LIBRARY OF WELLES LEY COLLEGE
PURCHASED FROM
Bimting F-und
Serbo-Croatian Folk Songs
NUMBER 7 IN THE COLUMBIA UNIVERSITY STUDIES IN MUSICOLOGY
Serbo-Croatian
FOLK SONGS
Texts and Transcriptions of SEVENTY-FIVE FOLK SONGS
from the MiLMAN Parry Collection
and a Morphology of SERBO-CROATIAN FOLK MELODIES
by BELA BARTOK ^W ALBERT B. LORD
with a Foreword by George Herzog
COLUMBIA UNIVERSITY PRESS
NEW YORK -1951
2'm'y } '
^3
COPYRIGHT 1951, COLUMBIA UNIVERSITY PRESS, NEW YORK
PUBLISHED IN GREAT BRITAIN, CANADA, AND INDIA BY GEOFFREY CUMBERLEGE
OXFORD UNIVERSITY PRESS, LONDON, TORONTO, AND BOMBAY
MANUFACTURED IN THE UNITED STATES OF AMERICA
TO THE MEMORY OF MILMAN PARRY
Digitized by the Internet Archive
in 2011 with funding from
Boston Library Consortium IVIember Libraries
http://www.archive.org/details/serbocroatianfolOObart
"BOARD OF EDITORS
OTTO KINKELDEY
PROFESSOR EMERITUS OF MUSICOLOGY, CORNELL UNIVERSITY Honorary Chairman
PAUL H. LANG
PROFESSOR OF MUSICOLOGY, COLUMBIA UNIVERSITY Executive Chairman
ARCHIBALD T. DAVISON
PROFESSOR OF CHORAL MUSIC, HARVARD UNIVERSITY
GEORGE S. DICKINSON
PROFESSOR OF MUSIC AND MUSIC LIBRARIAN, VASSAR COLLEGE
CARLETON SPRAGUE SMITH
CHIEF OF THE MUSIC DIVISION, NEW YORK PUBLIC LIBRARY
Foreword
MiLMAN Parry, Homeric scholar at Harvard University, had the inspired thought that if we wanted to form a picture of the genesis of the great Homeric chants and how they were per- formed, we should observe the life of folksong where it has best survived to the present day, on the Balkan peninsula. The heroic epic songs of Yugoslavia, the so-called "men's songs," come nearest in that region to the type of tradition which was probably the foundation of the Homeric epics. So Professor Parry made in the thirties prolonged collecting trips to Yugoslavia and with the assistance of Dr. Albert B. Lord gathered a vast collection of songs, especially in the jagged mountains of Bosnia, Hercegovina, Montenegro, and South Serbia. In addition to the epics, many "women's songs" — lyrical songs — were collected; these are less outstanding in the historical and literary perspective, but their melodies have a wider musical appeal than the chant tunes of the epic poems. In order to make a lasting record of the songs in the form in which they are actually performed and to make their study possible also from the musical angle, the expedition recorded a large number of songs on disks. But soon after his return to this country, Mr. Parry died, in a tragic accident. Research on the music of the collection was for a long time hampered by the unavailability of a real connoisseur of the folk music of eastern Europe. This need was remedied when the late Bela Bartok came to this country. Bela Bartok is known to the musical world as one of the foremost com- posers of our time, and one whose creative inspiration was deeply rooted in folk music. In his earlier works, especially, he often uses a vocabulary inspired by the clear spirit of folk melodies, while later the folk elements became more and more integrated and absorbed in his own personal idiom, which they helped to expand. His settings of Hungarian, Slovak, and Rumanian tunes and dances, produced over a long span of years, reflect an intimate familiarity with this material. Time and again he came back to the challenging problem of translating it for the musical audience of the city, and he evolved a variety of solutions. But it is less well known that Mr. Bartok was also a devoted student of folksong, indeed, an inter-
[ ixl
Foreword
national authority in this field. Much of the folk music incorporated into his creative work was entirely unknown until he discovered it, in the course of laborious collecting trips to villages off the beaten track, to places that were "behind the back of God," to use a Hungarian phrase. The intensive cultural and national renaissance in Hungary at the onset of the twentieth century awakened a strong interest in folk life and folk- lore. The patient search conducted by Mr. Bartok and his colleague, the composer Zoltan Kodaly, soon revealed that what had been accepted as Hungarian national music, through the effusive Rhapsodies of Liszt and Brahms which followed in the wake of the sparkling "alia Ungharese" movements by Haydn and others, had little to do with the folk them- selves. It was merely an inflated elaboration of the popular entertainment music of the town, propagated by the Hungarian gypsies. But the country- side still possessed an entirely different body of old melodies, simple and sturdy, practically unknown in the Hungarian cities. It was a situation [much like that in this country when the ballads of New England and of the Southern Mountains had to be "discovered" for a public nurtured on Stephen Foster.
Under the influence of the Hungarian peasant tunes, Bartok and Kodaly, in the first decade of the century, began to develop a new national school of composition. But the regional confines were soon transcended, especially by Bartok. His scholarly interests also expanded, to embrace the folk music of the nationalities surrounding the Hungarian island. On his last recording expeditions the search for peasant music took him as far as Turkey and Algeria. Thus, by the time he came to this country, shortly before the war, the experience and the detailed knowledge of Eastern European folk music which he had amassed was equaled by no one. It was most fortunate, then, that through the warm interest of the Music Department at Columbia University arrangements could be made with Harvard University for the musical study of the Parry Collection to be delegated to him. On his part, it gave him deep satisfaction that while he was finding his roots in a new country, which he knew chiefly as a modern technological civilization, he could once more immerse himself in a musical dialect from one of those regions of Europe where old rural culture has not yet fully disappeared. This book, appearing posthumously, is one of the results of this work.
Mr. Bartok's contributions to our knowledge and understanding of folk
[xl
Foreword
music are the fruits of an indefatigable activity and industry. In the days of the cylinder phonograph he preserved, on records or by writing them down by ear, thousands of folk melodies. To their collating and study he devoted a scholarship so painstaking and minute that it is not easy to fathom how he could have found the time and the energy for it, in the midst of a fully creative life as composer, pianist, and teacher. The monu- mental corpus of Hungarian folk music which he prepared with Mr. Kodaly was to have been published by the Hungarian Academy of Sci- ences, when the war intervened. He was the author of the monograph Hungarian Folk Music, a model of methods of analysis, and of basic investigations on Rumanian and Slovak folk music, besides many shorter studies and theoretical articles. In a territory charged with old political, national, and cultural aspirations and tensions, these writings are dis- tinguished by an objective and international outlook — one important essay traces the cross-influences between Hungarian folk music and that of the neighboring peoples. He liked to stress the fact that there are no true folksongs in this region that express national prejudices or antag- onisms of the peasant groups, and he felt that such attitudes were kept alive artificially by ofiicial cliques and school-taught songs. He took equal pride in being a citizen of the world, taking an interest in music the world over, and in being a member of a particular national culture, for the music of which he had the special attachment of one born in a small town, not too far from the "soil."
A combination of creative musical genius of the very first rank with persevering and detailed scientific activity is rare. Bartok looked upon the tasks and responsibilities of the collector and of the composer as entirely distinct. Those of the collector, he felt, were to gather and present the material exactly and faithfully, without any patronizing emendations or collated versions of the sort which were current in Europe and are still popular in this country. In his settings of folk melodies he often submitted also the notations of the tunes as they were recorded from the folk singers, so that the reader could compare the two and see the material in its pris- tine form. Characteristically, perhaps, he gave no opus numbers to those compositions which were based chiefly on folk tunes. At the same time Mr. Bartok manifested some of the same traits both as composer and as scholar: integrity of purpose, a complete lack of capacity for com- promise, subordination of the subjective element to what he felt were the
[xi]
Foreword
dictates of his material, and a careful workmanship with regard to details which, no matter how large the framework, was so exacting as to result almost in self-abnegation. He made it his task to acquire an intimate familiarity with the general folklore of the national group in which he worked and sufficient knowledge of the language for the arduous task of recording and studying thousands of song texts.
It is, perhaps, characteristic of his mode of workmanship that in order to provide the general observations in this book of seventy-odd melodies with a broader basis, Mr. Bartok made a systematic analysis and a melodic index of thousands of Yugoslav melodies, practically all that have appeared in print so far. Also, in his introductory essay he has presented a unique and searching discussion of the difficult problems of notation, classification, and analysis in the study of folk m^usic. In his methods, Bartok was influenced by the so-called Finnish school of folklorists, who attempted to cover their national folklore systematically and in great detail. The large amount of material which they amassed called for devel- oping effective techniques for analyzing, classifying, and indexing tunes, and these have been used fruitfully in the study of Eastern European folk music. Thus, detailed consideration is given in this book to matters such as the metric structure of the text and its relation to the tune, the struc- ture and tonal range of the melody, and various other features. Musical rhythm is treated more briefly, since its patterns in Yugoslav folk music are free, very rich, and even diffuse; the basic structures are overlaid by a free, rubato performance and luxuriant ornamentation. But the details of this ornamentation and of the plastic fluctuations of time and of intona- tion were caught and notated by the author, as in his other transcriptions of folk melodies, with an exactitude practically photographic. This was possible only for a musician with an unusually discriminating and well- trained ear.
Doubts have been raised about the value and the reliability of this type of notation, since no musical notation can supplant for the reader the actual auditory experience and many small details of delivery vary in a truly alive folk tradition from performance to performance, even by the same singer. But a simplified notation, which is the alternative, robs a folk melody of its true flavor; it unavoidably introduces subjective judg- ments as to what is important and what may be dispensed with, so that at least for scientific purposes only a detailed and faithful notation is
[xii]
Foreword
satisfactory. It can be combined, however, with a simplified picture of the melody, as is done here by means of the skeleton notations. The more exact transcriptions expose the construction of a melody in microscopic detail, but to follow it along places a demand on the reader, whom the author envisaged as a serious student. Much of the discussion, too, couched in a formulistic style, calls for work before pleasure. It may be kept in mind, however, that this is the first book written by the author in English; the publishers felt that his personal style and organization should not be altered appreciably after he could no longer authorize changes. Many of the findings in this study are of interest primarily to the specialist, but it is of more general interest in that it is the first sys- tematic discussion of Yugoslav folk music. (The standard collection of Yugoslav folk melodies published in the eighteen eighties by F. X. Kuhac has many of the faults of the early folksong collections, and excellent recent compilations, such as those of V. R. Djordjevic, contain little musical analysis.) The student of folksong and folk music will find this method of analysis full of suggestive leads, and the frequent comparative remarks on folk music are very valuable. The general reader will be rewarded by the interesting melodies and poems.
South Slavic folk music has an especial appeal. This may well be due to the contrast between the essential simplicity of its basic materials and the pulsing quality of life achieved through an abundance of expressive devices, including the ornamentation. This ornate treatment is partly due to an old, general European mode of folk singing; partly, no doubt, to the various Oriental influences which impinged on the Balkan peninsula, the Turkish domination being only the latest. This mixture of simplicity and of artistic elaboration pervades also the poetry, as may be seen in the decorative arrangements and repetitive patterns in the second part of the book. While the latter was planned so as to be subordinated to the musical study — detailed treatments of the literary side of this tradition will be published separately — yet we are given here a full and valuable background for the musical analysis; a picture of the recording expedition at work and of the singers, a collection of annotated authentic texts which will be welcomed by the specialist, and translations which are as pleasant to read as they are unpretentious and faithful to the original songs. Studies devoted to the heroic epic poetry in the Parry Collection of Harvard University and its music, on which Mr. Bartok
[ xiii ]
Foreword
worked intensively, will appear in the future. Meanwhile, this book repre- sents an important addition to the study of South Slavic folklore and to our knowledge of the rural culture of southeastern Europe ; it is a modest memorial to a great artist and a sympathetic scholar.
Indiana University george herzog
[ xiv ]
Preface
THE MAIN OBJECT of Professor Milman Parry's journeys to Yugo- slavia in 1934-35 was to study, to collect, and to record as many as possible of the Serbo-Croatian heroic poems still in existence. He recorded about 350 heroic poems, collected from 90 different singers in 23 villages, on some 2,200 double-faced disks. His interest was not limited to this type of poem only. He recorded also: approximately 205 Serbo-Croatian so-called "women's songs," on about 210 double-faced disks; about 14 in a Macedonian-Bulgarian dialect, on 9 double-faced disks ; about 30 Turkish and 1 1 Albanian songs, on about 40 double-faced disks; and 16 instrumental pieces on 8 double-faced disks. The Parry Collection, including the amazingly large number of heroic poems, towers high above all other recordings of Serbo-Croatian folk music in the world, being larger than aU the rest combined.^
This publication is based on part of the Serbo-Croatian "women's songs," which I was privileged to transcribe and to study in 1941 and 1942, under the auspices of Columbia University and with the permission of Harvard University, the owner of the Parry Collection.
Two hundred songs are not many to represent an area as large as the Serbo-Croatian linguistic territory. However, their importance is enhanced by the fact that every song, regardless of its length, has been recorded in full — a procedure not very often used in recordings of Eastern European folk music, for various reasons (such as the high costs and lack of time) . From this point of view the Parry Collection ranks above all European collections known to me; these generally do not contain more than the first three or four stanzas of the longer folk poems.
The material published in this book is based upon transcriptions of recordings of Serbo-Croatian folk music and is, in this respect, the first publication of its kind.^
^ According to the statements in Folklore musical, Institut International de Cooperation Intellectuelle, Paris, 1939, pp. 205-9, these consist of: 466 (probably single-faced) disks, 348 cylinders, and 150 other enregistrements (we do not know what kind of enregistrements they are, since disks and cyhnders are separately mentioned) scattered in several European coun- tries. Hereafter this pubHcation will be cited as Folk. mus.
^ According to Folk, mus., pp. 210 ff., where no clear mention is made of material published on basis the of records.
[xv]
Preface
The records are excellent, on the whole. The singers, with a few excep- tions, were very well chosen. Their performance, especially concerning the timbre of voice, tone production, and ornamentation, is absolutely natural and exactly what we should expect from peasant singers whose performance is not yet marred by urban influences.
I feel privileged to have had the opportunity to make this study. It is a part of a larger investigation which was undertaken under the auspices of Columbia University in 1941-43 and was based on the extensive Parry Collection of unique materials on Yugoslav folk song and folk music deposited at Harvard University. I wish to express my appreciation to all those who helped me in this research. Harvard University gave its full co-operation by making the records and texts available, and by giving its permission to publish the results in the present book. Mr. Albert B. Lord, who participated in the late Professor Parry's collecting trips in Yugoslavia and after Parry's death carried on some of his research, gave assistance in many ways, and contributed the section on the texts, includ- ing an introduction and texts and translations of the poems. The Archives of Primitive Music, Department of Anthropology, Columbia University, furnished duplicates of the original recordings, so that the latter were saved from the damage due to the wearing down attendant on the labo- rious process of transcription into musical notation. I am obliged to Dr. George Herzog, of the Archives of Primitive Music, for his helpful assist- ance, including going over the manuscript. Credit is due to Mr. Wolfgang Weissleder for the great care he exercised in producing the excellent music copy. The Columbia University Press co-operated in every way in its task of producing a publication somewhat burdened with technical detail. Above all, however, I am indebted to Professor Douglas H. Moore, the Department of Music, Columbia University, and the Ditson Fund, whose active interest made the study and its publication possible.
New York bela bartok
February, 1943
[ xvi ]
Contents
FOREWORD, by George Herzog ix
PREFACE, by Bela Bartok xv
PART ONE : BY BELA BARTOK
INTRODUCTION TO PART ONE 3
MORPHOLOGY OF THE SERBO-CROATIAN VOCAL FOLK
MELODIES 21
REGIONS, VILLAGES, SINGERS, AND OTHER DATA 88
CHARACTERISTICS OF SOME OF THE SINGERS 89
EXPLANATION OF THE SIGNS USED IN THE MUSIC NOTATIONS 89
WORKS CITED IN PART ONE 93
MUSIC EXAMPLES 95
NOTES TO THE MUSIC EXAMPLES 231
PART two: BY ALBERT B. LORD
INTRODUCTION TO PART TWO 247
TEXTS AND TRANSLATIONS 257
WORKS CITED IN PART TWO 426
INDEX OF FIRST LINES 427
[ xvii ]
Part One
by BELA BARTOK
Introduction to Part One
METHOD OF TRANSCRIPTION
THE TRANSCRIPTION of recordings of folk music should be as true as possible. It should be realized, however, that an absolutely true notation of music (as well as of spoken words) is impossible because of the lack of adequate signs in our current systems of notation. This applies even more to the notation of folk music. The only really true notations are the sound-tracks on the record itself. These, of course, could be magnified, photographed, and printed instead of, or with, the usual notation. But this complicated procedure would not be of much use, in view of the all- too-complicated nature of the curves in the tracks. The human mind would not be able to translate the visual signs into tones. It must have as visual impressions conventional symbols of drastic sim- plicity in order to be able to study and to categorize sound phenomena. These symbols are what we call "notation" of music. When applying them to the transcription of folk music, we may add supplementary diacritical signs, in smaller or larger numbers, devised for our special purposes, in order to represent certain phenomena which occur in and are characteristic of folk music.
In spite of these additional signs, the current notation, when used to transcribe folk music, has intrinsic limitations. These limitations, how- ever, can be overcome to a certain degree, according to our purpose and to our well-weighed choice. Our choice will take into consideration the perceptive abilities of the human mind and their limits.
In transcribing folk music only two dimensions can be assumed : pitch (as the vertical one) and the rhythm (as the horizontal one). The con- sideration of the third dimension, having relation to intensity and color of sound may well be discarded. We have no adequate signs for marking intensity (except for the well-known but too-general signs of dynamics) and no signs at all for tone color (timbre) . Fortunately, change of inten- sity is no important factor in folk music (at least that of Eastern Europe) , the performer's intention being to achieve evenness. And if the perform-
[3]
Introduction to Part One
ance is nevertheless anything but even, this results from purely mechan- ical factors. Notes sung in high pitch sound louder; those in lower pitch, softer. Very rarely, indeed, do we hear single notes accentuated inten- tionally by the performer (and then more frequently in instrumental music than in vocal melodies) or groups of notes produced purposely with changes of dynamics. In such cases, of course, the conventional signs for dynamics must be used. Creation of new signs for color would be not only too complicated but also probably useless; the reader could never catch the right idea of the color, however elaborate the signs or description used. In this regard, in addition to a few descriptive words, all that can be done is to refer the reader to the record itself. So we must impose exactness only with regard to pitch and rhythm.
Eastern European folk music is generally based on the diatonic system of our art music. Exceptions are few; rather obvious among them is the music of the Serbo-Croatian heroic poems. The reader must bear in mind, however, that the intonation (that is, placement of pitch) of the respective degrees (whether altered by accidentals or not) is much less exact than in art music.^ Nevertheless, these deviations, since they show a certain system and are subconsciously intentional, must not be considered faulty, off-pitch singing. This is the essential difference between the accidental off-pitch singing of urban amateurs and the self-assured, self-conscious, decided performance of peasant singers.
The first problem with which we have to deal is with what degree of exactitude we shall transcribe these deviations in pitch. My method is as follows: if the deviation is perceptible enough, arrow signs above the notes are used. An arrow pointing downward means lowered pitch, almost reaching the quarter tone below; pointing upward it means raised pitch, not quite or almost reaching the quarter tone above. These arrows may be used in conjunction with both flats and sharps, too, where they will be valid as long as are the accidentals themselves. Some special signs could conceivably be used for differences of exactly one quarter tone. They were almost never employed here because of the great difficulty in determining by ear whether the difference is exactly a quarter tone or only approxi- mately. Obviously these arrows and comparable symbols used by others are signs of approximate value, and are, perhaps, the only practical means to be used concerning pitch deviations in Eastern European folk music.
^ It has been shown that even the intonation of trained professional singers is seldom absolutely exact.
[4]
Introduction to Part One
As a matter of fact, a system has been established and instruments in- vented which enable us to measure the vibrations of each scale degree of a given melody. The system is the "cent" system. In this system every tempered half tone is divided into 100 equal cents, and it is possible to express every tone in cents instead of its number of vibrations. In many cases, especially in melodies with typical tone deviations of degree, such measurements would give highly interesting and mathematically exact data about the deviations, as well as about their constancy. Some musicol- ogists, for example, Erich von Hornbostel and Idelsohn, frequently used this procedure. I am sorry to say that I never had occasion to use it myself.
Other special signs for certain phenomena connected with pitch are the various types of glissando}
The second problem is the degree of exactness to be used in transcribing the rhythm of folk melodies. Here, again, it must be borne in mind that an absolutely rigid rhythm never prevails even in so-called "rigid" dance rhythms.^ Whether it be the latter or the so-called parlando-rubato , free rhythm, with which we often have to deal in Eastern European folk music, limits must be set to the exactitude of rhythm transcriptions. As a general rule, the deviations in "dance rhythm" which are not noticeable to the trained observer in playing the records at ordinary speed should not be heeded. They are scarcely perceptible when playing the records at even half speed. A certain limit should be set even for transcribing devia- tions which become rather noticeable when magnified by this procedure. This limit should correspond to the ability of the human mind to perceive differences of rhythm. Speaking in practical terms, it is about the value of a sixty-fourth at a speed of 120 beats per minute. For example, if the singer performed an eighth dotted four times instead of a quarter note, it would be best to ignore this very slight deviation. Even for an eighth dotted three times it is perhaps more advisable to write a quarter note with a half-circle over it.'* The same limit to exactness can be set when dealing with parlando-rubato melodies. Of course, their irrational rhythmic formulas can by no means be called "deviations," since these melodies never present regular rhythmic patterns from which to deviate.
^ See explanation of signs, p. 91. ^ The same can be said of art music perfonnances.
* This haK circle is used to denote a slight shortening of value; see explanation of signs, p. 91.
[5]
Introduction to Part One
It is advisable to use the ordinary time-signature in rigid dance rhythm only, when no deviations need to be introduced into the transcription at all. If deviations occur, then the general time signature may be inclosed in parentheses, and no changes in signature need to be introduced. Changes in time signature are to be used only when the change does not result from an occasional deviation, but is an essential rhythmic feature (alternation of different measures). In transcriptions of parlando-rubato performances no kind of time signature should be used, for such a device would give the reader little help, and it would only bewilder him to find in almost every measure different and most unusual signatures, such as 11/16, 15/16, 31/32.
Summing up, the transcriptions of records of folk music and their publi- cation ought to be, as far as the described limits permit, in the manner of a so-called Urtextausgahe (that is, critically revised edition, in which the texts are presented according to the presumed intentions of the author or composer) . In other words, nothing should be changed by the transcriber except those parts in which the performer made an obviously uninten- tional mistake. In critically revised editions of higher art works, obvious mistakes of the author or composer appearing in the autograph or in the (authentic or nonauthentic) first edition must be replaced by the form which we have sufficient reasons to believe is in accordance with his original intention. This, of course, must not be done without referring to and carefully describing the mistakes and misprints or without ample explanations concerning the editor's reasons for the change. Similarly, absolutely unintentional mistakes made by the performers of folk music should not be entered in the published transcriptions. For instance, if when starting on a certain note the performer's voice fails or slips so that he or she has to try again to reach the note, the results of this mischance should not be entered, much less published. The same applies when he or she drifts into a wrong pitch, but becoming aware of it, stops, and begins the whole period again. However, off -pitch notes which were not corrected by the singer, periods or bars sung (exceptionally) some degrees higher or lower than they obviously should be, excess syllables (with the corre- sponding notes, of course), even if seemingly added accidentally, should be transcribed and published or at least mentioned. Syllables, notes, and rests evidently omitted by accident should not be restored by the tran- scriber, and so forth. In some such cases, however, marginal notes may refer to the supposedly correct form. In any case, every change intro-
[6]
Introduction to Part One
duced by the transcriber should be mentioned and explained (except the unintentional errors described above).
Some scholars may have a different opinion on this subject and would perhaps include these slips and failures in the transcription as character- istic of certain singers or areas. Even so, I am against including them, for (1) they show on the records, and whoever wants to study them may find them there ; (2) I consider these imperfections analogous to physical abnor- mities and should prefer, as far as possible, to keep apart well-shaped melodies from unintentional desultory forms.
However, in connection with this whole question, it may be borne in mind that decisions as to the cause of deviations on the part of the singer are often difficult to reach and can be made only on the basis of very intensive experience. A false judgment can change the picture of what the performance was and what the singer intended. Thus, when in any doubt at all, an alternative method may be used, by giving in the transcriptions what the investigator thinks was intended, but giving in notes all changes and errors, obvious or not, so that the reader can evaluate them as he wishes.
SETTING OF BARS^ AND CHOICE OF VALUES
Both problems, although they do not interfere with truthfulness in the notation, are rather important. The right setting and choice will help greatly in understanding the structure of the melodies and will make for greater consistency in editing. Some scholars say bars should not be used in parlando-ruhato melodies. Probably their point is that in these melodies no regularity of rhythm can be observed and that therefore the periods must not be divided into measures, since measures refer to certain regu- larities of rhythm. However, when we consider the original meaning of the bar (it means an articulating accent on the value following the bar), then we can easily acquiesce in the setting of bars in these periods. We can, moreover, appreciate them as very useful means of giving a clear idea of the articulation of the melody. Their place should be determined generally by the metrical structure of the text line, in the case of vocal melodies. In a style in which the meter is based on syllable count, lines of eight syllables with 4 -\- 4 structure may have bars after the fourth and
5 "Bar" will be used in this book in its original meaning, that is, to signify the vertical lines dividing the piece into small parts; "measure" will signify the part between two bars.
[7]
Introduction to Part One
the eighth syllables, respectively; those with 3 -\- 2 -\- 3, after the third, the fifth, and the eighth syllables, and so forth. If we accept this as a basis for the determination of rhythmic structure of the melody, it will be extremely important to have the text in front of us when transcribing the melody. If the transcription of a recorded text is missing and if we cannot make out the text by listening to the record, we may be unable to deter- mine even the periods of the melody, much less the sub-periods (measures) . In transcriptions of parlando-rubato melodies bars are used for the above purpose, and broken bars are added to measures which are rather compli- cated and of considerable length, in order to indicate subdivisions of the measure. Of course, any other sign (comma, short line, short double line, and so forth) in place of continuous or broken bars will serve the purpose as well. It would be desirable, however, for as many editors as possible to reach an agreement concerning the use of uniform signs.
In melodies having a more or less rigid dance rhythm or a rhythm that sounds like a parlando-rubato rhythm, but is actually the transformation of an originally rigid rhythm, the use of bars is generally accepted. The metrical structure of the text lines is here too, the reliable basis in deter- mining the place of the bars. In some special cases, however, we must par- tially abandon this basis and follow the rhythmical pattern of the melodies of rigid rhythm (for instance iJJJjo |J JJ|o |,in spite of having 2 -\- 2 -\- 2 -\- 2 as the metrical structure of the text lines; see music example No. 21) or the "declamatory"^ accents of melodies in parlando- rubato rhythm.^
The structure of the melodies should be determined by the structure of the text; that is, the single sections (in German, Melodiezeile) of a given melody may be expected to correspond to the respective lines of the text (or of the stanza). This seems obvious, since the sections of melody and text generally tally with each other. However, in some cases found in Eastern Europe, especially in Serbo-Croatian folk melodies, the structure of the melodies seems at first hearing to contradict the respective portions of the text. Without the text at hand, one might very often be misled by some peculiar features of the melody (rests, climax notes, repeats) and might determine its structure erroneously, that is, in contradiction to the text structure (for example, our No. 14).
* "Declamatory" means here: conforming to the rhythm and inflections of the language as applied to the melody.
^ This occurs chiefly in the Yugoslav heroic poems.
[8]
Introduction to Part One
It may be asked whether musical structure should be determined on the basis of extramusical factors. An affirmative answer is dictated by the following facts. The texts of Eastern European folk songs have almost without exception full word meaning (or at least pretend to have one^) ; texts consisting entirely of meaningless syllables do not exist (as is possible in primitive folk music). The primary purpose of the performers is to convey this meaning to the listener; the melody is a secondary factor and serves only to facilitate this conveying, to decorate the procedure, and to enhance the impression. Of course, they cannot do without melodies; texts of folk songs are never performed independently of singing. The sentences and periods in an Eastern European folk song text always cor- respond with its metrical sections and subsections; overlapping does not occur. Considering this fact and the previously mentioned psychological factor, we must admit the necessity for a very strict correspondence between the structure of the melody and that of the words. Obviously the end of the sentence or period in the text marks the end of the correspond- ing section of the melody as well.
As to purely instrumental melodies, generally those in rigid dance rhythm do not offer much difficulty, in view of the usually decisive rhythmical performance, on one hand, and the clearly symmetrical struc- ture of the periods, on the other hand. In some cases, however, intricacies appear; for instance, when a constant ' counter rhythm" is applied by the solo player on the second and fourth eighths of a 2/4 measure, as a substitute for the counter beats of an imaginary accompanying second instrument. In this case the listener may misinterpret these counter rhythm accents on the upbeats for downbeats. Only a very careful exam- ination or eventual comparison with variants will enable us to eliminate this misinterpretation — that is, wrong placement of the bars.
Less frequent are instrumental dance melodies in which the periods contain repetitions of phrases of irregular length, for instance, 2/4 a6l(;a||6c|(/e||.^ Almost invariably the listener is tempted
to interpret the ah c as a 3/4 bar and thus the whole period as 3/4 ah c\ah c\ 2/4 d e\\. Re -examination is likely to reveal that examples of this kind are nothing but 2/4 melodies with "shifted" rhythm in their phrases.
' This may happen in certain kinds of jesting texts, where frequently words are used with no sense at all, to imitate the sound of a foreign language, and so forth. ' Each letter stands here for a 1/4 value.
[9]
Introduction to Part One
Much greater difficulties present themselves in determining the bars and the structure of instrumental parlando-ruhato melodies. No guidance is given by text lines ; accents in the performance are often misleading. If a vocal variant is at our disposal, these difficulties, of course, disappear. If there are no such variants for comparison, we may find some guidance through comparison with instrumental variants, or with other instru- mental (or vocal) melodies of the same type, or through examination of the structure of the melody itself. In extreme cases, however, the position of the bars and the determination of the structure will remain matters of guesswork.
Another problem is the choosing of the values most suitable for tran- scribing a set of melodies. When editing single melodies, it matters little which values are used if extremes (unusually long or unusually short values) are avoided. Very long values make the transcription look too heavy; very short ones render it difficult to read. But if one works on a set of melodies belonging to the same or to kindred types, one obviously should try to attain a certain consistency in choosing the value units. It is regrettably inconsistent (but unfortunately rather frequent in ama- teurish publications) to transcribe melodies of similar rhythm and the same tempo in values differing in the proportion of 1 to 2 ; for instance,
8 Ir (* f p f Iffrr " ^^^^ for some of the melodies, ^\-p ^ » »\» • m m'\
for others. One must bear in mind that everything, even the slightest detail, in any publication of folk music should be in accordance with well- weighed principles, for the establishment of which the editor should be able to give his reasons.
There would be no particular difficulty in choosing the value units, had the melodies of a certain type and of similar rhythm everywhere the same tempo. The facts, however, show that the tempo of melodies of the same type (especially of vocal melodies) is subject to great variation, even when performed by the same singer. The variation may exceed the double, or the half of a certain supposed standard tempo. The problem arises whether we ought to keep strictly to the values chosen for the standard tempo and change only the M. M. figure or whether we ought to set upper and lower limits and, in case the tempo passes these limits in some of the melodies, to change the values (using half or double values). Both pro- cedures have their conveniences and inconveniences, therefore no abso-
[ 10]
Introduction to Part One
lute preference can be given to either of them; our decision should always be governed by the specific circumstances.
The parlando-rubato melodies present more difficulties in connection with the value choice. The best method seems to be as follows. By listening to the melody several times, we aim to catch the prevailing (most characteristic) length of the notes in the melody and to choose the value best suited for it. Once this is done, according to our judgment, all the values which differ from the prevailing one have to be measured and expressed by corresponding values.
In many instances even the prevailing lengths are not constant, that is, they change from section to section. In such cases, it is preferable to change the M. M. figures from section to section rather than adhere rigidly to the value established in the first section. In other words, it is preferable in such cases not to express the deviations in the subsequent sections in proportion to the standard value of the first section. For example, if the prevailing values of the first section are eighths and those of the second, dotted sixteenths, M. M. J^ = 150, it is much more pre- ferable to use eighths, M. M. J^ = 200, in the second section. This pro- cedure of giving each section and stanza its own metronome value is imperative when by adhering to the original M. M. figure the changes could be expressed only by too-complicated value signs, for example, a six- teenth dotted three times, or a quarter note tied to a thirty-second.
If we neglect this precaution, it may well happen (as it sometimes happened to me) that arriving at the first section of the second stanza we are compelled to use completely different values for exactly the same note lengths, for instance a quarter note in the second stanza instead of an eighth, which was used in the first stanza, although both note values have in fact the same metronomic length. These are the general require- ments which help to obtain consistency in the notation of values.
In addition, there are some problems concerning details. For example, a triplet of three eights is amost equivalent to three dotted sixteenths. Substituting one for the other for the sake of a clear, more legible notation means, of course, neglecting the difference of a thirty-second note. In my opinion this type of very slight deviation from exact notation is admissible in parlando-rubato melodies, especially when a 1/4 note is not slower than
M. M. 100 (that is, } = M. M. 800). The rhythmical formulas J^^ and
[ 11]
Introduction to Part One
S\^ , or the formulas ij- » •!• > *p and 4«f , have exactly the same
value. Their application should depend entirely upon the individual case — in making a choice our purpose should be to attain the clearest and most legible reading. One thing, however, should be avoided — the arbitrary interchange of these formulas without any reasonable cause. This last problem, of course, belongs already to the orthography and aesthetics of transcription.
Signs in the current notation which have a more or less vague meaning (the hold, the trill sign, the mordent, comma, and so forth) should never be used ; instead of the hold and a comma, the exact measurement of the value, and instead of embellishment marks, the exact pitch of the orna- mental tones should be used in the notations. Perhaps only four excep- tions are permissible: the use of a wavy line above the note (for the vibrato), of semicircles for slight lengthening or shortening (see p. 91), and of the short grace-note. An attempt exactly to measure the duration of the latter would be unnecessary hairsplitting. As a general rule, the quarter note is suggested for values of approximately M. M. 50-130 dura- tion; the eighth for approximately 100-200 duration. This suggestion does not pretend to be anything but a broad principle, as has been indi- cated already by the fact that these suggested figures show a certain over- lapping. Deviations from this rule are therefore unavoidable, and each exceptional circumstance is to be dealt with on its own merits.
CERTAIN PECULIARITIES APPEARING IN THE PRESENT PUBLICATION
I always use beams instead of the old-style flags for a succession of notes sung to different syllables ;^° to melismatic groups (that is, to differ- ent notes sung to the same syllable) I apply a slur. The usage of flags (or hooks) in vocal music, still prevalent in our day, may have historical and logical foundations. However, sticking to this usage in folk music purely because of tradition is, in my opinion, rather senseless. A profusion of these hooked notes, especially when they have mixed values, are decidedly confusing to the eye. Moreover, if hooks are used, articulation remains unexpressed; j3 J3 and JJJJ , although they have different meanings;
^" Z. Kodaly applies the same method; and we both (as well as A. Schonberg, from his Op. 18 on) use it also in our original vocal works.
[12]
Introduction to Part One
become undistinguishable if hooks (^J^^^) are used. I cannot afford to discard the valuable advantage in articulation afforded by the other method.
As to key signatures, only such sharps and flats should be used which refer to degrees actually occurring in the melody, and which are valid throughout the whole melody. It happens frequently, even in recent and serious publications, that accidentals of the key signature refer to degrees which are not contained at all in the melody. The usage of such accidentals in the key signature is senseless and ludicrous. This unjustified habit is probably the result of a major-minor complex — a sad inheritance from the nineteenth century. According to this complex, the only existing possible and permissible scales are the major and the minor scales of the higher art music of the last two centuries, and every other scale has to be projected on these two scales and considered as an accidental deviation from them. People obsessed by this wrong theory forget that even J. S. Bach used the Dorian key signature when he meant to write in the Dorian mode.
The consistent employment of the above-described and only permissible procedure sometimes leads, of course, to rather peculiar key signatures; for instance, b\? c^, or a\} eb, or b\} a\? d\? cb ! However, there is a practical advantage in this procedure ; the reader perceives at a glance which scale is used in the melody.
PITCH OF MELODIES
In principle, melodies ought to be published in the original pitch as sung or played by the performers. In practice, however, we have to make a compromise in order to attain certain goals. One of these goals is to make the survey of the material as easy as possible. The most suitable method by which to attain this is to transpose all the melodies to one pitch, giving the melodies a common ''tonus finalis"; to place them, so to speak, over a 'common denominator." In collections where this method is used, one glance at a melody is frequently sufficient to determine its relationship to others. If we indicate, in addition, the original pitch of the tonus finalis by some procedure (see explanation of signs, p. 91), we fully comply with the requirements for recording the original pitch. Nevertheless, some scholars stick obstinately to the original pitch in
[13]
Introduction to Part One
publishing folk music, in spite of the many inconveniences it presents. For instance, approximately half the melodies may bear key signatures crowded with accidentals, at times each accidental decorated with arrows for pitch deviations. What procedure shall be followed, however, if the performer gradually changes the pitch during the performance? Let us state an example : the last note of the first stanza of a melody is a quarter tone or a half tone higher, in comparison with the first tone, than it should be; the same happens in the second stanza, and again in the third stanza, and so forth. (See music example No. 6c and other pieces sung by the same singer.) These changes are introduced gradually, not suddenly at a certain point in the stanza. Therefore it is not possible to transcribe any of the stanzas consistently with the original pitch; one has to resort to explanations in footnotes, thus breaking the principle cherished by these scholars. This obviously shows that their principle cannot be fol- lowed with consistency. Incidentally, these gradual changes must not be regarded as off -pitch singing; probably they are characteristic of certain regions (for example, in the Rumanian-inhabited parts of the Banat dis- tricts^) and therefore are to be observed with meticulous care.
METHODS IN SYSTEMATIC GROUPING OF FOLK MELODIES
Unfortunately, most editors of folk music have not used any musical principles in grouping; the need for it did not even occur to them. They group their material either according to texts or according to geographical units (villages, districts) , or they simply do not group them in any way. The first large collections grouped according to systematic principles derived from the characteristics of the music itself were published by the Finns ;S^ others were later published by the Ukrainians.^^
There are two predominant principles which differ in purpose and partially in method. According to the first, the "lexicographical," a more or less rigid system is to be used (similar to the one used in lexicons and dictionaries) in order to make it possible to locate each melody by com- paratively simple mechanical means. This has its advantages; its disad- vantage consists in the fact that frequently variants (sometimes even
" See Bart6k, "Rumanian Folk Music," Vol. II, Nos. 15c, 67f, 176e, 198a, 202g, and others. (Manuscript deposited at the Music Library of Columbia University.) ^ Ilmari Krohn and others. ^' Philaret Kolessa and others.
14]
Introduction to Part One
slightly differing variants) will have to be placed far from each other. This disadvantage is even more disturbing when it comes to melodies that belong to the same style, origin, or stock. Thus, in this grouping a given melody may be easily located, but the reader cannot very well get an idea of the different styles, structures, and even of variant groups prevailing in the material.
The system according to the second principle, which I would call "grammatical," groups all melodies belonging to the same family, or being of similar structure and representing the same style, as near each other as possible, and presents all members of the variant groups together. In order to attain this object, a rather complicated grouping system must be constructed and used, a system having all the elements of the lexico- graphical system intercrossed by various and numerous additional ele- ments. A systematically grouped collection based on this system offers rather clear boundaries between the single melodic styles, structures, and families. In addition, one can locate single melodies, but in order to be able to do so one must have some previous familiarity with the material and a thorough knowledge of the rather complicated grouping system.
The question is, which is of higher importance — to be able to locate melodies easily and almost mechanically or to get a clear idea of their relationship? I am in favor of the second objective, the "grammatical" principle, in spite of its imperfections. For it has imperfections, as well as any other conceivable system. (Perfect systems for grouping can never be invented.) Therefore the system used in the present work (as well as in all my recent publications) is based on this principle.
When creating a system, various aspects of the melodies have to be considered: (1) section structure (a "section" being a portion of the melody corresponding to one line of text); (2) metric structure (metric units such as feet or the number of text syllables in each section) ; (3) rhythmical character (for example, free rhythm, called parlando-ruhato or, in short, " parlando") ; (4) melodic structure (referring to the final notes of the sections, or other significant melodic features); (5) range (ambitus); (6) scale system; (7) form, or content structure (referring to the content of the single sections); and in addition various subaspects, which cannot be enumerated here.
According to the lexicographical system the order of importance of these aspects is approximately as follows: 1, 4, 2, and 5 (3, 6, and 7 are
[15]
Introduction to Part One
rarely considered in the Eastern European material). According to the grammatical system: 1, 2, 3, 4, 5 (6 and 7 are rarely considered).
In order to simplify matters, symbols consisting of figures, letters, and the like, are to be devised and used (a procedure more or less similar to the one established in mathematics) . For instance :
I I indicates a principal section of a melody, in general
[jj indicates a specific section of a melody, having a^ as final note _j indicates a section preceding the principal one, in general v] indicates a specific section preceding the principal one, having d^
as final note L indicates a section following the principal one, in general y indicates a specific section following the principal one, having 6^1? as a final note.
6, (that is, 6 followed by a comma) indicates an isometric stanza struc- ture with 6 syllables for each section
7, 7, 8, 7, indicates a heterometric stanza structure with 7 syllables for the first, second, fourth section, and 8 for the third section
VII— 5, a range from/^ to d^; and so forth (see description of grouping
system, p. 28)" For instance, 11, ll LAJ li 1-8, A b b a is the figure and letter symbol
of a well-known Hungarian type of melody, comprising hundreds of
specimens.
A general rule must be heeded in connection with the second grouping system : variants should never be scattered in different classes (subclasses and so forth), whatever the difference in their structures. If some or all the variants of a group differ in structure, these differences should be carefully examined; preference should be given to the structure which seems to be more ancient or simpler or prevailing, and so forth ; melodies of this selected structure should be put at the head of the variant group, and the whole group put into the class (or subclass and so forth) indicated by the structure of this dominant variant.
Another rule (not without exception) in grouping melodies is to proceed from lower pitch or figures to higher pitch or figures. In doing so, figure and letter symbols are of great help.
" See explanation and enumeration of symbols on pp. 90-91.
[16]
Introduction to Part One
THE PROBLEM OF VARIANTS
To assemble all variants is not as easy as it may seem to a layman. Even if we succeed in locating all of them mechanically or by memory, it remains to be determined when a melody ceases to be the variant of a kindred melody. There remains primarily the question: What character- izes a melody as a variant of another? I should say that variants are melo- dies in which the pitch relation of the various principal tones to each other shows a certain similarity; or, in other words, in which the contour line is entirely or partly similar. But even on this basis immense difficulties will present themselves when we deal with hundreds or even thousands of melodies belonging to a characteristically homogeneous style. In such a long chain of melodies, in which each of two neighboring melodies can be considered beyond any doubt and at first sight as a variant of the other, and the first one in the row is so entirely different from the last one that it should not be called a variant, one has to break up these melodies into two or more separate groups. Where shall one group end, where shall the new group begin? Which of the more or less similar melodies shall be considered variants; which not? Again subjective decisions will have to guide us in the solution of all these questions.
ORNAMENTS {GRACE NOTES)
Two or more notes of different pitch sung to one syllable constitute ornaments in vocal melodies. Usually one of the notes in an ornament group can be regarded as the principal one (the one which would remain if the ornament group were stripped of the ornamental notes), and the rest as supplementary ornamental notes. The latter are frequently sung with a much lighter tone emission and ought to be transcribed with small note heads, in order to distinguish them from the heavier principal tone. In some areas, however, especially of Serbo-Croatia, ornamental groups with equally heavy tones and with sharp, definite rhythmic and melodic patterns are more frequent. I should call them "heavy" ornaments. Com- pared with the performance of Hungarian, Rumanian, Slovak, and probably the Ukrainian m.elodies, the Serbo-Croatian folk melodies (probably Russian and Greek folk melodies, too) are characteristically rich in heavy ornaments. As the heaviness or lightness of ornamental
[17]
Introduction to Part One
notes is an important characteristic of various territories, it is obviously- important to discriminate between them in the notation by using full- size and small notes. Unfortunately, in many of the published collections no discriminating means are used.
It is rather difficult, of course, to find the principal tone in a heavy ornament group. Comparing variants and examining upbeat and down- beat relations may give useful hints for solving this problem.
OBJECTIVITY AND SUBJECTIVITY IN TRANSCRIBING AND GROUPING FOLK MUSIC
Evidently objectivity is what we must strive for. This is the ideal, an ideal which, however, can never be fully attained. Objectivity could be furthered by the use of more accurate measuring devices but in practice they would probably prove too complicated and the results would cer- tainly be too cumbersome for the reader. If we do not employ such means, our eyes and ears serve as measuring apparatuses — rather imperfect apparatuses. Thus, through our imperfect senses many subjective ele- ments will get into our transcriptions, rendering them that much less reliable. This applies especially to the most minute details of the per- formance (very short grace notes and the like); divergences in the per- ception of these details are not only frequent among well-trained tran- scribers, but even in transcriptions made by the same person. For one reason or another, it very rarely happens that the same person has tran- scribed twice the same piece of folk music. But when it happens, these double transcriptions, especially if separated by some period of time, are very valuable documents — the differences between them (while they concern mostly only slight and unimportant details) will indicate the degree of reliability of the transcriber in question.
Even if we do employ objective measuring instruments for determining pitch and speed, there are no imaginable implements to guide us in making the judgments which enter into our determination of structure, range, and scale of folk music. Lajrmen will say: "Why implements? Nothing is simpler ; look at the last note of each section, count the syllables of each, and so forth, and you will have established structure, range, and scale." Few persons know, indeed, how many intricacies enter into the solution of these questions and provide hundreds of occasions for subjective judg- ments. A few examples: The last note of a section or even of the melody
[18]
Introduction to Part One
is not always the structural "final" note {tonus jinalis) . In a style in which the metric system is based on syllable count, the number of syllables does not always correspond to the number of the structurally essential sylla- bles. Sometimes certain sections or portions of the melody are repeated; we have to determine whether these are structurally essential repeats (that is, whether they are independent portions which enter into the structural form) or nothing but casual or playful ad hoc repeats. A melody in rigid dance rhythm is sometimes performed in free parlando-ruhato rhythm, but it should, nevertheless, be grouped as if it were in rigid rhythm. Sometimes it is difficult even to determine the boundaries of a single section.
There certainly are established rules and a tradition for deciding these issues, but they are not always applicable. There, again, we must turn to subjective judgment for aid. Needless to say, on these questions the difference of opinion between scholars is even greater than on those of a purely technical nature.
VARIABILITY IN FOLK MUSIC
I want to call the reader's attention to one of the most conspicuous features of folk music, namely, its continuous variability. This has been said so often that it has almost become a truism. Nevertheless, since lay- men still persist in believing that folk melodies are something rigidly established and unchangeable, it is pertinent to stress over and over again the fallacy of this belief.
Some linguists assert that each uttered sound of speech is a unique phenomenon which has never occurred in the past and will never occur again in the future. At first hearing, this seems to be a bewildering state- ment. However, one will be more inclined to take it as an evident fact if one considers a parallel and more obvious phenomenon which does not even need to be proved ; everyone will agree that each individual of every living species of animals or plants is a unique phenomenon. The same is true concerning folk melodies — a given performance of a folk melody has never occurred before and will never occur again in exactly the same way. We must not say or think that a folk melody is what it appears to be in a given performance ; all we can say is that then and there, at the time OF THAT PERFORMANCE, it proved to be such.
Some beHeve that the essential difference between art music and folk
[19]
Introduction to Part One
music is the continuous variability of folk music as against the rigid stability of art music. I agree with this statement, but with a qualifica- tion : the difference is not one of contrast, but one of degree — that is, the performance of folk music shows an almost absolute variability, while art music varies in a far lesser, sometimes in only an infinitesimal degree. We must realize that even performances of the same work of art music by the same performer will never occur twice in absolutely the same way. Of course, the notes of art music — because of their fixation by notation — ^must never be changed, whereas in folk music even notes are subject to change. And yet, if we reconsider well the first part of this statement, we see that even some of the notes in art music are subject to change : I mean the particles of embellishments (notes of trills and the like).
This eternal changeableness gives life to music, be it folk or art music, whether the changes are considerable or scarcely perceptible. These intrinsic characteristics of music seem to be in contradiction with the contemporary trend of producing music more and more by mechanical means, according to which music would be compressed into a frozen and never-changing form.
In closing these introductory remarks, may I point out to the reader once more their leading idea : that although perfection cannot be attained in transcribing and classifying folk music, we must always endeavor to approach an ideal of perfection, an ideal which in itself is still but dimly perceived. We should never tire of improving and changing our methods of work in order to accomplish this task as well as is humanly possible.
[20]
Morphology of the Serbo-Croatian Vocal Folk Melodies
SEVENTY-FIVE "women's songs" (in Serbo-Croatian, zenske pjesme) are published in this book in fifty-four variant groups.^ With few exceptions, no selection has been made, could not even be attempted since when the material was compiled for this publication the remain- ing one hundred(approximately)^ songs were not yet transcribed.
What are "women's songs" in Serbo-Croatian folk poetry? Almost all songs except the "heroic poems" (called also "'gush songs," because they are to be performed with the accompanim^ent of the one-stringed gusle) . They are sung without instrumental accompaniment and as a rule, though not invariably, by women, hence the designation.^ The text of most of these songs consists of ten-syllable lines, although lines of seven, eight, nine, and more than ten syllables may also occur. As to their subject matter, they have lyric or ballad texts, the latter generally with ten- syllable text lines. There are, however, a few "ballads" among the women's songs of our collection which have eight-syllable lines and a peculiar stanza structure (Nos. 20, 34-36; see pp. 65-66). It is not quite certain whether these should be regarded as women's songs, for they were all sung by men and are distinguished from the bulk of the women's songs by their strange stanza structure.^ Some of the "women's songs" were sung with instrumental accompaniment (with tambura: music examples 27a, 44b, 39b, and even with gusle, examples 7, 34). The singers are, however, in all these cases, men.
^ Referred to by number and letter, the letters indicating variants of a given basic melody.
2 There are, m addition, about thirty songs sung by an old woman {-Bida Dizdarevic) whose performance was not reliable concerning pitch and could not be studied for musico-scientific purposes.
^ We do not know whether this designation is of rural origin, or whether it has been intro- duced by the early Yugoslav scholars. See Part Two, pp. 247-249.
* For more details on this subject, see p. 254.
[21 ]
Morphology
COMPARISON OF THE MATERIAL PUBLISHED IN THIS BOOK WITH SOME OTHER AVAILABLE MATERIAL
Seventy-five melodies are a rather limited number. They would give a somewhat narrow view of the Serbo-Croatian folk melodies if published without any reference to other Serbo-Croatian material. Therefore, I deem it necessary to include a general survey of the most important published collections of Serbo-Croatian folk melodies as a background and to treat the pieces of the Parry Collection against this background. Fortunately, the main published collections of Serbo-Croatian folk melodies, the grouping of which I began some years ago when still in Europe and concluded in the United States, were at my disposal. Some unpublished melodies from a collection made by L. Kuba and some from my own collection have been added to the printed ones.
The following collections have been incorporated into this survey.
PUBLISHED COLLECTIONS
1. Bosiljevac, §. Album bosansko-hercegovackih pjesama (Folk Songs from Bosnia and Hercegovina). Zagreb, St. Kugli. Fifty melodies. Cited as "Bosiljevac. "5
2. Busetic, Todor. Srpske narodne pesme i igre s melodijama iz Levca (Folk Songs and Folk Dances from Levac). Beograd, 1902. Srpski Etnograj'ski Zbornik, III. Eighty-eight melodies. Cited as "Iz Levca."
3. Dordevic, Vladimir. Srpske narodne melodije; Juzna Srbija (Serbian Folk Melodies; Southern Serbia). Skoplje, 1928. Four hundred and twenty-eight melodies. Cited as "Juz. Srb."
4. Srpske narodne melodije (Predratna Srbija) (Serbian Folk Melodies
from Prewar Serbia). Beograd, 1931. Six hundred melodies. Cited as "Dordevic."
5. Narodna pevanka (Folk Song Book). Beograd, 1926. Approximately
three hundred and eighty melodies. Cited as "Dordevic Nar. Pev."
6. Kacerovski, B. U kolo (Round Dance). Sarajevo, J. Studnicka & Co. Eighty-seven melodies. Cited as "Kacerovski."
7. Kuba,Ludvik. "Pjesmei napjeviizBosneiHercegovine" (Songs and Melodies from Bosnia and Hercegovina), Glasnik Zemaljskog Muzeja u Bosni i Hercegovini, XVIH (1906), 183 ff., 354 ff., 499 £f.; XIX (1907), 103 ff., 273
* Citations are always by the number of the melody, except item 5 ; for this latter work page citation had to be used, the melodies not being numbered there.
22
Morphology
jl., 405 &., 629 £F.; XXI (1909), 303 &., 581 £f.; XXII (1910), 513 JEf.; Sarajevo. Nine hundred and sixty-five melodies. Cited as "Kuba, B. H."
8. Slovanstvo ve svych zpevech (The Slavs in Their Folk Songs).
Pardubice, Podebrady, Praha, 1884-1929. Vol. IX, sixty-four Croatian melodies; Vol. X, sixty-one Dalmatian melodies; Vol. XI, seventy-one Montenegrin melodies; Vol. XII, sixty Serbian melodies; Vol. XIII, sixty Bosnian and Hercegovinian melodies; Vol. XIV, thirty-one melodies from "Old" Serbia (that is, the southern part of Serbia [north of Macedonia]). Cited as "Kuba IX," and so forth.
9. Kuhac, Fr. §. Juzno-slovenske narodne popievke (Yugoslav Folk Songs). Zagreb, 1878-1881. One thousand six hundred and twenty-eight melodies. Cited as "Kuhac."
10. Manojlovic, Kosta. Muzicke karakteristike nasega juga (The Musical Characteristics of Our South), Sv. Cecilija (Zagreb), V (1925), 138 &.; VI (1925), 174 fif. Approximately twenty melodies. Cited as "Manojlovic."
MANUSCRIPT COLLECTIONS
11. Kuba, Ludvik. The continuation of "Pjesme i napjevi iz Bosne i Herce- govine" (Songs and Melodies from Bosnia and Hercegovina), not pubhshed in Glasnik Zemaljskog Muzeja u Bosni i Hercegovini, going from No. 966 to No. 1125. One hundred and sixty melodies. Cited as "Kuba, B. H., MS." in Tab. of Mat., otherwise without MS.
12. Bart6k, Bela. Serbian Folksongs from the Counties of Torontal and Temes (in prewar Hungary, now Rumania). Five melodies. Cited as "Bart6k." Unpublished.
The data on published collections are mostly in accordance with the statements of the semiofficial bibliography of Serbo-Croatian folk music, in Folk. mus. (pp. 210 ff.) Item 4, however, has been omitted there, apparently by an oversight, although it is one of the most important col- lections of Serbo-Croatian folk melodies ; the numerical indications of the respective volumes in item 8 are missing ; and different numbers of melo- dies are given for items 4, 8, and 9; in items 1, 6, and 10 no figures are given for the number of melodies. The only Kacerovski publication mentioned in Folk. mus. has as title Bosanske sevdalinke za klavir; we are not sure whether it is identical with his U kolo. As to Manojlovic, it is uncertain whether the publication used by me is identical with the one cited in Folk. mus. as Muzicke karakteristike nasega juga.
There is a large collection not included in the material as enumerated
[23]
Morphology
above, although it has been at my disposal, that is. Dr. Vinko Zganec, Hrvatske pucke popijevke iz Medumurja (Croatian Folk Melodies from the area between the Rivers Mura and Drava, at their confluence) , Zagreb, 1924. I had well-weighed reasons for the omission. This area consists of about three hundred square miles, formerly belonging to Hungary proper. The material collected there by Dr. Zganec, 638 melodies, a rather considerable number for such a small area, shows an overwhelming Hungarian influence — about two-thirds of them are borrowings from the so-called "Old" and "New" Hungarian musical stock (as I pointed out in "La Musique populaire des Hongrois et des peuples voisins," in Arch- ivum Europae Centro-Orientalis, Budapest, 1936) ; and even the remaining one-third have a more or less international (or more correctly "inter- Central-East European") character. Therefore I prefer for the time being to exclude them from the above detailed material; they would only add many foreign traits to the Serbo-Croatian stock and would change the proportion between autochthonous and foreign components too much in disfavor of the former. As to the material enumerated under items 1-10, I had to leave out
(a) the few instrumental pieces contained in some of the publications;
(b) the melodies deriving from territories outside Serbo-Croatia (that is, the Slovenian^ and the Bulgarian melodies in item 9) ; (c) the Macedonian melodies deriving from an area which is clearly marked as Bulgarian, ethnically as well as linguistically, on maps dating before 1914^ (m.ost of items 3 and 10, and some of 5 belong to this category) ; (d) melodies of lower art music (urban music of a certain artificial kind which never could have been sung by rural people), to be found mostly in item 9.
A number of songs are not Kuhac's own collection, but are taken from older manuscripts or printed collections. Some of the same sources are used in item 5, thus it happens that the same melody appears in
" Serbian and Croatian are practically the same language; the two designations are due to historic reasons, and signify political, religious, and cultural differences. The Serbs adopted the Greek Orthodox religion, the Cyrillic alphabet, and a cultural orientation towards the south-east; the Croats, the Roman Catholic religion, the Latin alphabet, and a cultural orien- tation toward the west. The Slovenian language, however, is unanimously regarded as an independent language by linguists, though closely related to the Serbo-Croatian.
^ There are differences of opinion regarding the language of this area, the Serbo-Croats asserting that it is an archaic form of Serbo-Croatian, the Bulgarians declaring it a Bulgarian dialect. Two facts, however, seem to be clear to neutral observers: first, that it is not Serbo- Croatian; second, that it is closer to the Bulgarian than to the Serbo-Croatian language.
[ 24
Morphology
different publications and in exactly the same notation.^ Most of the melodies in Kuba XIII are published in item 7 as well. We find in item 5 some melodies of (the later published) item 4, These duplicate melodies will, of course, count as single ones.
Having made these exclusions, the following number of melodies remain for our purposes: 49 in item 1; 75 in item 2; 30 in item 3; 560 in item 4; 143 in item 5; 50 in item 6; 1,116 in items 7 and 11; 255 in item 8; 1,162 in item 9; 4 in item 10; 5 in item 12; total, 3,449.^
A "Tabulation" of this material has been made, containing in the first column the current number in the grouped material, in the second the number in the original edition, and in the remaining columns, the respec- tive symbols. Since the publishing of this "Tabulation" did not seem feasible, it has been deposited at the Music Library of Columbia Univer- sity for the use of readers interested in details. Hereafter the "Tabula- tion" will be referred to as "Tab. of Mat."
Materials not available to me were, according to Folk, mus., as follows.
PRINTED MATERIAL
Sirola, B., and M. Gavazzi. Obredne popijevke (Ceremonial Songs). Narodna Starina, Vol. XXV. Zagreb, 1931. Twenty-six melodies.^''
Kuba, L. Narodna glazbena umjetnost u Dalmaciji (Folk Music in Dalmatia). Zbornik za Narodni Zivot i Obicaje Juhiih Slavena, Vol. III-IV. Zagreb, 1898, 1899.11
Dobronic, A. Ojkanje. Ibidems. Vol. XX, pp. 1-25. Zagreb, 1915.
CatineUi-Bevilacqua-Obradovic (No title is mentioned!). Twenty-five melodies.
Stankovic, K. (No title is mentioned). Twelve melodies.
Some other titles are cited without the number of melodies; some of the melodies in these collections have been reprinted, however, in items 5 and 10.
* Kuhac even had the misfortune to publish unintentionally one of his songs twice — in two different places and with two different numbers (Nos. 851 and 871).
^ All this material will be referred to as "published material," for brevity's sake, although 11 and 12 are still in manuscript.
^° This is, however, not mentioned in Folk. mus.
1^ Only the second part of this article was at my disposal, therefore, I could not include its melodies in the list of Published Material. Besides, some of them are included in Kuba X. The others, not included anjTvhere, represent such important data that I could not disregard them. The article will be quoted later as "N. g. u. D." (see p. 61).
[25]
Morphology
UNPUBLISHED MATERIAL
There are about 2,900 melodies in various, probably private, deposi- tories.^^ Therefore it can be stated that our material for comparison com- prises little less than half of all melodies existing in notation.
Items 1 and 6 are rather amateurish publications, which have no scientific claims at all. The most important publications are items 9 (Kuhac) , 7 and 8 (Kuba) , and 3 and 4 (Dordevic) .
The oldest of these is Kuhac's collection, made probably in the sixties and seventies of the nineteenth century. He seems to be rather mediocre as a transcriber of melodies and as an observer. In addition, he could not resist the then prevailing fashion of publishing many hundreds of folk melodies with piano accompaniment — sheer waste of time and space, even if the accompaniment were better than it is. The customs of his time may be his excuse, although his Czech contemporaries K, J. Erben, Fr. Susil, E. Peck, and Fr. Bartos, whose works Kuhac as a Slav should have known, rejected this nonsensical method. Their publications are on a much higher level than that of KuhaC in every respect.
The next explorer, in chronological order, is L. Kuba, a Czech painter and well-trained musician, who did his research work on Serbo-Croatian territory around the turn of the century. His work towers high above that of Kuhac, and his contribution of ca. 1,400 melodies to the stock of col- lected Serbo-Croatian folk music material is indeed invaluable, in spite of some idiosyncrasies in his notation. He has a keen sense of observation for certain very characteristic phenomena which almost entirely escaped the attention of Kuhac (line or syllable interruption, "swallowing" the last syllable of a line). In his largest publication, published in Sarajevo (without his consent and collaboration, he says), he tried to group the material on the basis of musical features; in this case, on the basis of scale structures. This material consists of 1,125 melodies, 965 of which have been published. The remaining 160 melodies are still in manuscript. These
^2 There are rumors that unpubHshed material collected by KuhaC exists somewhere and that it is almost as extensive as the published collection. The Folk. mus. gives the following rather obscure information on this question (p. 204) : "Fr. Kuha^ a recueilli un nombre im- posant d'airs populaires: ses oeuvres completes pubhees en partie seulement sont si nombre- uses — leur seule enumeration remplirait une brochure — que leur publication integrale consti- tuerait un corpus de quelque 8,500 pages, grand in -8, sans compter les 3,700 pp. in folio de ses sept livres in folio (dont 4 publics) de la Collection Juznoslovenske Narodne Popievke." This would mean about 1,200 unpubHshed melodies.
[ 26
Morphology
he kindly placed at my disposal in the summer of 1938 by sending me a copy thereof in his own handwriting.
Dordevic, who made his research apparently in the first decades of the twentieth century, obtained excellent results. His collections, items 3 and 4, seem to be the best publications so far. The very short collection by Manojlovic is approximately on the same level. Item 5, of Dordevic, however, seems to have been intended for the general public and there- fore complies less with scientific demands than do the other two.
The modest publication by Busetic is on the whole good, in spite of the indirect transcription of the melodies. He was not a musician; he learned the melodies from the peasants and later asked a musician (Mokranjac) to do the notation from his own singing.
In aU these publications names and data concerning singers, dates of notation (with rare exceptions), and so forth are completely missing. Some of them (items 1,5, and 6) do not even give the names of the villages from which the melodies are derived. All these circumstances, as well as the fact that the notations therein were made by ear alone, must be kept in mind when evaluating their reliability.
SYMBOLS AND FORMULAS
Before being grouped, melodies shall be brought to a common denom- inator. For this purpose they will be transposed in such a way that they get a common tonus jinalis (final tone). As such, g^ appears to be the most suitable (see p. 13).
To facilitate the grouping work, the following system of symbols has been devised.
1. For scale degrees:
have as corresponding symbols: i i lii ]i?'Yii'rii i 23 4 5678^ lb
Accidentals will be placed before the symbol (b3 = 5^b). Auxiliary symbols for designation of caesura (the original conception of caesura, that is, the point between melody sections,^^ will be extended to cover the final note of a given melody section, except, of course, the last one) :
^^ A portion of a melody corresponding to an entire text line, or to its repeated fragment or fragments (see p. 8), will be called a "melody section."
[27]
Morphology
LJ designates the main caesura; J, the one preceding it; L, the one following it. Therefore, 11 Lkll li means a melody of four sections hav- ing c^ for final tone in the first and third sections, h^\? in the second (main) section (mus. ex. No. 41) ; lH] U means a melody of three sections having &^b for final tone in the first (main) section, g^ in the second (mus. ex. No. 18) ; il vkA means a melody of three sections having c^ for final tone in the first section, a^ in the second (main) section (mus. ex. No. 22) ; Iv^l means a melody of two sections having /^ for final tone in the first section (mus. ex. No. 6). The same degree-symbols are used to designate ambitus (range) ; for example : VI 1-4 for a range from f^toc^b (mus. ex. No. 5) ; 1-8 for a range from g^ to g^ (No. 28c).
2. For syllable numbers : italicized Arabic figures, each followed by com- ma; 8, 5, 8, 5, means a melody of four sections, its first and third with eight syllables, the second and fourth with five (mus. ex. No. 41). For isometric melodies (melodies with sections each having the same syllabic number, contrasting with heterometric melodies having sections of unequal syllabic number) the use of one figure will suffice. In case of the splitting of text lines (see p. 37), an additional figure in parenthesis indicates the syllabic number of the original proper text line : {10) 10, 6, 10, 6, means that the text of the six-syllable melody section is a portion of the original ten- syllable text line. In some melodies one (sometimes two) of the sections is composed of two metrically equal (so to speak, "twin") parts; these will be regarded as "double" sections (that is, virtually as one section). This circumstance will be indicated in the formula by having two (equal) figures connected by the plus sign : 8, 8, 8 -\- 8, 8, ox 5, 8 -\- 8, 5, the first one for a four-section melody with the third section double, the second formula for a three-section melody with the second section double (for instance, the melody quoted from Kuba on p. 62).
3. Similarity or difference in the content of the sections will be indi- cated by small-capital letters — the same letter for similar content, dif- ferent letters for different content : A A b a means a four-section melody in which the first, second, and fourth sections are the same, the third different. Some special symbols used in connection with the small-capital- letter symbols are as follows.
A^ (a^, a^), a fifth (fourth, third) higher than a.
As, a fifth lower than a
a^, a variant of a (that is, A slightly changed)
[28]
Morphology
Aa (s = sequentia), A one degree lower
A*, A one degree higher.
Ag A33 Aggs, A repeated three times in a downward sequence,
4. For isometric yet heterorhythmic melodies (heterorhythmic = of unequal rhythm; isorhythmic = of equal rhythm) small letters will be used. The same letter indicates the same rhythm; different letters, differ- ent rhythms: aaba, abab, aabc, and so forth, symbolize hetero- rhythmic four-section melodies.
5. For symbolizing the proportion of syllabic figures in heterometric melodies, the letter ^ will be used in two (resp. three) sizes; ZZ^Z indicates the metrical structure of any four-section melody in which the syllabic number of the first, second, and fourth sections is equal and lower than that of the third section. Thus, this same formula may indicate either 5, 5, 6. 5, or 5, 5, 7, 5, or 6, 6, 8, 6, or 8, 8, 10, 8, and so forth, syl- labic structure; j^Z^, either 10, 6, 10, or 10, 8, 10, or 8, 6, 8, syllabic structure; the formula for this last type but with one double section will be: ^Z + ZZi' Here we shall restrict ourselves to these few examples and must refer to the "Tab. of Mat.," where the reader will find tabulated all the ^ formulas occurring in this material.
The melodies will be grouped, successively, according to the following five principles.
(a) Determination of the sections, that is, the number of sections in each melody. It will be seen that our material has five classes : class A , melodies with one section; class B, with two sections; classes C and D, with three sections; class E with four sections; class C contains melodies with the main caesura at the end of the first section; class D, at the end of the second.
(b) Determination of the length, or rather the proportionate length, of the section. The length of a section is determined by the number of corresponding text syllables. We find that some of the melodies have sections of equal length (isometric structure); others have sections of unequal length (heterometric structure). On this basis we shall divide the B, C, D, and E classes each into two subclasses: I. Isometric melodies; II. Heterometric melodies.
(c) Further grouping of subclass I melodies according to the length of the section. This gives us groups in which each melody section of the song
[29]
Morphology
has a) five syllables, h) six syllables, c) seven syllables, d) eight syllables, e) nine syllables, /) ten syllables, g) eleven syllables, h) twelve syllables in each melody section, and so forth. (Class A will be divided into groups in the same way.) The groups of subclass I will be further divided accord- ing to the rhythmic structure of the sections into two subgroups: 1. iso- rhythmic (all sections having the same rhythm), 2. heterorhythmic (two, three, or four different rhythms for the sections) . Here small letters will be used as S3nnbols (as mentioned under Section 4 of Symbols and Form- ulas, p. 29). Subgroup I. 1. is further divided according to the metrical structure of the section rhythm; more natural or frequent metrical struc- tures will precede complicated ones. For example: 4 -f 4 = 5", precedes 3 + 2 + 3 = 8, (the latter marked in this book as 8b, ") ; 4-\- 2 = 6, precedes 2 -\- 4 = 6; and so forth. ^^ Subgroup I. 2., is divided according to the small letter formulas, and after having established for the latter a certain order of succession: aaba, aaab; aabb, abab, abba; aabc, abac, abcb. The grouping of subclass II is a much more complicated task. It follows the proportionate length of each section. In order to facili- tate the procedure we will use the ^symbols (as mentioned under Section 5 of Symbols and Formulas, p. 29), the heterometric melodies will be grouped according to their formulas. The order of the formulas is to be seen in the "Tab. of Mat.," pp. 39-50, 56-79, 103-135. Melodies with identical ^ formulas will be grouped according to the first syUabic number and then the second: 5, 5, 6, 5, will be placed before 5, 5, 7, 5,; this before 6, 6, 7, 6, and so forth.
(d) Divisions and subdivisions of the subgroups are formed according to the final tone of each section (caesura), the number and bracket S5an- bols (shown under section 1 of "Symbols and Formulas," pp. 27-28) being used in this connection. The order of succession is determined by the following rules: (a) the lower tone will precede the higher; {h) melodies are grouped first according to the final tone of the Q (that is, main) caesura (this gives us divisions) , then according to that of the preceding J, and finally to that of the following L caesura (these give us subdivisions). For example: 3 QJ [2 (mus. ex. No. 40) is placed before 3 HH li (No. 41).
1* b stands here for "Bulgarian," since this metrical structure is very characteristic for Bul- garian folk-poem texts.
1^ Throughout this book, and in the tables also, an arable numeral with a comma after it is used consistently as the symbol for the syllable number of a metric line.
[30]
Morphology
(e) If in a subdivision several melodies show identical S3rnibols, they are grouped, according to their range, into "families." Therefore VII-\}3 comes before VII'4, and the latter before VII-5 (or 6); but VII-4 (or -5, or -6) comes before 1-2 (or -bJ). Should "families" with identical symbols (including range symbols) appear, their order will be arranged according to the circumstances, since so far no systematic method for these has been established. It would lead too far afield to give detailed indications about these; it may suffice to mention two of them. In this material melodies with more lower tones are placed before those with fewer lower tones. For certain types of melodies in another collec- tion, that is, in my "Rumanian Folk Music", a more complicated system was devised.^^
PROBLEMS CONNECTED WITH GROUPING
1. The Serbo-Croatians use 5, 6, 7, 8, 9, 10, 11, 12, 13, and 14 syllabic text lines proper. It is not difficult to recognize and determine them. Two questions, however, arise. The first concerns the thirteen-syllable line. Shall we consider it as a single line, or as two lines, that is, 8, 5,? Disre- garding the melody, these thirteen syllables constitute metrically without any doubt one single line. Their melodies, and analogies in the comparable material of some north Central-Eastern European peoples, lead us to divide such lines, when applied to melodies, into two lines; for example, 8, 5,}'^ To quote an example, the most frequent melody-stanza structure connected with such a line is 8, 5, 8, 5, (some others are 8, 5, LJ and 8, 5, 5, □ D; yet not more than one iJ, H □ E (= [<? + 5] X 4) stanza structure exists in the Serbo-Croatian material.-^^ The other ques- tion concerns what shall be regarded as one text line when parts of the text lines are repeated, for example, when the ten-syllable line appears
^^ See the respective melody groups in Vol. II of said collection, Nos. 53-134, and also the respective description of the system.
^^ In publications consisting only of texts, in which texts are stripped from their melodies (as, for instance, in Part Two of this book) such lines may be either unbroken, as thirteen- syllable lines, or broken up into two lines, one eight-syllable and one five-syllable line.
1* "Tab. of Mat." 1354 (Kuba IX, 2). In the Hungarian material, apart from 8, 5, and 8, 6, text lines, there are real thirteen- and fourteen-syllable text lines, four of them forming a thirteen (or fourteen) Zl dH d melody-stanza structure. A 4 X i4 syllable stanza structure is represented by six melodies in the Serbo-Croatian material: Tab. of Mat. 1355 (Kuha£ 1122), 1356 (Kuhac 1385), 1358a. (Iz Levca 22), 1361 (Kacerovski 6), 1363 (KuhaC 86), 1364 (Kuba XIV, 17).
[31 ]
Morphology
in a fourteen-syllable form as a result of a repeat of its first four syllables? We shall regard these fourteen syllables as two lines {8, 6,), the first of them formed by the repeated first four syllables of the original dekasyl- labic line, and the second by the remaining fifth to tenth syllables.
2, The next problem concerns the occasional text refrain syllables.^^ What shall be regarded as a text refrain? The application of the generally known definition, "recurring phrase or line especially at the end of stanzas," will not be satisfactory if we have to decide what shall be the lower limit of the number of refrain syllables. Three kinds of refrains exist: the content has a certain relation to the content of the text lines proper; the content has no relation to them, yet has a definite sense; the refrain has no recognizable sense at all. The length of types 1 and 2 will generally be not too restricted; type 3, however, may present as short a form as one-syllable or two-syllable words. According to my opinion, a limit must be set : one syllable added in the guise of a refrain should never be called a refrain; two syllables having the same function may or may not be a refrain, according to the given situation.
Syllables of refrains, or of additions in the guise of a refrain, although they do not belong to the text lines proper, are an essential part of the text and are to be included in the syllable count of the text. This will not be the case with nonessential additions of one or two (sometimes even more) syllables; they are "nonessential" because their use is most fre- quently not constant, though in some cases they may appear as such. Thus, they are not to be counted when establishing the syllable number of the sections. The most common of the nonessential syllable additions, and very easily recognizable as such, is the interjection Ej {Hej, Aj, Haj,
Oj, Hoj) at the beginning of a line, for example: J J is changed into
Ml-Ja
J Jj ^"^ (music example No. 9, 1st St., 2d section) .^^ More difficulty is
A), mi-la ^ f } } J J
caused by other, generally two-syllable, additions, put into various places (end of a line or between the metrical parts of a line). These are very characteristic of the Serbo-Croatian and Bulgarian folk songs. The most common are the following three: more (interjection having no specific
^' About Serbo-Croatian refrains see more details in B. Bartok, Rumanian Folk Music, manuscript deposited in Columbia University Music Library, III, xxxix-xli and 316-331.
^^ It is frequently used in Slovak folk songs; somewhat less frequently in some types of Hungarian folk songs; even less often in Rumanian; and very rarely in Turkish material. In this publication these additional syllables are marked by the sign i — i.
" Our music example No. 3 Id presents a very strange case: the repeat of the first essential text syllable is substituted for Oj {Ru-, ruznu instead of O;', ruznu, and so forth).
[ 32 ]
Morphology
sense), aman (Turkish, "mercy, favor"), dzanum (from camm, Turkish, "my soul"). It is not always easy to decide whether they are really non- essential additions or not. We shall consider them nonessential when they appear, so to speak, cursorily, in a more or less hurried way, on compara- tively short tone values; especially when transforming an original J J
into J J J J , and the like. When, however, they appear on heavy
^ — , di^a-num
tones of longer values, in repeats or cumulation (that is, when the corre- sponding part of the melody seems to be an integral portion of the mel- ody), then they are to be regarded as essential syllables. In the first case their use is more or less inconstant; they may be dropped from the follow- ing stanzas and reappear again. In the second case their constancy elevates them to the rank of a refrain.
3. Is the succession of the various points of view, based on the supposed importance of the various aspects of the melodies and indicated on pp. 15- 16, completely satisfactory? So far I have accepted it as satisfactory, other- wise I should have chosen a different one. What I want to point out is this: perhaps later investigators will discover some other points of view or will give a different interpretation to the degree of importance of the already established aspects. The range, for example, as an aspect of the melodies, is placed here at the very end of the rank. However, there are a certain number of melodies of a rather limited range {VII-1, VII-\?2, VII-2, 1-2, l-\?3) which seem to be in very close relation to each other, although their features established on the basis of the remaining aspects are rather different (see pp. 52-54). One would almost be tempted to separate them from the rest of the melodies and treat them as a special class. This procedure, however, would destroy the unity of the grouping system, that is, would intermingle two contradictory systems. Besides, the transi- tion from these narrow range melodies to those with wider range is not at all sharp; the separating of the former from the latter would be arbi- trary in many instances.
4. Instead of dividing the melodies into two-, three-, and four-section melody classes, and then these single classes into isometric and hetero- metric subclasses, one could first divide them into an isometric and a heterometric class and then divide the two classes into two-, three-, and four-section melody subclasses. This order of succession would also be justifiable.
5. The last problem concerns the occurrence of repetition of given sections in the melody. We must determine whether the repeat is an
[33]
Morphology
essential structural feature (a a) , or whether it is but a result of some individual whim. In many instances it is not so easy to determine this. Sometimes wearisome comparisons with related melodies will lead to the right decision; sometimes tradition will give valuable indications.
6. Finally I must mention in this chapter a negative feature, the absence of scale aspect as a determining factor in the grouping system. Many people who attribute great importance to this factor will be sur- prised by the statement that it cannot be used at all as such in the Serbo- Croatian material, as well as in some other Eastern European material, because of its inconstancy. The pitch of certain degrees, especially that of a^ and b^, is subject to such fluctuation, even in the performance of a single person, that sometimes it is impossible to determine whether the scale purports to be a minor or a major scale. Neutral degrees are abund- ant; they themselves may be fluctuating in intonation. Variant groups representing three or four different scales are frequently found.^^ Besides, the overwhelming majority of the Serbo-Croatian material shows only scale fragments, that is, a pentachord or tetrachord or even less, due to the very restricted range of the majority of the melodies (see pp. 52ff .) . For instance, what kind of scale could a "scale" of g^ a^\? degrees be called P^ All this does not mean that aspects of the scale should not be investigated; it means it cannot well serve for grouping purposes. More important than scale is the relation between range and final tone, for example, a range of VI 1-4, with main accents on the /^, a^(b), and c^ tones and with final tone g^, is very characteristic of Serbo-Croatian and Bulgarian folk music (see pp. 52-53).
GENERAL CHARACTERISTICS OF THE SERBO-CROATIAN FOLK SONGS
STANZA STRUCTURE IN THE TEXTS
When we proceed from Slovakia^'* and Hungary toward the southeast, that is, to Rumania, we perceive a rather peculiar change. The texts of folk
22 "Tab. of Mat." No. 123 has in the 1. variant (Dord. 173) f g^ a^ t^b, in m. (Dord. 153) /' gi a^b b\ in o. (Dord. 301) /« g^ a^ b^\,, in v. (Iz LevCa 72) g^ a^ b^\? cH^) d^.
23 In grouping the folk-song material of some other peoples, for example, the Hungarians, the scale aspect may become a determining factor in some cases, though even then it will be less important than the range aspect.
2* Here and in the following text, "Slovakia" stands for "Slovakia and Moravia," for brev- ity's sake (Moravia is situated between Bohemia and Slovakia) ; "Rumania" stands only for
[ 34 1
Morphology
songs in the first-mentioned two countries present almost without excep- tion^^ decided stanza structure (that is, a more or less symmetrical, recur- ring structure of two, three, or four text lines) connected with the use of rhymes; the text stanzas are predominantly composed of four lines, cor- responding to the four sections of the melody; a substantial part of those melodies which are not sung except on special occasions (weddings, har- vests, and so forth) have no special group features.^^ In Rumania, sud- denly, the text-stanza structure disappears (except in songs of urban or semirural origin) though rhymes are still in use ; the four-section melody structure is not predominant, since a rather important part of the mate- terial has a three-section and a not negligible part a two-section structure ; the melodies divide into primary classes according to the different occa- sions at which they are sung, each of these primary classes having their specific musical characteristics.^^ Going farther south, that is, to Bulgarian and Serbo-Croatian territory, we find again unequivocal changes: no more text-stanza structures and no rhymes (except in songs of urban or semirural origin) ; two-thirds of the Serbo-Croatian melodies have two- section or three-section structures, much of the remaining third — with four-section structures — is of urban, semirural, or foreign origin. Here, as against the Rumanian material, no primary musical classes would result from a grouping of the melodies according to their function. All melodies, however, have stanza structure. When we use the expression "text stan- zas" or "absence of text stanzas," we refer to the form of the texts stripped from the melody. If the lines of the song text keep a symmetrical, recur- ring arrangement, then obviously there exists a text stanza structure (which generally tallies with the melody stanza). But if, stripped from their melodies, they present no regular arrangement and only structurally independent text lines remain, it would be senseless to call them text stanzas. In the latter case the single text lines will be adapted to the
those parts of that country which belonged to Hungary before the first World War. The folk music material of this region is fairly well known from various publications, whereas the very rich material of the remaining sections of the country is stored on records, for the time being, in various museums of Rumania, not having been published yet (with very few exceptions).
^ As an exception, the texts to the Hungarian mourning songs may be mentioned: im- provisations based on certain patterns with text lines being rather in prose or reminding us of free-verse structure, having no rhjrmes at all.
''^At variance with this rule, the melodies of the Hungarian mourning songs, children's plays, and winter-solstice songs, as well as of the Slovakian children's plays and some harvest songs, and so forth, have distinctive characteristics.
^' Songs performed on certain occasions will be called "Ceremonial."
[35]
Morphology
melody stanzas in various ways, being originally probably only one text line in various forms of repeats to a given melody stanza.
All folk poetry of Western Europe, Slovakia, Hungary, and perhaps Ukrainia has texts in stanza structure; absence of text-stanza structures is, however, characteristic of Rumanian, Serbo-Croatian, and Bulgarian folk poetry .2^ The folk song editors of the latter peoples clearly perceived this phenomenon and accordingly published their text material, when stripped from the melodies, without breaking it up by stanza subdivisions. The first impression created by these facts is that text stanza structures originated in Western Europe, perhaps in its urban poetry, and then spread gradually towards the East, pervading Slovakia and Hungary, and that absence of text-stanza structure has its origin in the Near East, perhaps in the ancient Greek folk poetry (as for instance in the ancient Greek heroic poems). Such an assumption, however, is at variance with the result of investigations in Turkey. My rather short research work in middle and southern Asia Minor and that of others has shown that the Turkish folk poetry knows no other form but a very definite text-stanza structure, with rhymes and generally with four-line stanzas. This obvi- ously could not be ascribed to Western European influence; evidently it belongs to the nature of the ancient Turkish folk poetry of central Asia. Therefore the Hungarian text-stanza structure of the so-called "Old" Hungarian melodies may not be considered as of Western European origin; it is, indeed, a remainder of Old Turkish cultural influence.^^
It remains an open question for the time being whether the absence of text-stanza structures in the folk poetry of the above-mentioned peoples has its origin in the ancient Greek folk poetry or in some other sources.
MELODY-STANZA STRUCTURES
As has been pointed out, almost three fifths'" of the Serbo-Croatian material belong to the two-section and three-section melody class. Com-
28 This is probably true concerning Greek and Great-Russian folk poetry, too.
^^ Hungarian folk poetry has also some other traits in common with Turkish folk poetry; the whole old stock of the Hungarian melodies originates from an old Turkish musical culture of central Asia, as recent discoveries prove. See B. Bartok, "Turkish Folk Music from Asia Minor," manuscript deposited at the Music Library of Columbia University; pp. VI-VIII, XX-XXI, XXIII-XXIV, XXVII-XXX.
^° Ca. 2,000 (ca. 1,200 two-section and ca. 800 three-section melodies) out of ca. 3,450.
[36]
Morphology
ing from Hungary and Slovakia, we carry with us the idea that two-section or three-section melodies are generally fragments of some known or unknown four-section m^elodies; this seems to be justified by the com- parative rarity of these shorter structures there and because investiga- tions based on comparison within the material show that some of them ARE IN FACT fragments of longer structures. This assumption cannot be maintained when we get farther to the southeast, especially to Serbo- Croatian (and Bulgarian) territory. Even there we may find that shorter forms are fragments of longer ones ; the great bulk of the two-section and especially of the three-section melodies, however, are evidently inde- pendent formations. In dealing with this question we run into some diffi- culties. For instance, repeats of both sections of a two-section melody or of only one of them are frequent. If we regard these as structurally essen- tial repeats, then the melody in question will have to be considered as a four-section or three-section melody. Comparisons and analogies may guide us in deciding this question.
Isometric structures are overwhelmingly common in the Hungarian folk melodies, in both the so-called "Old" and the ''New" styles (in the latter, original isometry is frequently transformed into heterometry) . The same is true regarding Rumanian folk melodies, except the so-called Colinde (winter-solstice songs) — and some melody structures of the Banat area adjoining the Serbo-Croatian territory to the north. The apparently oldest part of the Slovak melodies and the Turkish melodies have the same trend. The Serbo-Croatian melodies, however, have a leaning toward heterometric f orm-S ; as a matter of fact, the structures most characteristic for Serbo-Croatian folk m.elodies are heterometric. This trait, among others, draws a sharp line between the autochthonous melody classes of the Hungarians, Rumanians, and Turks on the one hand and the Serbo- Croatians on the other. And it is this trait which is in innermost connec- tion with a most peculiar treatment of the text lines — the splitting of the isometric text lines in order to make them applicable to heterometric melody stanzas. ^^ A similar phenomenon has not yet been observed else- where, or only sporadically, as far as I know, except in the Bulgarian material which has many traits in common with the Serbo-Croatian.^^
'^ This feature has been already observed and mentioned as an outstanding characteristic by B. Sirola in his article in Zenei Lexikon ("Music Lexicon"), Budapest, 1931, under Szerb- horvdi-szlavon zene ("Serbo-Croatian and Slovenian Music"), Pt. II, p. 557, col. 2. Incidentally, there are cases in which line fragments are applied to isometric melody stanzas, too.
^' A sharp line can hardly be drawn between the folk music of these two peoples.
[37]
Morphology
Before describing this most important characteristic in extenso, I must give a short survey of the metrical peculiarities of the various text lines. Serbo-Croatian and Bulgarian folk poetry (as well as Hungarian, Slovak, Ukrainian, Turkish, and, in a different way, Rumanian^^) uses text lines having metrical attributes which are based on the recurrence of syllable groups as metrical units, according to fixed schemata; the first syllable of these units always bears a verse accent which does not neces- sarily TALLY with the spoken word accent. This syllabic verse structure completely disregards the natural or positional length of the syllables, a
SPECIAL FEATURE IS THE ABSENCE OF UPBEATS.^^
The main schemata are as shown on page 39. The order of the syllable-group schemata in each section (marked with a),b), and so forth) indicates their frequency, proceeding from the more frequent ones to the less frequent.
There are rarely more than fourteen syllables; when found at all, they are usually in urban texts.
According to Dr. R. Jakobson's investigations^^ there are certain rules concerning the metrical structure of the dekasyllabic line in the Serbo- Croatian heroic poem, 4\\ 4\ 2, g)l. The principal rule deduced by him states : the main caesura never divides a word (that is, the fourth syllable of the line must be the last syllable of a word and the fifth of a line the first of a word). The secondary caesura (after the eighth syllable) does not follow any rule in the cutting of words, although the corresponding musical section has a. 4 -{- 2 structure almost without exception. This Jakobson rule applies on the whole to the dekasyllables of the lyrical and other texts as well, therefore, no further investigations are needed con- cerning these. I tried to examine, however, the simple lines b)l., c)l., d)7., e)l., e)2., f)l., h)l., and i)2. in a certain number of melodies (that is, in Dordevic and in our music examples), with the following results:
in e)l. 8 syllables = 4\\ 4, the caesura never cuts into a word (in 139
texts of Dordevic, and in 13 of our texts) ; in d) 1 . 7 syllables = 4\\ 3, the caesura never cuts into a word (48
Dordevic, 1 of our texts) ;
'' The Rumanian folk poetry shows only eight- and six-syllable text lines proper (that is, refrain text lines excluded).
^^ Some exceptional examples for upbeats exist in the Serbo-Croatian heroic poem texts and in the Rumanian material. — See also p. 80.
^ R. Jakobson, "tJber den Versbau der serbokroatischen Volkepen," Archives Neerlan- daises de Phonetique Experimentale, Amsterdam, 1933, Vols. VIII-IX.
[38]
a) b)
Morphology
Table 1
Metric Schemes oe the Lines
Simple Lines
4 syllables (very rare)
1. 5 syllables = 3
2. 5 syllables = 2
3. 5 syllables = 4
c) 1. 6 syllables = 4
2. 6 syllables = 3
3. 6 syllables = 2
d)l. 7 syllables = 4
2. 7 syllables = 3
3. 7 syllables = 2
e) 1. 8 syllables = 4 2. 8 syllables = 3
f) 1. P syllables = 4
9 syllables = 3
g) \. 10 syllables = 4
2. 10 syllables = 4
3. 10 syllables = 5
h) I. 11 syllables = 4
2. 11 syllables = 4
i) 1. 12 syllables = 4
2. 12 syllables = 4
3. 12 syllables = 4
J)
Composite Lines
4\\5^ 3\\4\3
k) i4 syllables = 4\ 4\\4\2
13 syllables = 4 13 syllables = 3
2-
3
1
2 3 4
3
4 (rare)
3\\2 (only one example: Tab. of Mat. 816)
J^
2\\3 (rare) 4 1 2 (rare)
4 I 2 (very frequent)
31 3
5'
3
4\1
4\2^
2
4\ 4 3\\4\1
" The double line indicates the position of the main caesura, the single line of the secondary. '' The scheme e)2. is very frequently found in Bulgarian, less often, but still frequently, in Serbo-Croatian material.
* g)3. is, in fact, a doubled five-syllable Une and becomes a ten-syllable line only if connected with such a melody-section.
^i)l. probably is a doubled six-syllable line, though this is not always discernible.
• Group 5 inj),l, may be subdivided as the five-syllable hnes, b) 1.-3.
[39]
Morphology
in h)l. ii syllables = 4 | 4 || J, neither of the two caesuras cuts into a
word (11 Dordevic, 3 of our texts); in e)2. 8 syllables = J | 2 || J, the main caesura never cuts into a word,
the secondary seldom.
The proportion concerning the secondary caesura is: in 71 Dordevic texts, approximately 2 percent of the text lines have a wordcut at the secondary caesura. The words in the respective text portion are mostly
2 -\- 3 syllable words (the musical structure remains, nevertheless, 3 -\- 2); in three text lines 4-\r 1 syllable, and in one single line 1 -\-4 syllable words; in all five texts in our collection, neither the m.ain nor the secondary caesura cuts into a word.
In b) 1 . 5, = J 1 1 ^, and q)\. 6, = 4\\ 2, there is no prevailing rule for the placement of words, since 3 -\- 2, 2 -\- 3, 4 -\- 1 in b)l., and 4 + 2,
3 -\- 3, 2 -\- 4 in c)\. are intermingled without any regularity, although there is a slight tendency toward 3 -{- 2 in b)l., and especially toward
4 -\- 2 inc)\. For the remaining few formulas no positive results could be obtained because of the scarcity of examples; two f)l. 9, = 4\\2\ 3, and one i)2. 12, = 4 \\4\ 4, which were at our disposal show no wordcuts at the caesuras.
If we extend our examination of the rules and absence of rules in the 5- to 11-syllable lines, we perceive certain parallels. The six-syllable lines seem to possess exactly the same metrical qualities as the second half of the dekasyllable (syllables 5-10); the eleven-syllable lines may be con- sidered an extension of the 8, = 4\\4 syllable lines or of the seven-syllable lines; the five-syllable lines show a certain analogy with the first half of the ^, = J I 2 1 1 J syllable lines (although the latter has much more regularity) :
11,=
7, =
The composite text lines oil3 {= 8 + 5), 14 {= 8 + 6), and 15(8 +7) syllables were examined only in our collection. The components show the same rule or irregularity as the corresponding simple lines: 8 and 7 invariably with word limit at the caesura; 5 and 6 irregular. In 13, = 8 -\- 5 there is a tendency for 4 -f i in the five-syllable part (word limit between the fourth and the fifth syllables, the fifth syllable representing a one-
[40]
Morphology
syllable word of a certain weight; for example, mus. ex. No. 43, melody St. 1-6) . This is evidently called forth by the rhythm of the cor- responding portion of the melody which is very frequently J3T3 [ J || .
The same rules concerning the cutting of words are observed in Bul- garian, Slovakian, Czecho-Moravian, and Ukrainian folk song texts. But if we turn to the Hungarian texts, we see that although there is a tendency toward ending words at the caesura points in seven-, eight-, and eleven- syllable lines, there are so many exceptions that we cannot assume the existence of established rules. This irregularity is still more prevalent in Turkish texts. What is the explanation of this phenomenon? One possible guess — at least concerning the 8, = 4\\ 4 lines — is that these were gener- ally sung to parlando-ruhato melodies, where the caesuras are less marked than in melodies with rigid rhythms. The weak point of this guess is that it does not very well apply to the 7, = 4\\ 3 and 11, = 4\4\\ 3 syllable lines, which are generally sung in rigid rhythm.
There remains another possible guess, a linguistic one. Both the Hun- garian and the Turkish languages are agglutinative languages and there- fore are apt to use comparatively longer words (especially the Turkish in its agglutinated infinitives and participles) than the inflecting Slavic languages.^^
The Rumanian text lines, based on a somewhat different metrical sys- tem, cannot be used for comparison.
The aforementioned splitting of text lines will occur at one of the caesural points (marked with | and 1 1 , the letter for the main caesura) .^^ The separated line fragment is repeated at various places, according to the structure of the melody. Hundreds of combinations exist.^^
The addition of refrain lines to the isometric text lines proper is another means of making the texts adaptable to heterometric melodies. This course is com.mon and rather well known in the folk-song material of other
^ See more details in "Turkish Folk Music," pp. xxiii-xxiv.
^^ There are rare although very curious, almost baffling exceptions: No. 446 and 1396 of "Tab. of Mat." (Kuba XI. 5, 11, Dordevic 207, 251 and Kuba XL 19. B. H. 853, 854); the latter has : Poseta-, | posetala !
^* Such line fragment repeats may well be distinguished from refrains proper. The fact that A — in our case a refrain — may be interchangeable with B — in our case the repeat of a Hne fragment — does not result in a development of similar features in A and B. Indeed, refrains and text-fragment repeats are actually antithetical features: the first represents never chang- ing repetitions, the second ever changing repetitions. The use of the first involves repeats occurring not within the melody stanza but permanent during the whole piece (that is: not changing from melody stanza to melody stanza). In the first case parts of the main text operate; in the second, text parts which do not belong to the main text.
[41]
|
Morphology peoples, therefore I shall not dwell upon its description. Instead, I shall give a short tabulation of the combinations which are the result of text- line splitting and occur more than once. Frequently the two procedures are combined, that is, refrains are added and the lines are split. In Table 2 schemata showing such combined procedures are also included. The table refers to the printed material and to our group of songs. Table 2 |
|||||
|
Syllables of the per- manent text-lines |
One- Section Melodies |
n Isometric Melodies |
n Heterometrio Melodies |
3 O Isometric Melodies |
n C Heterometrio Melodies |
|
(7) |
1.) ii,=4+7e [two] |
2.) 10, = 7+31 7+3 n [five] |
|||
|
(8) |
3.) 12,=8+4I8+4|I [two] |
6.) 12,8,=4+8l80 [three] |
|||
|
(8b) |
4.) 11,=8+3|8+311 [two] |
7.) 6,5,=3+3l50 [two] 8.) 8,5,=815|1 [two] 9.) 11,8,=8+3|80 [four] |
18.) 8,8,5.=8i8l5B [or (2r.)+3] [two] |
||
|
(10) |
5.) 6,=4+2(l.-2.)l6|| [two] |
10.) 8,6,=4+4l6ll [sixteen] 11.) 10,6, = igl6|| [twenty-three] 12.) 7,10,=IllOll(sic!) [two] 13.) 8,10,=4+4I10|1 [five] 14.) 8,10,=4+(r.4)|10|l [six] |
17.) 6, = 4+ (1st r.2) 04+ (2nd r.2) 1 611 [two] |
19.) 10,10,6,=101110|60 [two] 20.) 8,6,6, =4+4ll6l6D [twenty-nine] 21.) 8,6,6,=4+(r.4)fl6l6D [two] 22 ) 10;6,6,=10lli+(r.2)l6D [two] 23 ) 10',8,6,=10D4(l.-4.)+4(l.-4.)|6! [four] |
|
|
(11) |
15.) 8,7, = 8l4(5.-8.)+_3|| [three] 16.) 8,7,=4+4l70 [two] |
||||
|
Total of types |
1 |
4 |
11 |
1 |
6 |
|
[42] |
Morphology
Explanation : — The first figure (in parentheses) gives the syllable num- ber of the (isometric) text lines. The second figure or figures (each followed by a comm.a) give the syllable number of the melody sections. Then follows the schema showing the treatment of the text lines, by splitting them up and repeating the fragments. The signs | or 1 1 indicate the end of the sections. The underlined figure or figures indicate the entire
Melody Stanza Structures
|
2 D Heterometric Melodies |
jnC Isometric Melodies |
DDL Heterometric Melodies |
Total of types |
|
2 |
|||
|
2 |
|||
|
29.) 6,5,6,5,=3+3l5l|34-3|5l| [nine] |
6 |
||
|
24.) 7,7,10.=4+(l3tr.3)14+(2ndr.3)||10ll [two] 25.) 10,6,10= 10l6Dl0i [thirteen] 26.) 8,6,10,=4+4l6|110| [nine] |
28.) 14,=4+1014+10||4+10|4+100 [two] |
30.) 10,10,13,13,=10|10[|(r.7)+6|(r.7)+6|l [two] 31.) 8,6,8,6,=4+4l6l|4+4l6|| [thirty-nine] |
18 |
|
27.) 8.6,ll,=8l3+3|ina [two] |
32.) 8,6,8,6,=8l3+3||8|3+3i leTght] 33 ) 8,7,8,7,=4-|-4l704-l-4l7|I [four] |
5 |
|
|
4 |
1 |
5 |
33 |
[43]
Morphology
main text line which can either fall within one melody section or may be distributed over two sections. The not underlined figure or figures pre- ceding an underlined figure indicate the syllables of a repeated line frag- ment which appear before the entire main line and have been taken from its first syllables. The not underlined figure or figures following an under- lined figure indicate the syllables of a repeated line-fragment appearing after the entire main line and taken from its last syllables. If these line- fragment syllables are taken from a different portion of the main line, this circumstance is shown by figures in parentheses which immediately follow and show the position of the syllables in the main line (that is, '' 10\\4 (1-4) + 4 {1-4) I (^ II" means that after the entire main line \10] its first four syllables are twice repeated).
Refrain syllables are put into parenthesis and bear the discriminating letter r. The number of melodies or variant groups in which the schema occurs is indicated in brackets. The combinations which occur only once are enumerated in Appendix I of "Tab. of Mat."
Relations between some of these schemata can easily be discovered. For example, 11.) {10 \6\\) may be considered as a shortened form of 25.) {10_ \6\\ 10), or 25.) as an enlarged form of 11.); a similar relation exists between 32.) and 4.); 32.) is the double of 7.), 31.) of 10.), 33.) of 12.).
Some interdependence appears between the figures of the first column and some of the schemata, too: 31.) originates from {10) text lines, 32.), however, from {11), although the syllabic symbol for both is the same: 8, 6, 8, 6,.
Some of the schemata have an influence on the musical content of the melody sections, inasmuch as the repeated text fragment carries with itself the repeat of the respective section fragment of the melody. This happens for instance in almost all melodies of schema 25.). The melody- fragment may be repeated either without any change, as in melodies having the same final tone in sections 1 and 2, or with a change of cadence (that is, of the final tone).
This leads us to the description of an extremely strange group of schemata (not quoted in Table 2) most characteristic for Serbo-Croatian melody-stanza construction. In this group the arrangement of the text line fragments runs parallel to that of the melody-section fragments, that is, the former are determining factors of the melodic content structure, and originate a rather peculiar form of melody stanza. The syllabic
[44]
Morphology symbols of the most frequently occurring structure of this kind are
{10) 4,0,6,4, , the symbol of the melodic content-structure a :^ a (or
some variants of this) . The sign ^-v-- means one text line. That is, a dekasyl- lable ("^ 1 1 <^) is sung to the first half of the melody, and then the same dekasyllable is sung to its second half, this time, however, in inverted order: first comes the portion of the line standing after the main metrical caesura (syllables 5-10), and then follows the portion preceding it (syl- lables 1-4). The repeat of syllables 5-10 and 1-4 coincides with the repeat of the respective portions of the melody section, either without essential changes or in a more-or-less altered form. Therefore this peculiar treatment of text lines is an originator of a melody-content structure. Most of the less frequent structure schemata of this kind are in relation to the afore-described one, being its abbreviation or extension ; a common characteristic for all these is the placing of the repeat of the first text syllables (that is, of the first metrical unit) at the end of the melody. In view of their importance it appears useful to enumerate here all instances of these structures, published elsewhere and in this book:
34.)39 {10) ^AM,: Kuhac, 540, 736, 795, 892, 1208; Kuba B. H.. 193.
208, 295, 339, 349, 360, 406, 429, 493, 499, 702, 716, 723, 724, 872, 873, 875, 928, 1003, 1093-1096; Kuba, IX, 52; XI, 1, 30, 31, 36, 51; Dordevic, 365, 450, 483; Dordevic Nar. Pev., pp. 29 (second melody), 108 (second melody), 186 (second melody); Bosiljevac, 16, 37; our examples 31 a.b. c.d., 32, 33. [42 -\- 6 melodies in 15 -j- 2 variant groups.]
35.) {11) 4-\-4 + 3^3, 4,: Kuba B. H., 392. 36.) {13) 4 + 4 + 5,4,: Kuba B. H., 480.
37.) {8) 4^4^y, (^+Z^^^^ and ^^' ^^^^^' ^^ Y
^^-^3^-x^ Dordevic, 546; Kuhac, 879, 880, 881.
38.) {8b) 5 + 3 + 3,5,: Kuhac, 660; Kuba, B. H. 467.
39.) {8b) f+1^5,: Dordevic, 307.
^^ This and the following similar numbers continue the numbering from Table 2.
[45]
Morphology 40.) (Sb) /+5^,: Kuhac, 1271.
41.) (8) 2-\-2 + 4,2,: Kuhac, 1069.
42.) (10) f+J^2,: Dordevic, 305.
The overwhelming majority of these extremely characteristic melody types come from Bosnia, Hercegovina, and Serbia; this is one proof of the assertion that these territories — and in addition Montenegro and Old Serbia — are the real sources of autochthonous Serbo-Croatian folk songs. It is significant that Kuhac contributed to this type a comparatively small number of melodies (eleven, as against thirty-one from Kuba and nine from Dordevic) ; it looks as if he neglected in his research, for mysterious reasons, to investigate the areas richest in aboriginal material.
Schemata 10.), 11.), 20.) and 25.) of the heterometric two- and three- section melodies are still more numerous in the above-mentioned areas, and a great majority come from there. They must, therefore, also be regarded as characteristic Serbo-Croatian melody types. The number of melodies is as follows :
10.) (10) 8, 6,: 4: melodies from Kuhac, 31 from other collections, in 16 variant groups
11.) (10) 10, 6,: 4 melodies from Kuhac, 46 from other collections, in 23 variant groups
20.) {10) 8, 6, 6,: 4: melodies from Kuhac, 57 from other collections, in 29 variant groups
25.) (10) 10, 6, 10,: 9 melodies from Kuhac, 68 from other collections, 9 among our music examples; in 13 variant groups
Total : 21 melodies from Kuhac, 204 from other collections, in 81 variant groups
Here, Kuhac 's contribution is even smaller!
It seems that the original procedure was to use only one text line (one composite text line in structures like in 8, 5, 8, 5, or 8, 6, 8, 6, and so forth) for the entire melody stanza: the text line, or its parts, being repeated as m.any times as was required by the melody stanza. This pro- cedure is still used in the majority of cases. Later, however, it became the fashion to use a given text line only once. The two kinds of procedure frequently get mixed, and in this case the repeats or nonrepeats of a text line by no means follow established patterns, but seem rather due to the impulse of the moment. This appears clearly when pieces are recorded
[46]
Morphology
twice by the same singer : the place of text line repeats is not the same on the two records (see the last Note to mus. ex. No. 12a., p. 234).
PRINCIPLES OF rOEM CONSTRUCTION
No form construction occurs in one-section melodies. In fact, most of these (Nos. 1-35 of "Tab. of Mat.") may be fragments of structures of more than one section which have not yet been discovered. New material will no doubt clarify many of these cases (for example, since the "Tab. of Mat." was completed it has been discovered that our mus. ex. No. 17 is a complete two-section form (a a^) of No. 34 (Dordevic 8) in the ''Tab. of Mat."). Nevertheless, the statement that one-section melodies do not exist would not be true: our music example No. 1 seems to be such a melody, because one section is repeated (as long as the text lasts) without
CHANGES THAT WOULD LEAD TO GROWTH OE EORM. This is SO although
there are two-section melodies which are so similar to this example that they may well be considered variants of this one: "Tab. of Mat." No. 146 (Dordevic 168, 183, 211) and No. 310 (Kuba B. H. 319, 321, 825). One must keep in mind, however, that because of the imperfections in the notation, the real form of the melodies frequently is not sufficiently clear in transcriptions made by ear.
Two-section melodies are frequently created by repeating a section, with minor changes (aA(v) from a). This structure is especially charac- teristic of Serbo-Croatian autochthonous melodies. There are various devices for effecting a change in the second section. The simplest is to prolong the rest at the end of the second section, whereas the two sections remain essentially identical (mus. ex. No. 8, 10, 16a) .^° The published collections do not disclose this method, because the collectors neglected to measure even approximately the rest at the end of the stanzas. Without recording melodies nobody could ever discover this rather peculiar prin- ciple of form building.
Another method, always connected with the first one, consists of a slight change in the second section at any place except the last tone ; for example, changing a |JJJJ bar into |J J J^ (mus. ex. No. 17) or interpolating a rest somewhere (mus. ex. No. 14).^^ This method, too, is
*" The same method is sometimes used in order to differentiate the two halves of four- section melodies (Nos. 37, 38).
" The same method is sometimes used in order to differentiate the two halves of four- section melodies (Nos. 40, 49).
f 47 1
Morphology
fairly typical. A third method, connected with the first also, consists of changing the ^M, E^, Ekll], \A\, and so forth, final tone of the first section into final tone 1 in the second section. This procedure reminds us of the method used in the construction of periods in art music. A fourth device is seen in many melodies in which no repeat occurs, the second section having an entirely new content (a b) (mus. ex. Nos. 3, 5, 7). This structure is less characteristic. Finally, there are melodies in which the content of the two sections is similar — the position of the second, however, is one degree lower or higher (sequential structure a Ag or a a^; for example, "Tab. of Mat." 161 (Kuhac 550), 229 d. (Kuhac 470) ; 1081 (Kuhac 1319). This structure is of western origin; the melodies are of urban or foreign derivation and frequently only fragments of longer sequential structures. Table 3 shows the content structure of two-section melodies in the material (including the melodies in this book). In this table groups 7,1 8b, I 10, I 11, I 12, 1 and 10, 6,\ deserve special attention as containing the largest number of A A(v) structures. The last of these is related to another typical form, the three-section form 10, 6, 10, (see p. 46, above) ; this explains the frequency of the A A^^^ structure in it (45, against 5 A b) (see Table 3, p. 49).
Isometric three-section melodies will show an enlargement of the above- described structural principles. We shall not dwell upon the structure ABC, less interesting from the point of view of form-developing methods. A B B(v) structures in isometric l_J L melodies and A A(^) b structures in isometric J l__| melodies may be interpreted as derivations from A b ; AAA(v) as derivations from aA(^), or inversely. The very rare a b A(^) (in four melodies) may be brought into relation with four-section a b A(v) b structures.
The statistical data are as follows (including the melodies in this book) : Of 57 isometric LJ L melodies, 16 have A b c; 31, A b B(y) ; 3, a A(^) a;
3, A B A^. Of 105 isometric J \Z\ melodies, 28 have a b c; 50, a A(v) b; 13, a a a^^) ;
1, A B A^. The less frequent a a^^^ a and a a A(v) seem to be the most character- istic; in fact fourteen of them come from Bosnia, Hercegovina, and Serbia.
[48]
Table 3
|
Syllables of the Sections |
A B |
AA(^) |
Sequential Structure |
A^A |
A^^A |
Total |
|
|
5, |
14 |
1 |
15 |
||||
|
6, |
30 |
22 |
13 |
65 |
|||
|
7, |
22 |
|22| |
1 |
45 |
|||
|
8, |
132 |
i69j |
3 |
1 |
205 |
||
|
Of 898[_ isometric |
8b, |
34 |
|54| |
6 |
2 |
96 |
|
|
melodies |
9, |
2 |
3 |
5 |
|||
|
10, |
131 |
|148| |
3 |
282 |
|||
|
11, |
29 |
|44| |
72 |
||||
|
12, |
5 |
|26| |
31 |
||||
|
13, |
1 |
1 |
|||||
|
Total |
399 |
390 |
25 |
3 |
1 |
818 |
|
|
8,5 |
9 |
5 |
14 |
||||
|
Of 192 hetero- metric melodies |
8b, 5, |
3 |
1 |
4 |
|||
|
8,6, |
53 |
7 |
60 |
||||
|
with Z^ Z struc- |
8, 7, |
19 |
4 |
23 |
|||
|
ture |
10,6, |
5 |
|45| |
50 |
|||
|
All others |
25 |
16 |
41 |
||||
|
Of 66 1 1 hetero- metric melodies |
7, 11, |
9 |
2 |
11 |
|||
|
8,10, |
17 |
10 |
27 |
||||
|
with Z^ struc- |
8, 12, |
6 |
6 |
||||
|
ture |
All others |
9 |
13 |
22 |
|||
|
Total |
155 |
103 |
258 |
[49]
Morphology
Beyond any doubt some of the heterometric LJ L structures are Western European, especially the groups 2)2.-6., 11. of "Tab. of Mat." (pp. 64-66) : of 59 examples, there are 28 from Kuhac, 3 from Kuba IX, XI, that is, probably more than the half came from the western parts of Serbo-Croatia. Structures showing complicated syllabic proportions (groups 3)2.-6. of ''Tab. of Mat.," pp. 67-70) probably are of foreign origin, too; the same may be said about melodies of similar structure in class _l LJ.
Syllabic structures 8, 5, 5, | 8, 6, 6, \ 8, 7, 7, \ and 10, 6,6, |, mostly, of course, with a b B(^) content structure, may be regarded either as enlarge- ments of 8, 5, \ 8,6,\ 8, 7, I and 10, 6, \ or as shortened forms of 8, 5, 8, 5, \ 8, 6, 8, 6, I 8, 7, 8, 7, \ and 10, 6, 10, 6, |.
Heterometric examples in class J[ |_J contain more autochthonous structures. The most important of these is the already mentioned syllabic structure 10, 6, 10, (p. 46), mostly with aa^b content structure. Of 75 melodies belonging to only 14 variant groups 6 have: a a^, k^2', 3 have A B c ; 65 have a a^ b (Kuhac contributed 1 1 melodies) .
The most popular melodies belong to this group: No. 983 of "Tab. of Mat." comprises 36 variants (in our melodies 5 more: No. 27); No. 984 has 11 (in our melodies 3 more: No. 28); No. 989 has 8 variants. All the 175 melodies come from the above-described "autochthonous" territories, except for Kuhac 703 (from Backa) and 1510 (from Croatia).
Among the other structures, 8, 8, 5, \ 8b, 8b, 5, \ 8, 8, 6, \ 8, 8, 7 , \ and 10, 10, 6, I seem to bear a relationship to the corresponding four-section structures which is quite similar to that between the afore-mentioned four groups in class LJ L and the latter.
It might be well to mention at this point some data concerning the main caesura tones in the two- and three-section melodies: about 34 percent have UJ, 17 percent Lkll, 16 percent Eklil, and 14 percent lYIll; the other 17 percent are split between VIA, \AA, LiJ, and so forth. The pre- ponderance of the four lower degrees, first mentioned, is in close connec- tion with the narrow range of the melodies, see pp. 52-53.
The class of four-section melodies offers the greatest number of bor- rowed foreign melodies, especially the heterometric subclass. The rich
[50]
Morphology
contribution of Kuhac to these classes, especially to the isometric one, is rather significant: of approximately 550 melodies, 280 are from Kuhac. An important part of these melodies come from the western portions of the Serbo-Croatian territory. Only a few of these will be described here.
Some ten-syllable isometric melodies (No. 1310 of "Tab. of Mat.," comprising 20 variants, 15 of them from Kuhac) seem to be "heroic- poem" motifs, misrepresented and wrongly squeezed into a form of four or more sections by editors. Thus, perhaps they should not have been included in the analysis of the nonheroic material. However, while the guess that heroic motifs are involved may be well founded, it is not suffi- cient basis for omitting this group.
The group of melodies with syllabic structure <?, 5, 8, 5, (about 125) may as well be regarded as 13-sy liable two-section isometric melodies (see p. 31). Several musical reasons, however, speak for splitting the 13 syllables into two parts {8, 5,) and considering them, at least musically, as individual sections. ^^ The relation of this structure to simpler structures has been already mentioned on the previous pages. A remarkable feature is that their texts seem to be of urban origin, in many instances even with text-stanza structure and rh3nrLes, whatever rural aspect their melodies may have (mus. ex. Nos. 39a.b., 40, 42, 45). On the whole, in spite of their large number, they do not belong to the characteristic Serbo-Croatian types.
More tj^ical is the group of 8, 6, 8, 6, especially {10) 8, 6, 8, 6,. The former comprises 38 melodies, the latter 85, and in this, of course, the characteristic splitting of text lines appears. For their relation to simpler structures see the previous pages.
The syllabic structures 8, 7, 8, 7, and (11) 8, 7, 8, 7, are represented each by 19 melodies (in 16 and 7 variant groups), of which, of course, the latter is more characteristic.
The reader may bear in mind that the syllable structures 8, 5, 8, 5, | 8, 6, 8, 6, I and 8, 7, 8, 7, \ are very well known from the material of the neighboring peoples. Structures 11., 13.-16. bis. of "Tab. of Mat." (pp. 127-30) which are related to the afore-mentioned 2)2.-6., 11. (p. 50), are certainly of foreign origin, as well as the latter.
The reader may be reminded that the groups 20.-24. of "Tab. of Mat." (pp. 132-35), with their more or less artificial syllabic structures, are probably of urban origin. In addition, many of them are represented by
^ See footnote 17, p. 31.
[51]
Morphology
a single melody; this circumstance makes it doubtful whether they ever lived a prolonged life, at least in the form in which they were notated. For, as long as only a single example of a certain peculiar structure is known, doubts concerning its authenticity are well founded; one never knows whether this or that particular feature may not be due, perhaps, to some fault in the performance (imperfect remembering, haphazard changes, and so forth), even if it is taken for granted that the collector has not introduced any arbitrary changes.
Our investigation has followed the principle of proceeding from the shortest structures to the longest. This was imperative because of the almost overwhelmingly greater number and importance of the shorter ones. It may be mentioned as an evidence of the great difference between Serbo-Croatian material on the one hand and Hungarian and Slovak on the other, that the investigation of the material of the latter peoples must proceed in the opposite direction. There the four-section structures are in overwhelming majority, and the three- and two-section structures are accessories only.
RANGE (ambitus)
At first glance one discovers a great preponderance of the comparatively narrow ranges. Here follows Table 4, enumerating the most important range formulas. Only range formulas from VII upward are separately quoted, the rest being less important.^^ Accidentals are omitted: VI 1-3, VII-b3, ^VII-\>3, are merged in VII-3.
The formulas form two parallel groups, those with VII as the lowest tone, and those with 1 (see Table 4, p. 53).
This table shows that {a) more than two-thirds of the material has a range between F7/ and 5; {b) almost half of these, that is, one-third of the material, has a range of VII-4 or 1-4 — a pentachord or a tetrachord in a certain position ; (c) there is a very marked descending trend in both groups; the range is widening toward the class of the four-section melo- dies, that is, toward such melody types as were collected mostly in the western parts of the Serbo-Croatian territory (a rich contribution by Kuhae).
Among the less common range formulas the extra-narrow VII-1, VI 1-2, 1-2, and 1-3, comprising 181 melodies, deserve special attention
" These may be studied in Appendix II, of "Tab. of Mat."
[52]
Table 4 Range
|
Range |
I One- sect. |
II a Iso- metric |
III D Hetero- metric |
IV DC Iso- metric |
V Iso- metric |
VI DC Hetero- metric |
VII in Hetero- metric |
VIII Special |
IX DDE Iso- metric |
X JUL Hetero- metric |
Total |
|
I-l to VI-10 |
2 |
51 |
19 |
8 |
6 |
26 |
15 |
76 |
129 |
332 |
|
|
VII- 1 |
3 |
1 |
1 |
5 |
|||||||
|
VII- 2 |
4 |
33 |
14 |
3 |
5 |
2 |
4 |
65 |
|||
|
VII- 3 |
4 |
110 |
51 |
1 |
8 |
32 |
11 |
9 |
10 |
17 |
253 |
|
VII- 4 |
5 |
189 |
41 |
10 |
20 |
88 |
84 |
14 |
66 |
138 |
655 |
|
VII- 5 |
80 |
23 |
4 |
12 |
41 |
31 |
3 |
58 |
139 |
391 |
|
|
VII- 6 |
8 |
3 |
7 |
8 |
8 |
36 |
57 |
127 |
|||
|
VII- 7 |
15 |
1 |
2 |
3 |
1 |
18 |
24 |
32 |
96 |
||
|
VII- 8 |
1 |
7 |
10 |
18 |
36 |
||||||
|
VII- 9 |
1 |
1 |
|||||||||
|
VII-10 |
1 |
1 |
|||||||||
|
1- 2 |
1 |
3 |
3 |
2 |
9 |
||||||
|
1- 3 |
7 |
59 |
9 |
2 |
4 |
6 |
7 |
1 |
7 |
102 |
|
|
1- 4 |
18 |
119 |
57 |
2 |
9 |
39 |
8 |
10 |
24 |
31 |
317 |
|
1- 5 |
3 |
103 |
36 |
12 |
18 |
43 |
17 |
7 |
62 |
83 |
384 |
|
1- 6 |
1 |
38 |
6 |
5 |
10 |
20 |
13 |
58 |
66 |
217 |
|
|
1- 7 |
15 |
2 |
6 |
5 |
9 |
11 |
3 |
27 |
23 |
101 |
|
|
1- 8 |
1 |
7 |
4 |
6 |
2 |
6 |
38 |
35 |
99 |
||
|
1- 9 |
1 |
2 |
9 |
4 |
16 |
||||||
|
1-10 |
1 |
1 |
4 |
1 |
7 |
||||||
|
1-11 |
1 |
1 |
|||||||||
|
Total |
45 |
828 |
266 |
60 |
107 |
328 |
230 |
59 |
507 |
785 |
3,215 |
[53]
Morphology
for the following reasons: (a) most of them come from the "autochthon- ous" territories, especially from Serbia (Dordevic) ; (b) the greater part of them are ceremonial songs (hay-gathering, harvest, Lazarus's day, cradle songs, and so forth) ; (c) although they belong to different groups with various stylistic aspects, they seem to be the remnants of a unified and ancient style; see also p. 60.
This group of narrow range melodies, in spite of their comparatively small number, is very important. In the Slovak material many supposedly very old ceremonial songs have melodies of a similar narrow range .^^ Most of them are four-section melodies, but some have three- or two-section structure. Their primitivity seems to indicate an archaic origin. Are these narrow range melodies in both m.aterials, perhaps, the remnants of a common ancient Slavic style? Unfortunately we do not have for this theory proofs as convincing as there are for the common origin of the Old Hungarian-Turkish melodies.
Other illustrations of such primitive style are found in a different region, in the narrow range Arabic peasant melodies, as the following example shows ;^^ it is a wedding song from the vicinity of Biskra (Algeria), sung by two women about twenty-five years old, in June, 1913. The author's collection; transcription from record, which was later lost. The second song quoted for comparison is No. 388 from Kuba, B. H. ("Tab. of Mat." No. 165b.). Both may be compared with our mus. ex. No. 2.
Parlando, J= ca.80
0 de-vDJ-ko, dii - dKo mo - ja, 0 de-voj-ko, dJU. - d£omo-ja!
It remains an open question whether there is a connection between the narrow range Serbo-Croatian and the narrow range Arabic melodies.
The rather rare wide range melodies VII-8, 1-8, 1-9, (151) also deserve attention, especially when connected with §] [bs] |3^ or J LLJ L
** See Bart6k, 44 Duets for 2 Violins, Universal Edition, Vienna, 1933, Nos. 11, 33. Reprint by Boosey & Hawkes, New York. ^ There is an essential difference between Arabic peasant (rural) and urban music!
[54]
Morphology
caesura, or a^B'^a b content-structure, or with a fragment of such struc- tures. These structures are characteristic of the Old Hungarian-Turkish (rural) style, therefore, should indicate either Hungarian or Turkish influence. The Hungarian alternative is not probable: the few examples of this type are found in the center of the autochthonous Serbo-Croatian areas, not in the northern border territory, and they seem to have no connecting link with the Hungarian territory. On the other hand, the centuries-long Turkish occupation may well be responsible for a rural Turkish origin.^^ These rare melodies would be the only vestiges of Turk- ish influence on Serbo-Croatian folk melodies. A detailed enumeration follows; the first number refers to the "Tab. of Mat." 657 (Kuba B. H. 194): 10, H H, 1-8, a b b,.
696 (Kuba B. H. 249, 250) : 6", 3 [H, 1-8, a b c; Turkish var.: "Turkish Folk Music from Asia Minor," No. 7, with <?, I] B H, 1-8 structure, from Avlik (Kadirli vilajet); Hungarian var.: Hungarian Folk Music, ^^ No. 17, with <?, H EH S, 1-8 structure. 828 (Kuba XII. 18) : 12, 8, 8, Ul G, 1-\?10, abb; —Turkish var. : "Turk. Folk M. from A. M." No. 17 a.b., with 11, 9 U] El, 1~\^10 structure, from Kara Isali (Seyhan vilajet). 1217 (Kuba B. H. 196) : <?, II [3 li, 1-8, a b b c; Hungarian var.: Hun- garian Folk Music, No. 8, with 6", 11 CI] LI, VII-8 structure. 1231 (Dordevic 116): <?, 3 [3 Q, 1-7, abcd; Hungarian var.: Hun- garian Folk Music, No. 10, with 5", 3 Ll] Q, 1-8 structure. 1247 (Kuba B. H. 1010, 1017; Dordevic Nar. Pev., p. 10, first melody; our mus. ex. 36) : <?, H ED EI or k^ CD Ik^ or 3 S IyS; with VI 1-7, A B B c, or with 1-8, abcb; Hungaran var.: Hungarian Folk Music, No. 14, with <?, 3 EU H, 1-8 (with d^) structure. 1268 (Dordevic 113): 5', 3 [I] ^, 1-8, abcd; Turkish var.: "Turkish Folk Music from A. M.," No. 3, with <^, D EI] E, 1-\?10 structure, from Tiiysiiz (Seyhan vilajet), Hungarian var.: Hungarian Folk Music, No. 19, with 5", D EH CI, 1-8 structure.
*^ The reader may keep in mind that there is an essential difference between rural and urban Turkish music, too. The former is of Central Asiatic origin, its main characteristics being wide- range four-section melodies, pentatonic scale structure (or one derived from pentatonic) and "descending" contour-line of the melody.: The urban Turkish music derives, on the contrary, from Arabic urban music, showing some unessential adaptations.
^^ Bartok, Hungarian Folk Music, London, Oxford University Press, 1931.
[55 ]
Morphology
1269 (Kuba B. H. 110, 1049): ,?, I] [I] 0 or I] [U H; i-bP, aa,b c; Turkish var.: "Turkish Folk Music from A. M.," with 11, 3 HI 1^, 1-9 structure, from Hiiyiik (Qorum vilajet).
1270 (Dordevic Nar. Pev., p. 153, first melody) : <?, U CI] B, 1-8, a^b^a^b^. 1291 (Dordevic 176): 8b, § [I] i, 1-8,aa,-b c; Turkish var.: "Turkish
Folk Music from A. M.," No. 14, with 11, 3 H] [I, 1-8 struc- ture from Tiiysiiz (Seyhan vilajet). 1313 (Kuba B. H. 186, 220, 224-27): 10, kH [vnl U, VII-7, a b c b,; Hungarian var.: Hungarian Folk Music, No. 69, with 10, ^ Iv^^l k^, VI 1-8 structure.
If we extend the study of these melodies to include the Bulgarian material, another theory suggests itself for explaining their presence in the Serbo-Croatian material. For in the former there are 278 melodies of this type, that is, with wide range, pentatonic scale-structure, "descend- ing" contour-line of the melody; this means 3 percent of all the Bulgarian material available in print (which is a little less than 10,000 melodies), whereas the above-mentioned Serbo-Croatian 21 melodies of the same type represent not more than about 0.6 percent of the investigated Serbo- Croatian material.
There were, and still are, many purely Turkish settlements in Bulgaria; in Serbia-Croatia, none. This circumstance may account for the fact that there are proportionately six times more such melodies in the Bulgarian material than in the Serbo-Croatian. The question arises whether the presence of those 2 1 melodies in the latter is not due rather to an infiltra- tion from Bulgaria than to a direct Turkish influence.
Besides this obviously Turkish influence, some rare melodies are actu- ally borrowed from the so-called New-Hungarian material. These melo- dies penetrated into the stock of Bosnia and Hercegovina in fragmentary form (except the last example). There are eight of them in the printed collections (the first number in the single items of the following list refers to the "Tab. of Mat.") :
(1) 516 (Kuba B. H. 410), (2) 1364 (Kuba B. H. 199), (3) 1392a.b. (Kuba B. H. 23 and 24), (4) 1751 (Kuba B. H. 281), (5) 1333 (Kuba Xm 12), (6) 1342 (Kuba XII 45), (7) 859 (Bosiljevac 36), (8) 1809a.b. (KubaB.H. 20 and 19).
Their Hungarian variants, reduced to the fragmentary form which the borrowed melodies show, are as follows :
[56]
Morphology
i^' ¥ 1 UU ^
[lSj\^^-^ II
1st (and 4th) section [2d and 3d missing]
B
2)f"f^r-fr r irrrr irrT r IF g
3d section
4th (and 1st) section [2d missing]
^)^\^m^l Ip-gHf r I;}n71|j^^^
1st Cand 4th) section [2d and 3d missing]
*)^U — r — r f ir r ^^
1st section
m
h r r r^
2d section
Ej, liaj, g"6n-d6r a ba - bam ^j, vaj, rin-dir i ba - baj
3d section
^efr)
Wr±tu I i^ii-- i J
5ze-rel i^ az en - gem i- g"a - zaa.
4th section of the Hungarian variants; missing in the Serbo-Croatian variant
t
5)^°:lr c?r r ir r r r ir r ^
3d section
f pr r ir— r-f ^U^fL-^^ j (on
4th (and 1st) section [2d missing]
[57]
L
^
Morphology
1st fand 4th) section
l)Y ip"*- f)r ir ^-^ "(^
2d (and 3d) section
^
)^=p=pF=^
-f'' — |-p 19=
3d section 4tli fand 1st) section [2d missing]
^^
II
f f S
^
4 r r r ^^
a
The missing sections in the Hungarian variants are as follows: in (1) and (3) unknown; in (2) and (5) similar to the third section, but with d^ as final tone; in (4) as given after the melody; in (7) similar to the first sec- tion, but one fifth higher. In (4) the Serbo-Croatian text has the senseless refrain rindiri baba — a distortion of the Hungarian original, for ''my sweetheart has curly hair." Our music example No. 49 probably belongs also to this group.
A possible explanation for this borrowing lies in the occupation of these areas by Hungarian troops between 1878 and 1918. In the last quarter of the last century Hungary's younger generation were already neglecting the "old style" Hungarian-Turkish melodies and fully adopted the "new
[ 58 1
Morphology
style" which began to flourish at about that time. Therefore, old-style melodies could not have been taken over from these troops.
Another possibility is the infiltration of Medwnurje material saturated with "New-Hungarian" melodies (see p. 24) into Bosnia and Herce- govina; this, however, seems to be less probable.
As a negative feature there must be mentioned the absence of the so- called "long-drawn" melody in the Serbo-Croatian (as well as in the Bulgarian) material (see p. 77, footnote 77). We know this melodic style, which originates without any doubt from Persian-Arabic sources, from the Ukraine, Rumania (Maramures and all territories of prewar^^ Ru- mania) ,^^ Turkey (only a few specimens), Persia, Iraq, and Algeria (Djelfa). So there is a gap in the continuity of the spread of this type in the south-northern direction (Bulgarian and Serbo-Croatian territory), another gap in its spread in the east-western direction (Syria, Egypt, Tripolis, Tunis). That the type exists in Serbo-Croatian territory is not very likely, since even Kuba, whose eagerness, zeal, and ability in collect- ing folk music is well known, could not discover any trace of it. Either it never existed in this territory, or it must have died out before the end of the last century.
SCALES (cf. pp. 52-56)
Owing to the preponderance of narrow-range melodies, scales extending over an octave (either diatonic, seven-degree scales, or "defective" anhemitonic pentatonic scales) are comparatively rare in the Serbo-Cro- atian material (at variance with the Old Hungarian-Turkish and New- Hungarian material). The majority of melodies have a diatonic scale fragment consisting of the first four, five, or six degrees of an / major or minor (or some other variety) ^° with the second degree (that is, g) as final tone (the pitch given according to our system of transposition) .
It must be emphasized that this is not a Dorian or Aeolian scale on g, turning at times to its lower seventh ( VII = /) . The difference is this : the first has/ a(b) c as main (accentuated) degrees, the two latter, gb{\i) d. Before the final tone of such a melody (appearing as accentuated tone only at the end) is reached, anybody would interpret its degrees as the
*8 "War" means "First World War."
*^ The Hungarian so-called Rdkoczi nota (Rak6czi song) from which the "Rdk.6czi March" originates, seems to belong to this type, too. ^" For example, with minor third and augmented fourth, or minor third and diminished fifth.
[59]
Morphology
pentachord (or tetrachord, hexachord) of a major or minor scale. Where this scale is common, the peasant singers, with a steady feeling for pitch, have a subconscious sense of modal octave-segments, even when they are presented as fragments, such as a pentachord or tetrachord. If they sing in succession two melodies, the first of which has a normal major, minor, or Dorian scale, and the second the above-described kind of scale ending on the second degree, they will sing them with one degree difference in pitch, that is, the second melody will be performed a whole tone lower.^^ Why do we not transpose these melodies ending on the second degree of the scale so that the final tone becomes a^, as some editors have done? The only answer is that deviations from the principle of setting the final tone in g^ would lead to confusion and to lack of consistency in our system of transposition.
In narrow range melodies the characteristics of this scale will, of course, disappear. It is spread over Bulgarian areas and appears in some Ruman- ian areas^^ also. The Hungarian material does not have it.
Much speculation may be indulged in concerning the antiquity of this scale. My hypothesis is that originally only narrow range, probably two- section, melodies existed in the Serbo-Croatian material. As foreign (western) influence arose, a gradual extension of range and borrowing of major or minor pentachord melodies could have occurred. While these melodies were in the process of infiltration, a change took place : the feel- ing for ending a melody became shifted, and the roles of the perfect and the imperfect cadences were exchanged. One tentative explanation for this phenomenon would be that Western European four-section melodies generally have the imperfect cadence at the end of the second section (the end of the fourth section, that is, the end of the melody, presents, of course, a perfect cadence). On the other hand, there is a marked trend among the Serbo-Croats (and among some other peoples, too) to borrow only fragments from foreign melodies, especially if in their feeling the melodies are too long. Accustomed to two-section structures, the borrow- ing of the first half of the foreign melodies with the imperfect cadence
" They will observe other modal relations as well. The melodies in this book do not contain examples, but in my Rumanian Folk Music, Vol. II, there are several very enlightening dem- onstrations for this phenomenon, for example :
No. 301, in Mixolydian mode, with final tone b, and No. 270, in Aeolian mode, with final tone c^^, both performed attacca by the same two women;
No. 321 in Dorian mode with final tone b, and No. 124a. in Phrygian mode with final tone c^^, both performed attacca by two other women.
62 Especially in the Banat area.
f 60 1
Morphology
would suit their musical sense. Such incomplete structures, with the imperfect cadence, may have spread over the territory and, developing various characteristic structures, led to new formations. The imperfect cadence, however, remained unchanged.^^ If this were true, the melodies with "final stop on the second degree" would be later developments.^^
There is another scale, a major scale, represented by melodies mostly of western origin, with the final stop on the third degree. There is a differ- ence between this and the Phrygian modal scale, which is analogous to the difference between the two afore-mentioned scales: the main degrees in the first are e^b g^ b^\> (e^b); in the latter, (d^) g^ b^\? d\
Augmented second intervals occur frequently in the minor scale, with the final stop on the second degree, as fl^b b^^, and in the normal minor as b^\? c"^^. Whether they are signs of oriental (Arabic) influence (perhaps through Turkish mediation) is still an unsolved problem.
A very peculiar scale formation is the above-mentioned /^g^a^b 6^b c^b, not very common (for example, Dordevic, 236), and the still less frequent g^a^iV) b^\f c^d^\? e^{\?)P (our mus. ex. No. 36).^^ The former is intermediate between the diatonic and the "chromatic" scales. In spite of the small number of these "chromatic" melodies, a complete list of them is given here, which includes also the examples in Kuba's article N. g. u. D. (p.
^^ There is at our disposal an example of a rather recent borrowing of a melody which shows that this hypothesis is not as fantastic as it may seem, perhaps. See the Hungarian melody (6) on p. 58, its second and third section with a^ as final tone. The Hungarian variant has of course g^ (first degree) as final tone of the entire melody; the Serbo-Croats, however, took over only its first and second sections; therefore, this borrowed fragment has a^ (the second degree) as final tone.
" There are examples of variant groups in which some of the variants have their final stop on the second degree, some on the first degree. Closer examination of these examples will perhaps help to solve this question. For example,
"Tab. of Mat." No. 158 a.-g.: Dordevifi Nar. Pev. p. 77, second melody, Ka^erovski 40a), Bosiljevac 47, Kuhai 878, Kuba B. H. 581 and 913 have imperfect cadence, Kuha£ 877, perfect;
"Tab. of Mat." 177 a.-d.: Dordevic Nar. Pev. p. 24, first melody, Iz Levca 23, Dordevic 449 have imperfect cadence, Kuba B. H. 175, perfect;
"Tab. of Mat." 810 a.-c: Kuba B. H. 817, 818 have imperfect cadence, Kuba B. H. 439, perfect;
"Tab. of Mat." 1002 a.-h.: Kuba IX, 3, KuhaC 653 a)b), 1083 have imperfect cadence, KuhaC 1039, 1040, 1099 and Dordevid Nar. Pev. p. 69, second melody, perfect;
"Tab. of Mat." 1010 a.-c: Kuba B. H. 507 and 688 have imperfect, Kuhac 178, perfect;
"Tab. of Mat." 1440 a.b.: Kuba B. H. 578, has imperfect, Kuhac 1129, perfect;
"Tab. of Mat." 1591 a.-c: Kuhafi 1111 b) has imperfect, KuhaC 408, Ka^erovski 4 b), perfect;
"Tab. of Mat." 1693 a.b.: Kuba B. H. 41 has imperfect, KaCerovski 68, perfect.
'^ The Rumanians know the former, especially in their Songs of Mourning, the latter in some of their cdntec ("song") melodies. Both are entirely lacking in the Hungarian material.
[61 1
I
Morphology
25) because of their importance; after the enumeration No. 62 is quoted entirely. Whole note in the scales means final tone;/^ in parenthesis, that this tone appears only in the lower part of two-part songs.
N. g. u. D. No. 61;
"Tab. of Mat." No. 233 (Kuba B. H. 1000) ]
"Tab. of Mat." No. 302 a.b. (Kuba B. H. 1002, 999) > : "Tab. of Mat." No. 1043 (Kuba B. H. 1003) )
|>- 1]" ^*
-&■
jygtSurb'
^—
"Tab. of Mat." No. 323 c. (Kuba B. H. 1004);
"Tab. of Mat." No. 308 (Kuba B. H. 1006);
"Tab. of Mat." No. 328 (Kuba B. H. 1007): |
"Tab. of Mat." No. 303 f. (Kuba B. H. 1018)^ ^
N. g. u. D. Nos. 62-66 >: ^ U)g\^«\'M!»-
N. g. u. D. No. 69 ) ^ ^
L. Kuba, N. g. u. D. No. 62.
5,8+8,5 _0]
f 4 f^ ^^
Tra - vc
T"
- ^e - na!
jxa. - vu HD - SB - na « _
I& 6@ ie-be, ira-VD.ko-iSi'l.kad ia Wdem sab-lu no-sit
g-^ J J
f <^^, I ^^
^
Ira - VD
U. Zagvozdu; devojke
KD - ^e
1^2:
na!
The second example represents an Arabic chromatic melody from my own collection, the third a Dalmatian two-part melody in chromatic style
(seep. 72) .^6
^6 By courtesy of Professor George Herzog, the owner of the Dalmatian record (transcribed by Bartok).
[62]
|
dZ 1; i; 1! 1:1 li: ■I ..= .tz f: M • |
5i "8 "B "1 i IS 1. k. !' \i \[ \i ]\ w [ •) (II Eli |
fn
N)
N
Morphology
As we see, these "chromatic" scales are narrow range scales; the melo- dies have very few ornamental tones. Chromatic changes affect the second and third degrees, both appearing as naturals and flats in the same melody (except in one case, where only the third degree has both forms) . One may get the impression that our current notation is inadequate for these scales: the use of flats and naturals in connection with the same degree implies a certain interdependence of the two and could mean here that the flat degree is a "colored" transitional form of the same natural degree, leading to another degree through the colorization (chromatiza- tion) ; that is, that the two forms differ only in "color," but not concerning their function and degree. Yet this is not the case with these melodies: it appears that the single tones, whether or not they show the same degree position through the inadequate notation, are treated as independent tones, a^\? and 6^b having nothing in common with a^b^ and b^}:\, respec- tively; each tone is used as an independent degree. Therefore these scales ought to be regarded rather as a segment — a kind of pentachord — of a twelve-hemitone scale; but, of course, with a fixed tonality because of their ever recurring, never changing final note.
As to the example quoted from Kuba's article (see p. 62), it is a clew to very significant findings. It has many variants, all of them in diatonic scales and generally with the same text or with text variants. These are as follows: ("Tab. of Mat." 835) Kuhac 979 (from Istria, the closest variant), 977, 978, 980, 981, Kuba X, 40, B. H. 556, 557; Slovak variants: Slovenske Spevy" I, 194, III, 36, 62, 286; Moravian variants: Peck^« 121, 183, 209; Cernik^^ 69, 282; SusiF" 780a). Hungarian variants (rather remote) : Bartok Hungarian Folk Music, No. 299 a.b.c.
This proves that the Serbo-Croatian melodies in question are borrow- ings from northern foreign material and that Kuba No. 62 has simply been adapted to that peculiar "chromatic" scale. A similar process of adaptation is perceived in the case of "Tab. of Mat." 303 f. (Kuba B. H. 1018) when we compare it with the rest of the variants in the same group (Kuhac 250, 251, Kuba X, 34, B. H. 387, 391). If we try, as an experi- ment, to spread the remaining 14 "chromatic" melodies over a diatonic scale, that is, changing the succession g a\} ab{b\? b^ into gat] b{\}) c d,
" Slovenske Spevy, Turi^ianky Svat^^ Martin, 1880, 1890, 1899, Vols. I-III.
*8 E. Peck, Valasske ndrodne pisne, Brno, 1860.
"^ J. Cemlk, Zpevy Moravskych Kopanicdru, Praha, 1908.
8" Fr. SuSil, Moravske ndrodne pisne, Brno, 1859.
[ 64 ]
Morphology
most of them will appear in this new and artificially widened garb as very well-known diatonic melody types. Therefore this "chromatic" system is not a separate independent system, with new melody formations; it is rather a transformation of the diatonic pentachord (or tetrachord), the medium range diatonic pentachord having been compressed into a nar- rower range "chromatic" pentachord (if one is permitted to call the latter a pentachord) . The same chromatism appears in those two-part songs of Dalmatia in which the interval between the parts is generally a major second (see below, p. 72, and second example on p. 63). It has not yet been discovered in any other European territory or in northern, central, or eastern Asia; certain types of narrow range Arabic rural melodies, however, show a slight resemblance to it (see first example on p. 63).
These scales remind us of the chromaticism in the heroic song motifs. Either their origin is to be sought in the influence of that chromaticism, or both their chromaticism and that of the heroic song motifs are, though perhaps independently of each other, due to Arabic or other Eastern influ- ence. Decisive answers to these questions cannot yet be given.
RHYTHM FORMATION
The largest part of the published material is unreliable in this regard, and therefore must be excluded from our investigations. Furthermore, all rhythm formations found in four-section melodies of probably foreign (Slovak, Hungarian, and so forth) origin will also be discarded as an irrelevant incumbrance. Therefore, the results of investigation are based mainly on the seventy-five melodies included in this book.
For purposes of analyzing rhythm formations, melodies shall be stripped of all embellishment tones, and the remaining skeleton form shall be used as a basis for investigation, as shown in most of the melodies published in this volume, below the first stanza.
The scarcity of real parlando-rubato rhythm is remarkable. Because of their peculiar "free stanza" structure, a small group of melodies with this rhythm gain a certain importance. The melody sections of these consist mainly of J J jj| jJjJi|, some of the eights sometimes prolonged. Their "free stanza" structure is based on the text: its "chapters" (portions having more or less rounded content and no constant number of lines)
65 ]
Morphology
will determine the melody stanzas, one text chapter corresponding to one melody stanza (see mus. ex. Nos. 20, 34-36).^^ This melody stanza contains three (No. 20) or four (Nos. 34-36) more or less different pat- terns of sections which may be repeated irregularly; the last, however, generally does not appear except at the end of the stanza. The text lines are not repeated (as is the case in heroic poems) except the last line of the stanza in Nos. 20, 36 (both sung by the same singer, Meho Jaric). No. 35 presents a rather fragmentary form of two or three sections in its stanzas, except the first stanza. If we strip these melody stanzas of the repeats caused by the text stanza, there remains a normal four-section or three- section melody. Thus, the music has essentially an extended four-section or three-section structure. Such subordination of the melody stanza to the longer text chapters is rather peculiar and deserves fullest attention. No trace of it can be discovered in the published collections, where the editors tried to squeeze every melody into a two- three- or four-section form. ^2 So few examples of this feature occur in the collection that it is impossible to ascertain whether it is an individual and haphazard forma- tion or a wide-spread usage. In the latter case, the discovery of this phe- nomenon would be exclusively due to Professor Parry's method of record- ing songs in their entirety.
The rest of our group consists mostly of melodies in non-parlando, fixed rhythm, though sometimes the rhythm may appear to be somewhat flexible or, more rarely, may take the aspect of parlando rhythm. Exam- ining the various rhythmic patterns found in this material, we start by assuming the pre-existence of simple original rhythmic patterns as fol- lows:
a)
1) nn I JTO I for 8-syllable lines
'; |-«« »>ii». ,1 I I for 10-sylIable lines, or the same patterns with double l)J77}\ni^U J I values
3.) /TO|JTO|J^J I for 11-syllable Hnes
We see a)l. in No. 1 ; Nos. 16 a. and 29 have the same with double values and extended to twelve-syllables (JJJJ|JJJJ|JJJJ|).
" The second stanza of No. 20 presents an exception inasmuch as its last repeated hne is not the last line of the respective text chapter.
^^ Some signs in the Bulgarian pubHshed collections seem to indicate that a kind of free stanza structure exists in the Bulgarian material, too. See also Bart6k "Rumanian Folk Music," Ms. deposited in the Music Library of Columbia University, II, pp. 25, 26.
[66]
Morphology
Then we proceed to determine what kind of changes happen to these simple patterns. We shall see that usually only their first two measures are subject to changes of any consequence. The main variations are as follows :
b)
1. MJiJJMJ:JJ:|| No.2
2. J J jJJlJJidJlJJ'vl)
4. |:J1J;|;3J iL J J :|K
5. .r773l/3j J>|J J K
^)
2. JTn I d J : J J I J J K
3. JJ J J IJ JiJ-H I J. J.J. M
jjjj icijv^.g..j.|j. j.«oiiiNo.i4
4. /m IJ J J J I )
J J J J IJ J J J |J J J J |>No.l6b.
J J J J |J J J iyr, II )
5. /m ij J J J I J
|:J J J J U J iJ o ^ I :||}No.l8
6. rm N J jjijj J^7^| )^ ,- /m ijj j:3ijjj^yv^iip-^^
[67]
Morphology
d)
J JiJd |o. ( i--
4. |:Jd iJd iJd iJd ^
J d iJ d lo ( ^ :
5. |:Jd iJdHJJ Jl
/T-n u J d
m<i d o- w
No. 3
No.391>. (a. •• riibalo)
ND.43.44a.b.
No. 45 (rubato :=— )
No. 4 7
e)
I /] d i d I J. J^J J
|(vac «n J J) 2. -HJ Id I J JJJ
((vac J J /I)
3.( JJJ iJH J JJJ
Lap. J.J>) (varJ. J>)
^ I J JJJ
(var. J. J^
J..PJ J k J JJJ
4. 1:J3J ioH J J J J
d J , > I J d. ^
5. J-J'J id H J J J J
d-Jid d 1« - . ^^d s Id d iJ^d
d- Jid d 1 w ^
6. |: J J d lo 1 J J d I
rm I J. ;» o(
I:
7. J JiJ d IJ J d J^
d d. i.^ I d J J J. J" J J I o ^
d o. /f\ \
Jo ^ d o ffv
d d ^ ;(vacJ J)
► No. 40
No. 41
51
No.6a.b.c.
No. 12 a. -e.
>No.26,27a.-e.,28a.b (28d. •• rubato)
No. 30
[68]
Morphology
0
|
1 j J J. J 1 j J J j |
k J. i. 1 W.J. J> ) |
|
|
J J d oC J J J J |
/(van J. J" ) |
|
|
2. J J J J 1 J J J J 3. J c^ J. J 1 j j j j |
1 J J-l J. J 1 |
|
|
4. J.d J J 1 |
J.j J J |
1 J J 1 1 J J 1 |
>Ko.31a.b.<3.
No. 32 No. 33
No.31d.
g)
1 =^N J J IJ J IJ Jj J :|| )j^4
3. J J J IJ J I J
J J o^lJ J I J J J lo ^ I
Group b) consists of patterns in which the value of one eighth (or quarter) note in one or both of the first two measures is doubled. The doubling may affect any one of the four eighths (or quarters) . Variety of rhythm is attained by placing the change on different notes in the two measures (2. No. 7, both sections), by using the device in only one of the measures (3. Nos. 8 a. b.; 5. No. 19), and, finally, by using a different pattern in the second section (2. No. 7; 5. No. 19).
Group c) consists of patterns in which the values of the second measure in the first section are doubled. This appears clearly in 4. 5. 6. (Nos. 16 b., 17, 18). In 1., 2, (Nos. 9, 11) there occurs a quadrupling of the value of
the first and second eighths in the second measure : 1172 changed to J J J J 5 in 3. (No. 14) j j J J | becomes JJo J| . As to the second section, it has either the rhythm of the first section with a slight change (3. 6., Nos. 14, 17), or an altogether different pattern (1. 2. 5., Nos. 9, 11, 18).
[ 69 ]
Morphology
In group d) there are patterns which use the combination of J J and J d (or J J^ and J>J) rhythms, in one or in both measures of the first sec- tion; the combination may be either the same or different in the two measures. Heterometric stanza constructions appear in 2.-5. (Nos. 39b, 43, 44 a. b., 45, 47). The rhythm in the first measure of the shorter second and fourth sections is built on the same principle in 2.-4.
Group e) contains rhythmic patterns in which the four values of the first measure in the first section are articulated as J + i instead of the common 2 -\- 2 division, showing the characteristic J^ J : d rhythm or some variation of it; the last section (in two-section or three-section melodies) has the almost unchanged original rhythm { a)2. in double values), the second section of three-section melodies is simply a repeat — rhythmically as well as melodically — of the second and third measures of the first section (1. 2. 3.). In patterns 4.-7. the structure of the second and third (fourth) sections differs somewhat.
Group/), with melodies belonging to those with "special structure" (pp. 44-46), show first lines in which the first or both the first and the second bars are transformed from J J J «l into j J jcl- J or J J i J J (in 4., however, JJJJ into J.?jj) . No special comment is needed for this group. The same can be said concerning group g), in which the 8b, metrical line appears.
As the reader may note, great rhythmical variety is attained by the use of comparatively simple means.
PART-SINGING
Here follows a complete list of two-, three-, and four-part songs from the printed material and the Parry Collection.
|
Kuha B. H. |
"Tab. of Mat." |
Number of Paris |
|
337 |
500 a. |
2 |
|
338 |
500 b. |
2 |
|
456 |
283 a. |
2 |
|
474 |
752 c. |
2 |
|
829 |
1553 |
2 |
|
1005 |
115 b. |
2 |
|
1006 |
308 |
2 |
|
1019 |
285 |
2 |
|
1018 |
303 f. |
2 |
[70]
|
Morphology |
||
|
Kuba IX |
"Tab. of Mat." |
Number of Parts |
|
3 |
1002 a. |
2 |
|
5 |
321 g. |
3 |
|
7 |
1162 |
3 |
|
8 |
690 a. |
3 |
|
15 |
1296 |
2 |
|
16 |
588 |
2 |
|
21 |
473 b. |
3 |
|
23 |
1372-73 |
2 |
|
46 |
610 e. |
2 |
|
49 |
1892 |
2 |
|
52 |
1024 a. |
2 |
|
iTMfeaX |
||
|
12 |
624 |
2 |
|
13 |
1506 f. |
2 |
|
14 |
249 |
4 |
|
22 |
743 |
2 |
|
23 |
1381 c. |
3 |
|
24 |
227 |
3 |
|
ZZ |
229 c. |
2 |
|
43 |
1486 |
3 |
|
45 |
1093 |
3 |
|
48 |
1154 |
3 |
|
54 |
383 b. |
3 |
|
59 |
1758 |
2 |
|
Kuhai |
||
|
370 |
487 |
2 |
|
1088 |
1729 |
2 |
|
1470 |
485 a. |
2 |
|
Borievit |
||
|
218 |
498 |
2 |
|
570 |
508 |
2 |
|
Bartdk |
1020 a. |
3 |
|
1223 |
2 |
0«r Afjw. Ea;. 1
[ 71 ]
Morphology
This list shows that part singing occurs mostly in the northern and western border territories and in Dalmatia. Some of the two-part songs have simply a drone in the lower part, following the rhythm of the melody (mus. ex. No. 1). The majority of the part songs have adopted the simpler harmonies and intervals of Western European art music (melodies having a scale with final stop on the second degree receive a regular imperfect cadence of tonic-dominant chord progression at the end) . A peculiarity in their performance was pointed out by L. Kuba.^^ It consists of the fol- lowing usage: the main part — representing the melody proper — enters singly; the other parts join later, sometimes in the middle of a measure or even of a word. Data on the spreading of part singing and the extent to which it was considered "obligatory" in these border territories are very scarce. I learned from my Serbian peasant singers in the counties of Torontal and Temes, now in the Rumanian Banat area, that if several persons gather and want to sing they instinctively take to part singing. If an outsider asks them to sing in unison, they may do so, but they will not feel at their ease.^^
However, in Dalmatia, besides the "normal" part singing, there is an extremely peculiar kind of "two-part" singing — in major seconds. L. Kuba devotes one part of his afore-mentioned article to a detailed descrip- tion of it.®^ There are available (or at least there were before the Second World War) some commercial disks with such two-part songs of Dalma- tia.*^ Listening to those records, one may hear the major seconds some- times "degenerate" into minor thirds. Nevertheless, since these deviations seem to be more or less due to chance, this kind of performance may be called "part singing in major seconds. "^^ It is used, at least in Dalmatia, in connection with chromatic-style melodies (see p. 64).
" In his article N. g. u. D., second part, pp. 1-6 (see also "Tab. of Mat." 1223, a melody recorded by Bart6k) .
^* I obtained similar information from the Rumanian inhabitants of a near-by village, Nagyszentmikl6s (Sannicolaul-Mare), the only village where Rumanians spontaneously cultivate part singing, probably as a result of Serbian influence (see "Rumanian Folk Music," 11,44-45).
« Pp. 12-15.
"Readers who are interested are referred to the following disks: Edison Bell Z 2104, Z 2041-42; Z 2105, Z 2043-44; and others in the same series, issued in Yugoslavia. For the transcription of the first one, see p. 63.
'^ A proof for this is found in the instrumental preludes on the commercial disks, which for technical reasons invert the melody played on two sopela (a kind of oboe), that is, there appear minor sevenths instead of major seconds. These minor sevenths are never changed into any other interval.
[ 72 ]
Morphology
It is difficult to determine the origin of the normal as well as the major- second part singing. An intrinsic feature of peasant song melodies is their performance in unison. However, three great centers of part singing in Europe are exceptions : a) the German southeastern regions, h) the Great- Russian, and c) the Serbo-Croatian northwestern border territory — all three sections having normal part singing except Dalmatia. Does this usage originate in Western European^^ art music? This is possible, since the means used in both art music and peasant music are similar. So far, however, no evidence establishing this origin exists.
Explanation of part singing in major seconds is even more difficult. Analogous phenomena are entirely unknown in folk music anywhere, according to our actual knowledge.®^ Is it a transformation of a kind of normal part singing, or does it originate in the antique style of an extinct art music? Or perhaps the major seconds are simply "compressed" normal thirds. The latter theory seems to be the most acceptable explanation, since this part singing in major seconds is closely connected with the "chromatic" style melodies. And if the chromaticism of the latter is the result of a horizontal compression of diatonic scale segments, then the major seconds could be interpreted as having resulted from a vertical compression of thirds.
PECULIARITIES OF PERFORMANCE
Ornamental tones. — The ancient looking narrow-range melodies, the dance melodies, the parlando-ruhato melodies to eight-syllable text lines with free stanza structure, and many melodies of foreign origin are com- paratively poor in ornamental tones, and therefore differ little in their performance from most melodies in the material of neighboring peoples to the north and west. It is quite different with the other melodies, espe- cially in the "autochthonous" areas. There we find an abundance of orna- mental tones, the majority constituting "heavy" ornaments. These are, as pointed out on p. 17, ornament groups (melismatic groups) of equally heavy tones, with clear-cut, definite rhythmic and melodic patterns.^" When we look at the skeleton form of the melodies as given in this book below the first melody stanza of most melodies, we get the picture of rather
"8 "Western European" is used in an extended sense, including German territories. "' Except in a few examples from Bulgarian territory.
""> See our Nos. 2, 10 b (two examples of simpler ornaments), Nos. 4, 6 c, 10 a., 11, 12 e., 18, 22, 27 b., c, 28 c, 31 c, 41, and many others; the last is exceptionally rich.
[73]
Morphology
simple, less appealing melody formations. Reading the melody as it was actually performed, we get the impression of having an incredible wealth of melodic design before us.
The Rumanian material of certain areas (for example, the Bihor area), is rich in heavy ornaments, too. There is, however, a difference between the latter and the Serbo-Croatian. The ornament tones in the Rumanian material usually lead from one principal tone to the following one, filling out the gap between intervals with a succession of degrees. Or they embel- lish the principal tones by circling around them or by giving them a sharper pulse (similar in effect to the mordents and inverted mordents) . They are only accessories, having no individual patterns or life. The orna- ment tones in the Serbo-Croatian material, on the contrary, are forma- tions of very expressive patterns and of a marvelous variety. These pat- terns deserve special study. Unfortunately, even an attempt at this would lead too far afield in this book.
L. Kuba grasped the peculiar character of the heavy ornaments at the very outset, and in some of his papers he emphasized their distinctive significance long ago.
Line-, word-, and syllable-interruptions. — In this paragraph we again come to an extremely characteristic feature of the Serbo-Croatian (and Bulgarian) folk melodies, that is, the line-, word-, and syllable-interrup- tions effected by shorter or longer rests, not for articulation's sake, but for decorative — one might almost say for expressive — purposes. This astonishing custom will invariably deceive even the best-trained musicians when they first listen to this kind of interruption. They will without fail interpret the rests erroneously, that is, as section caesuras.
There is a peculiar usage, at least in the Gacko area, of putting an extremely long rest at the end of the melody stanzas. ^^ This rest, however, serves a decorative as well as an articulating purpose. As for the non- articulating rests, there are three kinds, referred to above; the difference between line interruption and word interruption, however, is not essen- tial, because the length and the placement of the words used in the text line will determine whether there will be a line interruption or a word interruption.
'^ Since the Parry Collection contains very few samples from other areas and the earher collectors never measured the rests between melody stanzas, we do not know the geographical extent of this usage.
[74]
Morphology
Line (or word) interruptions have been observed by the editors of the printed material in 105 melodies (generally not in the first melody sec- tion) :
74 by Kuba (two of them have two interruptions, the first of which takes place in the first melody section; three of the simple interruptions appear — as an exception — in the first melody section also) ; 27 by Dordevic (two of them with double interruptions, the first of which
is in the first melody section) ; 4 by Kuhac (one double, the first of them in the first melody section) ;
15 melodies of our collection have it also: Nos. 5, 9, 11, 14, 15, 21 a. b., 22, 23 a. b., 24, 30, 40, 41, 47, 49. Of the latter, Nos. 5 and 14 have two interruptions, the first of which occurs in the first section; No. 14 deserves special attention because of the peculiar and exceptional mode of placing the interruptions, and because of the shifting of the first one, in some stanzas.
From this description it may be seen that interruptions generally avoid the first melody section. The placement of an interruption is not at all connected with the metrical structure; as a matter of fact, an interrup- tion may appear following any of the syllables. Interruptions following the first or preceding the last syllable of the section are especially mis- leading, because generally there are no rests between the sections. For example, the melodic stanza of "Tab. of Mat." No. 1163 b. e. g. has the pattern 6, 6, 1 -'- 5, 6, (that is, it has the interruption after the first
syllable of the third section). Inexperienced listeners, however, will assume the syllable structure to be 6, 7, 5, 6, instead of 6, 6, 6, 6,. Simi- larly, "Tab. of Mat." No. 788 d. g. h., with pattern 8, 5 -'- 1, 6, (inter- ruption before the sixth syllable of the second section) , will be interpreted as 8, 5, 7, instead of 8, 6, 6,?^ In some cases the interruption is no intrinsic ingredient of the melody, as is shown by variant groups in which some of the variants have no interruption (variant a. of No. 1163), while others have it. One of our examples (No. 14) shows a shifting of the interruption in the first melody section from one syllable to another in some stanzas of the same song.
The less frequent syllable interruption has been observed in :
16 melodies by Dordevic (one has it in the first melody section) ;
^2 See, in addition, the respective examples in this collection.
[75]
Morphology
2 by Kuba (one has two interruptions, the first of them in the first melody
section) ; 1 by Bosiljevac (two interruptions, the first of them in the first melody
section) ; and in 1 melody in this book: No. 28 c.
The comments given for the line interruptions apply equally to the syllable interruptions.^^
Analyzing the material of the neighboring peoples, we find that both kinds of interruption are known on Bulgarian territory as well as on Serbo-Croatian. They occur in a considerable number of Rumanian melo- dies, the interrupting rests, however, are less decisive, shorter, and there- fore they do not give an impression of being used for decorative purposes, but rather as a mode of taking breath. On the whole, their use seems to be due to Yugoslav influence. The Hungarians and Turks do not know them at all.
In this connection a very important discovery has been made on Slo- vak territory. In a village in Zvolen County one single melody was found, in 1916, which has a permanent syllable interruption on the fifth syllable of the last section. ^^ It is a very ancient looking six-syUable four- section cradle song. Considering the apparently very great age of the melody, and the spontaneity with which all singers of the village, old or young, performed this song in the same way, we are fully satisfied that this interruption is a remainder — perhaps the only one on the entire North- Slavic territory — of an extremely ancient usage. Even on the basis of this single case, it is perhaps not too daring to submit the hypothesis that this phenomenon is an Old-Slavic usage, which although almost extinct among the Northern Slavs, still flourishes in Yugoslav territories.
Swallowing the last syllable of a section. — This has been observed in 20 melodies by Kuba and in 1 by Kuhac, the swallowing affects the last syllable of the last section in these 21 melodies ;^^ in 2 melodies collected by Bartok ("Tab. of Mat." 39 d., 1259 b.); and in 3 in this collection: No 3, (the swallowing at the end of each stanza), No. 16 a., and b. (the swallowing occurs, however, after the third syllable of the second section in 16 a., and at the end of the second section in 16 b.).
^3 Readers who are interested in further details are referred to the last column of the "Tab. of Mat." Line- (word-) interruptions are marked there as 4 -'- 6 ( = 10) ; syllable interruptions as ■^ -f I 5. I + 1 { = 6), I 5. I stands for "fifth syllable of the section, interrupted."
^* Published in B. Bartok, Five Village Scenes, Universal Edition, 1927, Vienna, No. 4.
'^ More details in the last column of "Tab. of Mat.-^
[76]
Morphology
This feature is known in the Bulgarian material, too. The same phe- nomenon is even more frequent in the Rumanian material.^® The Hun- garians of Transylvania have it, too, probably as a feature due to Ruman- ian influence. We have no data about its existence anywhere else. Its purpose is not quite clear. Its use, however, is by no means the result of a haphazard or individual mannerism; the performers employ it in the most deliberate way.
Clucking sounds. — There are two kinds of this manner of singing — (1) deliberate and (2) incidental. Clucking sounds deliberately used for ornamental purposes are all of extremely short duration and have a kind of a falsetto color produced by a peculiar technique of singing. They are very rare in the printed material, or to be more exact, they disappear there under cover of a common trill sign. Their only known precise tran- scription is shown in our music example No. 1; the variant of Kuba quoted on p. 231 has the trill sign as a substitute. To do Kuba and the other collectors justice, it should be mentioned that it is almost impossible to attain an exact notation of these strange sounds without studying them on records. Serbo-Croatian scholars seem to know this peculiarity very well; according to them it is spread over a certain part of Dalmatia. The Rumanians use somewhat similar sounds in their so-called "Canteclung" (long- [drawn] song).^^ They are entirely lacking in the Hungarian and the Slovak material.
2. The other type of clucking sound is the result of a physiological process, of a certain function of the sound-producing organs when they are used to their fullest extent, that is, in "full-tone" singing (see mus. ex. Nos. 12 b., 15, 19, 23 a., 27 d., and others; 12 b. is exceedingly rich in such sounds). These sounds are never used wilfully, and they appear either at the end of a tone or before it is attacked. They may appear as a phenomenon accompanying full-tone singing in the practice of any of the peoples of central and southeastern Europe. In spite of their unin- tentional nature, they ought to be carefully observed and transcribed, for they give a peculiar color to the performance.
Both kinds of clucking sounds appear mainly — in some areas exclu- sively— in the performance of femal singers; both are missing in the pub- lished collections.
" See Bart6k, "Rumanian Folk Music," Vol. II, Nos. 534 b., 639 a. c, and others. ''''Ibid., Vol. II, No. 613 a.-h., and Bart6k, Volksmusik der Rumdnen von Maramures, No. 23 a.-h.
[77]
Morphology
Emphatic rendition of syllables.- — ^The most frequent means of achieving this effect in the Serbo-Croatian style is to split a vowel into two or more identical vowels by interpolating one or more h sounds (mus. ex. No. 12 b.). The use of these dividing h sounds is generally accompanied by some happenings in the melody, too — either a clucking sound appears when h is introduced, or a change of degree, or both.
This kind of emphatic performance has been observed by Kuba (B. H. Ill, 472, 534, 600) and by Kuhac (955). The same mannerism occurs in the Turkish materiaP^ and in Arabic art music. In the latter, rather long ornament groups with accented tones sung to the same syllable account for it. In the former, its use is probably due to an influence of Arabic urban music. Perhaps this is true concerning the Serbo-Croatian emphatic performance, also. It is, however, possible that it originated in some usage of Western art music that disappeared long ago.^^
Another means of achieving emphatic performance (perhaps of jocular character?) is to interpolate meaningless syllables such as gi-ga-ga, go-gi- go, either inserted between the broken-up particles of the vowel or simply added to the unbroken vowel. Unfortunately our collection contains only one example (No. 54) ; since this has been recorded only once and has an undeterminable structure, no conclusions can be drawn from it. The printed collections contain 15 examples, 11 observed by Kuba, 2 by Kuhac, and 2 by Dordevic:
"Tab. of Mat." Kuba B. H.
265 d. 257
572 152
809 969
869 3
890 824
905 d. 583
1023 475
1619 e. 117
1826 987
1878 34
" See Bart6k, "Turkish Folk Music from Asia Minor," Nos. 20, 24.
^^ The Hungarian designation "hehhentes," still in use a century ago, for melismatic groups in art music, is significant; the h sounds in this onomatopoetic word point to a possible use of h sounds in the performance of those groups.
[78]
Morphology and in addition Kuba N. g. u. D. No. 64.
''T ah. of Mat." Kiihab
905 e. 1101
1084 590
"Tab. of Mat." BorSevit Nar. Pev.
867 p, 86, first melody
1755 bis p. 15, first melody
The purpose and character of this kind of emphasis is not clear. Accord- ing to our present knowledge the device is absolutely unknown elsewhere.
"Stuttering" entrance of words. — This is a rare phenomenon, observed in three melodies of the published material: "Tab. of Mat." 627 (Dordevic 360) and 905 c. g. (Kuba XI, 47, X, 11). It consists of the anticipated repetition of the first syllable in a word : Vise se-, sela ( < Vise sela) ; Kisa pa-, pade (<Kisa pade); Udri kir, -kisa {<Udri kisa)f° although there the repeat is functionally a substitution for the exclamation Oj. It seems
to be used for humorous effect. We mention it in spite of its rarity, since the Rumanian material contains some specimens, too: Bartok, Rumanian Folk Music, Vol. II, Nos. 428, 600. This either shows Serbo-Croatian influence on Rumanian material, or it indicates that the phenomenon is a characteristic trait of Balkan folk mAisic. We have no data about its existence in the m.aterial of other peoples.
Syllabication^^ of consonants. — This is the result of an emphatic pro- nunciation of a single (mostly syllable closing) consonant, and is achieved by adding the sound 9 (more-or-less similar to the Rumanian a or the Bulgarian n, or to the blurred vowel in unaccented English -er) to the consonant. ^^ The so-called liquids /, r — sometimes also j, m, n — do not generally need this addition (for instance, music ex. No. 14, syllable 1 : Si-td; No. 21 a., measure 11 : po-dd-. For syllable /, r, see mus. ex. No. 21 a.,
St. 2, measure 3: di-l-he-r; No. 22, St. 2, next-to-last measure, vi-ha-r;
in measure 2, 5, however, vi-ha-rd] etc.). The new syllable thus created
must, of course, never be regarded and counted as an independent sylla-
8" Cf. footnote 21 on p. 32.
^^ This word is used here in the sense of "making a syllable out of" a consonant.
^^ This is not to be mistaken for another well-known phenomenon in the Slavic languages, that is, the use of r (in Serbo-Croatian, for example, in hrt), or of r and I (in some other Slavic languages, for instance, Slovak smri, vlk), as syllabic consonants.
[79]
Morphology
ble. No trace of this pecuharity can be discovered in the printed collec- tions, except in Kuhac No. 812, where he explains that "praviti" was sung as ''piraviti" (more correctly, pd-raviti?).
A deviation from the official Serbo-Croatian syllabication rules must be mentioned in this connection. The singers seem to be inclined to divide words into syllables in a more natural way; for example, instead of ka-kvo, we have kak-vo in mus. ex. No. 28 a., St. 6, measure 5; instead of la-dna, we have la-dd-\-na in our mus. ex. No. 39 b. measure before the last;
i-gd-\-re in No. 33, St. 4, measure 3, instead of i-gre-, and even u-gd-\-le-\-
da-\-o in No. 8 b., St. 2, measures 2, 3, instead of w-|-g/e-|-^a-|-o. When
a word begins with two or more consonants, it may happen that the first will become syllabified and will be thrown back to the end of the preceding note value, so that it will function as an upbeat to the following value; for example, mus. ex. No. 27 e., St. 3, bars 8-9; No. 6 a., St. 3, bar 4; No. 24, St. 5, bar 1; No. 23 b., St. 4, bar 1. As these examples show, the syllabified consonants receive an individual ornamental tone in the melody.
This usage is spread over all central-eastern Europe and probably over many other territories.
SOME OTHER PECULIARITIES OF PERFORMANCE
1. Structurally essential upbeats are unknown in Eastern European folk music (except in Great-Russian and Greek) . The beginning of a sec- tion, especially after a rest, seems to need some short "preparation," however. This is provided by structurally unessential, usually very short upbeats, sung to syllables not belonging to the text proper. Such elements in Serbo-Croatian songs are: 9 (see p. 79), dj, h9j, i, n, m, hn, sometimes ej, oj (the latter not to be mistaken for ej, oj, used on downbeats: see p.
32), and so forth.
Anticipated repetition (doubling) of word-beginning "liquid" con- sonants may substitute for these syllables, or other syllable consonants taken from a word-beginning group of consonants. Examples in our melodies, Nos. 14, St. 1,2,5, 6, and 31 b., St. 2. These upbeats may appear even after line and syllable interruptions (for example, No. 14, St. 2-5).
[80]
Morphology
This usage is very common in the folk-song performance also of other Eastern European peoples.
2. Hiatus-filling consonants are used in Serbo-Croatian folk songs rather rarely (at least, in our melodies, the other editors having neglected this phenomenon), except for the two final syllables of the participle perf. ending, io, which generally appears as ijo. J is used in other instances, too (for instance, mus. ex. No. 28 b., St. 4, measure 2). Hungarians, Slovaks, and Rumanians use j for the same purpose, Rumanians also d; Turks prefer d; these peoples, however, use hiatus-filling consonants much more frequently.
RELATION BETWEEN TEXT AND MELODY
When a great many melodies of identical structure exist in a body of folk-song material, it is very improbable that each of them will have its own text: texts and melodies will be more probably interchangeable.^^ Although this might have been the case in the Serbo-Croatian material, too, in the course of time some of the texts have become linked to some of the melodies. Examples: Among our variant groups, only Nos. 21, 23, 31, and 44 contain identical texts, for two variants in each of these groups. In the subgroup "Tab. of Mat." 111-201 {8, ==4 + 4) there are 43 variant groups; among these 11 contain two (or, in a few cases, more than two) melodies with the same text. Examining the largest variant groups — "Tab. of Mat." 983 a.-ll. and our No. 27 a.-e. — we find not more than ten variants (out of 41) of which several have the same texts: five (g., 1., bb., ee., ii.) with U Omera vise Sarajeva, three (c, bb., our No. 27 b.) with Beg Aliheg (or other name) na kuli (or somewhere else) sjedase, and two (p. t.) with the same lullaby text.
These data prove that in the majority of the songs text and melody have remained interchangeable, within the limits given by the differences in the metrical structure of the text lines (and excepting songs of urban origin) .
Ceremonial songs and songs for special occasions. — The Parry Collection contains few songs designated by the singers as connected with the per-
^' This is the actual situation in the Rumanian rural material, for instance, within the limits set by the difEerences in the subject matter of the various ceremonial categories, as well as of the nonceremonial songs (but not in songs of urban origin) .
[81 ]
Morphology
formance of ceremonies, or with special occasions, distinguishable from others by style or subject matter.^* The printed collections, especially those of Dordevic and of Busetic (Iz Levca), contain many ceremonial or other designations. An enumeration of these is given below, in various groups and subgroups.
Songs performed in connection with : I. Outstanding events
1. Svatovska^^ (wedding song); 2. Naricanje (song of mourning) II. Events of e very-day life
1. Uspavanka (lullaby); 2. Na ranilu (morning song)
III. Work
a) Work in the fields
1. Zetvarska (harvest song); 2. Kad se here seno (hay-gathering song); 3. Na kopanju {kad se kopa loza; pri kopanju kukuruza) (hoeing song [when hoeing vine; when hoeing maize]); 4. Kad se ide u branje zdravca (at "zdravac"^^ gathering); 5. Pri radu (when at work) ; 6. Kad se vraca sa rada (when returning from work [in the fields])
b) Work in or around the house
1. Kad se vlaci vuna (when combing wool) ; 2. Cilimarska (carpet- weaving song) ; 3. Na prelu (spinning song) ; 4. Pri huckanu (but- ter-whipping song)
IV. Recreation
1. Sedeljka (evening gathering of girls for some kind of light work) ;
2. Pripev^'' (song to match a girl and a boy) ; 3. Pozdrav devojaka na saboru (girls' greeting song); 4. Ljuljaska pesma na Durdevdan (swing song for St. George's day); 5. Slavna (drinking song)
V. Dance and play songs
1 . Kolo {zensko-, musko- and so forth) , round dance [for women, for men, and so forth]; 2. Poskocnica (jumping dance); 3. Pljeskavica (clapping dance); 4. Caruzanke; 5. Balkanka; 6. Paracinka (4.-6. are designations of various dances); 7. Vanjkusac (pillow [or cush- ion?] dance); 8. Oro (round dance); 9. Igra {Decja igra) (play or dance song [children's play song])
8^ See Part Two, p. 253.
^ Various forms of the same designation are used by the various editors; only one form is given here. *• Crane's bill (geranium macrorhizum). *^ This word means also "refrain."
[82]
Morphology
VI. Songs connected with certain days of the year
1 . Lazaricka (Lazarus's day song) ; 2 . Kraljicka ([Song] of the "Queen" [on Trinity Sunday?]) ; 3. Kralja na Spasovdan ([Song] of the "King" on Ascension day) ; 4. Na Svetog Nikolu (St. Nicholas's day song) ; 5. U oci Jeremijinog dana (on the eve of Jeremy's day); 6. U oci Durdevdana (on the eve of St. George's day) ; 7. "Pevaju se prilikom maskarada o Beloj Nedelji" (on Carnival Sunday); 8. Krstono^ka (Rogation day song); 9. Uskrsnja (Easter song); 10. Na premlaz {hacija) (At the milking festival) ; 11 . Koleda (winter-solstice [or Christmas] song); 12. U vreme mesopusta (song for the Lenten season) VII. Songs connected with rainfall
1, Dodola (rain-begging song) ; 2. ^^ Pevaju deca kad pada kisa" (chil- dren's song when it is raining)
REMARKS
I. 2.: Represented, unfortunately, by only two melodies (Dordevic, Iz LevCa) ; despite the fact that a scarcity of songs of mourning is scarcely conceivable.
IV. 1.: Almost every song not connected with some other ceremony or occasion bears the designation sedeljka in Dordevic and in "Iz Levca." The designation is apt to mean that they are generally sung at such gath- erings as the most suitable occasion for community singing.
IV. 5.: Sung on name-day festivals.
V. 8.: Only in Kuhac {Oro seems to be the Bulgarian equivalent for Kolo).
VI. 11.: Represented by only two examples. Would this mean that the usage is dying out or that the collectors had neglected this category of songs?
VII. 1.: Represented by four examples, although rain-begging songs probably are known in every village.
Very frequently the designations apply only to the content of the text.
A text in the Parry Collection which contains conversations among singers concerning songs, has references, in addition, to "songs for love- making" {pjesme za asikovanje), to different kinds of kolo dance songs {pjesme za kolo), to game-songs of children and adults, humorous songs {smijesne pjesme), songs of auto and bus drivers (soferske pjesme), and,
[83]
Morphology
among Mohammedans, songs for the circumcision rite. According to this source, there are no special songs chosen for evening gatherings {za sijelo). Some songs are designated as shepherds' songs (cobanske pjesme). (A.B.L.)
Probably all these categories of songs exist also in the Bulgarian mate- rial. The Rumanians have wedding songs and songs of mourning (in every village, at least one kind); lullabies; harvesting and hay-gathering songs; winter-solstice or Christmas songs (their material is extremely rich in "Colinde," as they call the Koleda)f^ rain-begging songs (only in the southwestern areas); and many types of dance melodies (most of them only in instrumental form). The Hungarians have wedding songs and songs of mourning; harvest songs; name-day songs; songs to match a girl and a boy ; a kind of winter-solstice or vassailing song (called regosenek) ; and in addition a type of summer-solstice song which is apparently lacking in the Serbo-Croatian material. The Slovaks have the same categories as the Hungarians except the winter-solstice song, and in addition they have lullabies, "christening," and some other songs. ^^
As has been pointed out on p. 35, ceremonial songs in the Serbo-Croa- tian material have no specific group features in their melodies except that dance m.elodies are less ornamented, and the rather ancient looking melo- dies of certain categories (for various special days, such as Lazarus's day, St. Nicholas's day, and others) have a narrow range. These traits, how- ever, do not justify placing them in independent classes.
Only one kind of Serbo-Croatian song, the heroic (or gusle) song bears special melodic characteristics, and it has to be treated apart; this cate- gory, however, does not belong to the subject-matter of this book.
A category quite apart and very well known in the Hungarian, Slovak, German, and French material is that of the "children's play songs." These melodies often have no symmetrical form, but seem to be a potpourri of shorter or longer fragments of a certain kind. The Turkish material has this kind of melody in its rain-begging songs ; the Rumanians do not know it at all. It exists in the Serbo-Croatian material, too; their melodies, besides being real children's play songs, serve also as dance songs, name- day songs, festival songs, and (in one case) as wedding songs. Since they have an undeterminable structure, they are placed in a separate category ("Tab. of Mat.," p. 135).
^^ See Bartok, Melodien der Rumanischen Colinde, Universal Edition, Vienna, 1935; 484 melodies. ^^ It is not pretended that these enumerations are complete.
[84]
Morphology
CONCLUSIONS
After our intensive analysis of Serbo-Croatian folk-music we can state its main characteristics as follows :
In melody: Heterometric syllabic structures are characteristic; they result from the so-called splitting of text lines (see pp. 37ff .) . The prevalent structures of melodic content may be expressed with the formulas a A(v), AAA(^) or A A(^) A, and a b B(^) or a a^^) b (pp. 47-50). The main caesura in the melodies, which consist of two or three sections or phrases, is prevalently on the first degree of the scale, or on the (major or minor) second above. The characteristic range is narrow, between a fourth and a sixth ; very often from a second below the first degree of the scale to the fourth (or fifth) above, or from the first degree to the fourth above (p. 52). The characteristic scale is a pentachord formation, VII-4 or VII-5, with/^, a^ or a^ flat, and c^ as its main degrees and final stop on the second degree, g^ (pp. 59-61).
As for text structure, there is no stanza structure. Ten-syllable lines are preponderant.
As for mode of performance: heavy ornamentation is frequent (pp. 73- 74). Rests are used not so much for articulating as for decorative, expres- sive purposes (p. 74). The performance of many ornamental tones is emphatic (p. 78) .
Melodies with these characteristics are spread mainly over the "au- tochthonous" territories of Serbia, Old-Serbia, Montenegro, Hercegovina, and Bosnia.
Foreign influences are found chiefly in the remaining territories, espe- cially in Croatia, the Croatian seashore, Istria, and Dalmatia. Slavonia, Banat and Backa are, perhaps, transitional territories in this regard. The very small area of Medumurje is almost completely saturated with foreign melodies (p. 24). Foreign influence is shown by the preponderance of four-section melody structures, wide-range melodies, A^ b^ A b or similar content structures, and text-stanza structures. It is not strong in the "autochthonous" territories. As for the few older melodies there which suggest foreign influence, it originates probably from Turkish sources, either directly or indirectly through Bulgarian mediation (pp. 54-56) ; as for the numerous more recent melodies in the western areas, it origi- nates from central or western Europe. The very few recent^° borrowings
"" "Recent" means here, within the last sixty or seventy years.
[85]
Morphology
in the "autochthonous" territories come from New Hungarian material (p. 56-59).
INFLUENCE OF THE SERBO-CROATIAN AUTOCHTONOUS MATERIAL ON THE FOLK MUSIC OF OTHER PEOPLES
On Hungarian material there has been no Serbo-Croatian influence.
On Rumanian material some influence can be traced in the Banat area (including the counties of Torontal, Temes, Arad, Karas-Severin, and, perhaps, Hunedoara). This influence appears in the frequent use of the scale described above, with the final stop on the second degree. The very- characteristic Serbo-Croatian heterometric structures, however, did not exert any influence. Some melodies of urban origin are common in Serbo- Croatian, Rumanian and Hungarian material. This fact, however, is irrelevant when it comes to the "autochthonous" material of either of these peoples. Instrumental Rumanian dance melodies have probably undergone some Serbo-Croatian influence. The scarcity of published Serbo- Croatian instrumental melodies unfortunately prevents satisfactory com- parative investigation.
RELATION OF SERBO-CROATLAN AND BULGARIAN FOLK SONGS
The relation seems to be very close (p. 37). Since I have not yet been able thoroughly to study the great bulk of Bulgarian material which comprises almost 10,000 published melodies and in manuscript another 10,000, I must be content with the results of a superficial examination,