustvedt, for instance, akin to that used at the Deutsches Volksliedarchiv, relies on the melodic content of the first and last line of each song, perhaps excessively so, since no analysis of the overall structure of the tune is attempted. Consequently melodic events occurring in the middle section of the song are not taken into account. Hustvedt applied this method to the Child ballad tunes. [13]
A modified version of the lexicographic approach was also used, as previously mentioned, by Bartók and Koddly. Their adaptation consists mainly in adding to the picture the following factors: 1. cadential structure, indicating the line-ending notes (labelled with Roman and Arabic numerals); 2. the number of syllables per line, 3. the range of melodies. The songs are also arranged in ascending order, from lower cadential notes to higher ones, from small numbers to larger numbers of syllables and, finally, from narrow range to broader range.
A more profound adaptation and improvement on the lexicographic method was devised by Bela Bartók in his later years. He named it the 'grammatical method' [14] since it tendsto point out some formal and structural characteristics of the repertoire being dealt with. These were not easy to single out in the lexicographic method. Bartók makes use of such characteristics later in order to break down the body of songs under consideration into style groups which he links to hypotheses about their genetic and historical development. The grammatical method was formulated while dealing with Serbo-Croatian material and was used by Bartók ever after, whereas Kodály always favoured the lexicographic approach. The grammatical method, besides considering the melodic aspects of songs already taken considered by the lexicographic system, introduces new elements, namely: 1. section structure; 2. metric structure (metric units or number of syllables per line); 3. rhythmic character of the lines (Parlando rubato or tempo giusto); 4. cadential structure of the strophe; 5. range 6. scale; 7. melodic content of the sections. Both the lexicographic and the grammatical methods have been employed by other Hungarian scholars besides Bartók and Kodály - particularly by Laslo Vikár and Lajos Vargyas.
In Western Europe folksong scholarship has developed so far in a much less systematic fashion than it has in the Eastern countries of the continent. The problem of classification, consequently, was felt as a less pressing one, since the task of collecting and documenting the existing traditions remained a priority. However, in a number of cases, especially in the Anglo-American area (the best studied in the Western tradition), there have been attempts to adapt the Bartók system to the nature of local national repertoires. This was the case of George Herzog [15] for instance, who studied folk tunes from the Mississippi area (using a collection by Arthur P. Hudson), and for Jan P. Schinhan, who edited the songs and ballads of the «Frank C. Brown Collection of North Carolina Folklore». [16]
In recent years, as has already been mentioned, several efforts have been made to use the computer as an aid in classifying folksongs. Scholars currently interested in this approach are Benjamin Suchoff, curator of the Bartók Archive in New York, who has been working on its Bartók material, Wolfram Steinbeck, who has been working on German material, Alicia Elschekova who has been dealing with some general theoretical problems and, of course, many others.
The history of folksong classification seems to prove the great difficulty of devising a methodology that can be applicable to several repertoires on a cross-cultural basis. So far every scholar has been improving on the work of his predecessors, with a particular eye to the nature of the material he is studying and to the specific scope of his investigation. It appears, therefore, that an all-purpose scheme of classification is impossible, just as, in another sense, an all-purpose folksong collection is usually ineffective. Each needs to be aimed at a particular goal. Otherwise, even the modest intent of merely describing a tradition may not be effectively achieved. Significant and insignificant features have to be distinguished before a description can be undertaken, so that when a specific problem is focused on, one has an empirical basis from which to determine such a differentiation.
It is hard to say whether computers will help the scholar in achieving what he has been unable to achieve so far with non-technological means. It is perhaps conceivable, however, that a computer theoretically capable of taking into consideration an unlimited number of parameters and of finding correlations between each and every one of them may be able to detect patterns that would otherwise escape human attention. Such patterns might then be used in bringing together variants in an objective way. [17] Once that is done, scholars would have a much better chance of studying how musical traditions interact and influence each other and, which is putting it in another fashion, to what extent they are similar and therefore compatible.

NOTES


1. At that time the discipline was referred to as 'vergleichende Musikwissenschaft' (comparative musicology).

2. About the concept of 'tune family' cf. Samuel P. Bayard, 'Prolegomena to the Study ofthe Principal Melodic Families of the British-American Folk Song', «Journal of American Folklore», LXIII (1950), pp. 1-44.

3. The study of folksong developed first within the discipline of folklore, a discipline which, in its beginnings, did not arouse the interest of musical scholars. It is worth pointing out in this connection that musicology itself owed its beginnings more to the contributions of psychologists, physicists, linguists, and folklorists than to those of professional musicians.

4. «The English and Scottish Popular Ballads», 1882-1898 (reprint, New York, Dover, 1965).

5. Oswald Koller, 'Die beste Methode, Volks- und volksmassige Lieder nach ihrer melodischen Beschaffenheit lexikalisch zu ordnen,' «Sammelbände der Internationalen Musik-Geselischaft», IV (1902-3), No. 4, pp. 1-15; Ilmari Krohn, 'Welches ist die beste Methode, um Volks- und volksmassige Lieder nach ihrer melodischer Beschaffenheit lexikalisch zu ordnen,' ibidem, pp. 643-660.

6. I. e. music not showing the melodic-harmonic characteristics of modern tonality (a tonal centre and a leading note, as found in that of the so-called 'common practice period,' from the late sixteenth century to the beginning of the twentieth).

7. Cf. «English Folk songs: Some Conclusions», London, 1907.

8. Cf. «English Folksongs from the Southern Appalachians», London, 1932.

9. Princeton University Press, 1959-72.

10. Percy A. Grainger (1882-1961), Australian composer and folksong collector. Collected mostly British but also some Danish material, in collaboration with Evald T. Christensen.

11. Cf. Suchoffs recent edition of Béla Bartók, «The Hungarian Folk Song», Albany, State University of New York Press, 1981.

12. The Archive was founded in Freiburg im Breisgau in 1914. Its card catalogue consists of five sections: a catalogue of melodies, texts, catchwords, authors, and of sources. With the exception of the First, they are all arranged alphabetically.

13. Sigurd B. Hustvedt, 'A Melodic Index of 'Child's Ballade Tunes', in «Language and Literature», Publications of the University of California at Los Angeles, Berkeley, 1936, 1, No. 2, pp. 51-78.

14. Béla Bartók and Albert B. Lord, «Serbo-Croatian Folk Songs», New York, Columbia Press, 1951.

15. Cf. Arthur P. Hudson, «Folk Tunes from the Mississipi». Edited by George Herzog, National Service Bureau Publication No. 2, December, 1937 (reprint Da Capo Press, 1977).

16, Cf. Vols. IV and V of the «Frank C. Brown Collection of North Carolina Folklore», Durham, North Carolina, Duke University Press, 1957-1962; these two volumes are under the musical editorship of Jan P. Schinhan.

17. The reader interested in the use of computers in folksong research and classification will find useful information in the following publications: Edmund Bowles (ed.), «Computers in the Humanistic Research», Prentice-Hall, Englewood Cliffs, New Jersey, 1967; Harrv B. Lincoln, 'The Current State of Music Research and the Computer,' «Computers and the Humanities», 1971 (1971), No. 1; Harry B. Lincoln, 'Use of the Computer in Music Research: A Short Report on Accomplishments, Limitations, and Future Needs,' ibidem, VII (1974), No. 4; Benjamin Suchoff, 'Some Problems in Computer-Oriented Ethnomusicology,' «Ethnomusicology», XIII (1969), No. 3; Benjamin Suchoff, 'The Computer and Bartók Research in America', «Journal of Research in Music Education», XIX (1971), No. 1; Evan Stein, 'The Use of Computers in Folklore and Folk Music: a Preliminary Bibliography', «Library of Congress, Archive of Folk Song», Washington, D.C., 1979.