Chapter 3. The Formula

There came a time in Homeric scholarship when it was not sufficient to speak of the "repetitions" in Homer, of the "stock epithets," of the "epic clichés" and "stereotyped phrases." Such terms were either too vague or too restricted. Precision was needed, and the work of Milman Parry was the culmination of that need. The result was a definition of the "formula" as "a group of words which is regularly employed under the same metrical conditions to express a given essential idea." [1] By this definition the ambiguity of "repetitions" was eliminated; we were henceforth to deal with repeated word groups, not with repeated scenes, although Bowra uses the term "formula" still to apply to both. [2] At the same time, Parry's definition broadens "formula" to include within its scope more than the repeated epithets. Furthermore, the opprobrium attached to "clichés" and "stereotyped" has been removed.
Students of epic have now willingly applied themselves to the study of the repeated phrases by textual analysis, by counting repetitions, classifying similar phrases and thus extracting the technique of composition by formula manipulation. Yet in following this method they tend to treat all texts alike, whether by the same singer or not, whether sung or dictated, whatever, indeed, the circumstances of their collection may have been. Much has been gained from this type of analysis, and from it surely much more remains to be learned concerning the details of the process in any given tradition. Yet it seems to me that in confining ourselves to this method we tend to obscure the dynamic life of the repeated phrases and to lose an awareness of how and why they came into being. Are we not conceiving of the formula as a tool rather than as a living phenomenon of metrical language? In this chapter we shall attempt to look at the formula not only from outside in terms of textual analysis, but also from within, that is, from the point of view of the singer of tales and of the tradition.
The stress in Parry's definition on the metrical conditions of the formula led to the realization that the repeated phrases were useful not, as some have supposed, [3] merely to the audience if at all, but also and even more to the singer in the rapid composition of his tale. And by this almost revolutionary idea the camera's eye was shifted to the singer as a composer and to his problems as such. [30]
At all stages in our musings about oral epic we find it necessary to recreate in our imagination not a general but a specific moment of performance. The singing bard must be our guide; and the singing bard is never a type, but an individual. Whenever we say "the singer does this or that," we must make it clear that our statement is based on experience with a specific singer, or on the combined experience of various singers. Our method will be to follow the developing career of the young singer, beginning even from the time when he starts to absorb the tradition by much listening to the songs about him and continuing with each advance of his own flight of song.
It may seem strange that we have very few texts from singers in the earliest stage of apprenticeship, as it were, in their art. But collectors seek the best singers, and the best singers are usually the older men. Their reputation is great; they are brought forward by those whom the collector questions. On occasion a younger singer in his twenties or thirties may be suggested, often because he has a good voice or a fine manner of singing, more rarely because he is a narrator of quality. Yet it should surprise us that it has not occurred to anyone to make a special study of the youngest group. It is a commentary, indeed, on the force of the belief that the songs are set and that younger singers have not had time to memorize a song as well as an older man. Perhaps exposing this belief as false will encourage giving more attention to songs of the youngest singers, imperfect though they may be.
Surely the formula has not the same value to the mature singer that it has to the young apprentice; it also has different values to the highly skilled and to the unskilled, less imaginative bard. We may otherwise think of the formula as being ever the same no matter from whose lips it proceeds. Such uniformity is scarcely true of any element of language; for language always bears the stamp of its speaker. The landscape of formula is not a level steppe with a horizon which equalizes all things in view, but rather a panorama of high mountains and deep valleys and of rolling foothills; and we must seek the essence of formula at all points in the landscape. Moreover, with the penetrating eye of the mind we must look for this essence backward through the centuries which formed the mountains and the valleys. For the singing we hear today, like the everyday speech around us, goes back in a direct and long series of singings to a beginning which, no matter how difficult it may be to conceive, we must attempt to grasp, because otherwise we shall miss an integral part of the meaning of the traditional formula.
Or to use another figure, the formula is the offspring of the marriage of thought and sung verse. Whereas thought, in theory at least, may be free, sung verse imposes restrictions, varying in degree of rigidity from culture to culture, that shape the form of thought. Any study of formula must therefore properly begin with a consideration of metrics and music, particularly as confronted by the young singer first becoming aware of the [31] demands of his art. Later we shall have to consider the question of why story becomes wedded to song and verse, to ask ourselves what kind of tale finds its expression in these very special methods of presentation. These are not problems that the contemporary singer of tales faces; for he has inherited the answers. The fact of narrative song is around him from birth; the technique of it is the possession of his elders, and he falls heir to it. Yet in a real sense he does recapitulate the experiences of the generations before him stretching back to the distant past. From meter and music he absorbs in his earliest years the rhythms of epic, even as he absorbs the rhythms of speech itself and in a larger sense of the life about him. He learns empirically the length of phrase, the partial cadences, the full stops.
If the singer is in the Yugoslav tradition, he obtains a sense of ten syllables followed by a syntactic pause, although he never counts out ten syllables, and if asked, might not be able to tell how many syllables there are between pauses. In the same way he absorbs into his own experience a feeling for the tendency toward the distribution of accented and unaccented syllables and their very subtle variations caused by the play of tonic accent, vowel length, and melodic line. [4] These "restrictive" elements he comes to know from much listening to the songs about him and from being engrossed in their imaginative world. He learns the meter ever in association with particular phrases, those expressing the most common and oft-repeated ideas of the traditional story. Even in pre-singing years rhythm and thought are one, and the singer's concept of the formula is shaped though not explicit. He is aware of the successive beats and the varying lengths of repeated thoughts, and these might be said to be his formulas. Basic patterns of meter, word boundary, melody have become his possession, and in him the tradition begins to reproduce itself.
In the months and years of boyhood, not very long indeed after he has learned to speak his own language, the future singer develops a realization that in sung stories the order of words is often not the same as in everyday speech. Verbs may be placed in unusual positions, auxiliaries may be omitted, cases may be used strangely. He is impressed by the special effect which results, and he associates these syntactic peculiarities with the singing of tales. Moreover, the linking of phrases by parallelism, balancing and opposition of word order become familiar to him; the verb, which occurs, for example, just before a syntactic pause, is repeated at the beginning of the next phrase or is balanced by a verb just before the following stop: (The verbs in the passage are italicized.)
Đe sedimo, da se veselimo,
E da bi nas i Bog veselio,
Veselio, pa razgovorio!
Where we sit, let us make merry,
And may God too make us merry,
Make us merry and give us entertainment!
In these pre-singing years, together with a sense of new arrangements of ideas and the words which express them, the boy's ear records the repetitions [32] of the sounds of the words. His instinctive grasp of alliterations and assonances is sharpened. One word begins to suggest another by its very sound; one phrase suggests another not only by reason of idea or by a special ordering of ideas, but also by acoustic value.
Thus even before the boy begins to sing, a number of basic patterns have been assimilated in his experience. Their form may not be precise—the precision will come later—but it can be truly said that in this youth the idea of the formula is in process of becoming. What we shall soon designate as melodic, metric, syntactic, and acoustic patterns are forming in his mind.
The chief reason, of course, that the formula does not take precise shape at this stage, is that only the necessity of singing can produce a full-fledged formula. The phenomenon of which it is a manifestation arises from the exigencies of performance. Only in performance can the formula exist and have clear definition. Besides, not all the singers whom the boy hears in his family or community have the same formulas for a given idea or the same manner of treatment of formulas. There is no rigidity in what he hears.
What has been described so far has been an unconscious process of assimilation. Consciously the boy has been thinking of the stories themselves which are related in this unusual way. But when he begins to sing, the manner of presentation comes for a long time to the fore. Then the formula is born for him and his formula habits are acquired.
One of the first problems for the young singer from the very beginning is to learn to play the instrument which accompanies the song. This is not a really difficult task, since most of the instruments which accompany chant are not intricate. In the Yugoslav case, the boy has to learn to bow a one-stringed instrument, the gusle, the range of which is open string plus four fingers, an ambitus of five notes. The rhythm is primary; the grace notes are ornamental. Some older singer may show him how to finger the instrument, or the boy may simply imitate his elders by himself in private. He may make a small gusle for himself, because the grown-up size is too big for his hands, or his father or mentor may make one for him. He imitates the fingering, the melody, and the manner of his elder. Rade Danilović in Kolašin has told us how his father, Mirko, used to put the boy's hand on his own as he fingered the string (Parry 6783).
Thus begins the stage in which the rhythmic impressions of the earlier period of listening are fitted to the restrictions of the instrument and of a traditional melodic line. Usually the rhythms and melodies that the youth learns at this period of initial specific application will stay with him the rest of his life. He may acquire others from singers of great reputation or striking manner of performance, but they will be in addition to the earlier ones or, at most, they will only modify, not replace them.
At the same time, the boy is trying to sing words. He remembers the phrases he has heard, sometimes whole lines, sometimes only parts of lines. [33] From now on, for a considerable period of time, he will listen to his elders with more attention to the lines and phrases. He will pick them up from any singer whom he hears. As he practices singing by himself he realizes the need for them and he uses them, sometimes adjusting them more or less consciously to his own needs, sometimes unconsciously twisting them. They are not sacred, but they are useful. In this way he acquires the formulas of his elders and establishes his own formula habits. He is doing what all singers before him have done.
The most stable formulas will be those for the most common ideas of the poetry. They will express the names of the actors, the main actions, time, and place. Thus in the line, Vino pije Kraljeviću Marko, ‘Kraljević Marko is drinking wine’, Kraljeviću Мarko presents the hero in a complete second-half-line formula. Kraljević, properly a title, ‘king's son’, or ‘prince’, is treated as a patronymic. In another line, Sultan Selim rata otvorio, ‘Sultan Selim declared war’, the title ‘Sultan’ makes it possible to name Selim in a four-syllable initial formula. The young singer learns that patronymics, titles, and indications of city of origin, for example, od Qrašca Tale, ‘Tale of Orašac’, are of great use in naming his heroes. Epithets are not so frequent in this tradition because the shortness of the line does not present a need for them that cannot be fulfilled by title or patronymic. They come into usage either when there is no title or because the make-up of the line does not allow a long patronymic, or when the singer wishes to express the actor in a whole line, frequently a vocative, as in Sultan Selim, od svijeta sunce, ‘O Sultan Selim, light [sun] of the world’.
The most frequent actions in the story, the verbs, are often complete formulas in themselves, filling either the first or the second half of the line, as in Govorio Kraljeviću Marko, ‘Kraljević Marko spoke’. If the verb is a syllable short, a conjunction often completes the formula, as in Pa zasede svojega dorata, ‘Then he mounted his brown horse’. The length of the action formula is naturally in part determined by whether the subject is expressed in the same line and by the length of the subject. The singer finds that he can say, ‘Marko said’, in the first half of the line with subject expressed, Marko kaže, or in the second half line, govorio Marko, or in the whole line, govorio Kraljeviću Marko. Obviously here the length of the subject is influenced by the length of the verb. If the subject is not expressed, if the singer wants to say merely, ‘he said’, govorio does very well for the first half of the line; the addition of a conjunction and the personal pronoun come to his aid in the second half line, pa on govorio, as does also very frequently a change of aspect of the verb, pa odgovorio, ‘then he replied’. But in order to accomplish this in a whole line, the singer must repeat the idea in the second half of the line: Govorio, riječ besedaše, ‘He spoke, he uttered a word’. This example illustrates that the object of a verb forms an integral part of the verb formula, and shows as well how and why pleonasm is so common in oral style. Many of the formulas for [34] the second half of the line are made up of verb and object: rata otvorio, ‘opened war’; knjigu napisao, ‘wrote a letter’. By a change of tense this last formula is often expressed in the first half of the line as Knjigu piše, ‘writes a letter’. In both cases the other half of the line is left for the subject.
A third common set of formulas indicates time when the action occurs. A typical example, with Homeric overtones, is: Kad je zora krila pomolila, ‘When dawn put forth its wings’, or Kad je zora i bijela dana, ‘When it was dawn and white day’, or Kad je sunce zemlju ogrijalo, ‘When the sun had warmed the earth’.
The singer must learn another category of common formulas indicating the place where an action occurs. ‘In Prilip’, for instance, can be expressed in the first half of the line U Prilipu, in the second half of the line by u Prilipu gradu, and in the whole line by U Prilipu gradu bijelome, ‘In Prilip, that white city’. Similarly, ‘in the tower’ can be expressed in the first half of the line by A na kuli, with the conjunction a as a filler; in the second half line by na bijeloj kuli, ‘in the white tower’, and in the whole line by Na bijeloj od kamena kuli, ‘In the white tower of stone’.
The formulas represented by the preceding examples are the foundation stone of the oral style. We have seen them from the point of view of the young singer with an essential idea to express under different metrical conditions. Their usefulness can be illustrated by indicating the many words that can be substituted for the key word in such formulas. For example, in the Prilip formulas above, any name of a city with a dative of three syllables can be used instead of Prilip: u Stambolu, u Travniku, u Kladuši. Instead of a u kuli, ‘in the tower’, one can say a u dvoru, ‘in the castle’, or a u kuči, ‘in the house’. These formulas can be grouped together in what Parry, when studying the traditional epithets in Homer, termed "systems." [5] It is often helpful to write them as follows:
    ⎧ kuli
au⎨ dvoru
    ⎩ kući
Such a substitution system expresses graphically the usefulness and the relationship of a group of formulas.
A style thus systematized by scholars on the foundation of analysis of texts is bound to appear very mechanical. Again we may turn to language itself for a useful parallel. The classical grammar of a language, with its paradigms of tenses and declensions, might give us the idea that language is a mechanical process. The parallel, of course, goes even further. The method of language is like that of oral poetry, substitution in the framework of the grammar. Without the metrical restrictions of the verse, language substitutes one subject for another in the nominative case, keeping the same verb; or keeping the same noun, it substitutes one verb for another. In studying the [35] patterns and systems of oral narrative verse we are in reality observing the "grammar" of the poetry, a grammar superimposed, as it were, on the grammar of the language concerned. Or, to alter the image, we find a special grammar within the grammar of the language, necessitated by the versification. The formulas are the phrases and clauses and sentences of this specialized poetic grammar. The speaker of this language, once he has mastered it, does not move any more mechanically within it than we do in ordinary speech.
When we speak a language, our native language, we do not repeat words and phrases that we have memorized consciously, but the words and sentences emerge from habitual usage. This is true of the singer of tales working in his specialized grammar. He does not "memorize" formulas, any more than we as children "memorize" language. He learns them by hearing them in other singers' songs, and by habitual usage they become part of his singing as well. Memorization is a conscious act of making one's own, and repeating, something that one regards as fixed and not one's own. The learning of an oral poetic language follows the same principles as the learning of language itself, not by the conscious schematization of elementary grammars but by the natural oral method.
Any thorough grammar of a language notes exceptions to "rules," dialectal differences, "irregular" nouns and verbs, idioms—in fact those divergences from the systematized rules that arise in usage and in the normal organic change constantly in operation in a living spoken language. If we analyze oral epic texts that are recorded from actual performance rather than texts taken from dictation and normalized to some extent, we can observe the oral poetic language in its pure state, with its irregularities and abnormalities arising from usage. Then it is clear that the style is not really so mechanical as its systematization seems to imply.
The value to us of drawing up a number of substitution systems is that we immediately begin to see that the singer has not had to learn a large number of separate formulas. The commonest ones which he first uses set a basic pattern, and once he has the basic pattern firmly in his grasp, he needs only to substitute another word for the key one. The actual basic formulas which any given singer may learn first would be practically impossible to determine; it would vary from singer to singer. Probably if the first song learned by the singer concerned Marko Kraljević, Marko's name and the varieties of it used in making lines would set the basic pattern for similar names, which would fall into a four-syllable plus two-syllable pattern. The fundamental element in constructing lines is the basic formula pattern. There is some justification for saying indeed that the particular formula itself is important to the singer only up to the time when it has planted in his mind its basic mold. When this point is reached, the singer depends less and less on learning formulas and more and more on the process of substituting other words in the formula patterns. [36]
Although it may seem that the more important part of the singer's training is the learning of formulas from other singers, I believe that the really significant element in the process is rather the setting up of various patterns that make adjustment of phrase and creation of phrases by analogy possible. This will be the whole basis of his art. Were he merely to learn the phrases and lines from his predecessors, acquiring thus a stock of them, which he would then shuffle about and mechanically put together in juxtaposition as inviolable, fixed units, he would, I am convinced, never become a singer. He must make his feeling for the patterning of lines, which he has absorbed earlier, specific with actual phrases and lines, and by the necessity of performance learn to adjust what he hears and what he wants to say to these patterns. If he does not learn to do this, no matter how many phrases he may know from his elders, he cannot sing. He does this in performance, not before an audience at first, of course, but by himself. This style has been created and shaped in performance; it has been so with all singers since time immemorial, and it is so with him. The habit of adjustment, the creation of lines in performance, this is acquired from the moment the boy begins to try to sing.
What is meant by "adjustment" can best be comprehended in terms of the establishment of various kinds of patterns and rhythms of expression. These the boy has picked up in his pre-singing years and he now finds his own means of forming them naturally and readily. We may begin again with the melodies of the singing itself. The boy learns that there is a special pattern for the opening of a song, with its own beginning and cadence. There is at least one oft-repeated melodic pattern for sustained narrative. Sometimes in the course of his life the singer acquires from one to three variations of this most important pattern. It is quite possible that he has discovered that by changing the melody he rests his voice. On occasion, but by no means regularly, the melodic pattern shifts for dramatic emphasis. There is a modified version of the singer's main pattern for stopping before а rest and another somewhat modified version for reprise after a pause. The song also has its concluding cadence. An example of these patterns can be seen in the appendix to Volume I of Serbocroatian Heroic Songs in the musical transcriptions of the "Captivity of Đulić Ibrahim," sung by Salih Ugljanin in Novi Pazar with music notations from the records by Béla Bartók.
From these musical examples one can see also the rhythmic patterns, generally trochaic. Here the play or "adjustment" between melody and meter can be observed in operation. We note the inadequacy of our texts without music in presenting a picture of epic song. The line is syllabic, or better, syllabo-tonic, a trochaic pentameter with an invariable break after the fourth syllable. It is simple, yet subtlety has entered from the interplay between melody and text. There is a tension between the normal accent and the meter. The accent of the meter does not always fall on the normal prose accent, [6] nor are all five stresses of the same intensity. The ninth syllable [37] is the most prominent, has the strongest beat, and is held longest; the seventh and eighth are the weakest. The tenth may be lost entirely, completely swallowed, or hopelessly deformed. It may be carried over to the beginning of the following line, [7] or it may be an ordinary short beat. The first and the fifth syllables tend to be of the same intensity because they are the initial beat in the line and the first after the break; but when a proclitic stands in these positions, as is very common at the beginning of the line and not unusual in the fifth syllable, the first and third feet are sometimes iambs rather than trochees, and the melody follows this rhythm. Occasionally the first foot, sometimes even the second or third foot, is a dactyl in the regular practice of some singers; and they have sets of formulas adjusted to this rhythm. [8] In these cases the extra syllable is often supplied by a word without meaning.
It is noteworthy also that Serbocroatian maintains a pitch accent, rising or falling, and pays much attention to long and short vowels. The subtlety of the rhythms is, of course, further complicated by these characteristics of the language. The metric differences here demonstrated required at an early stage an adjustment of formula by the singer, or perhaps were called into being because of an adjustment. Individual variations in melody and rhythm are greater than one might expect, and only when the actual melodies of recorded songs are published will this fact be properly realized. [9] Some idea of the range of variation can be obtained from sample lines from three singers (see pages 39–41). [*]
Under the pressure of rapid composition in performance, the singer of tales, it is to be expected, makes occasional errors in the construction of his lines. His text line may be a syllable too long or a syllable too short. This does not trouble him in performance, and his audience scarcely notices these lines, since they have an understanding of the singer's art and recognize these slight variations as perfectly normal aberrations. The singer himself adjusts his musical line to the text by making a dactyl out of a trochee or by holding one syllable for two rhythmical heats rather than for one.
An additional set of patterns, related to the rhythmic patterns, which the singer must learn to control in these first years, is that of word boundaries, or more properly, length of accentual groups (that is, a word plus proclitics and enclitics). This need is especially important to the singer because the feeling for the mid-line break is very real. An accentual group cannot, and in practice only very rarely does, bridge the fourth and fifth syllables, although neither the melodic nor rhythmic patterns show this. When listening to the song one hears no pause at the break. The end of the line is very clearly marked, and run-on lines are few. In the first half of the line the most common word-boundary patterns are 2–2, 1–3, and 4: vino pije, ‘he [38] drinks wine’; pa govori, ‘then he says’; Kraljeviću, or a vodi ga, ‘and he leads him’ (where a is proclitic and ga is enclitic).
In the second half of the line the most common patterns are 2–4, 4–2, and 3–3: juriš učinio, ‘he made an attack’; zametnuše kavgu, ‘they started strife’; and besedi serdaru, ‘he says to the sirdar’. Most of the formulas that the singer hears are in these patterns, and he will make new ones on the basis of them.
Closely allied to the word-boundary patterns, to no small extent helping to form them, are the syntactic patterns of the formulas. The order in which the parts of speech appear, hence the relation of ideas, is involved. In a style in which actions or things are added one to another in series, the conjunction plays a large role, and the most common patterns for the beginning of the line naturally begin with a conjunction. In fact conjunction-verb in the first half line is very frequent. For example: [41]
A ćesar se na me naljutijo,
Pa na mene naljetljemu dao,
Pa me danas surgun učinijo, [10]
A prati me k tebe u Bagdatu,
And the emperor was angered at me,
And he inflicted outlawry upon me,
And today he has exiled me,
And sent me to you in Bagdad,
(II, No. 1:1194–1197)
There are many initial formulas beginning with a conjunction, especially when an uncompounded form of the verb is used, for example, the narrative present, the imperfect, or the aorist. In the case of compound tenses, the auxiliary appears in the first half of the line and the participle or infinitive in the second. In the latter half of the line one finds most of the noun-epithet combinations: knjigu šarovitu, ‘well-writ letter’; visoku planinu, ‘high mountain’; gradu bijelome, ‘white city’. [11]
A tasevi od srme bijele,
A sinđiri od žežena zlata.
Ej, Spanula bagdatska kraljica;
The cups were of white silver,
And the chains were of 'fined gold.
Ej, Then appeared the Queen of Bagdad.
(II, No. 1:1143–1145)
Such are the syntactic patterns which the boy now begins to store in his experience and to use as a basis for new phrases.
The second half of the line is dependent not only syntactically on the first, but is also to some extent suggested by the sound patterns with which the line opens. There are a number of lines that have become set through the pattern of internal rhyme: Kud god skita , za Aliju pita, ‘Wherever he wanders, he asks for Alija’; Zveknu halka, a jeknu kapija, ‘The knocker resounded, and the gate echoed’. The importance of alliteration is apparent in such a line as Kazaše ga u gradu Kajniđu, ‘They pointed him out in the city of Kajniđa’, in which the k-g alliteration is arranged in chiastic order, k-g-g-k. Nothing would seem to have hindered the singer from using u Kajniđu gradu in the second half of the line, but he appears to have preferred the chiastic order, in part also perhaps under the influence of the a-u-a-u assonance in the middle of the line. The singers have a sensitivity to proportion and completeness of form even within the limits of a single line. Whatever feeling for such sound patterns the boy has absorbed in his pre-singing days is crystallized when he begins to perform.
This period in his training is pre-eminently one of learning to produce lines. Part of the process is accomplished by remembering and using phrases heard from other singers. This constitutes one element in the continuity of oral epic style. The phrases help to establish in the singer's experience a series of patterns, and these patterns are also an element in the continuity of the style. At the same time, by necessity, because he does not remember all the phrases which he needs, he is forced at the moment of his private performances to form phrases on the basis of the patterns. Since they follow the traditional patterns, they are indistinguishable from the other phrases that he has remembered, and may unconsciously be actually identical with them. To him the first matter of importance is certainly not the source [42] of the phrase but the phrase itself at the critical time. For anyone, however, who is trying to understand how a particular style comes into being, it is necessary to note that there are two ways by which a phrase is produced; one is by remembering it, the other is through creating it by analogy with other phrases; and it may well be impossible to differentiate between the two. While both remembering and creating (in the sense of making, not necessarily "originating") play important roles, the latter, creating, is especially significant. The singer cannot, and does not, remember enough to sing a song; he must, and does, learn to create phrases. Hence the most important elements in the style are the basic patterns which we have illustrated, and which are established at this period.
In the course of time and of much practice, the need for a particular phrase arises over and over again. Whether it is one remembered from other singers or one created anew (and perhaps re-created several times as the need recurs), a phrase becomes set in the poet's mind, and he uses it regularly. Then, and only then, is the formula really born. The remembered phrase may have been a formula in the other singer's songs, but it is not a formula for our singer until its regular use in his songs is established. The remembered phrases from other singers are more numerous, of course, in the early years of training, and decrease gradually as the ability to make phrases is developed, although both processes continue during the singer's lifetime. The phrases for the ideas most commonly used become more securely fixed than those for less frequent ideas, with the result that a singer's formulas are not all of the same degree of fixity. Indeed, the creating of phrases continues always as well. I believe that we are justified in considering that the creating of phrases is the true art of the singer on the level of line formation, and it is this facility rather than his memory of relatively fixed formulas that marks him as a skillful singer in performance.
The very fact that the practice of oral narrative song has endured so long is proof enough that it can absorb new ideas and construct new formulas. But the process of building formulas is so quiet and unspectacular and so slow that it is almost imperceptible. Since the patterns of thought and the rhythm of presentation remain unchanged, the new words in the formulas are not noticed except when the ideas behind them are in striking contrast to the surroundings in which they occur. Thus proper names, recent foreign or international words, and the inventions of a mechanized age, when they find their way into the songs, as they do and must, provide us with the means of studying new formulas. It would be nonsense to suppose that the singer in whose songs these novelties are found is their originator. He may be, but the chances are against it.
New formulas are made by putting new words into the old patterns. If they do not fit they cannot be used, but the patterns are many and their complexity is great, so that there are few new words that cannot be poured into them. Salih Ugljanin's song of the Greek War (I, No. 10), a song [43] which he claimed to have made up himself, contains some new ideas. He uses the word Avropa in the sense of ‘the rulers of Europe’, Avropa me odi zatvorila, ‘Europe imprisoned me here’, and Avropa me is only a variation of Ibrahim me or Mustafa me. The Queen of England, although a newcomer, is perfectly at home in the line Misir daše ingliskoj kraljici, ‘They gave Egypt to the Queen of England’; we are familiar with both the moskovska kraljica, ‘the Queen of Moscow’, and the bagdatska kraljica, ‘the Queen of Bagdad’. When, however, we come upon Ti načini sitne teligrafe, ‘Prepare short telegrams’, the newness strikes us in the face. Salih is singing of a new age and he has simply substituted the new means of communication for the old type of official document, the bujruntija. Тi načini sitne bujruntije was his model. But when he tries to use the three-syllable nominative singular teligraf he runs into difficulty. The nominative singular bujruntija has four syllables, and the other most common missives, knjiga and ferman, have two. Formulas for communication have heen built with either four- or two-syllable words in mind. He is thinking of Od sultana brže knjiga dođe, or Od sultana brže ferman dođe, ‘A ferman came swiftly from the sultan’, when he sings Od sultana brže teljigraf dođe. In the last appearance of the word in his song he has solved the problem and found the right pattern: Pa kad takav teligraf dolazi, ‘When such a telegram arrived’.
Even in a song of olden times new words have crept in. Avdo Međedović uses terms that he must have picked up when he was in the army. In Parry Text 12389, the action of which, at least in Avdo's imagination, is placed in the days of Sulejman the Magnificent, we find Moja braćo, moje dve kolege, ‘My brothers, my two colleagues’ (line 415), O kolega , Fetibegoviću, ‘O my colleague, Fetibegović’ (line 2376), Ja sam na tο riskirao glavu, ‘It is for that that I risked my life’ (line 1570), A na njima careva niforma, ‘They were wearing imperial uniforms’ (line 4085), and Sve soldata , sve pograničara, ‘All soldiers, all men of the border’ (line 6794). One can thus observe that the Yugoslav tradition was still very much alive in 1935 and still receptive to new ideas and new formulas. [12]
We have seen a bard's formulas coming into existence from the earliest period of his singing and we have noted the significant fact that they are not all alike either in their genesis or in their intensity of "formulicity." We have also suggested that the formulas themselves are perhaps less important in understanding this oral technique than the various underlying patterns of formulas and the ability to make phrases according to those patterns.
In order to avoid any misunderstanding, we must hasten to assert that in speaking of "creating" phrases in performance we do not intend to convey the idea that the singer seeks originality or fineness of expression. He seeks expression of the idea under stress of performance. Expression is his business, not originality, which, indeed, is a concept quite foreign to him and one [44] that he would avoid, if he understood it. To say that the opportunity for originality and for finding the "poetically" fine phrase exists does not mean that the desire for originality also exists. There are periods and styles in which originality is not at a premium. If the singer knows a ready-made phrase and thinks of it, he uses it without hesitation, but he has, as we have seen, a method of making phrases when he either does not know one or cannot remember one. This is the situation more frequently than we tend to believe.
Thus far we have attempted to show the way in which the formulaic style enters into the consciousness of a young singer as he learns to use it for the telling of tales. Such a living art, so closely united to individual experience, cannot help but leave its peculiar stamp upon the songs and their texts. Because of this mark left upon them we can with a high degree of certainty determine whether any text that is before us was formed by a traditional bard in the crucible of oral composition.
Formula analysis, or even more generally textual analysis, must begin with a scrutiny of a sample passage in order to discover the phrases in it that are repeated elsewhere in as much of the work of an individual singer as there is available. In doing this we are following Parry's example. He took the first twenty-five lines of the Iliad and of the Odyssey and underlined those groups of words which he found repealed elsewhere in Homer. One needs only to glance at his charts [13] to see how many formulas there are in those samples. Chart I does the same for the Yugoslav material.
From Volume II of the Parry Collection we have chosen a passage of fifteen lines from the "Song of Bagdad," which was sung for phonograph recording by Salih Ugljanin in Novi Pazar in 1934 (II, No. 1). The singer was an old man at the time of recording and an accomplished performer with a large repertory, which he claimed included one hundred songs. His style, therefore, is not that of a beginner. The sample has been selected from the middle of the song rather than from the very beginning, because many of the Yugoslav songs open with an invocation which can be used for any song. Most Yugoslav epics are shorter than the Homeric poems, and we have had to use several of Salih's songs for corroborative purposes, rather than just two, in order to have sufficient material for analysis.
We have attempted, moreover, to choose a passage that did not contain one of the more frequently recurring themes such as those of letter-writing or of the arrival of an army on the field of assembly. In other words, the sample has been selected with an eye to making the experiment as valid as possible and to anticipating any objection which might be brought that the passage is of a sort that would be more formulaic by the very nature of its position or of its contents. For a similar reason, we have not admitted as [45] supporting evidence for establishing a formula any repetition which occurs in the same passage in the two other versions of the same song by the same singer which are included in the material analyzed.

Chart I [14]
Jalah reče, / zasede đogata;
––––––––    ––––––––––––
With "By Allah" she mounted her horse;

Đogatu se / konju zamoljila:
–  –  – –   –  –  –  –  – –  –  –
–  –  – –   –  –  –  –  – –  –  –
She implored the white horse:

"Davur, đogo, / krilo sokolovo!
– –  –  –  –  –    – – –  –  –  –
– – – – –  –      –––––––––––
"Hail, whitey, falcon's wing!

Četa ti je / o zanatu bila;
–   –  –  –   –  –   –  –  –
Raiding has been your work;

Vazda je Mujo / četom četovao.
–  –  – – – –   –   – –  –  –  –  –
–  –  – – – –   –   – –  –  –  –  –
Ever has Mujo raided.

Vodi mene / do grada Kajniđe!
– –  – –   –   – –  –  –  –  – –  –
– – –  –  –   ––––––––––––––
Lead me to the city of Kajniđa!

Ne znam đadu / ka Kajniđi gradu."
– –  –  –  –  –  –  –  – –  –  – –  –
–––––––––––    –––––––––––––
I know not the road to the city of Kajniđa."

Hajvan beše, / zboriť ne mogaše,
–  – –   –  –  –   –   – – –  –  –  –
– – –  –  –  –   ––––––––––––––
It was a beast and could not talk,

Tek mu svašta / šturak umijaše.
–  – –  –  –  –    – – –  – – –  –
–  – –  –  –  –    – – –  – – –  –
But the steed knew many things.

Ode gljedat' / redom po planini.
– –  – –  –   – –  –  –  – – – –  –
–––––––––      –––––––––––––
He looked over the mountains

Uze đadu / ka Kajniđi gradu,
And took the road to the city of Kajniđa,

Pa silježe / planinama redom,
–––––––   –––––––––––––––
And crossed one range after another,

Pa ga eto / strmom niz planinu,
–  –  – –  –   – –  –  – – – – – –
–––––––    –––––––––––––––
Until lo he rushed down the mountain, 

I kad polju / slježe kajnićkome,
–  –  –  – –    –  –  – – –  – –  –
–  –  –  – –    –  –  – – –  – –  –
And when he descended to the plain of Kajniđa,

Kome stati / polje pogljedati,
––––––––   – – –  – –  – – –
Were anyone to look out over the plain,

In Chart I we have underlined the four-, six-, and ten-syllable phrases found more than once in the perusal of about 12,000 lines from the same singer. The chart is designed to show that in relation to 12,000 lines of diverse material from a given singer a certain number of phrases in a given passage are formulas. Twelve thousand lines is the approximate length of the longest of songs and will serve as a basis for comparison with the Homeric poems and others. These 12,000 lines constitute eleven different songs, three of which are recorded on the phonograph discs, four recited, [46] but not sung, for the records, and four taken down from dictation. They give a good cross section of the more than 30,000 lines available from this singer.
From the chart we can see at a glance the number of repeated phrases that without any hesitation can be called "formulas." These phrases we know by demonstration that the singer has come in time to use regularly. Even within the limited number of lines used in the experiment, that is, 12,000, one quarter of the whole lines in the sample and one half of the half lines are formulas. It is most significant that there is no line or part of a line that did not fit into some formulaic pattern. In certain instances the pattern was a very common one and there was no difficulty in proving the formulaic character of the phrase. In a few instances the evidence was not so abundant, but it was still sufficient to make one feel certain that the phrase in question was formulaic. A number of the formulaic expressions could very easily have been classified as formulas, had we relaxed our established principles and standards. For example, davur đogo in line 791 misses being a formula because the evidence lists only davur šturan and davur doro. But đogo, šturan, and doro are all terms for horses. We could thus have easily increased the number of formulas.
Had we gone beyond 12,000 lines, the number of formulas would have continued to mount, and had we included material from other singers it would have increased still further, until it became clear that almost all, if not all, the lines in the sample passage were formulas and that they consisted of half lines which were also formulas. In other words, the manner of learning described earlier leads the singer to make and remake phrases, the same phrases, over and over again whenever he needs them. The formulas in oral narrative style are not limited to a comparatively few epic "tags," but are in reality all pervasive. There is nothing in the poem that is not formulaic.
Moreover, the lines and half lines that we call "formulaic" (because they follow the basic patterns of rhythm and syntax and have at least one word in the same position in the line in common with other lines or half lines) not only illustrate the patterns themselves but also show us examples of the systems of the poetry. Thus, although the beginning of line 790 was not found repeated exactly in the material analyzed, it belongs in a system of initial formulas made up of a three-syllable noun in the dative followed by the reflexive. Another example of the system is junaku se. The system would be written:
đogatu ⎫
            ⎬ se
junaku ⎭
Similarly, in line 791 davur đogo belongs in a system with
davur ⎨
          ⎩ šturan [47]
Any two-syllable word for a horse can fit into this system with davur. Finally, around the second half of the first line in the chart a lengthy system can be formed:
                  ⎧ đogata
                  ⎪ kočiju
                  ⎪ dorata
                  ⎪ paripa
zasednu  ⎫⎪ hajvana
zasedem ⎪⎪maljina
zasede    ⎬⎨ binjeka
zasedi     ⎪⎪mrkova
zaseo      ⎭⎪vranina
                  ⎪ menzila
                  ⎪ šturika
                  ⎪ zekana
                  ⎩ eždralja
Since the singer learns his art from other singers and in his turn influences them, there are many formulas which are used by a large number of singers. For example, the following formula, line 789 from Chart I, is to be found in the songs of other singers from Novi Pazar: [15]
Jalah reče, zasede đogata.
"By Allah," she said, she mounted the white horse.
Sulejman Fortić
Jalah reče, posede đogina.
"By Allah," he said, he mounted the white horse.
(II, No. 22:433)
Jalah reče, posede hajvana.
"By Allah," he said, he mounted the animal
(II, No. 23:308)
Đemail Zogić
Jalah reče, sede na dorina.
"By Allah," he said, he mounted the brown horse.
(II, No. 24:746)
Jalah reče, posede hajvana.
"By Allah," he said, he mounted the animal.
(II, No. 25:31)
Sulejman Makić
I to reče, posede dorata.
And he said this, he mounted the brown horse.
(Parry 677:714)
Alija Fjuljanin
A to reče, zasede hajvana.
And he said this, he mounted the animal.
(Parry 660:435)
One should not conclude, of course, that these singers learned these formulas from Salih or he from them. Salih learned them bit by bit from the singers whom he heard, and they from all whom they heard, and so [48] forth back for generations. It would be impossible to determine who originated any of them. All that can be said is that they are common to the tradition; they belong to the "common stock" of formulas.
Although the formulas which any singer has in his repertory could be found in the repertories of other singers, it would be a mistake to conclude that all the formulas in the tradition are known to all the singers. There is no "check-list" or "handbook" of formulas that all singers follow. Formulas are, after all, the means of expressing the themes of the poetry, and, therefore, a singer's stock of formulas will be directly proportionate to the number of different themes which he knows. Obviously singers vary in the size of their repertory of thematic material; the younger singer knows fewer themes than the older; the less experienced and less skilled singer knows fewer than the more expert. Even if, individually, every formula that a singer uses can be found elsewhere in the tradition, no two singers would at any time have the same formulas in their repertories. In fact, any given singer's stock of formulas will not remain constant but will fluctuate with his repertory of thematic material. Were it possible to obtain at some moment of time a complete repertory of two singers, no matter how close their relationship, and from that repertory to make a list of the formulas which they know at that moment of time, there would not be complete identity in the two lists.
What is true for individuals is true also for districts. Differences of dialect and vocabulary, of linguistic, social, and political history will be reflected in thematic material and in formulas. The songs of Christian groups will have themes and formulas distinctive from those of Moslem groups, and vice versa. The formula stock of the Serbocroatian speaking district as a whole will be the sum total of the formulas known to its singers, but not all the singers will know all the formulas. One is ever being forced to return to the individual singer, to his repertory of formulas and themes, to the quality of his practice of the traditional art. One must always begin with the individual and work outwards from him to the group to which he belongs, namely to the singers who have influenced him, and then to the district, and in ever enlarging circles until the whole language area is included.
There would, however, be a large group of formulas known to all singers, just as in any speech community there are words and phrases in the language known to and used by all the speakers in that community. Even as these represent the most common and most useful ideas of the community, so too the stock of formulas known to all practitioners of the art of traditional narrative poetry represents the most common and most useful ideas in the poetry. Again they can be correlated with the thematic material. This common stock of formulas gives the traditional songs a homogeneity which strikes the listener or reader as soon as he has heard or read more than one song and creates the impression that all singers know all the same formulas.
The question whether any formula belongs to the common stock of formulas cannot be decided merely on the basis of its relative frequency in [49] the songs of any given singer. In order to find the answer we must know its distribution among the singers of the tradition. For work of this sort a formula index is necessary, but this is a labor of many hands over many years. Only by compiling such an index could we determine with any degree of accuracy the frequency and distribution of formulas and the number of different formulas within a tradition. It would readily show us what formulas comprise the common stock of two or more individual singers, of a given district, or of a group of districts, and of the language tradition as a whole. This would do for formula study what the great motif indexes have done for thematic study.
Once a singer has solved a particular problem in verse-making, does he attempt to find any other solution for it? In other words, does he have two formulas, metrically equivalent, which express the same essential idea? Parry has shown how "thrifty" Homer was in this respect. Bowra has indicated that this thrift is not found in other oral poetry. [16] What facts can we deduce from our Yugoslav songs in the Parry Collection?
In order to test the possibilities, we have taken one of the formulas in Chart I and traced the instances in some nine thousand lines of Salih Ugljanin's songs of the essential idea of the formula. The purpose was to discover whether Salih had only one formula to express that idea under any one set of metrical conditions or whether he had several. This would show his "thrift," if any. The essential idea chosen was that of the second half line, zasede đogata (line 789), ‘he [or she] mounted his [or her] white [or black, or gray, etc.] horse’. Horses play a very large part in Yugoslav traditional poetry, and the action of mounting them is frequently mentioned in Salih's songs.
In 3–3 rhythm in the last half line, with another clause ending at the break, and with a singular verb, Salih uses the following:
Jalah reče, zasede đogata
"By Allah," he said, he mounted his white horse.
(II, No. 1:789; No. 2:912)
Jalah rekni, zasedi đogata!
Say "By Allah" and mount your white horse!
(II, No. 1:1103)
Under the same conditions but in 4–2 rhythm, he uses:
Jalah reče, zasednuo vranca.
"By Allah," he said, he mounted his black horse.
(II, No. 18:795)
This change of rhythm was necessitated by the use of a two-syllable word for horse. In 2–4 rhythm, with a clause ending at the break, and with a plural verb, he uses:
Pa skočiše, konje zasedoše.
Then they leaped up, they mounted their horses.
(II, No. 17:323)
Here the first question arises. Since zasednuo vranca and konje zasedoše both contain a four-syllable word and a two-syllable word, why is the [50] rhythm of one 4–2 and of the other 2–4? There is a sound answer to this question. Zasednuo vranca is used in conjunction with jalah reče in the first half line, and the balanced chiastic pattern (object-verb, verb-object) of this common whole-line formula is in Salih's mind, so that jalah reče, zasednuo vranca follows along in the series with all the other instances of this full line. On the other hand, as we shall see shortly, when Salih uses konje he invariably puts it in this position in the line, and he is also following a different syntactic pattern. He has in mind such lines with skočiše as Svi skočiše, seljam prifatiše, ‘They all leaped up, they received the greeting’ (II, No. 2:248), where another balance of verbs prevails, namely subject-verb, object-verb, as well as internal rhyme. When the subject of the verb "to mount" is expressed, it must be put in the first half line:
Svi konjici konje zasedoše.
All the horsemen mounted their horses
(II, No. 1:880)
A svatovi konje zasedoše.
And the wedding guests mounted their horses.
(II, No. 4:1282)
Ta put hajduk šajku zasednuo.
Then the hajduk mounted his mare.
(II, No. 11:593)
A Mujo svoga pojaše đogata.
And Mujo mounted his white horse
(II, No. 11:694)
Jalah Suka sede na menzila.
With a cry to Allah, Suka mounted his post horse.
(II, No. 2:99)
The two instances of konje zasedoše here bear out what we said in the preceding paragraph. Salih always uses konje in this rhythmic pattern. But with šajku zasednuo we think back to zasednuo vranca and wonder why he did not say zasednuo šajku, following the same pattern in the second half of the line. First, however, we see that the syntactic pattern of the whole line is different from that of jalah reče, zasednuo vranca, the chiastic arrangement of which has already been indicated. Second, šajku usually occurs in this penultimate position in Salih's singing, especially in the common noun-epithet formula, šajku bedeviju, in the second half of the line. Added, then, to the pull of a whole-line syntactic pattern of subject, object, verb, is the influence of other formulas with šajku. In fact, such formulas begin in line 575, I sa šnjime šajku bedeviju, and continue with Helj da ostane šajka u aharu (576), No najprijem šajku izvodićeš (580), and Pa odriješi šajku bedeviju (584). In one of these cases the syntactic pattern of the second half line is the same as šajku zasednuo; namely, object, verb, šajku izvodićeš (580). Third, the two preceding lines end with the syntactic pattern, object, verb, and the rhythmic pattern, 2–4: Jedno hebe zlata napunili (591), and Pro konja hebe proturiše (592). Fourth, the vowel pattern discloses a chiastic order in the repetition of aj-u in the third to the sixth syllables, a-u-aj-u-aj-u-a-e-u-o. Šajku nicely repeats hajduk in the play of vowel sounds. Zasednuo šajku had no chance of breaking into such an aggregation of forces. [51]
With Mujo svoga pojaše đogata, the principle of vowel alternation is again operative. Although pojaše đogata and zasede đogata mean essentially the same thing and the metrical conditions are identical, the back vowels of the first half of the line and the "o-a" pattern of svoga, repeated in đogata at the end of the line, call forth pojaše rather than zasede. We can see that the two formulas are not real equivalents in the phonological context. The next line, Jalah Suka sede na menzila, is a peculiar one. Sede na menzila and zasede menzila both have the same meaning, although they are not true alternates, because the rhythm is different; the former is 2–4 and the latter, 3–3. The second half of the previous line was careva fermana, ‘imperial firman’; the 3–3 might have called forth a 3–3 in the following line, but not necessarily. The intrusion of the subject Suka in the first half of the line has caused a change in the line. Suka has taken the place of reče; the singer has in his mind jalah reče and also Suka reče and Suka sede. The two-syllable sede plays not only its own role but also that of reče, with the same vowel arrangement. The line is an irregular and awkward one. In line 242 of the same song, E! Jala, sede, krenu ka Budimu, ‘With a cry to Allah, he mounted, and set out for Budim’, one also finds the omission of the verb reče, and further adjustment in the line because of its absence. So far we have found no true alternates.
When a modifier is added to the idea "horse" or when an adverbial idea is to be added to the idea "mounted," the verb moves to the first half line. Or to state it in another way, if the verb is put in the first half line, some modifier must be added to the idea "horse" or an adverbial idea must be added to the idea "mounted." Thus we have:
Pa zasede krilata đogata.
Then he mounted his winged white horse.
(II, No. 1:1121)
Pa posede šajku bedeviju.
Then he mounted his bedouin mare.
(II, No. 11:627)
Eh, zasede njezina đogata.
Well, she mounted her white horse
(II, No. 2:862)
Zasedoše konje u avliju.
They mounted their horses in the courtyard.
(II, No, 4:1538)
Pa zaseše konje na jaliju.
Then they mounted their horses on the bank.
(II, No, 17:702)
Zasedoše dva konja menzila.
They mounted two post horses.
(II, No. 1:248)
Here there is only one violation of the principle of thrift. Posede and zasede are interchangeable. There is so slight a difference in meaning between these two perfective aspects of the verb that they can be considered as identical. Very likely the alliteration of posede with pa and bedeviju has played a role in its choice. Thus far there has been variation, but no clear-cut departure from the principle of thrift.
There are three more instances of mounting in the material studied: [52]
E! Jala sede, krenu ka Budimu.
With a cry to Allah, he mounted and departed for Budim.
(II, No. 2:242)
Đulić sede svojega dorata.
Đulić mounted his brown horse.
(II, No. 4:1541)
A gotove konje zasednuše.
They mounted their ready horses.
(II, No. 13:112)
In the first line Jalah reče zasede đogata has been telescoped into the first half of the line by omitting the verb "said" and the idea "horse," and by using the uncompounded verb sede, ‘he sat’. Strictly speaking, the "essential idea" is not the same as the one that we are investigating, because the idea "horse" is omitted, but even if it had been expressed, as in the line which follows it above, it would not break the principle of thrift, because the uncompounded verb is forced on the singer by the preceding two-syllable word. The following line bears this out. Nor does konje zasednuše affect our thesis. It is the same as konje zasedoše except that it uses the momentary aorist instead of the simple aorist. The singer undoubtedly had in his mind the verb krenuše, ‘they departed’, in the third and fourth syllables of the following line, so that the last two syllables of one line rhyme with the third and fourth of the following line.
When our judgment concerning thrift takes into consideration the acoustical context, there are few if any instances where substitution of one word for another even if they have the same essential meaning and metrical value is justified.
There has been a tendency to come to conclusions from an examination of all the songs in a collection regardless of whether they are from the same singer or even from the same district. Under such circumstances one would scarcely expect to find thrift. A singer's thriftiness is significant; that of a district or tradition less so (if it exists) for our purposes.
Indeed, it seems to me that the thriftiness which we find in individual singers and not in districts or traditions is an important argument for the unity of the Homeric poems. Homer's thriftiness finds its parallel in the individual Yugoslav singer, but not in the collected songs of a number of different singers.
Our brief excursion into the principle of thrift in actual oral composition among Yugoslav singers has served to emphasize the context of the moment when a given line is made. In order to understand why one phrase was used and not another, we have had to note not only its meaning, length, and rhythmic content, but also its sounds, and the sound patterns formed by what precedes and follows it. We have had to examine also the habits of the singer in other lines, so that we may enter into his mind at the critical creative moment. We have found him doing more than merely juggling set phrases. Indeed, it is easy to see that he employs a set phrase because it is useful and answers his need, but it is not sacrosanct. What stability it has comes from its utility, not from a feeling on the part of the singer that it [53] cannot or must not be changed. It, too, is capable of adjustment. In making his lines the singer is not bound by the formula. The formulaic technique was developed to serve him as a craftsman, not to enslave him.
In the foregoing, for the sake of clarity, we have spoken only of single lines and their parts. In actuality, lines cannot be isolated from what precedes them. The singer's problem is to construct one line after another very rapidly. The need for the "next" line is upon him even before he utters the final syllable of a line. There is urgency. To meet it the singer builds patterns of sequences of lines, which we know of as the "parallelisms" of oral style. As we have said, some sense of these is gained in the pre-singing period, but when the singer begins to practice and to train himself the patterns here too must become specific. Moving from one line to another is not merely, perhaps not even correctly, the adding of one ready-made phrase, or group of ready-made phrases, to another. Oddly enough, because of the variety of patterns for sequences of lines there is greater flexibility possible and greater skill is needed than in pure juxtaposition of formulas. The complexity and artistry of the result are often surprising to anyone who feels that illiterate singers can produce only simple structures. The passages below, chosen almost at random, will serve to illustrate the potentialities of the style.
In South Slavic song, the end of a line is marked by a pause for breath, by a distortion of the final syllable or syllables, frequently by an ornamental turn in the musical accompaniment. Since it is the close of a unit of composition, it is clearly emphasized. Very rarely indeed does a thought hang in the air incomplete at the end of the line; usually we could place a period after each verse. Of 2400 lines of Yugoslav epic analyzed, 44.5 per cent showed no enjambement, 40.6 per cent showed unperiodic enjambement (that is, the sense was complete at the end of the line, but the sentence continued) and only 14.9 per cent involved necessary enjambement. The greatest number of exceptions in Yugoslav epic involve a preceding subordinate clause, or a line consisting of a noun in the vocative case plus modifiers, [17] and even in these cases a thought, even if it is not the main thought of the sentence, has been presented whole by the end of the line. This absence of necessary enjambement is a characteristic of oral composition and is one of the easiest touchstones to apply in testing the orality of a poem. Milman Parry has called it an "adding style"; the term is apt.
In rapid, almost staccato, style the singer may add together a series of actions, moving the story quickly forward: (I have italicized the verbs.)
Kud gođ skita za Aliju pita. Wherever he went, he asked for Alija.
Kazaše ga u gradu Kajniđu. They said he was in the city of Kajniđa.
Kad tatarin pod Kajniđu dođe, When the messenger came to Kajniđa,
Pa eto ga uz čaršiju prođe, He passed along the main street,
Pa prilazi novom bazdrđanu, Then he approached the new shopkeeper,
Te upita za Alino dvore. And he asked for Alija's court. [54]
Bazdrđan mu dvore ukazao. The shopkeeper pointed out the court to him.
Kad tatarin na kapiju dođe, When the messenger came to the gate,
Pa zadrma halkom na vratima. He beat with the knocker on the door.
Zveknu halka a jeknu kapija. The knocker rang and the gate resounded.
  (II, No. 3:108–117)
Or he may break in on a series of actions with description, providing at one and the same time a more leisurely tempo and a richness of detail. The following passage has an almost Homeric touch:

Tevabije brže u podrume; The retainers went quickly to the stable;
Izniješe takum na đogata, They brought forth the trappings on the white horse,
Vas u srmi i u čisto zlato, All in silver and in pure gold.
Pa konjičko preturu oruže, Then they placed on the weapons for fighting from horseback,
S obe strane dvije puške male On each side two small pistols
Sa dva grla a zrna četiri. With two barrels which take four bullets.
Preložu hi surom međedinom, Over them they placed a brown bearskin,
Da mu rosa ne kvari oruže. That the dew might not rust the arms,
Pa preložu pulu abrahiju; Then they placed on a blanket with sequins;
Zlatna pera biju niz đogata. Its golden tassels beat against the white horse's flank.
Vezlje su je četiri robinje Four slave girls had woven it
U Dubrovnik za četir’ godine. In Dubrovnik for four years.
Pa udriše đema nemačkoga. Then they put a German bit into the horse's mouth.
Ej! Stasa đoga, žešće bit’ ne more! The white horse stood there, he could not have been prouder or fiercer!
  (II, 1:737–750)
This last line, beginning with a shout and sung in a different and cadential rhythm, marks the close of the passage. We have italicized the series of verbs which carry along the actions of caparisoning the horse, and also the lines which break this forward movement by providing ornamental, descriptive details that add color and poetry to the actions themselves. The vivid adornments may be added one to another: to the idea of the sequined blanket is added that of the golden tassels striking against the horse; then the blanket is made more glorious by the story of its creation by four slave girls; and finally, this detail is heightened by the fact that they were in the famed city of Dubrovnik and that they worked on this blanket for four years! When we reach the last line of the passage, we cannot but admit that the white horse "could not have been prouder and fiercer!" The method of addition seems simple; yet in the hands of a skillful singer it has a cumulative effect that is telling.
The total impact, however, is due to more than the adding style. The connections between the parts of lines and between lines and between groups of lines is far more intricate and subtle than that. The singer has a [55] strong sense of balance which is shown by the patterns of alliteration and assonance and by the parallelisms. [18] Take the first passage, for example. Note the positions of the italicized verbs. There is internal end rhyme in skita and pita in the first line; and an internal initial rhyme in zveknu and jeknu of the last line. The play of "k" alliteration, caused, no doubt, by the proper name Kajniđa is clear in the first three lines, which have two "k's" in each line. This same "k" alliteration recurs at the end of the passage, namely in the last three lines, where the key word is kapija, "the gate"; line 115 has two, line 116, one, and line 117, four "k's." The central part of the passage, lines four to seven, is dominated by "p," "b," and "z" alliteration around the dominant word bazdrđan. In lines three and four the verb comes at the end of the line; in lines five and six the verb is in second position after the conjunction; in lines seven and eight the verb is again at the end of the line; and in lines nine and ten, as in lines one and two, the verb is in the first half line. Moreover, in lines three and four, dođe and prođe rhyme, and the second half of line four is syntactically parallel to that of line three: Pod Kajniđu dođe and uz čaršiju prođe. Both have the pattern: preposition, noun, verb. This pattern is repeated in line eight, na kapiju dođe. Indeed, line eight is the same as line three in the first half of the line as well.
The singer in this passage is guided for his acoustic patterns not only by the alliteration but by assonance also. The vowel patterns are set by Aliju, Kajniđu, and kapiju. "A-i" and its opposite "i-a" predominate: but "a-u" plays a role also, influenced by the key words as well as by gradu and bazdrđanu. Its opposite, "u-a," is found, but is not so important. Dođe, prođe, and dvore establish an "o-е" pattern, which modulates via "u-o-e" and "o-o-e" to an opposite, "e-u" in the last line. But this pattern is subsidiary to the "a" and "а-i" patterns. The following table, columns one and two, will make the chief alliterations and assonances apparent at a glance.
K-D, G-Đ, sK-T-, Z-, -l--, P-T- u-o-I-A-a-A-I-ju-I-A 2–2, 4–2
K-Z-š-, G-, -, Gr-D-, K-n-Đ- a-a-e-a-u-a-u-AJ-I-u 4, 3–3
K-D, T-T-r-n, P-D, K-n-Đ-, D-Đ- a-a-A-I-o-AJ-I-U-O-E 1–3, 4–2
P-, -T-, G-, -Z, č-rš--, PR-Đ- a-e-o-a-u-A-I-JU-O-E 1–3, 4–2
P-, PR-l-Z-, n-v-m, B-ZD-Đ-n- A-I-A-I-o-o-A-R-A-U 1–3, 2–4
T-, -P-T-, Z-, -l-n-, Dv-r- e-u-I-A-A-A-I-O-O-e 1–3, 4–2
B-ZD-Đ-n, m-, Dv-r, K-Z-- A-R-A-U-o-e-u-a-a-o 4, 2–4
K-D, T-T-r-n, n-, K-P--, D-Đ- a-a-A-I-A-A-I-JU-O-E 1–3, 4–2
P-, Z-D-m-, h-lK-m, n-, Vr-T-m- a-a-r-a-a-o-a-A-I-A 1–3, 2–4
Zv-Kn-, h-lK-, -, -Kn-, K-P-- e-u-a-a-a-je-u-A-I-JA 2–2, 3–3
The singer is also influenced by the rhythmic and word-boundary patterns as he moves from line to line. The analysis of this passage continues, in the third column, with a list of rhythmic sequences.
The alternation of the patterns of the second half of the lines is worthy of particular notice. After the parallelism of 1–3, 4–2 in lines three and four of the passage, there follows a series of 2–4 alternating with 4–2, which is [56] not broken until the 3–3 of the last line. The pattern is too persistent and regular to be accidental. Moreover, it forms a nice counterpoint to the syntactic parallelisms; there is, indeed, a kind of syncopation between the syntactic parallelisms and the word-boundary patterns. Lines three and four are parallel both in respect to word-boundary patterns and syntactic patterns, but whereas the first half of five and six are both syntactically and rhythmically parallel, the second halves are 2–4 and 4–2 respectively, following an alternation beginning in line four with a 4–2 pattern. One has, therefore:
. . . . . . . . . dođe 4–2
. . . . . . . . prođe 4–2
. prilazi . . . . . . 2–4
. upita . . . . . . . 4–2
. . . . . . ukazao 2–4
. . . . . . . . dođe 4–2
. zadrma . . . . . 2–4
The syntactic and rhythmic parallelism of lines three and four modulates into a pattern of syntactic and rhythmic opposition in lines four and five, six and seven, eight and nine, at the same time that syntactic parallelism is kept between five and six, seven and eight. Had Ugljanin been a literate poet who sat down with pen in hand to devise these lines with their inner balances and syncopations, he could not have done better. One can even fancy the overliterate "interpreter of literature," innocent of Salih's ignorance of such matters, extolling the syncopation as the artful intent of the poet to indicate the zigzag search of the messenger for Alija!
A perfectly natural consequence of building passages by syntactic parallelisms and acoustic patterns is that passages so built tend to have a comparative stability, or better, a continuity in time both in the habit of the single singer and, to a lesser degree, in the current of a tradition. Just as formulaic lines with internal rhyme or with a striking chiastic arrangement have a long life, so couplets with clearly marked patterns persist with little if any change. For example:
Bez eđelja nema umiranja,
Od eđelja nema zaviranja.
Without the fated hour there is no dying,
From the fated hour there is no escape.
(II, No. 24:631–632)
A zečki je polje pregazio,
A vučki se maši planinama.
Like a rabbit he crossed the plain,
Like a wolf he ranged along the mountains.
(II, No. 24:41–42)
It seems preferable to keep such couplets in a class by themselves and not to call them formulas, reserving that term for the components of a single verse. Some singers, however, have a tendency to sing in couplets, and in [57] their songs the cadence really comes at the end of the second line; with them it would be perfectly defensible to extend the formula to the couplet.
There are, in addition, larger groups of lines which the singer is accustomed to use often, and through habit they are always found together. The repetition of these groups is sometimes word-for-word exact, sometimes not. Often enough the order of the lines is different. But these clusters of formulas or of lines, which are frequently associated together and are recurrent, also mark one of the characteristic signs of oral style. [19] They are useful to the singer; for they emerge like trained reflexes. The example (in Chart II) from Zogić's favorite song about the rescue of the children of the bey by Bojičić Alija will illustrate. The first passage is from a version sung and recorded in 1934, and the second is the parallel passage in the dictated version of the same year.

Chart II
Pa proklinje careva fermana:
–  –  –  – – –   –   – –  –  –  –
I proklinje careve fermane:
– –  – – – –   –  –  –  –  – –
"Bor ubijo careva fermana!
Ferman care od Stambola sprema,
"Ferman care od Stambola sprema,
Oprema ga Alibegu mome,
Pa mi bega traži u Stambolu.
Pa mi bega traži u Stambolu. 
Hasi mu se narod učinio, 
Hasi mu se narod učinijo;
Niť mu porez ni vergiju daje,
––––––––––   –  – –  –  –  –
Niť mu porez daje ni vergiju,
––––––––––   – – –  – – – –
Niť mu asker ni mazapa daje,
–––––––––     –   –  –  – –   –
Niť mu asker daje ni mazapa,
––––––––––   – – – – – –  –
Da bi l' malo narod umirijo. 
–––––  – – – –––––––––––
Da bi l' kako narod umirijo.
––––– – – – –––––––––––
Od fermana nema varakanja.
Od fermana nema varakanja.
Pa kad begu ferman degdisao,
– – – – –––– –––––––––––––
Kad mu begu ferman degdisao,
– – – –  –––– ––––––––––––––
Beg se spremi na bijelu kulu,
Beg se spremi na bijelu kulu,
A pripasa silah i oružje, 
Na njegova široka dorina,
                   ––––– – – – –
I opremi široka dorata.
              –––––  –  –  – 
Navali mu pusat i saltanet, 
Jalah reče, posede hajvana,
Isprati' ga do dimir kapije. 
Ode beže preko polja ravna,
Pa ga nagna preko polja ravna.
A zečki je polje pregazio,
– – ––––––––––––––––
Oh zečki je polje pregazijo,
– – ––––––––––––––––––
A vučki se maši planinama,
A vučki se maši planinama, 
Dok preskoči dvije tri planine,"
Preturijo dvije tri planine,"
(II, No. 24: 26-43)
(II, No. 25: 18-35)[58]
Chart II Translation  
Then she cursed the imperial firman:   And she cursed the imperial firmans:
"God destroy the imperial firman!    
The sultan sent the firman from Stambol,   "The sultan sent the firman from Stambol,
He sent it to my Alibey,    
And he sought my bey in Stambol.   And he sought my bey in Stambol.
The people were in revolt against him;   The people were in revolt against him;
Neither tax nor tribute do they give him,   Neither tax do they give him nor tribute,
Neither soldier nor sailor do they give him,   Neither soldier do they give him nor sailor,
That he might quiet the people a little.   That he might somehow quiet the people.
There is no avoiding a firman,   There is no avoiding a firman.
And when the firman reached the bey,   When the firman reached the bey,
The bey prepared himself in his white tower;   The bey prepared himself in his white tower,
    And girded on his belt and arms,
And mounted his broad-backed chestnut horse,   And prepared his broad-backed chestnut stallion.
    He put on him his arms and trappings,
    With a cry to Allah he mounted his beast,
I accompanied him to the iron gate.    
The bey departed across the level plain.   And he drove him across the level plain.
Like a rabbit he crossed the plain,   Like a rabbit he crossed the plain,
Like a wolf he ranged along the mountains,   Like a wolf he ranged along the mountains,
Until he had leaped over two or three mountains."   He passed over two or three mountains."
How persistent such a "run" may be can be seen from the same passage sung for the records in 1951, seventeen years after the two excerpts in Chart II.

Pa proklinje careva fermana:
"Bog ubijo careva fermana,
Što ni care ferman opremijo!"
Pa mi traži Alibega mlada.
Traži bega care u Stambolu.
Then she cursed the imperial firman:
"God destroy the imperial firman,
Which the sultan sent to us!"
He sought my young Alibey.
The sultan in Stambol sought the bey.
From Hasi mu se through the line Na njegova široka dorina in Chart II the 1951 text is word-for-word the same as the 1934 sung text in the first column above. Then it continues:
Krenu beže preko polja ravna.
The bey set out across the level plain.
The next two lines are the same in all three texts. And the last line in 1951 is:
Dok preturi dvije tri planine.
Until he passed over two or three mountains.
(Lord 200:21–37) [59]
Another excellent example of a cluster of formulas, or a "run," is afforded by the following six lines from Ugljanin's colorful description of the hero Tale and his horse:

Na kulaša sedla ni samara,
Sem na kula drvenica gola.
S jedne strane topuz od čeljika;
On ga tiče, on mu se spotiče
A na Tala od jarca ćakšire,
Dlake spolja; sva koljena gola.
On the mouse-gray horse was neither saddle nor pack-carrier,
But only a bare wooden frame on the mouse-gray.
From one side (hung) a steel mace;
It struck the horse and caused him to stumble.
Tale was wearing goatskin trousers,
The hairy side out; his whole knee was bare.
(II, No. 1:627–632)
This description is word-for-word the same in the song, "Ženidba Ćejvanović Meha" (II, No. 12:485–490).
If one takes two texts of the same song, as we have above, and underlines the verses that are common to both, one discovers a characteristic picture. There will be a series of lines unmarked followed by a series of underlined verses with occasional small breaks perhaps, followed in turn by another "clear" spot. If a singer sings a song many times the underlinings, as in Zogić's case, will be many, but this will not be the case with a song infrequently sung. One obtains thus a photograph of the individual singer's reliance on habitual association of lines and of the degree to which habit has tended to stabilize, without fixing or petrifying, passages of varying length. One might well contrast, for example, the comparative stability of Zogić's passage from his favorite song with the fluctuation and sparseness of underlining in the following passage (Chart III) from Halil Bajgorić's song of Alijaga Stočević, sung for the records in 1935 and again for the records in 1950 at Stolac, Hercegovina.
Chart III
Razbolje se Stočević Alija
Razbole se Stočević Alija
Usred Stoca grada kamenoga.
––––––––––  – – –  – –  – –  –
Usred Stoca grada bijeloga.
–––––––––– – –  –  –  –   –
Pa boluje za punu godinu.
Vazda misle age Stolačani,
Da j’ Alija svijet mijenijo.
Pa boluje za dvije godine,
– – –  – –  –  – – –  – –  – 
Te boluje za dvij' godine dana.
–  –  – –  –  – –  –  – – –   –   –
Pa boluje i treću godinu.
Vazda misle age Stočevljani,
Da j'Alija i umiro davno. [60]
Bože mili, na svemu ti fala!
No Alija nikoga ne ima,
Samo imade sestru svoju Fatu.
Niko ne zna u bijelu gradu,
Da li Alija boluje al' ne boluje.
Neko misli da ga tuka nema,
Da je Alija izgubijo glavu.
To se čudo na daleko čulo
Za to začu sivi Arapine
––––––––  – – – – –  –
Za to začu crni Arapine
––––––––  – – –  – –  –
Preko sinja mora debeloga,
– –  –  –  –  – –  –  –  – –  –
Preko mora sinja debeloga,
–  – –  –  –  –  – –  –  – –  –
Pa on jaše svoju bedeviju,
A crna je kako gavran crni,
Da je umro Stočević Alija,
I zakuči sebe i kobilu,
I ovako junak progovara:
"Hajte, sić' ću Stocu kamenome,
Tome Stocu na Hercegovinu.
Ima tamo Ijepih devojaka,
Kako čujem u bijelu Stocu,
A danas nema nikakova junaka,
Da će meni stanut' na mejdanu.
Ja ću sići u polje Vidovo.
U njemu ću čador razapeti,
I nametnut' namet na vilajet,
Svaku nojcu po jalovu ovcu,
I po kab'o preljetne rakije,
Sedam oka crvenoga vina,
Rujna vina od sedam godina,
I ljubiću svaku nojcu po jednu đevojku.
Kad se svane i ograne sunce,
Ja je ocu i materi spremam,
Ali drugu do večera tražim."
Što govori Arapine crni,
On je tako isto učinijo,
Te zapuči sebe i kobilu,
І on vodi četiri sejiza,
Što mu nose skute i rukave.
Silan Arap pa se posilijo.
'Оćе Arap da mejdana traži,
'Oće Arap da đevojke ljubi,
'Oće Arap pa da vina pije,
'Оćе Arap i rakiju da pije,
'Оćе Arap da je junak na mejdanu.
Pa eto ga Stocu kamenome.
– – ––––– –  – – –  – – – – –
Eto ga kamenome Stocu.
–––––  –  –  – –  – –  –  – 
Dođe Arap u Vidovo polje.
(Lord 83: 1-45) [61]
(Parry 6697: 1-15)
Chart III Translation
Stočević Alija fell ill
Stočević Alija fell ill
Midst Stolac, stony city.
Midst Stolac, white city.
He was ill for a full year.
Ever the aghas of Stolac think
That Alija has changed worlds.
He was ill for two years,
And he was ill for two years of days.
And he was ill also a third year.
Ever the aghas of Stolac think
That Alija has long since died
Dear God, thanks to Thee for all things!
But Alija has no one,
Except his sister Fata.
No one knows in the white city,
Whether Alija is ill or not.
Some think that he is not there,
That he has lost his life.
That marvel was heard afar.
A dark Arab heard of this
A black Arab heard of this
Across the dark blue sea, the deep,
Across the sea, dark blue, deep,
And he mounted his bedouin mare,
Black as a black raven she was.
That Stočević Alija had died,
And he secured himself and his mare,
And thus the hero spoke:
"Come, I shall go to stony Stolac,
To that Stolac in Hercegovina.
There are beautiful maidens there,
As I hear in white Stolac,
And today there is no hero,
To meet me in single combat.
I shall go to Vidovo plain.
On it I shall pitch my tent,
And impose tribute on the province,
Every night a gelding sheep,
And a bucket of fine brandy,
Seven pounds of ruddy wine,
Red wine seven years old,
And every night I shall love a maid.
When it dawns and the sun rises,
I shall send her to her father and mother,
But I shall seek another by evening."
What the black Arab said,
The same he did,
And he secured himself and his mare,
And he took with him four squires,
Who carried his sleeves and train.
The mighty Arab strengthened his might. [62]
The Arab will seek single combat,
The Arab will seek to love maidens,
The Arab will seek to drink wine,
The Arab will seek to drink brandy too,
The Arab will seek to be a hero in combat.
And lo, here he is in stony Stolac.
Lo, here he is in stony Stolac.
The Arab came to Vidovo plain.
It is clear that Bajgorić is actually re-creating the song with little reliance on habitually and frequently sung passages. The importance of these observations for the comparatist lies in their possible application to divergent manuscripts of the same song which we may be fortunate enough to have from medieval or ancient times. The answer to the question of how the divergences arose may possibly be found in some cases in the fact that one is dealing with two oral texts rather than with a text modified by a scribe or by a second poet working from an already written text.
All singers use traditional material in a traditional way, but no two singers use exactly the same material in exactly the same way. The tradition is not all of one mold. We can differentiate individual styles in the epic technique of oral verse-making. The significance of this for the Homeric songs is clear. It should be apparent that if we make proper use of our knowledge gained from testing the Yugoslav sample, we should be able at some time to answer with some degree of certainty the question of whether the Iliad and Odyssey are by the same singer.
We have three texts from Zogić (all of the same song), two from 1934 and one from 1951, totalling 3495 lines, and from Makić four texts (all of different songs) from 1934, totalling 2873 lines. One could be sure that these two groups are by different singers, in spite of many similarities, by noting that the formula series consisting of conjunction, plus evo or eto, plus a personal pronoun in the genitive, for example, pa eto ga, is used only twice by Zogić but twenty-two times by Makić.
Zogić Makić
Kad eto je hanka na kapiju Kad eto ga jedna sirotinja
(II, No. 24:370) (II, No. 26:585)
Pa eto je kafezli odaje Pa eve ga šarena kafaza
(Lord 200:157) (26:31)
Closely related:  
Eto ti je kafezlji odaji Pa eve ga na odaju dođe
(II, No. 24:649) (26:36)
Eto ti ga kafezlji odaje Pa eve ga na planinu dođe
(Lord 200:734) (26:582)
  A eve ga muhur sahibija
  E eve ga đadi dolazijo
  (26:584) [63]
  Ej eve ga do Bagdata priđe
  Pa eve ga na kapiju siđe
  Eh, eve ga kod vezira dođe
  Helj eve ga kralje Rakocija
  Pa eve je kod devljeta stigla
  Pa evo ga među đamovi [ma]
  Pa eve ga boja najgornjega
  (28:772, 29:325)
  I eve je nа noge skočila
  Pa eve je u avliju siđe
  (29:94, 128)
  Pa eve ga do kapije dođe
  Pa eve ga kod kočije dođe
  I eve ga pijanoj mehani
  I eto ga poljem zelenijem
  I eto je na kapiju prođe
  Closely related:
  On, eve ga pod ravnu Semen [tu]
The use of this formula in the material from the two singers is a clear and statistically measurable mark of a difference between the two men.
Another distinguishing mark between the same two singers is the form of the couplet expressing the idea "he who was nearby looked at the ground; he who was farther off pretended not to hear." Zogić says:
Ko je bliže, ka zemlji gledaše,
Ko je dalje, čini se ne čuje.
He who was nearby looked at the ground,
He who was farther off pretended not to hear.
(II, No, 24:463–464, 588–589; 25:297–299; Lord 200:440–441)
In No. 25 he inserts the line Kako raste trava na zavojke, ‘To see how the grass was growing in spirals’ between the two lines of the couplet, Makić's form of the couplet is:
Ko bi dalje, čini se ne čuje,
Ko bi bliže, zemlji pogleduje.
He who was farther off pretended not to hear,
He who was nearer looked at the ground.
(II, No. 29:260–261) [64]
The two lines are reversed, the aorist is used instead of the present, ka is omitted in the second line, and an aspect of the verb gledati is used which allows rhyme between pogleduje and čuje. Although we have only one instance of this couplet from Makić, it is of the kind that becomes fixed in a singer's usage, and one can be certain that he would not change it. These are but samples to illustrate one kind of distinguishing characteristic in individual formula styles.
The poetic grammar of oral epic is and must be based on the formula. It is a grammar of parataxis and of frequently used and useful phrases. Usefulness in composition carries no implication of opprobrium. Quite the contrary. Without this usefulness the style, and, more important, the whole practice would collapse or would never have been born. The singer's mode of composition is dictated by the demands of performance at high speed, and he depends upon inculcated habit and association of sounds, words, phrases, and lines. He does not shrink from the habitual; nor does he either require the fixed for memorization or seek the unusual for its own sake. His oft-used phrases and lines lose something in sharpness, yet many of them must resound with overtones from the dim past whence they came. Were we to train our ears to catch these echoes, we might cease to apply the clichés of another criticism to oral poetry, and thereby become aware of its own riches.
For while I have stressed usefulness and necessity in composition as essential considerations in studying formulas and the whole formulaic style, it may well be that these characteristics belong to the preservation and development of that style and of the formula rather than to their origins. It is certainly possible that a formula that entered the poetry because its acoustic patterns emphasized by repetition a potent word or idea was kept after the peculiar potency which it symbolized and which one might say it even was intended to make effective was lost—kept because the fragrance of its past importance still clung vaguely to it and kept also because it was now useful in composition. It is then that the repeated phrases, hitherto a driving force in the direction of accomplishment of those blessings to be conferred by the story in song, began to lose their precision through frequent use. Meaning in them became vestigial, connotative rather than denotative. From the point of view of usefulness in composition, the formula means its essential idea; that is to say, a noun-epithet formula has the essential idea of its noun. The "drunken tavern" means "tavern." But this is only from the point of view of the singer composing, of the craftsman in lines.
And I am sure that the essential idea of the formula is what is in the mind of the singer, almost as a reflex action in rapid composition, as he makes his song. Hence it could, I believe, be truly stated that the formula not only is stripped to its essential idea in the mind of the composing singer, [65] but also is denied some of the possibilities of aesthetic reference in context. I am thinking especially of what might be called the artistically weighted epithet: what later literary critics find "ironic" or "pathetic." Indeed one might even term this kind of criticism "the pathetic fallacy" in that it attributes to an innocent epithet a pathos felt only by the critic, but not acknowledged or perhaps even dreamed of by either the poet or his audience. Being part of the tradition, they understand its characteristics and necessities. Nevertheless, the tradition, what we might term the intuitions of singers as a group and as individuals who are preserving the inherited stories from the past—the tradition cannot be said to ignore the epithet, to consider it as mere decoration or even to consider it as mere metrical convenience. The tradition feels a sense of meaning in the epithet, and thus a special meaning is imparted to the noun and to the formula. Of course every adjective and epithet can be said to do this, but I am not thinking in this case about the surface denotative meaning of the adjective, but rather of the traditional meaning, and I would even prefer to call it the traditionally intuitive meaning. For it is certain that the singer means on the surface "drunken tavern" to mean a tavern in which men drink and become drunk, but it could well be argued that the epithet is preserved in the tradition because it was used in stories where the tavern was the symbol for an entrance into the other world and the drinking involved is the drinking of the cup of forgetfulness, of the waters of Lethe, and that the drunkenness involved is not that of the ordinary carousel, but is itself a symbol for consciousness in another world, perhaps even death. This meaning comes to it from the special, peculiar purpose of oral epic song at its origin, which was magical and ritual before it became heroic.
This sense of "drunken" becomes clear when one follows the various stories of Marko Kraljević and his brother Andrija, for example, in which Andrija is lured by a tavern maid into her tavern, where he is made drunk by a band of Turks and then killed. Some of the variants have him asking for water rather than wine because he has been contending with his brother to determine which could stand thirst the longer; and Andrija breaks a taboo imposed by his brother in that he dismounts from his horse although instructed not to do so, and enters the tavern. Other variants have Marko reporting his brother's death to their mother according to the elaborate instructions given by the dying Andrija, and saying that Andrija has fallen in love with a girl in a far-оff country who has given him of the waters of forgetfulness so he will not return. This last is from our earliest version in the sixteenth century; other examples can be found in the songs in Volume I of the Parry Collection. [20]
Webster may well be correct in regard to his tracing of the meaning of formulas, such as "ox-eyed Hera" and "bright-eyed Athene" to cult songs, [21] although it is not entirely clear what he means by them. These epithets do seem to refer to the epiphanies of the goddesses and thus to strengthen the [66] power of the invocation of the goddess by the repetition of the goddess in several different ways, that is to say, not only by invoking her by her name but also by her epiphany. I think we are safe in assuming that the repetition was there in two forms originally, not for the sake of meter, nor for the sake of convenience in building a line, but rather for the sake of redoubled prayer in its hope of surer fulfillment. The metrical convenience, or even better, the metrical necessity, is probably a late phenomenon, indispensable for the growth of epic from what must have been comparatively simple narrative incantations to more complex tales intended more and more for entertainment. This was a change concomitant with the gradual shift toward the heroic and eventually the historic. It is quite likely that the later stages could not have developed until the formula became a compositional device; yet because of its past it never could become merely a compositional device. Its symbols, its sounds, its patterns were born for magic productivity, not for aesthetic satisfaction. If later they provided such satisfaction, it was only to generations which had forgotten their real meaning. The poet was sorcerer and seer before he became "artist." His structures were not abstract art, or art for its own sake. The roots of oral traditional narrative are not artistic but religious in the broadest sense. [67]


[ back ] 1. M. Parry, “Studies in the Epic Technique of Oral Verse-Making. I: Homer and Homeric Style,” HSCP, 41:80 (1930).
[ back ] 2. C. M. Bowra, Heroic Poetry (London, 1952), p. 222.
[ back ] 3. Bowra still clings to this idea; ibid., p. 231. See also Tradition and Design in the Iliad (Oxford, 1930), pp. 87 ff.
[ back ] 4. A study of Serbocroatian metrics should begin with Luka Zima, “Nacrt naše metrike narodne obzirom na stihove drugih naroda, a osobito Slovena,” Rad jugoslavenske akademije znanosti i umjetnosti, 48:170–221 (1879) and 49:1–64 (1879), and the monumental work of Tomislav Maretić, “Metrika narodnih naših pjesama,” Rad, 168:1–112 (1907) and 170:76–200 (1907), and Metrika muslimanske narodne epike (Zagreb, 1936). Svetozar Mati's “Principi umetničke versifikacije srpske,” Godišnjica Nikole Čupića, 39:119–162 (1930) and 40:51–72 (1931) sums up all previous work; it is reviewed, together with other works on South Slavic poetry, by Roman Jakobson, Slavische Rundschau 4:275–279 (1932). André Vaillant, “Les chants épiques des Slaves du sud,” Extraits des Cours et Conférences, 30 janvier, 15 février, et 15 mars 1932 (Paris, 1932) devotes considerable space to metrics. It too is interestingly reviewed by R. Jakobson in Byzantinoslavica 4:194–202 (1932). See also R. Jakobson, “The Kernel of Comparative Slavic Literature,” Harvard Slavic Studies, 1:26 ff, (1953) and “Studies in Comparative Slavic Metrics,” Oxford Slavonic Papers, 3:21–66 (1952), which is reviewed at considerable length by Kiril Taranovski in Prilozi za književnost, jezik, istoriju i folklor, 12:350–360 (1954). Taranovski's own article, "O jednosložnim rečima u srpskom stihu," Naš jezik, 2:26–41 (1950) is of interest also, as is the book of Franz Saran, Zur Metrik des epischen Verses der Serben (Leipzig, 1934), which was published posthumously, Radovan Košutić's O tonskoj metrici u novoj srpskoj poeziji (Belgrade, 1941) was also reviewed at great length by K. Taranovski, "O tonskoj metrici prof. Košutića," Južnoslovenski filolog, 18:173–196 (1949–50). Finally Taranovski's book on Russian metrics, Ruski dvodelni ritmovi (Belgrade, 1953) is not lacking in interest here as the latest work of importance in Slavic metrics.
[ back ] 5. M. Parry, L'Epithète traditionnelle dans Homère, pp. 11–15, et passim.
[ back ] 6. See R. Jakobson, "Über den Versbau der serbokroatischen Volksepen," Archives néerlandaises de phonétique expérimentale, 8/9:135–144 (1933).
[ back ] 7. See R. Jakobson, "Studies in Comparative Slavic Metrics," Oxford Slavonic Papers, 3:27 (1952), where this phenomenon is explained by the "tendency to avoid closed syllables at the end of the line." See also Matija Murko, "Nouvelles observations sur l'état actuel de la poésie épique en Yougoslavie," Revue des Etudes Slaves, 13:31 ff. (1933).
[ back ] 8. R. Jakobson in his "Studies in Comparative Slavic Metrics," Oxford Slavonic Papers, 3:24 ff. (1952) has listed as the first metrical constant of the Serbocroatian epic decasyllable, "Isosyllabism: each line contains ten syllables." This is certainly an almost invariable norm. A check of the singing of Stanko Pižurica, the best of the Montenegrin Christian singers with whom Parry worked in Kolašin and from whom we have an abundance of material tends to uphold these findings that among such singers in Montenegro, the classical terrain (as it is often thought of) for South Slavic epic, the ten-syllable line is practically invariable in regard to the number of its syllables. Apart from mistakes made in rapid composition, the perfectly normal slips of the tongue, there are a number of regular abnormalities, so regular, indeed, that they have affected formula construction. A detailed study of this is out of place in this book, but the following may be noted. In Parry and Lord, II, the textual notes indicate all metrically abnormal lines. A sampling of the first three songs of Ugljanin sung for the records shows 42 eleven-syllable lines: 12 of these clearly begin with a dactyl—e.g. Vazda je Mujo četom četovao (II, No. 1:793), Ala ne boj se, pile sokolovo (No. 1:863), Pa ću je, care, tebe pokloniti (No. 2:826), Ibrahim paša četvrti je bio (No. 2:1152), Lasno je sići ka Zadaru gradu (No. 4:1748)—and 7 more are ambivalent, in that they might be considered to begin with an extra-metrical syllable; 13 have a dactyl in the second foot—e.g. Proj se vraga i bijela Bagdata (No. 1:46, No. 2:46), Viknu redom sve paše i vezire (No. 2:810), Ja ću bana i do tri đenerala (No. 4:1470); 8 have a dactyl in the third foot, that is after the break—e.g. Da se Bosna pominje do vijeka (No. 1:224), A sultanu koljko ga žao bilo (No. 2:1170), Stara majka sto i kusur godina (No. 4:307), Kako sam se grdna obradovala (No. 4:732), Ṧta je bilo banu i kapetanim' (No. 4:1556); and one strange case has a dactyl in the fourth foot. It is to be noted that a number of these instances bring together two vowels without elision.

In Bihać in northern Bosnia, Moslem singers accompany the songs on the plucked tambura. Ćamil Kulenović in Parry Text 1950 sang 55 eleven-syllable lines in the first 500 lines, most of which begin with a dactyl. Here are examples from this and another text of Ćamil’s: Ban ga otočki i ban šibenički (Parry 1951:11), Kako bi tude sva četiri bana (1951:18), Tako mi boga i zakona moga (1951:27), Ja sam ga jednom bijo ufatijo (1951:28), Lipu ga Zlatu timar tefterdara (Parry 1950:243), Jedno ga grlo četiri đerdana (1950:249), Dorat mu svezan kod gradske kapije (1950:152), Kolki sam godić u svilenu pasu, Tolki je Turčin u bijelu vratu (1950:148–149). Maretić pointed out this phenomenon in 1936 as a characteristic of Moslem epic, it being especially common in Volumes III and IV of the Matica Hrvatska Collection. See T. Maretić, Metrika muslimanske narodne epike (Zagreb, 1936), pp. 218–220. It may be that in northern Bosnia this initial dactyl is somehow related to the instrumental accompaniment of the tambura used by Moslem singers. In the Novi Pazar district this seems not to be true. It may be that the Moslems there, many of them of Albanian descent, are less respectful of the deseterac, and it is perhaps worth noting that the tambura (Albanian çifteli) is used in some parts of the north of Albania for epic. From my experience there in 1937, however, the tambura is used chiefly among the Christians for the shorter local and historical songs, purely Albanian and octosyllabic mostly. For remarks on the use of the decasyllable by Albanians in Albania in Albanian see Stavro Skendi, "The South Slavic Decasyllable in Albanian Oral Epic Poetry," Word, 9:339–348 (1953). This whole subject is worthy of more careful study.
[ back ] 9. Unfortunately, most printed collections are not very useful for this kind of study, because the texts have been subjected to strict editing out of abnormalities and little music exactly annotated is available. On the whole all of the admirable work that has been done has been on the basis of comparatively little recorded material; one would probably be shocked to know just how few recorded lines of epic verse have been used. No large body of songs on phonograph records was available for study before the Parry Collection was made.
[ back ] * A. "The Captivity of Đulić Ibrahim," sung by Salih Ugljanin (1, pp. 437, 439); B. "Osman Delibegović and Pavičević Luka," sung by Avdo Međedović (Parry 12389); C. "Junaštvo Đerđelez Alije," sung by Avdo Međedović (Parry 12379)
[ back ] 10. Učinijo is actually pronounced and printed in Parry and Lord, Serbocroatian Heroic Songs, in the passage quoted, as ućinijo, with ć instead of the normal č. It is thus that the singer sang the word, reflecting his own dialect. Here and in other instances henceforward I have normalized the spelling in such cases. The exact pronunciation can be found in the published volumes.
[ back ] 11. I see no reason to place the noun-epithet formulas in a separate class from all the other formulas, as Bowra does (Heroic Poetry, p. 222). They are not the only formulas that do not advance the narrative.
[ back ] 12. For other examples see Matija Murko, "Nouvelles observations sur l'état actuel de la poésie épique en Yougoslavie," Revue des Etudes Slaves, 13:43 (1933), and La poésie populaire épique en Yougoslavie au début du XXe siècle (Paris, 1929), p. 24.
[ back ] 13. M. Parry, "Studies in the Epic Technique of Oral Verse-Making. I: Homer and Homeric Style," HSCP, 41:118–121.
[ back ] 14. The supporting evidence is collected in my unpublished thesis in the Archives of Harvard University.
[ back ] 15. A check of the other lines in the chart would yield approximately the same results.
[ back ] 16. Bowra, Heroic Poetry, pp. 234 ff. Parry, L'Epithète, pp. 218 ff.
[ back ] 17. See A. B. Lord, "Homer and Huso III: Enjambement in Greek and Southslavic Heroic Song," TAPhA, 79:113–124 (1948), from which these figures are taken. See also M. Parry, "The Distinctive Character of Enjambement in Homeric Verse," TAPhA, 60:200–220 (1929). I follow the definition of enjambement given by Parry, ibid., pp. 203–204: "Broadly there are three ways in which the sense at the end of one verse can stand to that at the beginning of another. First, the verse end can fall at the end of a sentence and the new verse begin a new sentence. In this case there is no enjambement. Second, the verse can end with a word group in such a way that the sentence, at the verse end, already gives a complete thought, although it goes on in the next verse, adding free ideas by new word groups. To this type of enjambement we may apply Denis' term unperiodic. Third, the verse end can fall at the end of a word group where there is not yet a whole thought, or it can fall in the middle of a word group; in both of these cases enjambement is necessary … To know where there is no enjambement we must gauge the sentence. The varying punctuation of our texts, usually troublesome, will not do. I define the sentence as any independent clause or group of clauses introduced by a coordinate conjunction or by asyndeton; and by way of showing that this definition is fitting I would point out that the rhetoricians paid little heed to the sentence as we understand it: for them the unit of style was the clause, and the only group of clauses of which Aristotle speaks is the period." The statement of Jakobson, following perhaps a more widely known definition of enjambement, that there is no enjambement in Serbocroatian epic, is correct, although rare exceptions can be found.
[ back ] 18. For a fuller study of this passage, see A. B. Lord, "The Role of Sound Patterns in Serbocroatian Epic," in For Roman Jakobson (The Hague, 1956), pp. 301–305.
[ back ] 19. Bowra (Heroic Poetry, pp. 222ff.) puts all such repetitions together under the name of formula, distinguishing in formula only the noun-adjective combinations. This, it seems to me, is too cavalier a treatment. There are significant differences between the several groups.
[ back ] 20. V. Bogišić, Narodne pjesme iz starijih, najviše primorskih zapisa (Belgrade, 1878), No. 6, p. 18. In the songs of Salih Ugljanin, I have noted only one reference to a tavern maid (Ruža); this is in Parry 653 (Synopsis VI in I, p. 213) "Halil Hrnjičić Rescues the Daughter of the Vizier of Travnik." A more interesting instance is in Zogi's "Bojičić Alija Rescues Alibey's Children" (I, No. 24, pp. 255ff.). Another case in point is in Alija Fjuljanin's "Halil Hrnjičić and Miloš the Highwayman" (I, No. 31, pp, 309ff.).
[ back ] 21. T. B. L. Webster, From Mycenae to Homer (London, 1958), p. 94.