This volume presents the reader with songs that were never intended to be written down.

The forty texts and their multiforms included here were collected in 1934 and 1935 during Milman Parry and Albert Lord’s fieldwork in the former Yugoslavia. [1] The manuscripts from which they were taken are presently housed in the Milman Parry Collection of Oral Literature in Widener Library at Harvard University.

Any attempt to capture in writing the spirit of orally transmitted materials calls for certain sacrifices. This project has proved no exception. Beyond the obvious and unavoidable loss resulting from a written mode of presentation, there is the risk of isolating these songs from their historical and cultural context, of excluding the rich oral tradition from which they sprang. I attempt to indicate the breadth of this tradition by examining these songs in relation to relevant materials from the Parry Collection, as well as from several other published and unpublished sources. It is only within such a framework that the basic principles of composition and performance of these songs can be fully grasped. In selecting texts for the main body of the volume I have chosen to focus on a small group of singers whose songs exemplify these basic principles. The several interviews included in this volume cast additional light on the culture and society in which these singers lived, on their lives and family background, as well as on their art of singing.

I have made no attempt to provide a full concordance of the songs published here with those appearing in other sources of South {1|2} Slavic traditional poetry, but have rather limited my discussion to examples relevant to an understanding of principles of composition in the South Slavic traditional ballad. An exception to this restriction concerns songs published in Béla Bartók and Albert Lord’s Serbo-Croatian Folk Songs, since the present work draws on the same corpus of material and within it the same geographic region as that covered by Bartók and Lord. [2]

In light of the devastation in Bosnia and Herzegovina during the early 1990s one cannot but be profoundly grateful to Parry and Lord, whose interest and enthusiasm led them to preserve pieces of a tradition that even then was slowly dying and which in recent years has been nearly destroyed altogether. Systematic efforts to gather South Slavic traditional ballads began with the work of the legendary Serbian linguist and ethnographer Vuk Stefanović Karadžić in the early part of the nineteenth century, but he was hardly the first collector intrigued by these songs. They had attracted much attention already in the eighteenth century, when the Italian abbot Alberto Fortis published the famous ballad “Hasanaginica” (both in Italian and in the original language) in his Viaggio in Dalmazia of 1774. This song became particularly well known after Goethe provided his own artistic rendering of it (also in the mid 1770s). By 1778 Herder had already included it in his Volkslieder. [3]

Two of the greatest collections of traditional songs gathered in the South Slavic territory in the nineteenth century are undoubtedly those put out by Karadžić and by the Croatian publishing house Matica hrvatska, both of which contain several volumes of lyric, humorous songs, and ballads. [4] The methodology in both cases was similar, although the greater number {2|3} of collectors involved in the Matica series made it somewhat less uniform in approach. Karadžić, by his own admission, did not write down all the songs from dictation or singing, but rather noted some of them down from memory. Moreover, like the Matica hrvatska editors, he “improved” an occasional verse or omitted or added a line where he saw fit. [5] Unfortunately, there are no textual notes in either case indicating where such interventions took place. While the editors of the first collections of South Slavic ballads and lyric songs showed particular sensitivity to the question of multiforms, often including multiple versions of songs and (in the case of Matica hrvatska) providing whenever possible the name of the singer or the region in which the song was collected, the materials presented in these volumes are nonetheless burdened with flaws resulting from an inconsistent methodology and the lack of later technological advances. First and foremost, they hardly answer any of the questions pertaining to the singers and the performance circumstances, and they fail at times—it is impossible to know how often—in accurate representation at the textual level.

Commenting on the materials in the Parry Collection more than half a century after they were gathered, one can in this case too single out the problems that stem from a rather unsophisticated way of recording the songs, and this is particularly the case with the ballads and lyric songs. Despite these shortcomings (discussed below) it is still possible to isolate a considerable body of materials which, owing to the high scholarly standards of the collectors, stands in sharp contrast to the older collections of oral poetry from the South Slavic area. [6] Even though the songs in the Parry Collection performed by women were mostly written down, not recorded on aluminum disks, the breadth of the collected corpus is such that it is possible to establish with some certainty their place within the tradition. This factor is precisely what makes this collection of oral poetry so exceptional in terms of scholarly research. Of the 12, 554 texts contained in it, approximately 11, 250 are so-called “women’s songs” {3|4} performed by some 250 female singers from the Gacko district and about 50 male singers from Gacko and various other areas of Bosnia and Herzegovina. [7] While their assistants took down the vast majority of these songs from both dictation and singing, Parry and Lord themselves recorded approximately 250 songs (along with interviews with five singers) onto aluminum disks. It was from this body of recorded material that Bartók and Lord made a selection for their book, Serbo-Croatian Folk Songs. The rest of the South Slavic materials in the Parry Collection comprise about 1,550 men’s epic songs (half of which were recorded), and interviews with various male singers. [8]

When Parry and Lord set out on their journey to Yugoslavia in 1934, their goal was to collect a large body of epic songs that could be used to help explain principles of composition in Homeric epics. [9] Little did they know that the trip would become something of an odyssey in itself, one full of unexpected discoveries. Among these was a recognition of the overwhelming wealth and diversity of the women’s oral tradition in the area. Although they remained primarily interested in male performers, along the way they also collected as many songs performed by women as possible. For that purpose their primary assistant, Nikola Vujnović, helped them recruit and train two young men, Ibrahim Hrustanović and Hamdija Šaković, who had just completed their secondary education and who were willing, for a small fee, to visit villages in the Gacko area and write down songs sung by women. In addition to these two, many other local collectors, about whom we know little or nothing, submitted to Parry and Lord numerous notebooks filled with songs from the Gacko area.

It is clear that some of the notebooks in the collection were written down by children of illiterate singers. Understandably, Parry and Lord did not want to turn down any material that looked even remotely interesting. Something had to be done, however, to screen the process of collecting. They decided to ask Hrustanović and Šaković to supervise the gathering of songs, since some of the submitted manuscripts were illegible and some were clearly copies of various published song books (pjesmarice). [10] Some of the children who collected materials were close relatives and apparently shared their “treasures” with one another. As a result of this practice there are a few identical texts in the Collection. The fact that they share even the same metrical errors is a sure sign that the texts are not authentic. Despite the involvement of Hrustanović and Šaković, the quality of the written women’s songs in the Collection varies considerably. While some of the collectors, most notably Hrustanović and Šaković themselves, {4|5} followed the scholarly standards set by Parry and Lord, others were negligent, inaccurate, and inconsistent.

Determining the reliability of certain collectors’ work was thus an important preliminary step in selecting texts for the present volume. Equally important was the need to establish the identity of particular singers. Since the materials performed by women were collected in a relatively small geographic area, many of the singers came from the same villages, and it is not unusual to find that a number of them shared the same last name. Some were related to one another, and some even happened to have the same first name. This is the case, for example, with Almasa Zvizdić, whose talent and exceptional voice Lord singled out for praise. [11] Her maiden name was Sarić and she was married to Salih Zvizdić. They had no children and she was forty-five years old when the songs were collected. A second Almasa Zvizdić was also forty-five at the time of collection, but her maiden name was Duraković. Her husband’s name was Sujo Zvizdić and they had two daughters, Mejra and Rabija. Unfortunately, biographical information is scarce in the notebooks containing materials by these two singers, and one must consider carefully whatever information does appear there. In this case the collectors’ names helped in distinguishing the two singers: all the songs by the woman identified as “mother Almasa” were written down by her two daughters.

Approximately one-tenth of the songs sung by women in the Parry Collection was collected by Hrustanović and Šaković. Their own lengthy accounts about the process of collecting and about some of the singers provide valuable background information. The most interesting sections of these accounts are included in this volume. Especially noteworthy is a group of songs that they collected from their mothers, Hasnija Hrustanović and Emina Šaković, and from Hamdija’s aunt, Đula Dizdarević, all of whom were outstanding singers.

Partly on account of their reliability, I have chosen a majority of the songs from among those collected by Ibrahim Hrustanović and Hamdija Šaković. [12] Other collectors, whose work could be trusted and appears here, include Halid Dizdarević (six songs), Delva Hrustanović (two songs), and Ćamila Šaković (one song). Hrustanović and Šaković’s fieldwork is of particular interest not only because of its reliability, but also because they were Muslim and were well received in traditional communities that were otherwise inaccessible to outsiders. [13] The sheer number and quality of the songs by Đula Dizdarević gathered by Halid Dizdarević suggests that Halid might have been Đula’s son, or at least a close relative. With respect to the {5|6} remaining collectors represented in this volume, I would speculate that Delva Hrustanović was also a close relative of Hasnija Hrustanović since her name appears together with Ibrahim’s in one of the notebooks. Unfortunately, no information is available concerning the singers’ possible kinship with the last collector, Ćamila Šaković. Judging from her last name, it is conceivable that she was related to Emina Šaković’s family.

The question of which songs to select for this volume was, for a number of reasons, not an easy one to tackle. It was apparent from the very beginning that choosing a representative sample from the body of over 11,000 items would be an impossible task. Selection of the songs presented here was based on the quality of the available material, the reliability of collectors, the songs that have already been published, and the type of comparative analysis that would be possible were multiforms to be included. While Bartók and Lord placed a greater emphasis on music and accordingly chose the material for their book from among songs preserved in sound recordings, this volume draws, with the exception of two items, from the written corpus of songs. The intention was not to give “the best” samples from the Collection. However, as the three singers whose songs represent the nucleus of this book happen to be among the most talented performers, it is not surprising that their songs are often the most complete, dramatic, and polished. Contaminated materials and aborted songs are not a rarity in the Collection (although this is much more often the case with short lyric songs than with ballads), but these are used here only as additional sources for comparative analysis and are not included in the body of primary texts. [14]

One of the principal ideas behind this project was to present, inasmuch as possible, songs by a relatively small number of female singers. An initial plan to select only one singer was abandoned, since for purposes of analysis it seemed more productive to include materials by several singers from the same area. In the course of examining all of the relevant materials in the Parry Collection I found a large body of songs by three singers who were not only friends, but at least two of whom were also close relatives. These singers—Hasnija Hrustanović, Emina Šaković, and Đula Dizdarević—clearly felt at ease singing in front of their own children or relatives, and this is reflected in the exceptional quality of their songs. Multiple sessions and the absence of time constraints positively affected the number of songs and many versions of one song that could be taken down. Ibrahim’s mother, Hasnija, is the singer with the greatest number of songs in the Parry Collection—466 of them, over 10,000 lines of oral poetry in all. [15] Accordingly she occupies a central place in this volume and is represented by more songs than any other singer. The number of songs by the other two singers is no less {6|7} impressive: Hamdija’s mother, Emina, supplied 330 songs, and his aunt, Emina’s sister Đula, 338. [16] All three women were illiterate.

In most instances we know only the approximate age of the singers since birth records were non-existent and the year of birth was customarily remembered in relation to some important historical event. Ibrahim Hrustanović thought that his mother, Hasnija, was about seventy years old at the time of the interview, indicating a birth date of around 1865. In part of the interview with him not included here, he says that he himself was born in 1911 and that he was an only child. This would mean his mother was about forty-five at his birth; more likely she was somewhat younger than her son claimed. Đula Dizdarević was uncertain about her age and thought she might be about seventy-five, thus born around 1860. Elsewhere in the same interview Hamdija Šaković mentions that Đula (his aunt) was then about sixty-seven years old. Considering the age of her younger sister, Emina, who was fairly certain she was fifty-four (and thus born in 1881), I would guess that the age difference between the two sisters was probably less than twenty years, and that Đula might have been about seventy at the time of the interviews.

Further information on the unusual—even adventurous—lives of Hasnija and Đula turns up in the interviews with Hasnija’s son and with Đula herself. Some of the details included in them, while not strictly biographical, provide nevertheless a fuller picture of these women’s experiences and how they viewed the society in which they lived. Đula’s account of her life is especially interesting since it comes directly from the singer. Lord was well aware of the value of this particular interview: “She had lost her voice and all idea of pitch. Mr. Bartók found that her melodies could be transcribed only approximately; therefore they were not suitable for musical study. Her age, however, and the dignity of her family won her a place in this gathering. She remembered the old days very well and was always ready to talk about them. Moreover, the texts of her songs were good, even though her singing was not what it had once been. The Dizdarevići had been rich landowners in Korite, a small place on the road between Bileća and Gacko, during the Turkish occupation. Her stories of historical events at the end of the last century and the early part of the twentieth will be well worth publication when they are translated.” [17] There is not as much biographical information for Emina Šaković, but just as with the other two main singers, the interviews allow us some knowledge of her art of singing.

Songs by women other than these three are multiforms and are meant to provide at least some idea of the context of the tradition. The selection of these materials was based primarily on the existence of a relevant multiform to supplement those by the three primary singers and, as mentioned above, by the reliability of the collectors. Unfortunately, as the collectors supplied only the most basic information about these singers and there exist no interviews with them, {7|8} our knowledge of their lives and views on singing is far more limited. All of them were from the Gacko area, most were illiterate. It is not possible to determine with certainty whether any of them were related, although their last names suggest this may have been the case. In terms of their ages, seven of them were between thirty-five and fifty-five years old. The youngest, Hajrija Šaković, was in her early twenties; the oldest, Kana Tanović, said she was ninety. [18]

Most songs presented here have at least one multiform, the majority of them more than one. For about 98% of them a multiform exists in the corpus of songs from the Parry Collection. Numerous close or distant multiforms are found also in the Matica hrvatska edition Hrvatske narodne pjesme, or Vuk Karadžić’s Srpske narodne pjesme, or both. [19] Wherever possible multiforms performed by the three central singers are provided. The main reasons for incorporating multiforms by other singers are: to show either a close proximity or considerable distance between them; to observe the use of formulas and formulaic expressions in a group of related songs; to compare development of the plot; to follow a cyclical path of certain themes. Cycles 1 and 3 include more multiforms than the others, and while this could have been done in several other instances, limitations of space meant that many items worthy of inclusion had to be left out.

Owing to the nature of the materials on which this monograph is based, my approach is text-centered, although wherever possible I introduce broader contextual elements into the discussion of the selected songs. While the frameworks of tradition that produced these songs could at least partly be substituted through investigation of the multiforms and other related materials, the paralinguistic and performative aspects will in most of the presented cases remain unknown. [20]

The relatively recent move from a text-centered approach to a broader contextual examination of traditional performing in the South Slavic oral heritage is an important development, although unfortunately it comes at a time when the vast majority of oral traditional genres are in the process of disappearing or have already ceased to exist. [21] For many performances from this part of the world the only record we have is that of the text. Moreover, as already pointed {8|9} out, in numerous instances uneven policies of collecting and editing have rendered some of the texts unusable for certain types of analysis. In other instances (and this is the situation with the present volume), the texts have not been edited, but the non-textual level of the performances will in most cases never be available to us. What sets this group of ballad-texts apart from most collected in the South Slavic territory in previous decades, however, is the fact that we do have a considerable amount of ethnographic data at our disposal, enabling us to explain certain aspects of their composition that would otherwise remain unclear.

A sensitivity to issues other than the text or singing in a performing situation can be of crucial importance for understanding the function and meaning of these materials in the life of an individual performer, or of his/her community. As I demonstrate below, the type of performance circumstances typical for the South Slavic oral epic, for instance, differ greatly from those customary for the ballad. [22] In fact, they differ so much that, while it is indeed appropriate to refer to the singing of an epic song as a “performance,” one needs to bear in mind that the specific setting of an audience expecting entertainment of a particular type is for the most part non-existent in the case of the South Slavic traditional ballad. Although the singing or telling of South Slavic traditional ballads is referred to as performance for practical reasons, the researcher needs to be aware that in most cases these events were not presented as performances, but rather as an accompaniment to some other everyday activity. Consequently, the role of the “audience” (if any) is different, as is the type of communication that develops between the singer and the listeners. The function that the ballad has in a community also has nothing in common with that of the epic, nor does the position in society of a ballad singer correspond to that of an epic singer, and this is particularly true of the Muslim performers.

Some scholars stress the importance of not only the context, but also the process of contextualization, or the way in which “performers and audiences use poetic patterning in interpreting the structure and significance of their own discourse.” [23] The fact that Parry and Lord were deeply aware of this issue is reflected in the type of materials they collected. Their conversations with the singers and the members of their community testify to their attempts to provide data concerning the singers, as well as to establish the frames of contextualization for a specific group of performances. This is especially evident in their interviews with the South Slavic epic singers, in whose perception and understanding of the process of oral composition they were particularly interested. [24] In recent decades much emphasis has also been placed on an approach that does not perceive the singer as “an informant,” but rather has the researcher become a part of the traditional community and form his views about the collected materials {9|10} from within. [25] As many scholars attest, the influence that the outside researcher may exert on the process of collecting should not be underestimated. [26] It is evident that problems often arise even in those situations in which much effort is paid to the preservation of customary performance circumstances, such as during Parry and Lord’s sessions in which they interviewed some of the female singers from Gacko. Despite their arrangement to hold these sessions in the comfortable setting of a local house where women felt at ease enough to take off their veils and to communicate freely, and their attempt to gather a group of women who were not unknown to one another, the excitement caused by an unusual request from foreign academics to record their songs was enough for these women to become self-conscious and for their performances to be affected. [27] This becomes particularly evident if one compares some of the ballads recorded on phonograph disks during these interviewing sessions with the materials which were written down in the singers’ homes by the singers’ relatives and friends who worked as Parry and Lord’s assistants. Although the audio materials provide the types of data (music, pauses, pitch, interactions amongst singers, etc.) lacking in the written materials, it also shows how these exceptional circumstances caused most of the ballads to become truncated and to be delivered in a considerably poorer version than those performed for the singers’ relatives/friends in a customary setting. In quite a few instances the singers complained to Parry and Lord that they forgot how the song continues and would abruptly end their singing at virtually any point, but there is written evidence that under different circumstances they remembered the ending of the song quite well, or at least were able to finish a song in some logical way. If we remember the fact that these singers came from a conservative Muslim environment in which they had no contact with unknown men, it is not surprising that occasions such as these recording sessions would represent, despite the scholars’ best efforts, a highly curious event during which the women could not necessarily behave or sing in the same way they customarily would in their own households and with their own family members. [28] Keeping this in mind it is perhaps possible to say that, in light of a performance-centered approach, the written texts of the ballads from Gacko present in some way a more realistic {10|11} rendition of at least the textual side of the performance than do those recorded on phonograph disks.

Clearly there is a definite limit to how far we can go in terms of performance analysis of the materials collected without the advantages of contemporary technology and researchers’ attempts to maximally neutralize their own influence on the collecting process. It is doubtless therefore that some of the problems mentioned above will permanently taint the analyses of materials collected in previous decades. Modern computer technology, however, with its potential for establishing extensive databases and searching for specific types of traditional building blocks, enables us to familiarize ourselves relatively quickly with many of the facets of a traditional verbal art and to become aware of the context of tradition in ways unavailable to previous generations of researchers. Thus although some aspects of a performance may be permanently lost, others can be made accessible—provided, of course, that the body of material under scrutiny is large enough and that the circumstances of its collecting did not alter its quality. The existence of a large number of songs in the Parry Collection, most of them with multiforms, has enabled me to create such a database with the purpose of studying the types of transformations that happen both in multiple performances of a song by one singer, and in multiforms performed by other singers from the same region. Furthermore, such a wealth of material has made it possible to monitor the appearances and modifications of formulas, blocks of lines, and themes in various genres. [29]

Lord discussed the value and importance of “women’s songs” in the Collection and spoke about a long-term project to publish a part of the written corpus. This series was never realized, but some of the issues raised by Lord are germane to the present project and are worth quoting here: “If the women’s songs which were taken down from dictation are used with proper care and studied in relation to the recorded songs, they should make it possible to reconstruct the complete repertory of formulas, lines, and even songs of the oral lyric and ballad tradition of the Moslem population of Gacko; by comparing those from the younger members of the community or family with the same songs from the older singers, the reader will obtain some idea of changes from one generation of singers to another.” [30] The first part of this plan was accomplished with the publication of Bartók and Lord’s Serbo-Croatian Folk Songs. The present volume is intended as a further contribution to the series Lord envisioned. {11|12}

The process of selecting texts for this volume involved another complicated decision, one related to the question of genre. For practical purposes one could divide the songs performed by women into shorter lyric songs, humorous songs of various length, and longer narrative ballads. In reality this division is not always ideal since borderline cases, songs containing elements from all groups, and fragments are not rare. [31] The materials by the three primary singers are no exception to this. I came to realize during the initial stages of the project that, owing to such a diversity and volume of songs, the inclusion of more than one genre would seriously compromise the quality of the present volume and force me either to treat certain issues on a superficial level or omit them altogether. In addition, fewer multiforms could have been included, making the analysis of the stability of a given song problematic. Previous scholarship has placed a greater emphasis on lyric songs than on narrative ones, and the absence of a monograph in English concerning ballads from Bosnia and Herzegovina has created an imbalance which I hope to redress with this study. [32] More importantly, a realization of the compositional and cultural issues proper to the ballad, and the necessity of investigating these in order to situate the ballad within the context of South Slavic oral traditions, persuaded me to devote my full attention to this genre. A further motivation (and not the least among these) for focusing on ballads was the stunning artistry and beauty of some of these songs.

Mention of the term “ballad,” however, opens a Pandora’s box. Although the main purpose of this work is not to propose a redefinition of shorter verse genres in South Slavic oral traditions, a recognition of problems involving terminology is nevertheless in order. The guiding idea behind the discussion presented in this chapter is therefore not only to point out a number of potential terminological traps, but also to show how rigid labeling of this fluid material can lead to misunderstandings of its essential qualities. In addition, my intention is to demonstrate through several examples that the same principal structural elements constitute the basis for composition in different verse genres. What is implied by the terms “narrative song,” “ballad,” and “humorous song” often depends on the scholarly tradition in which they are used. Complicating things yet further, another term, “romance,” is frequently employed in the writings of South Slavic scholars. In the Western European tradition “romance” denotes a long narrative, often in verse, dealing with the adventures of chivalric heroes. In South Slavic traditions it designates instead a narrative song which, unlike the ballad, is light or even mocking in tone. Its focus tends to be love and the relationships within a family. In order to avoid confusion I have opted to refer to this group of songs as humorous songs. {12|13}

Definitions of the ballad are equally varied. In studies that focus on South Slavic materials there seems to be no agreement about whether the tragic ending of the ballad constitutes its most important feature. While some scholars follow a more traditional division and emphasize the tragic outcome as its hallmark, others stress components such as specific thematic frameworks, concise dramatic development with lyric overtones, dialogic form, and singularity of conflict as more important elements. [33] Older scholarship from both eastern and western Europe tended to view the elements of music and dance as one of its principal features, but considering that South Slavic ballads were both sung and recited (or simply told), classification of this type does not resolve the question of genre. [34] Since my intention is not to investigate the ballad from a theoretical viewpoint, for practical purposes I will work from the following definition: the ballad is a narrative song which conveys in concise and dramatic fashion a fragment of human existence, typically a conflict of a personal nature, giving particular prominence to its psychological and emotional aspects. [35] In my view, it is not necessarily the tragic outcome, but rather a solemn and emotional tone, along with the features listed above, which determine whether a song should be viewed as a ballad. Such a broader view of the term ballad provides a possible way forward in dealing with many of the terminological problems that presently exist. For example, in this volume there are songs which are clearly multiforms, but one ends on a sad note, another on a happy one. [36] According to a traditional division the former would be a ballad and the latter a narrative song, although we are essentially speaking of two multiforms. Furthermore, is not narration one of the primary features of the “tragic” ballad? Accordingly, there is no reason why we could not refer to it as “a narrative {13|14} song” as well. It is also often stressed that the ballad, on account of its tragic plot, typically carries greater emotional weight than the other folk genres. A narrative song with no tragic ending, however, can also convey a range of complex emotional and psychological states. [37] A good example of such a song is number 9 below (by Hasnija Hrustanović).

The customary division of South Slavic folk poetry into men’s (epic or heroic) songs and women’s songs (lyric songs, ballads, and humorous narrative songs) has caused perhaps even greater misunderstanding. As early as 1824 Vuk Karadžić recognized that the boundaries are far from clear: “All our folk songs are divided into heroic songs which people sing with the gusle*, and women’s songs which are sung not only by women or girls, but also by men, especially young men, most often in unison. Women’s songs are sung by one or two people for their own entertainment, while heroic songs are sung mostly for others to listen to; for that reason, in the performance of women’s songs more attention is given to singing than to the song, while in the performance of heroic ones the attention is turned mostly to the song.” [38] He mentions further on in the same text that heroic songs were mostly sung in Bosnia and Herzegovina, Montenegro, and the southern mountainous regions of Serbia, and that in those areas most households had a gusle. Moreover, he claims, most men knew how to perform with the gusle, and that many women and girls were endowed with this skill as well. It is hardly surprising, then, that Karadžić faced a dilemma when it came to grouping some of the materials he had collected, or that he thought that many of them appeared to be on the borderline “between women’s and heroic songs.” [39]

Basing his research on Karadžić’s two volumes containing “women’s songs,” Maximilian Braun concludes that the ballad is not represented in them at all. [40] In his view, the narrative songs from these two volumes rely heavily on motifs found in epics, modified so that women become the carriers of the narrative: “The ‘lyric-epic’ songs from [Karadžić’s] collection must be defined therefore as non-heroic epic songs, that is as narrative songs, which are closely connected to the formal procedure of the heroic epic, the content of which, however, cannot be assessed as a poetic expression of the heroic world view. For the most part these are ‘epic women’s songs’: they deal with either specifically female problems or general epic situations {14|15} from a feminine viewpoint.” [41] Similarly, Nada Milošević-Đorđević refers to these as “non-historic epic songs.” [42] Problematic in such arguments is their excessive reliance on content at the expense of a consideration of structural components. I believe that ballads do not simply follow the “formal procedure” of longer epics: while these two genres share many principles of composition and even a number of themes, they at the same time possess qualities (discussed below) that set them apart. Similar terminological problems present themselves in the editorial work of Nikola Andrić. In the titles of the various volumes of Hrvatske narodne pjesme he mentions romances, ballads, narrative songs, humorous songs, and love songs. [43] By narrative songs he understands those materials that “describe a ‘woman’s’ event without a tragic or sad ending. These are songs which do not fall into the category of ballads and romances, but by their character do not belong to heroic songs either.” [44] He, like Braun, proposes that this subgroup should perhaps be defined as “women’s heroic songs.”

This persistent mixing of genre and gender causes further confusion as, indeed, there exist heroic songs sung by women. Andrić was well aware of this fact since in another article he discusses the manuscript of songs by Kate Murat née Palunko from the island of Šipan near Dubrovnik. [45] This superb singer provided about 18,000 lines of oral poetry and was well versed in both short and long genres, including epics. [46] Andrić refers to Murat’s longer songs as “epics and longer women’s narratives” and speculates that these songs are “the best proof for how our heroic men’s songs evolved from women’s songs.” [47] In her critique of Andrić’s analysis, Tanja Perić-Polonijo comments that the main characteristic of the Šipan collection is “a permeation of genres and a fading of the boundaries between different types of songs.” [48]  {15|16} She only partly agrees with Andrić and claims that, while the songs by Kate Murat can indeed be divided by genre into epics, ballads, and transitional genres based on their formal features, with regard to the presentation of their subject matter only a small number of her epic songs can be called heroic. The reason for this is the insufficiently pronounced martial spirit, a quality which separates them from epics of the inland regions. While Perić-Polonijo disputes Andrić’s classification, according to which many songs from Murat’s collection are viewed as epics, she does agree that there exist a few items that fully exhibit features of this genre on the levels of both style and subject matter. In her opinion the ballads from the same collection are highly influenced by the epic style, especially in their form. She explains this atypical situation as having a geographical basis: the Dubrovnik area has always been a cross point of the continental epic tradition on the one hand, and on the other the balladic Mediterranean tradition. [49]

Compared with the songs of Kate Murat, the ballads in this volume—and in the Parry Collection as a whole—are much clearer cases of ballad narratives, which in their form and subject matter fall into the frameworks of the definition of this genre provided above. I have not encountered any songs by the relevant singers, or for that matter by any other female singers in the Parry Collection, that share both the stylistic features and the length of epics sung by Muslim men. Even the transitional cases, those with epic-balladic qualities, are extremely rare. The majority of the borderline songs (of which there are a considerable number) exhibit mixed characteristics of ballads and lyric songs. This is hardly surprising since both genres tend to be performed by women and are also related on the level of subject matter. What is surprising, though, is the fact that the expected interference of epic elements in the narrative songs performed by women is so minimal, especially if one takes into consideration the rich epic tradition of this area. [50] {16|17}

While many of the same principal structural elements (and sometimes even the same subjects) appear in both the epic and the ballad, the format of the songs and the performance circumstances differ owing to the distinct positions that male and female singers held in this traditional Muslim society. [51] This cultural environment, which was decidedly unlike that found in Dalmatia, is a factor that cannot be ignored. Despite the constrictions of the type of patriarchal society that existed until recently along the coast, the degree of freedom and contact with the outside world that women enjoyed there was considerably higher than in continental areas such as Gacko. Furthermore, the respective practices of the Muslim and Catholic religions defined differently women’s roles in society and in many ways determined their behavior, including the ways in which they could participate in oral traditions. It is conceivable that in societies in which women enjoyed more freedom they could, if they wished, perform songs customarily performed by men, and that they could even have had an audience made up of both sexes. The more contact they were allowed with men, the fewer obstacles there were to taking an active or passive role in what was traditionally viewed as men’s singing domain. As a result, the degree of influence, in particular on the structural level, of epic poetry on the ballad is more profound. In an environment in which women could not mix with men, or even go into public without being covered and usually escorted by a male family member, it is less likely that this type of permeation of genres could occur. In such traditional areas women sang for one another or alone while doing housework and taking care of children. There was no performative aspect to these songs, as was the case with epics. The female gatherings were significantly smaller than those of men, which would typically take place in kavane where they went for discussions of politics and for entertainment. [52] This kind of relaxed setting was naturally more conducive to the composition of the longest of oral genres.

“The term ‘women’s songs’ is almost a negative one,” argues Lord, “It means songs that are not men’s songs, songs that belong to the social milieu of the women of the group, and not songs which the old men would ever indulge in. (…) They are much wider in their scope and are found in much of the communal life of the village, almost everywhere, in fact, except among the old men.” He further points out that “[t]hey are the ‘popular’ songs of the village. It is therefore not to limit them to women, but to distinguish them from men’s songs that these {17|18} are called ‘women’s’ songs. They do not belong to the heroic masculine milieu.” [53] Indeed, it is not difficult to find examples of songs in genres typically viewed as “women’s” which are performed by men. A point in case is song number 33 by Nike Stjepović, entitled “Miloš čobanin” (“Miloš the Shepherd”), from the collection Narodne pjesme iz Luke na Šipanu. With regard to the materials from the Parry Collection, a good example is text 2304, a ballad about the frozen bride, by the male singer Avdija Ljuca, which has a shorter multiform in text 2852 by Hasnija Hrustanović. [54] As might be expected, the ballads in the Parry Collection that are performed by men maintain an imprint of the epic style. Often a ballad with the same subject matter is longer when rendered by a male singer than by a female one, as is the case in the example of Ljuca and Hrustanović. Men knew some of the ballads since they could easily hear female relatives singing in their own households, and many were raised to the tunes of these songs as sung by their mothers. Women, though, could not join the kavana gatherings. [55] Despite this restriction they did have limited exposure to epics at the occasional domestic gathering and were familiar with the formulas and plots from men’s repertories. [56] If a member of the family was a guslar* this exposure was considerably higher. In my view, their knowledge of epics is less the issue here than are the restrictions of the society that clearly defined a woman’s position. Unlike their Catholic and Orthodox counterparts, these women simply did not feel the freedom to sing in the genre reserved for men.

When Karadžić writes about encountering female guslars he does not specify whether this was the case throughout these regions or whether the singer’s religious affiliation played any role. I strongly believe that this kind of singing practice was more likely to occur in those areas in which a strict separation of women and men was not pronounced. As the Muslim regions of Bosnia had the most rigorous rules in this regard, it is not at all likely that women from those areas would have performed with the gusle, since this type of performance implies the degree of public exposure unacceptable to an observant Muslim woman. Mary Coote’s conclusions about the distribution of genres roughly confirm this division: “The distinction between heroic songs and women’s songs is less sharply drawn in many areas surrounding the central core of Serbian folksong, in the northwest in Croatia, Slavonia, and Dalmatia, and in the eastern South Slavic tradition in Macedonia and Bulgaria. It holds most consistently for the patriarchal traditions of old Serbia and Montenegro and the Moslem tradition of Bosnia and Hercegovina.” [57] Lord is far more cognizant of the religious/cultural component: “In a {18|19} society in which men and women are separated so frequently, where men and women assemble together almost never, but each group separately, it is inevitable that each group would have its own kind of entertainment.” [58]

Two singers from the Parry Collection, Đula Dizdarević and Halima Hrvo, claimed they were able to sing epics, though there is only limited evidence to support their statements. Interestingly, their respective fathers were recognized guslars. My suspicion is that in Đula’s and Halima’s perception any song sung by a man was qualified as “heroic.” Đula classified text 12 as “heroic” on account of having learned it from her father, who performed it with the accompaniment of the gusle. However, this song does not exhibit stylistic features of the epic. [59] The greater length and elaboration of ballads performed by men probably prompted Đula and other singers like her to distinguish these songs from those performed by women. Another gusle song in Đula’s repertory is indeed epic in style and although it could potentially stand on its own, many elements in it suggest it is an excerpt from a considerably longer epic song. [60] Similarly, it is conceivable that the two multiforms of a song qualified as “epic” by Halima Hrvo could potentially be rendered as an epic, but her particular performances do not fall into this category. [61] One of the multiforms is an aborted attempt, while the other could at best be considered a fragment from an epic song. Whether Dizdarević and Hrvo, or for that matter many other female singers from this area, possessed the ability to compose epics must remain uncertain. From the fragments available to us, however, I would venture to say they were capable of this. Had these singers lived under different circumstances, perhaps like their counterparts a few hundred miles away on the coast, they might more readily have tried their skill in what they perceived to be men’s genres.

In some South Slavic areas the divisions among genres are thus sharply drawn, while in others these are more fluid. In both situations the points of contact at the level of compositional elements and subject matter are not dependent on genre. These common components cross with equal ease boundaries of gender, region, ethnicity, and religion. [62] Even in those areas that maintained a strict division between the sexes there are stories that are popular with both male {19|20} and female singers, but the format in which they are rendered tends to be different. [63] Different, too, are the viewpoints expressed in them and the aspects of the stories that are emphasized. In such areas the gender of the performer expressing his/her experience of an essentially fe/male world—as well as an audience that shares his/her experience—influences the primary orientation of songs which overlap in subject matter but can appear in two or more genres.

A comparison of the basic plot of what Lord refers to as “return-rescue” epics with that of a group of ballads belonging to the rescue-and-wedding cycle illustrates this similarity of subject matter in different genres. The plot of these epics involves a hero rescuing his friend, sometimes with the help of the captor’s daughter. The daughter is subsequently abducted by or escapes voluntarily with the hero and marries him. Lord mentions as a typical example of these epics “Hasan of Ribnik Rescues Mustajbey” (performed by a male singer), but there are many more songs with the basic “return-rescue” plot in the Parry Collection. [64] In some instances it is the wife or the sister of the prisoner who come to his rescue. [65] Among the materials in the Parry Collection performed by women, one by Stoja Bjeloglav is of particular interest in this regard. [66] In this ballad, entitled “Marko Kraljević u azačkoj tamnici” (“Marko Kraljević in the Azak Prison”), the captor’s daughter herself tries to help the imprisoned man, thereby taking on the role of the “hero.” She begs her father to release Marko, but her attempts are in vain. Upon his request she brings Marko paper and ink so that he may write home of his unfortunate fate, at which point the song ends. There is no wedding {20|21} in this song and the abrupt ending suggests this may not be not the whole ballad, but only a fragment of it. Further evidence for this view comes from a group of songs on the same topic. In “Vladimir i Kosara” (“Vladimir and Kosara”), a ballad collected on the island of Korčula, Croatia, the maiden successfully persuades her father to release the imprisoned man, whom she then marries. [67] In two other multiforms (one from Knin, the other from Krajina in Croatia), an Arabian girl rescues a man from prison and converts to his religion after he promises to marry her. [68] There exists also a multiform by the Šipan singer Kate Murat that is much longer than any of the three multiforms just mentioned. [69] In it the captor promises freedom to the prisoner if he converts to Islam and marries his daughter. The man refuses, but the daughter, saddened by the fact that he will be executed, rescues him and escapes.

Yet another version of the rescue-and-wedding plot is found in a ballad collected by Vuk Karadžić. [70] In this song the maiden’s brother tries to marry her to a man she dislikes. During the night the maiden wakes up the servant and gives her word that if he escapes with her she will not betray him. In other words, she promises to marry him and save him from serfdom. Indeed, in the message she later sends to her mother she mentions that not only did she marry the serf, but also that she converted to his religion. Once again a woman converts to Christianity in order to marry the man of her choice. A twist on the basic plot of a maiden rescuing her beloved comes in a ballad by Sala Pošković, in which a sister rescues her two brothers from prison by promising to marry their captor. [71] After her brothers are released, she and the captor are married. As they are about to consummate the marriage, the maiden pulls out a knife hidden in her dress and kills him. Disguised in his suit in order to deceive his mother, she confiscates his treasure and reunites with her brothers. [72] The songs by Pošković and Bjeloglav, as well as some of the other multiforms mentioned above, exhibit numerous similarities in their descriptions of the conditions in prison, of the prisoners’ physical appearance, and {21|22} most importantly the idea of a “maiden rescuer.” All of these elements are indications that they belong to a group of closely related cycles. [73]

In the ballads performed by women that treat various rescue operations of men, it is women who remain the focus of attention and it is mostly their viewpoints that are expressed. Despite their general overlapping on the level of subject matter with a large group of epics, one should bear in mind that these ballads, by virtue of belonging to a different genre and therefore having different stylistic features, realize the rescue-and-wedding plot in a far simpler and less digressive fashion, with a greater emphasis on drama and individual psychology. The “return-rescue” epics feature an assortment of obstacles in locating the prisoner, battles with various enemies, and sometimes even several weddings at the end. Their equivalents in the ballad texts are much more straightforward in this respect, which is in keeping with the principal features of this genre.

Formulas, blocks of lines, and certain themes travel with even greater ease across genre boundaries. While the recognition of shared formulas in different genres is relatively straightforward, correspondences at the level of larger segments may pass unnoticed, in large part because these simply do not appear as frequently as do the formulas. Since it is evident from the excerpts included below that the overlapping of themes, and particularly of blocks of lines, includes the sharing as well of a certain number of formulas and formulaic expressions, it will be more productive for the present discussion to turn immediately to an examination of these larger segments.

Writing about the problem of stability in the oral traditional song, Lord concludes that “no matter what the size of the composition, the traditional singers work with comparatively small blocks of lines intermediate between the formula and the theme.” [74] In his view, these groups of lines of varying lengths (ranging typically between two and four verses) are “memorable,” but not memorized, and can be linked in different ways and modified depending on the context. Although Lord limits his observations to larger structural components within the same genre, such units can also be tracked across different genres. This is particularly valid in examining corresponding blocks of lines since, owing to their brevity, they are easily incorporated in all the genres. I have found that they can appear in short lyric songs, some of which are characterized by a relatively stable wording, as readily as in longer ballads, or even epics, genres that show little or no tendency towards maintaining a stability of wording.

A very good example of a shared block of lines is a passage found in, among other places, Avdo Međedović’s “Ženidba Vlahinjić Alije” (“The Wedding of Alija Vlahinjić”), an epic which in its recited version has 5883 lines and in the sung version 6048. The four lines of {22|23} interest describe a beautiful maiden for whose hand some thirty heroes compete, and who promises to marry the one capable of fulfilling her request. In the recited version of this song these lines read:

In the sung version this passage is slightly modified, which is to be expected since it is not memorized. Its basic structure and formulaic pattern, however, remain the same:

In ballad number 12 in this volume (text 11686) by Đula Dizdarević, there is a corresponding block of lines:

This block, which is one line shorter than the one in Međedović’s songs, is also part of a description of a beautiful maiden. In this song, however, she is preparing to run away with her beloved during a moonless night, but is tricked and stolen by another man. This same block appears as well in two lyric songs. In the first, by Emina Šaković, a maiden praises the beauty of her beloved using words similar to those in the two epics and the ballad:

In the second song, by Đula Dizdarević, a young man wishfully thinks about a maiden:

This list of examples, though not exhaustive, suffices to demonstrate the use of the same block of lines in songs of varying lengths and genres. It should be pointed out here that the formulas “pospuštala rese,” “popucaše rese,” “popuštao rese,” and “popustila rese” differ somewhat in their semantic range, but are very close phonetically. This, unfortunately, cannot be achieved in translation. The wording in these verses is not identical; this is true also of their renderings in different songs by the same singer. In addition, Dizdarević successfully employs these lines in songs belonging to different genres. Most importantly, the contexts in which they appear are quite distinct in all the songs. This is not a unique example in the South Slavic oral traditions, and it proves that the lingua franca of oral traditional poetry was so commonly known and widely used that it could accommodate not only the basic formulas, but also more complex segments. [79]

We will now make another step in the same direction by looking into examples of corresponding themes in two different genres, the epic and the ballad. [80] In both multiforms of Međedović’s epic “Ženidba Vlahinjić Alije” there is a theme of “prosidba,” or ceremony of asking in marriage. This theme appears also in ballads 1f by Zulka Tanović, 5 by Hasnija Hrustanović, and 5b by Đula Dizdarević, all included in this volume. [81] The basic grouping of ideas contained in it is the same in all the songs: a beautiful maiden has a number of suitors who are invited to her house for a ceremony lasting several days, during which they compete for her hand. [82] As a part of this festivity they are asked to submit whatever engagement presents they have brought along (customarily these consist mostly of rings and golden apples). The presents are placed on a tray which is then taken to the maiden’s room. Based on information provided by a close female relative and on the offerings from the men, the maiden is supposed to choose her future husband. The epic song contains in addition a sub-theme about the poorest of presents given by the bravest of the men, and the hero’s speech explaining the {24|25} reason behind it. In two epic multiforms the maiden first scrutinizes the suitors from the window in the company of her mother, who tells her about each of them. It is after this that the suitors offer their presents. In the ballad texts the maiden first chooses the ring. In 1f she never sees the suitors as she escapes from home before being forced to make the choice; in 5 she sees her future husband as well as the other suitors only when she is already in the wedding procession; and in 5b she actually never sees the other suitors. In ballad multiform 5 the maiden’s conversation with one of the bridesmaids about the men who competed for her hand takes 26 lines, while her conversation on the same topic with her mother extends to 373 lines in the recited epic multiform. Limitations of space do not allow me to include full excerpts, but the following lines will illustrate their basic organization:

There is no correspondence in wording between these two excerpts, but it is evident that they share the same question-answer structure: the maiden inquires about the suitor and her collocutor provides the relevant information. That the epic multiform of the theme possesses a {26|27} greater degree of elaboration and ornamentation is evident even from these few lines. The second part of this theme from Međedović’s song has its counterpart in ballad multiform 1f (text 2411) by Tanović. Again, the basic ideas encompassed in this segment coincide, but there is little similarity in the wording except for the two formulaic blocks of lines in boldface:

What constitutes one large theme in the epic is thus broken down into two separate, concise segments in two ballad cycles. However, they both still share the basic ideas with the epic. Each of the ballad cycles places more emphasis on different parts of the theme, but there is no question that all the excerpts have the same common denominator, despite the dissimilarity of the plots and the genres. “The Wedding of Alija Vlahović” focuses on a young man’s overcoming of various hardships in order to win a wife for himself. He is poor but nevertheless competes for the hand of a beautiful maiden against much wealthier suitors. It is his good fortune, though, that the maiden is looking not for wealth, but for bravery in her future husband. Knowing that only her beloved is a hero capable of meeting her challenge, she announces to the suitors that she will marry the one who avenges the execution of her young brothers. Alija volunteers and is successful in his mission. Overwhelmed with tiredness upon his return, he fails to do away with the killer, who is released by one of the rival suitors. Moreover, the rival then asks the killer to abduct the maiden. After many obstacles, Alija not only regains his maiden but takes yet another wife and finds a lost sister.

The two ballad plots that contain this theme differ not only from the epic, but also from one another. In 1f the maiden is sought in marriage by many suitors, and her mother, annoyed {28|29} with her daughter’s proposal, decides to take the control over the situation. She sends out letters to the suitors she deems appropriate and invites them to come and compete for her daughter. At the same time, the daughter sends a letter to her beloved asking him to come to her rescue. The suitors gather and bring the gifts, but while the rings are being collected, the maiden escapes from her room with her beloved. In the multiforms from cycle 5, this theme appears at the very beginning of the songs. In one of them the maiden regrets her greediness and her choice of the most expensive ring when she sees how old and unattractive her husband-to-be is. During the wedding ceremony she pretends weakness and eventually elopes with a younger man. In multiform 5a there is no such conflict, and the maiden happily marries the suitor she chooses.

Unlike shared blocks of lines, shared themes overlap to a far lesser degree in their wording. The theme is capable of withstanding more significant changes to fit both a variety of plots and the expressive demands of different genres. I believe it is the ideas rather than the verbal content that should be given prime consideration in defining a theme. [86] Instances of more or less fixed wording extending throughout the theme may occur in different performances of the same singer, especially if the theme appears in songs frequently found in his or her repertoire. [87] Even then, some modifications are not unusual. Variability is far greater in performances by different singers working in the same genre, and greater still if they are from different regions or if their songs have a common theme but belong to different genres. [88] Put differently, themes common to epics and ballads may share ideas, but their wording and organization differ more than if one were to conduct a comparison within the frameworks of a {29|30} single genre. [89] Furthermore, it is hardly surprising that a theme consisting of 26 verses in a ballad may become more stable in the course of many performances by one singer than will a theme in an epic that occupies 373 verses. Whether the same themes will share the same sets of formulas or even blocks of lines thus depends on a number of factors. If we recognize the underlying ideas that shape a theme as the basis for its definition, we may uncover in different traditions and parts of traditions tendencies that are sometimes dressed in different attire, but nevertheless have the same meaning and serve the same function.

This brief discussion of compositional elements and topics common to both epics and ballads helps demonstrate the coexistence of similarities and dissimilarities between these genres. Those subject matters that are of interest to singers of both sexes find their way into their songs, but their formal treatment may be influenced, depending on the area, by the cultural environment in which the singer lives. Similarly, themes, blocks of lines, and formulas that satisfy semantic demands of the plot are used indiscriminately by singers of both sexes in all the verse genres. While it is important for practical reasons to create a classification system that will accommodate these diverse materials, it is at the same time equally necessary to remember that points of contact are numerous and that they often exist on levels deeper than the division according to gender or genre. [90]

There is no doubt that an awareness of issues related to the gender of both performer and audience is essential to understanding the types of transformations that occur in the genres discussed here. However, the classification of oral traditional songs along gender lines poses its own serious problems. Defining a song as a “woman’s song” does not tell us much, as it may refer to practically any genre performed by a woman, and can even indicate a non-epic song performed by a man. Furthermore, a performer’s choice of a given genre or subject should not be seen primarily as a matter of preference, talent or ability, but rather as a decision determined largely by the social order, culture, and religion to which the singer belongs. With all of this in mind, it may be more appropriate to refer to “women’s songs” in general as “songs performed by women,” indicating thereby the sex of the performer without setting an automatic limitation in terms of genre. Accordingly, one could speak of “ballads performed by women” as opposed to “ballads performed by men,” etc. With regard to borderline genres, {30|31} coinages such as “epico-balladic” and “lyrico-balladic” could help eliminate certain terminological problems. [91]

An attempt at classifying ballads from the Parry Collection according to their subject matter is supplied here with two purposes: first, on a more general level, to provide insight into the breadth and type of topics appearing in ballads from the Gacko region; second, and more specifically, to indicate the position of particular songs presented in this volume in relation to other cycles from the same tradition. As many of the subjects are not unique to texts from the Parry Collection, this index to a certain degree encompasses also a spectrum of ballad topics typically found in other areas where Croatian and Serbian are spoken. The various attempts to treat the problem of classification of South Slavic oral genres (both in terms of subject and form) testify to the complexity and multiplicity of possible approaches to this issue. [92] A different set of criteria would doubtless have yielded a different type of classification, especially considering how entwined much of the material is. For example, certain subjects easily fit into more than one category, since some multiforms of a song stress one aspect of the plot more than another. Furthermore, some multiforms introduce new themes not found in other texts from the same cycle. In such cases I have listed the plot under two headings and provided cross references that point to the cyclical nature of the material.

Ballads from the Gacko area can be divided in groups along a temporal axis: pre-married life, married life, parenting, and death (either one’s own or that of a family member). The subject of death is often combined with that of an ominous dream. Dreams can also announce events belonging to other segments of the axis (such as marrying, conception, etc.), but in such songs the subsequent joyous events tend to become the focal point. The further subdivisions within larger groups are guided by the types of obstacles, conflicts, or wrongdoings encountered by {31|32} the main character. Accordingly, the main subject groups are marked by Roman numerals, the individual subjects by Arabic numerals, and the cycles by letters. [93] Although for reasons of space I have discussed at length only selected parts of this classification elsewhere in this introduction, it is my hope that the reader will be able to use it in conjunction with the songs provided in this volume for general reference. [94] The songs in this volume are organized with this classification in mind, and inasmuch as the idiosyncrasies of the material permit, they follow the order in which the subjects and cycles appear in it.

I. Premarital subjects

1. Unwanted marriage

2. Obstacles to marriage

3. Assault of a maiden

4. Sibling relationships

II. Marital subjects

1. Good/evil husband

2. Good/evil wife

3. Brother/sister-in-law’s intervention

4. Evil mother-in-law

III. Good/evil parents

1. Good parents

2. Evil parents

IV. Death

In his last book Lord devoted one chapter to the problems of textual stability and variation in traditional ballads with the purpose of determining the role of memorization in this genre. [95] Using as an illustration several multiforms of a Bosnian song he concluded that a sense of textuality is evident in South Slavic ballads, but noticed also that the components transmitted in ballad multiforms modulate. [96] While in some instances these include a type of verbal correspondence typically associated with lyric songs, in others they involve not the wording but the story, as in epics. [97] His observations eliminate the possibility of rote memorization and point to the fact that the singers “remembered” the ballad under investigation. In other words, they associated certain formulas and formulaic expressions with this song and had a clear sense of the plot. Lord’s examination of the opening theme from five ballad multiforms from the Gacko area is particularly relevant to the present discussion since they belong to the same cycle as song 40 in this volume. [98] The chapter on ballads, which Lord left unfinished at his death, contributes nevertheless in numerous ways to our better understanding of ballad transmission in general and the question of memorization in this genre. At the same time it leaves numerous paths for further analysis.

In addition to analyzing several ballads from the Gacko area on the microstructural level (through an investigation of formulas, clusters of formulas, and the wording of specific themes), this chapter also addresses macrostructural issues: the problems of composition, the linking of particular thematic groups, and variations on the plot level. The type of analysis which takes into account the development and combination of themes, such as is frequently found in discussions of South Slavic epics, has been conspicuously absent in studies of the ballad. It is not clear from the material Lord left us whether he intended to explore this problem, but the fact remains that virtually no work has been done in this area.

Textual stability in the ballad can be affected by a number of factors that cause some songs to become more stable while others remain more fluid. A large majority of ballads from the Parry Collection exhibit instability on the level of wording, but there exist a few interesting examples with an exceptionally high degree of correspondence. In these cases there is no doubt about the reliability of either the scribe or the recording method. In general, an increased stability—even fixity—is likelier to occur in songs by young literate singers, and {37|38} their repertories are for the most part smaller than those of the older singers. Such young singers relied on memorization to a degree that was entirely atypical for older illiterate singers. Examples of such unusually stable ballad multiforms are two songs by Hajrija Šaković: one is 79 lines long, the other 76, but the main difference between them is the omission or insertion of several verses. [99] Only in one instance is a formula modified: “Srete agu vjerenica ljuba” (“Agha was met by his faithful wife”) is replaced with “Sretila ga vjerenica ljuba” (“He was met by his faithful wife”). It is evident that Hajrija Šaković, who was in her early twenties at the time the song was recorded, committed this song to memory and did not compose it in performance. [100] We do not have a sufficient number of multiforms of other songs by this singer to establish firmly whether she generally relied only on her memory of fixed texts. An examination of Hajrija’s songs from the Parry Collection shows that contamination in them is not a rarity and that the number of well composed ballads is far smaller than is the case with more experienced singers. There are two likely reasons for this: first, she and other young singers were still apprentices in the art of singing, and second, education and gradual changes in styles of living contributed to the weakening of the tradition. It is this ossification of songs, resulting from a greater reliance on memorization, that suggests such a process was already under way in the South Slavic regions. Nevertheless, and despite the case of Hajrija Šaković’s song, a group of quite dissimilar multiforms by older singers included in this volume denies the possibility of textual fixity within the tradition as a whole.

The second case of exceptionally high stability involves two ballad multiforms (89 and 84 lines long, respectively) by Hasnija Hrustanović. [101] The sequence of themes in these two multiforms overlaps fully, and approximately half of the lines are repeated verbatim. Many of these are repetitive formulas involving exchanges between two characters. These verses are identical every time they appear, even in the course of one multiform. Five lines from one multiform do not have their equivalents in the other. In the remaining verses the differences consist of substitution of one word for another (rather than re-phrasing the whole verse), changes of the verb tense or aspect, and the replacement of various pronouns and conjunctions. However slight or small in number, these modifications are nevertheless indications that, despite an extremely high level of stability when compared with the multiforms of other {38|39} songs by this singer, this ballad was not fully committed to memory. If a singer is not comfortable composing in performance, s/he will try to avoid any changes since even the smallest modifications (particularly in an inflected language) can disturb the meter. [102] There is no question, though, that Hasnija had a vivid recollection of not only the sequence of themes but also their wording, which resulted in unusually high textual stability encompassing the whole ballad.

Through the example of epics Lord has demonstrated that singers can develop a preference for a particular wording of certain themes and that their sense of textuality becomes more pronounced the more frequently the song containing the relevant theme is performed and the shorter it is. [103] Among other materials he compares the opening themes in three multiforms of the epic “Bojičić Alija Rescues the Children of Alibey,” a favorite of the singer Đemail Zogić. The wording of these themes in two multiforms collected in 1934 and a further one in 1951 remained surprisingly stable (the lines in the first theme after the prologue coincide in about 50% of cases), despite a time lapse of seventeen years. The singer must have performed this song so often that he developed and remembered a preference not only for the sequence of themes but also for their wording. [104] In my view, it is this tendency that can be observed in the two multiforms by Hasnija, rather than a conscious attempt at memorization. A significantly higher degree of variation encountered overall in her other songs, as well as her unquestionable ability to compose in performance, both argue against memorization. A frequently performed song can become more stable and a sense of textuality more pronounced in it regardless of its genre. It is natural that this kind of occurrence is more pronounced in shorter genres. In the epic, pronounced textual stability can encompass a theme or several themes, in the ballad it can range anywhere from one theme to an entire song, and in the lyric song it can easily include an entire song. [105] {39|40}

The third set of corresponding texts differs from the preceding examples inasmuch as these were performed by two singers, Mulija Isović and Zulka Tanović, both in their early fifties at the time of collecting. [106] There are 128 songs in the Parry Collection by Zulka, who was illiterate, and only 4 by Mulija, about whose literacy no record was left. In terms of corresponding texts, Mulija’s multiform is 11 lines longer than Zulka’s. The only other differences between these two texts involve five verses, of which two have different word order and three have one word replaced with another. No fewer than 111 identical verses appear in both texts. In this instance we can no longer speak of textual stability, but of fixity. It is difficult to believe that Zulka memorized by heart all of the 128 songs she performed for the collectors, but at the same time there is no question that she was a much less accomplished singer than Hasnija, Emina, or Đula. Her frequent changes of residence, with long stretches spent in areas where another language was spoken, easily explain her weaker connections with the oral tradition of the Gacko region and, consequently, her smaller repertory and the poorer quality of her songs. [107] Our knowledge about the other singer, Mulija, is minimal: only her age was written down. The fact that so few songs were collected from her may well mean that her performing abilities were limited. [108] At all events, an occurrence of two nearly identical versions by two different singers points to memorization on the part of both women. In order for such a high degree of correspondence to appear in their texts, either both of them had to have heard many times a fixed version from a third singer, or else one of them must have frequently performed a fixed version of this ballad in the presence of the other. It is also conceivable that both singers heard this particular version on a number of occasions from a singer who learned it by heart from a written source and memorized it that way. With the onset of literacy among younger singers during the early part of the century and the circulation of inexpensive song books (pjesmarice), this possibility should not be overlooked. Unfortunately, in order to clarify this question we would need to have multiforms of several songs by both singers and these do not exist. The existence of other, quite fluid multiforms belonging to the same cycle in this region is again a sure sign that this example of fixity is the exception rather than the rule.

Although cases of high correspondence between texts such as those discussed above are relatively rare, they nevertheless need to be taken into account in an examination of textual stability in the ballad. The examples provided here indicate the complexity of the issue and the need to investigate the tradition that makes possible multiple versions of a song, but also the context that ethnological information can provide in order that we may better understand who the singers were and how they composed. {40|41}

Turning from these cases of extremely high verbal correspondence, we will examine the opposite part of the spectrum: newly composed songs that have no equivalent anywhere in the tradition. The ability of a singer to compose in performance on new topics is extremely important since it reveals how, in fact, traditional songs came into existence. At the time when the materials from the Parry Collection were gathered the skill of composing on an entirely new topic by female singers was on the decline. Mostly the singers learned already existing songs from one another, altering them in the process. Regrettably, no ballads were recorded for which it is possible to determine with certainty that the interviewed singers composed them entirely on their own. Parry and Lord had better luck with lyric and humorous songs, on one of which I will focus here. This song describes an amusing incident that occurred during one of their recording sessions. In order to welcome the guests, Fatima, one of the singers who performed for Parry and Lord, decided to bake a cake. In her excitement she forgot about her surprise, which, to her dismay, burned. Fatima’s friend, Almasa Zvizdić, composed on the spot a humorous song describing this event. [109] Moreover, she sang the song two times within a short interval. These two multiforms, while sharing the basic idea and a number of formulas, nonetheless differ in many ways. [110] Both songs open with the same lines:

But this is also where the overlapping in the wording of these two multiforms ends. They both tell of Fatima’s mistake (which consisted in putting too much hot ash on the lid under which the cakes were baking), her surprise when she saw the burned cakes, her tears in which the cake was soaked instead of honey, and Almasa’s trip to the kitchen where she learned about the mishap. In the longer of the two multiforms Almasa ends the song with her report to the “mister” (i.e., Parry) and his great surprise that this could have happened to Fatima. Basing these two compositions on formulaic expressions she knew from other songs and modifying them to fit the new context, Almasa composed on the spur of the moment not one but two closely related multiforms. While sharing the same “more or less stable core,” these multiforms differ in many details of wording. Almasa’s father was a renowned guslar and she reminisced that she would often sit around when he performed. Having grown up in such an environment, it is not surprising that she could compose a new song with such spontaneity. When asked to sing “the same” song again (in cases of both old and new topics) Almasa and other female singers reacted in the same way as the male singers who were interviewed: they {41|42} reproduced the same narrative outline, but did not use the same wording. They all claimed though that the song was “the same”—after all the subject matter was unchanged—and refused to consider, regardless of their scope, the changes in ornamentation, additions, or reductions as elements that make one oral performance different from another. [111]

The conversation brought on by the episode of the burned cakes took a humorous but revealing turn when Almasa was asked whether she had ever before composed a new song as she did that day. [112] Almasa repeatedly did not understand the question and kept connecting the two new multiforms to the specific event of baking cakes, rather than making a general distinction between the old songs and the new ones composed on her own. This failure on the part of a singer to answer the question about her ability to create songs on a new topic confirms that in the world of traditional singing the concepts of novelty and originality are not of particular concern to the singer. Songs may be learned from someone, or made in relation to an event, they can be performed in a better or worse way, but their attribution as a song created by a particular singer seems to be irrelevant for both male and female performers. Singers may remember from whom they learned a specific song, but will not know or care to know from where or whom that song originated.

In relation to this anecdote it might be of interest to recall songs by two male singers. The first is Milovan Vojičić of Nevesinje, who, like Almasa, composed in 1933 a 149-line long, epic-style song about Parry’s travels and work in South Slavic regions. [113] The other is Avdo Međedović, {42|43} who participated in an experiment conducted by Parry in Bijelo Polje in 1935. [114] Intrigued by Međedović’s unparalleled talent, Parry arranged for him to hear performed by another singer an epic of 2,294 lines that was previously unknown to him. Međedović was then asked to perform the song himself. This gifted singer created 6,313 lines, an epic nearly three times as long. Clearly, the stylistic features of the genres in question here differ significantly, as do some of the technical aspects of composing a long narrative versus a 10- or 15-line humorous song. Despite all the differences, these examples prove that in order to be able to compose in performance (and this becomes particularly obvious in compositions on a new topic), the singer must be fluent in the traditional language of oral poetry. This language has not only certain lexical characteristics that are different from everyday language, but also specific meter-dependent principles guiding the combination and selection of words, something which Lord refers to as “the ‘grammar’ of the poetry.” [115] Its acquisition starts at an early age through listening to other singers’ performances and is in many ways similar to the acquisition of one’s mother tongue. Just as in everyday language one creates new constructions and meanings on the basis of old paradigms, so can the singer compose new songs from components available in the traditional language. [116] The singers discussed here, regardless of the genre in which they worked, had this language very much at their fingertips.

John Foley correctly observes in Muslim epics from the same area that each individual performance is closely entwined with the tradition that invokes “a context that is enormously larger and more echoic than the text or work itself.” [117] The same could be said for the ballad. Moreover, the ballad even shares with the epic the context of the same tradition in the South Slavic regions. Foley demonstrates also that this “metonymic” quality is equally evident in both the formula and the story, which is one of the reasons why the traditional audience does not listen to the story for its “suspense.” [118] Again, a comparable situation exists in the ballad. His statement, however, reveals the view from our vantage point. As already stressed, among Muslim singers who performed for Parry and Lord in the thirties, the ability to compose in performance on a new topic was disappearing. This was the case not only with lyric songs and ballads but also, and to a similar degree, with the epic. A relatively small number of new songs {43|44} they recorded were among the last vestiges of a great tradition. “It is only when a tradition is dying that it begins to lose contact with the present and becomes a preserver of its own past rather than a continuator,” asserts Lord. [119] In a living tradition that actively produces new songs, a large body of plot patterns will indeed be known both to the singers and the audience, but new songs with new patterns that spring from the traditional base will also be created. In these songs the contextual or “metonymic” effect of the tradition is very much present, but primarily through the formulaic language and the use of certain ready-made themes. It does not, however, sustain the narrative aspect of the performance throughout the song. Regardless of how close two performances are in terms of their plot, the traditional audience, like participants in any ritual, tends to be interested more in the process than the outcome.

In the course of time some new songs may become part of the widely known traditional core, resulting in more stable plot patterns and, particularly in shorter genres, even a pronounced sense of textuality. In other words, in a living tradition there is a body of songs which at one time have no multiforms, but over a period of time can form fluid cycles of multiforms dealing with the same type of conflict situation or series of events. If the song is performed with regularity and disseminated in the area, its multiforms become progressively more set in their sequence of themes and possibly even wording. Creative singers may alter the pattern, but a majority of singers in the tradition will adhere to the commonly known basic plot. The shorter the song, the more pronounced this tendency becomes. In this group there will indeed be little or no suspense with regard to the outcome of the song. Needless to say, even in this category a short song can turn into a complex and long performance in the hands of talented singers such as Avdo Međedović, Almasa Zvizdić, Hasnija Hrustanović, and the others. Although we have no proof that any new ballads were recorded during Parry and Lord’s research trip in the thirties, the fact that a small group of songs stands alone in combining the known themes in unusual patterns testifies that at least some of the singers in this disappearing tradition were able, if not to compose new ballads, at least to preserve the old ones with much creative spirit.

The largest part of the ballad material from the Parry Collection thus falls in the middle part of the spectrum represented by the songs having plots widely known in this geographic region. Even in this largest of groups, in which neither fixity nor absolute novelty is encountered, we can differentiate between songs that exhibit greater or lesser degrees of textual stability. Although not entirely new, some songs are different enough from the other texts that they can be considered only their tangential multiforms. Put differently, they overlap only in one segment with the core of more stable multiforms from a given cycle. While not presenting {44|45} new types of conflicts or situations, they employ the known themes in combinations somewhat different from the rest of the tradition. The second group is comprised of the multiforms of songs that show relative stability in their plot and wording. In the same middle range of the spectrum belongs the third group. These are cycles with multiforms which, while not fixed, show a pronounced sense of textuality, as is evident in a consistent plot pattern and the steady inclusion of certain formulas. Such cycles became so popular that they often maintain stability of plot and some degree of verbal correspondence even in different regions and with singers of different ethnic or religious background. Into which group a particular performance will fall is determined by factors such as the length of time the song has lived within a certain tradition, its popularity in the area, and the individual singer’s talent. In this regard, Lord’s remark that some ballads appear closer to the epic, others closer to the lyric, is accurate.

Short ballads in particular tend to show pronounced textual stability characteristic of the lyric song. It is my conviction, one which I attempt to demonstrate in the penultimate chapter of this introduction, that some of these belong to the oldest layers of South Slavic oral traditions. This is particularly the case with ballads containing strong ethical components and narratives tailored to depict types of behaviors considered unacceptable by the community. A great number of recorded multiforms of such songs, not only in the Gacko region, but in the whole Bosnian, Croatian and Serbian territory, points also to their remarkable popularity. A good example in this volume is cycle 18, dealing with the subject of incest. Multiforms of this cycle are among the most stable in this volume and show significant correspondence on the level of both plot and wording. The number of lines in the complete Gacko texts ranges from twelve to seventy, but nearly all of them are about 20-25 verses in length. The basic plot pattern remains unchanged in all instances: the workers are harvesting, among them a young bey; his sister brings him lunch and he, blinded by her beauty, tries to take advantage of her; she curses him and runs away, while her brother is killed by lightning. Descriptive details may vary to a greater or lesser degree from performance to performance, but this basic narrative thread does not tend to change.

The strong sense of textuality that singers of this song share is evident in the texts from this volume, which include three multiforms by Đula Dizdarević, one by her sister Emina Šaković, and a recorded one by Raba Zvizdić. Both a consistency in narrative pattern and extensive overlappings in formulas are clear at a glance. Similarly, it is not surprising that, owing to their frequent performances of this popular song, singers such as Đula Dizdarević introduced little or no variation in some of the themes appearing in it. Despite this, there are no two texts that are exactly the same in the corpus written down or recorded by the reliable collectors. To make the picture clearer I have selected excerpts from several “incest at harvest” {45|46} multiforms that present the part of the theme describing the brother’s words to his sister. The core of this segment consists of a limited set of elements which are combined in several ways:

a. Text 11754 (18 in this volume) by Đula Dizdarević [120]

b. Text 1036 (18c in this volume) by Emina Šaković

c. Text 6501a (18d in this volume) by Raba Zvizdić

d. Text 2380 by Derviša Kažinić

e. Text 6894 by Vezirka Tanović

f. Text 5921 by Pašana Avdić

g. Text 7102 by Đula Čampara

h. Text 7490 by Dževahira Tanović

i. Text 8059 by Arifa Kurtović

Already this fragment of the theme shows that a relatively stable core exists in all of the multiforms, as would be the case in a lyric song. Despite a high occurrence of corresponding lines, it is not possible to predict exactly which ones will be employed in a particular performance or in which order. [121] In textually stable ballads like this one, the study of a large number of multiforms can provide a sense of the range of possibilities. [122] On a larger scale, though, while the components comprising a given theme may be reshuffled or omitted, the order of the key themes in ballad multiforms tends to show a certain consistency that preserves a narrative pattern. In the lyric song no such organizing factor guides the distribution of lines.

If we look at other parts of the “incest at harvest” multiforms, we note that while ideas expressed in some verses tend to recur, though not necessarily with exactly the same wording (as we have just seen), the ideas expressed in other verses cover a far greater range and vary much more in terms of their inclusion in the song. Some multiforms even introduce ornamental descriptions and new elements not found in other texts from this cycle, but which are common to other ballads from the tradition. Just as it is possible to distinguish core and {47|48} optional themes, similarly one can speak of two groups of verses that appear in a theme: the core verses which express the essence of the theme and have a high percentage of occurrence, and optional verses which express ideas that are not central to the development of the plot and consequently show up randomly. [123] The exact wording of the core lines cannot be known in advance but this is even truer of optional lines. Furthermore, the theme can be pared down to a handful of lines, or developed into dozens of lines which sometimes even contain sub-themes. A talented singer may opt to import and modify lines from other cycles, or compose new ones based on the formulas and combinations with which s/he is familiar. The brevity of most of the multiforms in the “incest at harvest” song allows for a systematic look into these two groups of verses. The core group indicates also the main themes of the song. These core verses found in most of the multiforms are often supplemented by any number of combinations from the optional column.

This listing is by no means complete. Not only does it include solely the multiforms of this ballad from one collection and one geographic area, but it is also based—as self-evident as this may be—only on performances that were taken down. Theoretically, each of the singers in each of the performances that have ever existed could have introduced new elements and made new combinations of old elements. If we had another fifty multiforms it is likely that the optional column would be considerably longer, while there would be little change in the core column. In other words, while there exist loose frameworks determining what can or cannot go into a song, these guidelines are not sacred and the singer is free to override them at any point.

In shorter multiforms the plot moves forward rather quickly and they have a minimal amount of material from the optional line column. In the longer ones the optional verse column is well represented and it sometimes branches out to form complementary themes. For example, in several multiforms the presentation of the sister’s sorrow or the mother’s desperation over her son’s death amount to well-rounded and subtle descriptions of a psychological drama. Although formulaic in language, there is nothing automatized or mechanical in these particular verses, which are refined enough to convey a range of nuanced emotional states.

In his examination of epics, Lord pointed out a “conservative urge” which preserves basic ideas found in a theme or group of themes. “Multiformity is essentially conservative in traditional lore,” he concludes. [124] The parsing of themes in the “incest at harvest” song demonstrates that this thesis is equally valid for the ballad. Conservative preservation of the ideas expressed in the core verses is precisely the element which allows freedom of variation elsewhere. In the same way in which, on a smaller scale, formulaic patterns represent a firm ready-made core around which the rest of the verse can be built in an instant, the core lines permit the singer to maintain the essence of the theme (or, on a larger scale, the plot) while exploring variation elsewhere. Conservatism is thus, in this case, not restricting but liberating.

The pronounced textual stability apparent in cycle 18 might have resulted from the general popularity of the song in the region over a long time, as well as from the importance of the moral beliefs expressed in it. [125] The 53 multiforms belonging to this same cycle make it one of {50|51} the best represented cycles in the Parry Collection. Such a high occurrence of one text indicates that the whole community was exposed with some regularity to performances of this relatively short ballad. Owing to such an exceptional degree of exposure the general knowledge of its themes and formulas was reinforced over time, resulting in an increased stability in individual performances. Fixity does not exist in these texts, but the singers’ sense of textuality in them is more pronounced than in most of the other cycles.

An entirely different situation with regard to stability is encountered in the multiforms from cycle 1 in this volume, dealing with the subject of the forced marriage of a maiden and her subsequent abduction by her beloved. Included in it are three multiforms by Hasnija Hrustanović, two by Emina Šaković, one by Đula Dizdarević, and two by two other singers from the Gacko area, Zulka Tanović and Zehra Šaković. The singers range in age from about forty (Zehra Šaković) to about seventy (Đula Dizdarević), thus representing two generations. With the exception of two multiforms by Đula and Emina, the other texts show remarkable diversity in details of both plot and wording. It is interesting, though, that Emina’s multiforms 1d and 1c differ more between themselves than do Emina’s 1d and Đula’s 1e. Multiforms 1d and 1e overlap not only in their selection and organization of the themes, but also coincide in wording in about two-thirds of the song. Fixity thus does not stretch throughout these texts, but it does encompass many of the themes. Curiously, at one point in 1d Emina even makes a mistake and refers to the maiden’s beloved as Mujo Orlanović (the name used in Đula’s 1e), although at the beginning of this multiform he was called Ahmet. If we look at multiforms of other songs performed by both singers, we can see that the degree of correspondence in wording is not any different than would be the case in performances by any two randomly chosen singers. In some cases differences between their multiforms are striking, such as in cycle 15, in which Đula’s 15a is 86 verses long while Emina’s 15 stretches to an astonishing 248 lines. As pointed out earlier, an increased correspondence in wording can sometimes be found in multiforms of a frequently performed song by the same singer. In this instance, however, one should remember that Đula and Emina were sisters and thus quite familiar with one another’s repertory. I would suggest that their frequent singing of ballad 1 in one another’s presence could explain the overlappings in their phrasing of certain themes. The two distinct multiforms by Emina confirm that the singer did not fully commit this song to memory, although she did have a preferred wording for some of the themes, especially those occurring in the first half of the ballad. The situation would be comparable if there existed another multiform of this song by Đula. In other words, a high degree of correspondence in the performances by two sisters reveals their preferences for the wording of some themes, but as some variation exists in other segments I would guess that similarities between these texts are unlikely to have been the result of conscious memorization. By contrast, Hasnija’s texts from this same cycle show a remarkable degree of variance. Not only does each open with a different theme, but they also continue to follow distinct, albeit related, paths throughout. If a comparison is made between these multiforms and her exceptionally stable multiforms of the {51|52} song discussed at the beginning of this chapter, one sees that a sense of textuality in the multiforms of different songs by the same singer can fluctuate considerably.

A close analysis of the central or conflict theme from cycle 1 will demonstrate that the textual stability is generally much lower than in cycle 18. This is manifested both in the wording and in the inclusion of certain themes. For example, the opening in these multiforms includes not one, but three different themes: the maiden weaves by the window (1); the maiden sings about her beloved and is overheard by a parent (1b, 1c, 1d, 1e); the maiden is sought in marriage by many suitors (1a, 1f, 1g). Despite different opening themes, all of the multiforms eventually share the theme presenting the central conflict situation in the ballad, a conversation in which a family member announces to the maiden that she will be married against her will to a man she dislikes. This is, in fact, the only theme that is common to all eight multiforms, but an extensive verbal overlapping exists only in the two multiforms by Emina and Đula, presumably for the reasons given above. Other texts, while sharing the same ideas and many of the same formulas, differ notably in their wording of this segment.

In Hasnija’s 1a, Zehra’s 1g, Đula’s 1e, and Emina’s 1c and 1d the conflict theme comprises the maiden’s conversation with her father, while in Hasnija’s 1 the maiden talks with her brother, and in Hasnija’s 1b and Zulka’s 1f the maiden’s collocutor is her mother. Unlike in other songs, in Hasnija’s 1 the maiden does not overtly reject her brother’s choice for her future husband, although her despair is evident from her questions and her general behavior. The wording of the exchange between brother and sister has nothing in common with other texts from this cycle. Similarly, the detail of the maiden’s interruption of her brother and her husband-to-be as they negotiate the terms of marriage is exclusive to this multiform. The theme ends with the maiden’s exclamation of protest—described as hissing “like a poisonous snake”—and her escaping to “her upper rooms.” In Hasnija’s 1a as well, the maiden’s anger is expressed by a similar formulaic expression that mentions a snake and escaping to “another room.” Her 1b concludes with the maiden simply escaping “to her room” and closing the door behind her. Although overall the wording of the conflict theme in Hasnija’s text 1 is as distant from her other multiforms as it is from those by the other singers, it concludes with a similar set of formulaic expressions describing the maiden escaping to her room, as do the other texts in this cycle. However, the formulaic expression that compares the maiden’s anger to the hissing of a snake is found only in one other multiform, Zulka’s 1f, in which the word “guja” (“snake”) is replaced by “zmija” (“serpent”).

The multiforms that share the same opening theme (Hasnija’s 1b, Đula’s 1e, Emina’s 1c, and 1d) also show, not surprisingly, a greater similarity in the presentation of the second theme. In these a parent asks the maiden about the man mentioned in her song and informs her of his/her own choice of the girl’s future husband. The conversation in all four multiforms opens with the same formulaic line (“My daughter, may a snake bite you!”). With the exception of the verb-adverb inversion in Emina’s 1c, her two multiforms, as well as the multiform by her sister Đula, continue with a question asked by the father (“What did you sing at the hearth this morning?”). Hasnija’s 1b has instead a different question asked by the girl’s mother, “What good is the judge of Travnik to you?,” inquiring similarly about the man mentioned in {52|53} her daughter’s song. All four multiforms proceed with two nearly identical formulaic lines stating that the parent has promised the girl to another man: “Yesterday I gave you in marriage, I gave you to Alaga Bojičić (the bey Crnčević).” In place of the time adverb “yesterday,” in Hasnija’s 1b an adjective describing the future husband is used: “I gave you to a better young man, I gave you to the bey Crnković.”

Emina’s 1d and Đula’s 1e continue to coincide throughout the part in which the girl responds to her father’s statement that she is already given to another man. The only differences between these two are two instances of inversion and a replacement of one noun with another. In the next part of the song, describing the father’s reaction to the girl’s protests, Đula’s multiform introduces two new lines, (“his hair stood up/as on a wolf on St. John’s day”), not found in Emina’s multiforms. Hasnija’s 1b closely follows the same narrative pattern, according to which the Muslim maiden protests being given to a Vlah, a man of a different religion (Orthodox) and ethnic background (Serbian). Despite numerous common formulas shared by Emina’s 1d and Đula’s 1e, the details of wording fully overlap in one line only: “Vuk will come to visit his sister.” In addition, 1b is also more elaborate in providing (unlike 1d and 1e) the parent’s verbal response to the girl’s angry words (not only the physical reaction), and a description of the girl’s further resistance. It is only then that the parent tries to hit her daughter, but unlike in most other multiforms, with no success. In 1d and 1e the father hits the girl immediately after her first (and only) words of protest. Both texts state, “He hit her with all his might/and knocked out two healthy teeth,/and moved another four from their place/dark blood spattered the maiden.”

In Emina’s other multiform, 1c, the maiden rejects her father’s order rather succinctly in words not found in 1b, 1d, or 1e (thus texts with the same opening theme): “I do not want any bey/as long as the alaybey* Ahmet is alive.” One of these two formulaic lines is found in Zulka’s 1f: “neither do I want any bey.” [126] The conversation in Emina’s 1c also ends with the father’s rage and the physical punishment of the maiden. In his violence he also smashes the maiden’s teeth, but the line describing how she was spattered with blood is omitted from this text. With regard to the physical punishment of the maiden in other texts, one of the formulas from this set is also employed in Zehra’s 1g. Although the verbal exchange between the maiden and her father in this multiform is entirely different, the line “and moved four of her teeth from their place” is included in it. By contrast, in Zulka’s 1f the mother hits her daughter leaving her teeth intact but splashing her with blood: “the dark blood spattered her.” Described in different words, but similarly soaked in blood, is another character appearing in the second theme of yet another multiform: the servant-messenger who receives punishment from the maiden in Hasnija’s text 1. Interestingly, in this text the maiden does not receive any physical punishment at all.

Hasnija’s 1a and Zulka’s 1f proceed to the conflict theme from a different opening theme, one that describes the maiden being sought by numerous suitors. They differ more significantly {53|54} from the group of multiforms just discussed since the conflict theme needs to be adjusted to what has already been stated in the opening section. In both texts the conflict theme begins with the parent’s aggravation caused by the frequent visits of numerous suitors. Text 1a is more elaborate and describes the father’s complaint about the suitors devastating his household by using up all of his supplies. To prevent his material ruin he pressures the maiden to decide whom she wishes to marry. [127] Multiform 1f is quite concise in this part and does not actually state the reason why the mother is vexed by the numerous suitors. The conversation between parent and daughter is similarly more elaborate in 1a than in 1f. Only two lines contained in this part are exactly the same in both multiforms (“Choose, my daughter, the one to your liking” and “to tell you in the room today?”). [128] Although worded differently, the pattern of exchange develops similarly in both texts: the parent points out that the girl has many excellent suitors and should make up her mind, and she in response asks permission to name openly the man of her choice. The parent consents and the maiden voices her choice. In Hasnija’s 1a the father responds to her choice by warning her that she may become a widow if she marries the warrior she loves and does not hit the girl. In Zulka’s 1f the maiden pleads that she be given to her father’s proxy. The girl’s mother foregoes further quarreling and responds to this only by hitting her daughter.

Finally, in Zehra’s 1g the father also does not give any reasons for urging his daughter to decide whom to marry and no aggravation on his part is mentioned. In fact, this multiform proceeds directly from stating the fact that the maiden is sought in marriage by many to the father’s brief question about her choice of future husband. The girl does not negotiate permission to state the name of her beloved, but says to her father that she swore to God that she will marry no other than Ibro Sokolović. Having heard this, the father curses her in the following words “Quiet, daughter, may a serpent strike you!” A curse with the same meaning, but different wording, appears also in 1a (“O Uma, may a snake strike you!”). As already established, yet another wording of this curse appears in 1b, 1c, 1d, and 1e (“My daughter, may a snake bite you!”). In Zehra’s 1g the father rejects the maiden’s choice for a reason not found in other multiforms: the maiden’s beloved is in prison. As discussed above, the girl’s contrariness earns her physical punishment in this multiform as well. The theme concludes with the father’s further objections to the status and property of his potential son-in-law. In all multiforms several sets of half-line formulas are combined in different ways to express that the maiden retreats to her room at the end of this theme. In 1g it is the father who steps outside after the argument. {54|55}

The theme presenting the maiden’s conflict with a close family member on account of her wedding encapsulates the essence of the song, and it is not surprising therefore that it appears in all of the multiforms. The degree of verbal variability found in this theme is considerable, but at the same time there is no question that the singers’ sense of textuality becomes most pronounced in this part of the song. Still, stability of this theme, as well as that of the entire song, is far lower than in cycle 18. In multiforms by the same singer the overlappings do not necessarily exceed those found in the multiforms by different singers. In a larger picture, taking into account the distribution and appearance of specific themes, it can be noted that a greater freedom in combining the themes in multiforms from cycle 1 results also in a greater variation on the plot level than is the case with multiforms from cycle 18. A higher degree of variance in wording and a lesser degree of uniformity in the plot bring multiforms like those forming cycle 1 closer to the type of low textual stability encountered in the epic multiforms.

In ways similar to the epic, in the ballad genre various themes can be combined and recombined to form a virtually inexhaustible well of patterns. While many singers tend to vary only the wording and create modest additions or reductions in a popular plot, some are able to weave new patterns on the basis of the old paradigms using the formulas, blocks of lines, and themes that are already available in the tradition. Furthermore, a singer’s “departure” from the customary rendering of the plot is sometimes not even the result of a conscious effort to improve or change the song, but rather a product of her misremembering of the plot or the order of themes she had heard from another singer. No song or multiform, though, is superior to another. [129]

In order to see what types of permutations can occur in a plot, let us maintain our focus on cycle 1. A summary of each multiform will allow us to identify the themes that typically appear in this ballad and observe the ways in which they are linked to one another. The information contained in the brackets reflects how the given theme is rendered in that particular multiform.

Multiform 1 by Hasnija Hrustanović

• weaving (the maiden weaves in her room)
• messenger (a servant brings her the message about her imminent wedding and is hit)
• maiden opposes the prearranged marriage (maiden’s conversation with her brother in which she learns who her future husband will be)
• letter (the maiden writes a letter to her beloved and orders the servant to deliver the letter; the servant seeks the maiden’s beloved in the mountains and delivers the letter) {55|56}
• abduction of the maiden (maiden’s beloved eavesdrops on the conversation between her brother and the suitor and sends her a message through the servant to ready herself; maiden is rescued by her beloved)

Multiform 1a by Hasnija Hrustanović

• suitors competing for the maiden (the maiden is sought in marriage by many suitors; she refuses all of them)
• maiden opposes the prearranged marriage (maiden’s conversation with her angry father in which he opposes his daughter’s choice of husband)
• letter (maiden’s father sends a letter to a suitor of his choice)
• wedding (the suitor arrives with wedding guests)
• maiden’s death (the maiden dies out of sorrow in a short time)

Multiform 1b by Hasnija Hrustanović

• overhearing the secret (the maiden sings about her beloved and is overheard by her mother)
• maiden opposes the prearranged marriage (maiden’s conversation with her mother in which she learns that a husband has already been chosen for her; mother tries to hit her daughter for opposing her)
• letter (the maiden writes a letter to her beloved and orders a servant to deliver it; the servant delivers the letter to the maiden’s beloved)
• death of the rival suitor (maiden’s beloved kills the suitor chosen by the maiden’s mother and marries the girl)

Multiform 1c by Emina Šaković

• overhearing the secret (the maiden sings about her beloved and is overheard by her father)
• maiden opposes the prearranged marriage (maiden’s conversation with her father in which she learns that a husband has already been chosen for her; father hits his daughter for opposing him)
• letter (the maiden writes a letter to her beloved and orders a servant to deliver it; the servant takes the letter to the maiden’s beloved)
• intervention of the third party (after receiving the letter, maiden’s beloved speaks with his father, who promises help)

Multiform 1d by Emina Šaković

• overhearing of the secret (the maiden sings about her beloved and is overheard by her father)
• maiden opposes the prearranged marriage (maiden’s conversation with her father in which she learns that a husband has already been chosen for her; father hits his daughter for opposing him)
• letter (the maiden writes a letter to her beloved and orders a servant to deliver it; the servant takes the letter to the maiden’s beloved, but his eyes deceive him and he initially does not recognize the maiden’s beloved)
• abduction of the maiden (maiden’s beloved sends her a message to ready herself since he is coming to her rescue and threatens to kill her father)

Multiform 1e by Đula Dizdarević

• overhearing of the secret (the maiden sings about her beloved and is overheard by her father)
• maiden opposes the prearranged marriage (maiden’s conversation with her father in which she learns that a husband has already been chosen for her; father hits his daughter for opposing him)
• letter (the maiden writes a letter to her beloved and orders a servant to deliver it; the servant refuses to take the reward for the service; the servant takes the letter to the maiden’s beloved, but his eyes deceive him and he initially does not recognize the maiden’s beloved)
• abduction of the maiden (maiden’s beloved sends her a message to ready herself since he is coming to her rescue and threatens to kill her father)

Multiform 1f by Zulka Tanović

• suitors competing for the maiden (the maiden is sought in marriage by many suitors; the maiden refuses all of them)
• maiden opposes the prearranged marriage (maiden’s conversation with her angry mother in which she opposes her daughter’s choice of husband; mother hits her daughter for opposing her)
• letter (maiden’s mother sends letters inviting suitors to compete for her daughter; the maiden writes a letter to her beloved
• suitors gathering (the suitors arrive bringing gifts, on the basis of which the maiden is to choose a husband)
• abduction of the maiden (maiden’s mother discovers that her daughter has escaped from her room with her beloved; the suitors debate whether to pursue the couple; maiden’s beloved takes the girl home)

Multiform 1g by Zehra Šaković

• suitors competing for the maiden (the maiden is sought in marriage by many suitors)
• maiden opposes the prearranged marriage (maiden’s conversation with her father in which he opposes his daughter’s choice of husband, who is imprisoned; father hits his daughter for opposing him) {56|57}
• disguising (the maiden goes to her beloved’s house and lies to his mother that she has found someone who will rescue her son; the maiden puts on her beloved’s clothing herself and deceives a servant in order to take her beloved’s horse) {57|58}
• the search (the maiden travels to the infidel’s land and finds her beloved)
• competing in a game (the maiden competes in a horse race in which prisoners are given as awards and wins her imprisoned beloved)
• revelation of identity (the maiden takes him away and reveals her identity in the mountain)

We have already seen that the multiforms from this cycle feature three different openings which all lead to the same conflict. Considerable diversity is found in the second half of these texts as well. Multiform 1g, for example, takes off in an entirely new direction immediately after the theme presenting the conversation between the maiden and her father. [130] It combines two subject matters—the forced marriage of a maiden, and the maiden rescuer (typically found in the rescue-and-wedding songs)—into one song and is therefore only obliquely related to the rest of the cycle. This is an excellent example of a song that is not new, but it employs well-known themes in an unusual way. By contrast, multiform 1f shows less distance from the other texts in this cycle, but it also incorporates a theme not repeated in any of them, the suitors’ gathering. The theme of abduction also includes the impact this event has on the suitors, a detail not found in other multiforms. The letter theme is modified in this multiform to the effect that not only does the maiden send a letter to her beloved, but her mother also invites suitors to compete for her daughter. In fact, the girl writes her letter only as a reaction to the summons sent out by her mother. The final theme of the maiden being taken away by her beloved, although worded differently, conveys the same basic idea as the equivalent themes found in other multiforms in this cycle. Multiform 1f thus meanders after the conflict theme only to return eventually to the main narrative stream and the customary ending.

Multiform 1a similarly diverges from the main stream after the conflict theme. In this text the letter theme is modified even further and the maiden never sends a letter to anybody. It is her father who invites the suitor of his choosing to come and marry his daughter. Unlike 1f this text does not funnel back to the final rescue theme, but remains on a separate track. The maiden is married off only to die of sadness in the end. Multiform 1c, while following the most common narrative pattern to a much greater degree than 1a, introduces nevertheless a new theme at the very end. After the maiden’s beloved receives her letter, he does not set off to rescue her, nor does he send any messages to her as in most of the other texts from this cycle. On the contrary, he remains at home and his father promises to help him. How exactly the maiden will be rescued remains unknown as the words of the father’s promise conclude this multiform. Texts 1a and 1c thus do not share the same beginning, but they flow together in their middle parts, only to branch out and introduce new themes in their final segments.

With the exception of the two opening themes, the remainder of multiform 1 does consist of the themes customarily employed in this cycle. Most of them, though, have additional elements which are different from or not mentioned at all in the corresponding themes from {58|59} other texts. For example, the servant does not simply deliver a letter to the maiden’s beloved, but he also has to look for him in the mountains. A different detail with the same function (a complication connected with the delivery of the letter) is found in texts 1d and 1e. In these the servant’s vision fails him and he does not initially recognize the maiden’s beloved. Further, in multiform 1 the detail of the maiden’s beloved eavesdropping on the conversation between her brother and a suitor while they negotiate the marriage is not found in the other texts. Rather, if the rescue is to happen the man usually just sends a message to the girl to ready herself. Even in the conflict theme, the detail of the suitor being present while the maiden is being informed that she will be forcibly married is an element not found elsewhere.

Finally, the subject of the forced marriage of a maiden is realized in strongly similar ways in multiforms 1b, 1d, and 1e. As already suggested, the extent of the overlap between 1d and 1e likely stems from the kinship of the two singers. Even in this situation of high verbal and plot agreement, one of the multiforms (1e) introduces details in the letter theme not found in the other. While in 1d the content of the letter is stated, the servant ordered to ready his horse, given money and asked to deliver the letter, in 1e there is an additional exchange between the maiden and the servant in which the servant refuses to take the money from the girl. In general, these three multiforms agree in their narrative pattern more than the other analyzed texts and could be therefore considered “the main stream” of this particular cycle.

It appears that the themes dealing with the introduction of the main conflict, as well as its resolution, tend to be much more varied and less predictable in their occurrence than those describing the central conflict itself. [131] In other words, the themes that are less critical for the development of the conflict that is the focus of the song may be more readily replaced by other themes, which themselves can be easily linked to this central segment. [132] This compositional principle is true not only for the Gacko corpus, but for South Slavic ballads in general, and confirms that multiformity is equally operative in the ballad as in the other oral traditional genres of this area. At the same time, this principle is the very reason for the classification problems discussed earlier in this introduction. The examples provided here vividly demonstrate that a tragic or happy ending in multiforms of the same song came about by a singer’s employing one or the other of two interchangeable themes or sets of themes. Most importantly, the use of either does not alter the stylistic features of the song in question.

The principle of multiformity in oral traditional poetry, which allows for one component to be replaced by another at virtually any level, is precisely what gives the contemporary reader a sense of the cyclical nature of this material. For example, the three different themes that are found at the very beginning of the multiforms from cycle 1 (weaving, suitors competing for {59|60} the maiden, and overhearing the secret) are also encountered in the openings of many other cycles. Although they may occasionally appear in other parts of the song, it seems that their primary function is that of an introductory theme. On the semantic level they are neutral enough to be linked with any number of themes. If we now turn and follow these themes in different cycles we will, in fact, very soon depart significantly from cycle 1. This type of investigation, in which particular themes are tracked down in various contexts, often resembles taking exits off a series of roads only to find oneself eventually on a road that very much resembles the one from which the journey originated. However, neither the road nor the landscape surrounding it will prove to be quite the same.

While one of the tasks of my analysis above was to determine the stability of particular patterns by investigating a number of multiforms, my primary goal here is to demonstrate how the themes can be modified and employed in a number of contexts. As discussed earlier, in multiform 1, by Hasnija Hrustanović, the first theme describes a maiden weaving silk in her tall tower in expectation of her beloved. Multiform 34b, by Mena Tanović, opens similarly with a depiction of a maiden embroidering, but no formulas are shared by these two songs. [133] The reason for this is, in all likelihood, their different metric patterns (10- and 13-syllable lines, respectively). While in text 1 this opening theme precedes the conflict regarding the maiden’s marriage, in 34b it sets the path for the theme of an evil mother-in-law who falsely accuses her son’s wife of flirting with her brother-in-law. Another example that uses similar verses in its opening is Emina Šaković’s song 32, which are followed by a theme presenting the problem of a young woman’s infertility. In this text the maiden does not weave but rather she embroiders, as in 34b (opening thus with the same half-line formula “Vezak vezla”— “[she] sat embroidering”). However, it shares most of its other formulaic expressions from this part with Hasnija’s 1. Furthermore, in the opening of Hasnija’s multiform 5 we find a maiden embroidering again, but this theme (significantly truncated) is now combined with one that describes the maiden being sought in marriage by many suitors. The ensuing theme depicts the maiden choosing a husband on the basis of presents sent by her many suitors. Finally, in Hasnija’s song 13 the opening theme of the maiden embroidering is not combined with elements from other themes, but instead it develops in yet another direction: the description of the attack of the “infidels” and the abduction of the maiden. The theme of the maiden embroidering or weaving thus serves as a road from which one can reach, to mention only a few, the theme of a forced marriage and abduction of the maiden; the theme of the evil mother-in-law accusing her daughter-in-law of flirting; the theme of a wife’s infertility; the theme of a maiden choosing a husband; or the theme of the maiden’s abduction as part of a raid. This widely employed opening is thus not exclusive to any particular plot and can therefore conveniently serve as an overture to otherwise dissimilar songs. In some instances it is {60|61} reduced to its core lines and clustered with other opening themes. [134] As we have seen above, this is only one of the several themes that have the function of a neutral opening. From the examples cited here it is apparent how one and the same theme can be creatively modified not only in songs by different singers, but even in different songs by the same singer. Indeed, the process of composition of a traditional ballad resembles in this regard very much that of composing an epic, inasmuch as the singer combines themes or clusters of themes having corresponding functions into different patterns. Themes occurring in South Slavic oral ballads are no less pliable than those occurring in the epic, and singers manipulate and alter them with equal ease and proficiency. [135]

Although the singer’s choice of themes and thematic elements may seem entirely random, this is not the case. [136] For example, if the maiden sings about a certain type of problem, she tends to be overheard by somebody, which further dictates the singer’s choice from a certain range of themes. If, on the other hand, she simply embroiders and waits, the direct intervention of another character is not necessarily called for. This does not mean, of course, that any theme that includes singing and being overheard must always be followed by exactly the same scenario. On the contrary, there still exist many different paths from which the singer can choose, but there are also those that she knows she cannot select if the song is to work on the semantic level. While certain choices that the singer makes are determined by the type of the story she wants to narrate, others are matter of personal taste, ability, and experience.

It has already been observed that in epics certain themes tend to cluster, owing to narrative demands and their customary association in a tradition, and that their association is not always linear. [137] In multiforms from different ballads that share the same clusters of themes, the corresponding themes may appear in conjunction with one another, but they may also be separated. In other words, they may be juxtaposed in a multiform of one song, but separated by other themes in a multiform of another. Moreover, it can happen that only one of two themes that tend to appear together is included in one multiform, while the other theme is found (also without its pair) in another multiform belonging to the same cycle. The following example illustrates what I have in mind. Multiform 5b begins with the theme of many suitors competing for one girl and proceeds immediately to the maiden’s choosing of her future husband on the basis of the rings offered by the suitors. In 1f we find the same opening theme about many suitors seeking one girl in marriage. In this text, though, the theme describing the offering of the wedding rings is found towards the end of the song. By contrast, multiform 5 opens with a different opening theme, that of the maiden embroidering in a truncated version. {61|62} Equally reduced is the next theme that mentions in a mere two lines the suitors competing for the maiden’s hand and is followed by the theme describing the offering of the rings. Finally, multiform 5a has the theme of many suitors seeking a maiden’s hand, but does not mention the offering of rings at all. This multiform actually employs the theme of a prearranged marriage (brothers giving their sister away without her consent to a suitor she does not like), which is a segment that corresponds to the situation in song 1. To make the picture clearer we could visualize this situation as follows:

Most multiforms in cycle 5 in the Parry Collection proceed to describe the maiden choosing the most expensive ring offered and, unknowingly, the least attractive suitor. Pretending sudden weakness the maiden makes it possible for one of the young suitors to approach and abduct her. In song 1f, on the other hand, the maiden escapes from her room with her beloved while the wedding rings are being collected from the suitors in the other room. The theme of abduction of the maiden by her beloved is shared by both cycles, but its context is quite different in their respective multiforms. Moreover, in multiform 5b the abduction never takes place and the maiden happily marries the suitor she chose. Similarly in 1a, as pointed out earlier, an abduction also never takes place, but there the maiden dies in the misery of a forced marriage.

I have mentioned that multiforms of the same song often conclude with different themes. Once again it is possible to observe this phenomenon from the opposite angle and say that the same concluding theme can accommodate a number of different plots. Such is the one describing the joint burial of a couple that features in a variety of ballads dealing with unfortunate lovers. Regardless of the cause of the characters’ demise, or the details pertaining to the tragedy, this romantic theme seems to accommodate an entire range of tragic plots. We find it in cycle 3, in which the ill groom-to-be gives his future bride to another man; in cycle 6, dealing with a mother’s forced marriage of her son; in cycle 33, describing the consequences of a husband’s infidelity, and in other cycles.

The list of examples and possible combinations of themes is far from being exhausted, since each cycle of multiforms has its own family tree in which themes can be imagined as relatives who connect themselves to other family trees. Themes clearly fluctuate from one cycle of songs to another within the same genre, but as demonstrated earlier in this introduction, they also travel among the narrative verse genres of one tradition or several linguistically related traditions much in the way formulas do. While in lyric songs a circular “network of traditional associations” (most of which average two lines in length) is employed as one of the {62|63} basic structural ingredients, in the ballad and other narrative verse genres themes (varying in length from a few lines to dozens of lines) circulate freely from one group of songs to another, resulting in a similar set of interchangeable and modifiable larger-scale components. [138]

The concluding part of this chapter investigates multiformity as caused by dialectal influences in the smallest component of oral traditional poetry, the formula. [139] The central South Slavic area presented to Parry and Lord in the 1930s a rare example of three cultures and religions existing in close proximity for centuries. Linguistically, too, the situation was unusual since, despite differences, mutual comprehension was rarely in question. As a result, oral poetry could travel freely from one ethnic group to another. The areas around Gacko and Dubrovnik (on which I will focus) provide a striking example of how the fluidity of oral poetic practices could traverse even very strong cultural borders. The Gacko region was technically a part of the Ottoman Empire until the Balkan war of 1912, when the full annexation of Bosnia and Herzegovina by Austro-Hungary took place. It had a mixed Muslim and Serbian Orthodox population and limited contacts with foreign cultures. Dubrovnik, on the other hand, was an independent republic until the arrival of Napoleon’s troops in 1806. It had a mostly Catholic population, highly developed trade, and western-oriented institutionalized art. In 1815 it became a part of the Habsburg Monarchy. Since in previous centuries Dubrovnik was the more prosperous of the two, people living deeper inland would regularly make trips to the coast in search of work. Despite their many differences, oral singing brought these cultures together since, along with labor and goods, formulas and plots appear to have been exchanged with equal zest.

Many formulaic expressions, even blocks of lines, undergo minimal changes when they cross from one tradition to another, particularly if these traditions share a broad linguistic base. Others, however, become colored by local dialects. In this type of exchange it is of particular {63|64} interest to observe the patterns of modification of specific formulas in order to see how they accommodate different lexical items conveying the same meaning within the same metrical frameworks. Let us look at an example shared by both cultures. A ten-syllable formula from a ballad recorded in the Gacko area, “Pošetala u šikli odaju” (“She stepped to the beautiful room”), [140] is transformed in the Dubrovnik region into “Pošetala b’jele u kamare,” again ten syllables with a similar meaning (“She stepped to the white rooms”). [141] In another set of examples from the same areas the correspondence is also obvious in the locative part of the line: “Ona ode u šikli odaju” (“She went to the beautiful room”) [142] and “Ane ide u bile kamare” (“Ane went to the white rooms”). [143] The stock epithet used most commonly in combination with “odaja” (“room”) in the Bosnian Muslim oral tradition is “šikli,” a word derived from French via Turkish meaning “beautiful, nicely decorated, ornamented with silver and gold.” [144] In traditional poetry from along the Croatian coast rooms are in a vast majority of instances described as white. It should be noted that white is the most frequently mentioned color in songs from Croatian- and Serbian-speaking areas and that it is most commonly used as an expression of the beauty of the described object. Although the semantic implications of stock epithets are valued differently from epithets in everyday language, these two adjectives actually do come very close in meaning. The equivalent of “šikli” is rendered in songs from Southern Croatia as “b’jela” or “bila,” depending on the area. A problem could have arisen since standard Croatian “bijela” has an extra syllable, thereby wrecking the metric pattern. In the Dubrovnik dialect, though, the reflex of “jat”—“ije” is pronounced not as two, but as one syllable. [145] The word “b’jela” thus perfectly fits the meter. [146] North of Dubrovnik, the Dalmatian (čakavian) rendering of the same adjective, “bila,” is used instead. {64|65}

In the same formula, “Ona ode u šikli odaju,” the word “odaja” (“room”) is of Turkish origin, and is transformed into the Italianate “kamara,” or more often plural “kamare,” in songs from Dalmatia. The Venetians held most of the Croatian coast, with the exception of Dubrovnik, until 1797. Along with several centuries of political and economic dominance, their linguistic influence left deep traces. Dubrovnik, though free, was surrounded by Italian culture and maintained heavy trade connections with Venice throughout the period. It is thus not surprising that the standard Croatian word for room, “soba,” which again does not have the right number of syllables, was superseded by the Italian word. The tendency to preserve the meaning and the metrical scheme in all these formulas is obvious. Although I have used examples from ballads to demonstrate the point, the same type of analogies exists in other genres, including the epic. These occurrences confirm once again that the seeming rigidity of the formula turns, in practice, into flexibility. It is precisely this pliability of the material at all levels that needs to be kept in mind when discussing the problems of textual stability in oral traditional song.

It is my hope that this analysis has demonstrated how a study of multiforms across different genres, regions, dialects (even languages), periods, cultural/ethnic groups, and genders may uncover, by means of comparison, ways of establishing otherwise hidden links and help us identify common compositional features in a large body of heterogeneous materials.

Oral traditional poetry and mythology have numerous points of contact in most traditional cultures since each expresses in distinctive ways many of the values of a given society. Since it would be beyond the scope of this study to investigate a broad variety of such connections, I will limit my discussion of mythological elements in the Gacko ballads to an analysis of a single formulaic distich. A focused study of this kind makes clearer the existence of subcutaneous associations and common meanings in several related folk traditions. Furthermore, it offers a fresh perspective on the dissemination of formulas and formulaic expressions and provides support for the thesis that many of these expressions occur in songs belonging to different genres and different regions.

The “incest at harvest” cycle (number 18 in this volume), in which a brother tries to take advantage of his own sister, contains in most of its multiforms some version of the following formulaic distich:

In addition to the multiforms gathered in the Gacko area, these verses appear also in a vast majority of multiforms from Croatia, Bosnia, and Serbia, as well as in several other cycles that deal with incestuous situations between a brother and a sister. In some songs the siblings do not even know they are related. They get married, but their kinship is eventually revealed to {65|66} them in the course of the song, as is the case in text 17 in this volume. In other, uncontaminated songs that do not present the subject matter of incest, a version of this formulaic distich with mythological roots appears always in connection with some other type of wrongdoing. This group of cycles is smaller than the incest group, but an interesting range of subjects appearing in it deserves a closer look.

In a number of songs various evil deeds are committed against a young woman (rather than a sister). A good example is Parry 2472, by Halima Hrvo, in which a maiden is forced to marry a man she despises. A sudden storm brings rain on this unhappy wedding procession. During the wedding night the man uses physical punishment to force the maiden to submission only to send her away in disgrace the following day. This song is in essence a multiform of song 3073 (number 4 in this volume) by Hasnija Hrustanović. A similar subject matter is found in the cycle on the frozen bride, whose multiforms also often include the sudden storm distich. [147] In this group an older and stronger man demands a young groom send him his horse and surrender his rights to his bride-to-be, or else fight for his life. Despite many disadvantages, the groom decides to fight, but, unbeknownst to him, his mother sends on the horse to his challenger. The abductor sends that very horse for the maiden along with the wedding procession that brings her to him. As he joins the procession and they cross the mountain, a sudden snow storm kills everyone but the frozen bride, whom the brave horse carries back to her beloved. While in all multiforms of this song a sudden storm involves the arrival of heavy clouds, only in some of them is there mention of the lightning that kills the abductor. [148] A more typical scenario is a deadly snow storm that descends from heavy clouds killing not only the abductor-groom, but also his guests, who presumably are to be considered equally guilty for their participation in the celebration of a forced marriage.

Multiforms of an interesting lyric song in which a man betrays his promise to a maiden and marries another woman exist in three different sources from different periods. [149] All of them contain elements from the same formulaic distich, but in no text is the young man killed. Rather, his punishment is called for by the betrayed maiden. Further, in one song by Kate Murat from Šipan, the sudden storm distich announces the punishment a man is about to receive for raping a young woman. [150] Elements from the sudden storm formulaic distich appear also in a handful of songs in which other types of sins are committed. To this category belong the songs that describe laboring on Sunday, dishonesty, stealing, and the desire of a mortal to marry a goddess. The distich is also used in dreams, serving as a premonition of somebody’s ill-wish, thus yet another type of wrongdoing. It further appears in one of the longest epics recorded in the South-Slavic territory, “Osman Delibegović i Pavičević Luka,” not in the context of incest however, but rather as a warning of an ill-fate that awaits the {66|67} opponent party. [151] Finally, a rare example in which not a man, but a woman is punished by a stroke of lightning is a song from the Matica hrvatska series in which a mother cruelly abandons her crippled children. [152]

Such consistent employment of these lines in connection with moral transgressions, in particular incest, or more rarely, other types of sinful behavior, could not be a coincidence. Its explanation necessitates an investigation into the possible common roots of the idea contained within it. Owing to its popularity in South Slavic oral traditions, the subject of incest between a brother and a sister has received considerable attention. Delorko traces this topic back to the literature of ancient Egypt and provides a brief study of Croatian folk songs dealing with it. [153] Milošević-Đorđević stresses that the subject of incest between a brother and a sister is far more widespread than any other type of taboo relationship in South Slavic oral poetry. She explains the popularity of this topic in Indo-European and other oral traditions, noting that within the matriarchal organization of primitive societies sexual relationships between siblings were allowed. In the period of transformation from matriarchal to patriarchal societies, male siblings fought for dominance over their female siblings. In support of this thesis she quotes, among other things, part of a folk story from a collection by Karadžić that deals with the subject of three brothers competing for their sister. [154] Finally, in her investigation of spatial references and their implications in the epic, Mirjana Detelić investigates manifestations of divine power on earth. Among other examples, she mentions the subject of brother/sister incest in the context of its ethical implications within the Christian world. [155]

It has been generally accepted that Perun, the god of lighting and thunder, is one of the principal deities in South Slavic mythologies. [156] Depending on the geographic area within the Slavic world the hierarchy may differ, but generally it appears that the throne was held by either Svantevid or Perun. In the description of Svantevid, some scholars recognize the god of the Sun, also known as Svarog or the White God. The god Perun is clearly a Slavic version of heavenly thunder masters such as Zeus, Jupiter, Thor, Thurr, or Donnar, who are all related to the Hindu deity Indra. Natko Nodilo draws the etymology of the word “Perun” from the proto-Slavic root “per-” from which the word “p’rati,” to strike, is derived. He also speculates {67|68} that the name Perun could have been derived directly from Parjányah, one of the names under which Indra appears in the Rig-Veda. [157]

The sudden storm formulaic distich that appears so regularly in the harvest ballad might very well be describing the wrath of the god Perun at the sight of an incestuous situation. [158] It is significant that the lines about an actual or possible punishment that strikes from the sky are found in the same types of songs from all three religious groups of the Balkans (Catholics, Muslims, and Orthodox Christians). Although some Christian elements are evident in a few Croatian and Serbian songs, as are influences from the Kur’an in several Muslim songs, the fact that these verses are used so consistently by singers from all three groups points to the possibility that their roots lie in the period preceding the spread of these three religions among the South Slavs.

In light of this thesis it may be of interest to look briefly at the multiforms of the “incest at harvest” song from different areas and different periods. As we have seen, the sister, who carries lunch to her brother, does not accept his advances in any of the Gacko multiforms. On the contrary, she begs her brother not to speak further and escapes home. In some texts she informs her mother about her brother’s actions. The sinful brother is eventually killed by a hand from heaven, a stroke of lightning. In some Croatian multiforms dating from the second half of the 19th century, the brother achieves what he desires, but the mother of the two siblings kills them both when she finds out what has happened. [159] In a Sarajevo multiform, dating from 1894, the sister initially consents to her brother’s demands, but eventually changes her mind and kills both him and herself. [160] In a smaller subgroup from this cycle, recorded mostly in Dalmatia at the end of the 19th century and the middle of the 20th, there is no mention of a harvest, but rather the brother watches after the cattle or horses and his sister brings him lunch. [161] He attempts to take advantage of her while she washes herself at the spring. In some multiforms, such as the one recorded by Karadžić, the brother’s occupation is not specified and he harasses his sister when she goes to fetch fresh water from the spring. [162] The endings in this subgroup vary from the sister escaping home, to the sister killing her brother, to their mother punishing them both. Finally, a rare Dalmatian example from the 1960s should {68|69} be mentioned. In it the sexual attraction between the two siblings is mutual, but curiously, there is no mention of the formulaic expression describing a sudden storm. [163] It must also be observed that this song is likely a result of conflation of the incest ballad and a humorous song in which the mother questions her daughter about her whereabouts the previous night. This is particularly apparent in the second half of the song when the balladic tone changes into a humorous exchange. As no guilt regarding inappropriate behavior is ever expressed, and as the ethical system is completely different from that in the ballad, it is to be expected that the formulaic distich on which we are focusing would not appear in this text.

In those multiforms in which the time of the action can be established, the season in question is summer. The signs indicating this are the harvesting of hay, cattle being taken to pasture, and the maiden washing herself at a spring. This temporal detail is relevant for a number of reasons. During this season of the year sudden storms with spectacular shows of lightning and thunder are common in south-eastern Europe. More importantly, the timing of the event described in the ballad places it approximately around the feast of Sv. Ilija (St. Elias), a saint believed by Christian South Slavs to be in charge of lightning and thunder. This feast, celebrated on August 2nd, is known amongst Christians in this part of the world as “Ilinden,” but it is also observed on the same day among Muslims as “Aliđun,” or “Ali’s day.” Moreover, several scholars have provided convincing evidence that Sv. Ilija is, in fact, a Christian incarnation of the Slavic thunder-god Perun. [164] The significance in the Gacko region of August 2nd, a holiday with clear pagan roots, was still considerable in the 1930s. In the interview with Đula Dizdarević, the singer mentions “Aliđun” and stresses that both the Muslims and the Christians of that area would regularly join in the festivities. [165] Although each of the more recent religions provides a valid explanation for celebration on this particular day, the sudden storm distich seems to confirm that it is the power of a pagan deity that has survived in this part of the world—all the way to the twentieth century.

An unusual fact about the Christian Saint Ilija is that he has a sister (typically referred to in South Slavic folk traditions as Ognjena Marija, Fiery Mary), who is also in charge of lightning, and a brother (known as Pantelija), who controls summer heat and fire. In some geographic regions it is believed that Sv. Ilija actually has two sisters, Blaga Marija and Ognjena Marija (Gentle Mary and Fiery Mary). Detelić observes that unusual familial relationships are quite atypical of Christian saints, while they are customary in pagan religions. [166] Confirmation {69|70} of this thesis is found in a few songs collected by Vuk Karadžić. In one of them Fiery Marija calls Thunder-Master Ilija her brother. [167] The song further discusses the division of powers given to various saints by God, according to which Ilija came to be in charge of thunder and his main helper, Marija, controlled lightning and thunder-bolts. Basing his conclusion on the same song from Karadžić, Nodilo also establishes that the mythic symbols of Fiery Mary are lightning and fire. [168]

Further evidence for Sv. Ilija’s connections with the Slavic pagan pantheon is found in a Croatian song belonging to a small group of songs that involve transgressions other than incest. [169] A sudden storm is indicated with the words “Navukla se tanka oblačina.” (“A large thin cloud covered the sky”). The formulaic distich discussed here underwent an obvious and significant transformation in this case, as did the context in which it appears: a group of harvesters do not observe the holiness of Sunday and are punished for this. The basic idea contained in these lines, punishment for a forbidden deed, is the same as in the incest songs. Curiously, the detail of harvesting is also maintained in this song, as is implicitly its temporal framework. Sinful harvesters are struck by thunder, ice, and lightning, respectively, from “prvi svetac gromovnik Ilija” (“the first saint, thunder-master Ilija”), “drugi svetac ledena Marija” (“the second saint, icy Mary”), and “a treći je Munja iz oblaka” (“and the third one, Lightning from the cloud”). The infiltration of Christian elements into a pre-Christian ethical system is visible not only from the reason for the punishment, but also from the saints’ names. However, a mythological residue is equally present as Sv. Ilija appears here with his pagan sisters, one of whom still bears her non-Christian name, Munja (Lightning). The other one, Icy Mary, is a version of the more common Fiery Mary, who has no connection with Jesus’ mother. The name “Mary” in songs of this kind is clearly of more recent origin. [170] In this song it is not Marija, but Sv. Ilija’s other sister-helper who is in charge of lightning.

Christianity, though, is not the only tradition associated with the sudden storm distich. In the Muslim version of the other incest cycle (text 17 in this volume), elements from Islam are encountered. In this song a widowed mother sends her only son to the school for pashas* and raises her nine daughters herself. For many years there is no contact between the son and the rest of the family. After having reached adulthood he unknowingly asks for the hand of one of his sisters. It is only in a conversation during the first wedding night that the two young people discover they are related. As they are about to consummate their marriage, lightning strikes on the bed between them, warning them of their potential sin. It is at that moment that they ask about one another’s background and recognize their kinship. The song ends with the brother promising to build “three white mosques” and feed and clothe three hundred poor people in order to receive pardon for his sins. {70|71}

It is worth mentioning briefly several more examples to illustrate how broad the spectrum of songs is in which the same set of formulas appear. A popular plot found in most Croatian, Bosnian and Serbian collections is structurally related to the Muslim song just discussed. It tells of a sister who was taken away by the Turks as a child and of her subsequent union with her brother. The way they meet varies: the sister is given by chance to her brother as a part of the yield from a plundering expedition; the woman is sold by her bankrupt husband and her lost brother happens to purchase her; the mother sends her son to look for her lost daughter and on the way the son abducts one of the maidens, not knowing that she is his own sister. [171] In all sub-groups the brother and sister recognize their kinship on the basis of body marks (moles, old wounds), either on their own or with the help of their mother. In this cycle of songs the occurrence of the sudden storm formula is lower than in the other songs discussed thus far, possibly because the focus shifts in most of the multiforms from the possibility of an incestuous relationship to the moment of recognition. This is obvious in the significant elaboration of the recognition theme in most of these instances. Still, the formulaic distich under investigation is used in a sufficient number of multiforms to confirm the role it plays in cycles dealing with the subject matter of incest. [172]

In a song entitled “Dušan Wants to Marry His Sister,” which Karadžić included in a volume of epics, an incestuous situation is brought about by the emperor, who, not wishing to see the land divided between him and his sister, insists on marrying her. [173] The sudden storm distich is re-phrased into a formulaic expression stating that the sky would burst open and the sun disappear sooner than she would consent to marry him. Finally, two examples of interest were collected on the island of Šipan, Croatia. One is a 133-line ballad by Kate Murat, a singer whose exceptional talent I have had occasion to mention several times in this introduction. [174]  {71|72} The other multiform is by another Šipan singer, Mare Fortunić, and is 160 lines in length. [175] The only connection that these songs have with the cycles discussed above is our mythological distich. The outline of the plot overlaps in both multiforms: a sister begs her brother not to go to war since she has had a dream warning that he will not return alive from battle. Determined to go, the brother criticizes her unusual affection and expresses irritation that she is the only maiden seeing a brother off. When she learns that her brother is wounded, the sister hurries to help him. Once she finds him, she needs to fetch water for him and uses drops of his blood to mark the stones showing her the way back. The angry heavens interfere by sending rain that erases the stone marks. She is lost in the mountains, and her brother dies. The sister apparently dies of sorrow and, at the very end of the song, is “dined on by wolves.” The incestuous relationship is only hinted at in this song through the brother’s criticism of her inappropriate attachment to him, but it may very well be the reason for the otherwise inexplicable tragedy. In one of the multiforms, the brother is wounded by a gun before he has a chance to reach the battle, in the other the cause of his wounds is not specified. Although the brother is the one who suffers physically, it is the sister whose emotional drama is the center of attention. The relevant verses do not mention lightning, but they do describe a clear sky turning cloudy and silent rain falling from the clouds. Yet another indication that the Šipan singers were probably hinting at an incestuous relationship is Kate Murat’s song “Vjerenica cara od Stambola na silu” (“The Maiden Who Was Forced to Marry the Sultan of Stambol”), in which the mention of the clear sky bursting open is unmistakably used in the context of incest. [176]

It is my hope that these examples demonstrate how essential it is for the interpretation of certain songs that a correlation with the mythological sphere be investigated. [177] In some materials, such as the ballad about the son-serpent (text 38) or the capturing of the vila (cycle 15), these links are abundantly clear. In others they may not be so obvious, but nevertheless the links to mythology open up new perspectives on the origins of some of the songs and can also cast more light on the relatively obscure pantheon of the Slavic pagan deities.

As with any type of orally transmitted material, an awareness of the key role played by the tradition and of the type of knowledge that comes with it cannot be stressed enough. Being a part of this same tradition, the singer’s audience shares not only the cultural knowledge, but {72|73} also the special language of oral poetry. This, however, is not necessarily the case with the reading audience who is, in addition, quite often dependent on translations. Even if one neglects the important factor of oral vs. visual perception of the material, a difficult question arises for anyone attempting to make an oral tradition available in translation: how can a broad spectrum of meanings, captured often in one phrase or even a word, be conveyed in an equivalent way in a different language to reading audiences who do not share the same cultural background? A few brief examples may illustrate the problems involved. The word “džeferdar” (derived from Turkish cevherdar), which is an archaic type of short gun lavishly decorated with mother-of-pearl and precious stones, has no true equivalent in English. The same can be said for most terms involving clothing and decorations. A particularly complicated term is “boščaluk” (from Turkish bohçalik), rendered in English here as “a present of fine attire,” which is indeed only an approximation. The full meaning of this word is a present consisting of garments wrapped up in a kerchief or a piece of linen (from which it took its name). A traditional “boščaluk” would typically contain underwear, a shirt, a kerchief made of the finest white cotton with branches embroidered in golden thread in its corners, and either a long cotton kerchief (also embroidered with golden thread), or a belt made of fine fabric for tying broad pants. It was customary for a bride to give this kind of present to the groom, his relatives and his most important wedding guests. Thus “a present of fine attire” is only a pale shadow of the original meaning. Similarly, certain features of style such as figura etymologica (cognate accusative) have been impossible to render truthfully in translation. For example, “lov lovijo” literally means “he was hunting his hunt,” while “vezak vezla” translates verbatim as “she embroidered her embroidery.” In the original such constructions are succint and allow for a smooth flow of the verse, but in translation they often appear cumbersome both in terms of semantics and rhythm.

In his consideration of the same problem in epics, David Bynum quotes an excellent example of several verses which depict a young man admiring a lovely maiden dancing. She is described as the one “who had sent him a letter” (“što je njemu knjigu opravila”). Bynum suggests that the actual meaning of this verse is “She who had sent him the letter proposing to elope and be his wife.” He points out that the proposed meaning of this “composite of traditional formulas” is valid “not only at this point in this narrative, but also at the same intersection of characters and events in dozens of other tales (many of them otherwise quite different from this). Any singer experienced enough to have sung this song in this tradition, and every one of his listeners who was experienced enough to understand him, must necessarily have heard numbers of those other tales before the singing or hearing of this one.” Bynum further suggests that because the singer’s words have a broader meaning than “their mere lexical import,” this needs to be reflected in the translation. [178]

A situation similar to that described by Bynum is also found in the ballads in this volume. Already in the very first text there is a typical scenario in which a maiden sends a letter to her {73|74} beloved. Unlike in some other multiforms from this cycle, in text 1 she does not state explicitly that she wants to elope with the man and become his wife. Rather, this meaning is implied. Similarly, multiform 1c mentions the maiden’s letter, but reveals nothing of its content. From the examples mentioned here the reader can gain an insight into how pregnant with meaning each formula or phrase can be, with much more implied than is actually stated.

The translations in this volume follow the original texts as closely as possible, with interpretive renderings of the original lines kept to a minimum. An attempt to render culturally specific terms in descriptive language would have proven not only extremely difficult, but would also have interfered with the principal features of the genre. Ballad is characterized by extreme conciseness and a high degree of drama. For that reason it was important to maintain the flow of narration and preserve the rhythm wherever possible. While a more descriptive approach is acceptable for the epics, which are by their nature more attentive to detail and more digressive, it is, in my view, ill-suited to the ballad.

Unfortunately, preserving the original prosodic features in translation was not an option. Most of the songs are decasyllabic with a caesura after the fourth syllable. Some of them are octosyllabic, however, such as the popular song about the execution of the Morić brothers (text 37), or the song about a wife’s adultery during her husband’s absence (text 29), and maintain the caesura after the fourth syllable. Others are in a thirteen-syllable meter with a caesura after the eighth syllable (seen in two multiforms from cycle 34). I have paid particular attention to formulas and formulaic expressions and made every attempt to translate these with exactly the same wording every time they occur. It is my hope that the reader who has had no previous contact with South Slavic oral traditions will be able to note these patterns and follow them in different contexts. Herein lies much of the beauty of these songs.

The problem of instability of personal names in South Slavic traditional poetry, recognized by other scholars, is present also in the songs in this volume. As Delorko pointed out, the role that a character is playing is more important than his or her name. [179] Thus, as can be seen in cycle 1, the maiden is called Duda, Uma, and Zlata. In each instance it is the same lyric persona that is in question. This fluidity in the use of names has also been noticed within a single song. For example, in song 11, the name of the maiden is Umihna Nazić, however, in line 19 she is called Naza. In song 1a, Dervišbeg Čengić, the maiden’s beloved, is renamed Dedaga Čengić in its second half. For clarity’s sake such changes have been indicated in the footnotes. As can be seen from the second example, titles such as agha and bey (aga and beg) are frequently used interchangeably. Since aga has two syllables and beg only one, this type of change often results from metric constraints.

To avoid confusion, in each text only one form of each character’s name is used in the translation. While to a native speaker it is clear that Melća, Mejra, and Melećhana are one and the {74|75} same person, this may not be the case with a reader of the English text. In terms of the original manuscripts, these are devoid of punctuation, which, in the interests of clarity, I have provided. I have normalized the capitalization of proper names, but have not altered the spelling of words from how they appear in the notebooks. In a few instances I have separated two words written as one; these changes are not indicated in the notes. Otherwise no changes have been made. As a result, expressions in ijekavica, ikavica and ekavica [180] can be found next to one another, as well as forms such as što/šta and đevojko/djevojko/devojko. [181]

The songs presented here provide only one facet of the extraordinarily rich oral tradition of Bosnia and Herzegovina. Despite an attempt to apply scholarly criteria in their selection, the final choice has nevertheless been tinged with subjective preferences. It is hoped that these outstanding examples of folk art will bring us a step closer to understanding the type of skill and talent necessary for their creation, and that they will serve at the same time as an important cultural document of a tradition that has all but vanished. {75|}