An eEdition of The Wedding of Mustajbey’s Son Bećirbey

Nikola Vujnović’s Resinging

The singer in question was Nikola Vujnović, a stone mason by trade, from Stolac in Hercegovina, who was drinking wine and putting the gusle in order when Kutuzov and myself arrived. He sang quite willingly for us the first evening. The first song he sang and so the first of the songs I heard was the beginning of the poem, Prva varnica svjetskog požara (The First Spark of the World’s Conflagration), of Aleksa Guzina. I learned later that he had bought the pjesmarica [songbook] of this poem — which is one of the poems issued by the publishing house of “Progres” at Nikšić which provides the “new” poems for singers — while he was serving his term in the army in 1927, at Avtovac.

Milman Parry, from his field notes (Ćor Huso, unpubl.),¹
on his first meeting with Nikola Vujnović, 7/19/33,
at the kafana of Pero Arbulić in Dubrovnik

Nikola Vujnović (NV) was himself a South Slavic guslar, a singer of tales. He was a member of the same oral epic tradition that Milman Parry and Albert Lord journeyed to the former Yugoslavia to study and collect, and it was not least for that very reason that he was able to participate so productively as their interpreter and co-fieldworker. Picture the ongoing scenario: the American scholars proceed from place to place, traveling to six major regions in pursuit of epske pjesme, implementing their large designs and posing their specific questions through NV. That their chosen intermediary could communicate with the singers as a native speaker was of course a prerequisite; that he could address these men as one of their own informal guild was determinative. With his unique combination of modest literacy skills and an insider’s knowledge of the epic tradition, NV had a foot planted in each world. On the one hand he could push his employers’ agenda in arranging song-performances and pursuing insights on the art of oral composition; on the other hand he could speak to guslari on their own terms. Indeed, it has not yet been fully appreciated just how crucial his double perspective was to the success of the overall research project and the manifold comparative extensions that were to follow in the decades to come.²

Just as NV was able to converse with other guslari as one of their own number, so his later transcriptions of the epic performances recorded in the former Yugoslavia reveal the shaping force of his singer’s mentality, the influence of his singer’s ear. When I began my own transcription of the Ženidba Bećirbega Mustajbegova (ŽBM) by Halil Bajgorić (HB), I naturally expected NV’s ear to be far more accurate than mine. While there would certainly be some inadvertent miscues, how could his interpretation not prove far more dependable, far truer to the acoustic reality of HB’s words than anything I could muster? After all, NV was not merely a native speaker; he was a guslar with an insider’s knowledge of the epic way of speaking. His text must stand only one small step removed from a performer’s transcription of his own performance.

But that, as I was soon to learn, was exactly the problem. Because NV relied on his ready fluency in epic phraseology, he sometimes remade rather than recorded what he heard. As the title of this section indicates, in many respects he resang the epic. Instead of reproducing syllable after syllable, line after line, with slavish verbatim loyalty to whatever was encoded on the aluminum records, he “heard” and transmitted the performance through the filter of his own competence as a traditional oral epic poet. To be more precise, he occasionally transferred the ŽBM from HB’s epic idiolect to his own. And although they shared roughly the same singing dialect, based as they both were in the Stolac region, their idiolects were not identical. As we shall see, NV’s transferral or remaking led on numerous occasions to what I call lapsus auris, a “lapse of the ear,” which produced corrections, mistakes, and viable substitutions. In the table that follows, I distinguish categorically between this phenomenon and both lapsus linguae (HB’s misspeakings) and lapsus calami (NV’s errors in writing).

The set of discrepancies that lapsus auris generated — with NV’s transcription diverging systematically from HB’s performance at various points — raises a question that a textual mindset will find strange and uncomfortable. Namely, are such incongruencies really mistakes? Are they really “wrong”? From one point of view, there is little question that NV sometimes failed to mirror what HB actually sang; in that sense his written codification is incorrect. But in another sense the transcriber is doing no more than any oral epic singer does when reperforming a song heard from another singer (or even over his own subsequent performances of the same song): he is refashioning the tale by manipulating the plastic compositional language that is its natural medium. He is putting his individual stamp on a shared idiom. So which is “right” — HB’s original or NV’s remaking? Most of us would opt for the original recording, since the present volume aims at providing an edition and translation of HB’s singing of the ŽBM. That performance must remain our first priority.

But it need not be our sole priority. In order to illustrate how a song travels from one singer to another — even when the subsequent “performance” consists of the writing of a text — I have constructed below a table that charts the journey and documents the changes en route by noting and describing every instance in which HB’s articulation and NV’s resinging do not match identically. The first column lists what HB actually sang on June 13, 1935, as nearly as I can determine from my own audio review of the digital and analogue tapes. The second column reports the corresponding but variant item from NV’s handwritten transcription, as gleaned from the small notebook in which he penned his text in a clear, legible hand. The third column, entitled “Explanation for Variance,” briefly identifies the root cause(s) of HB’s nonstandard articulation, NV’s divergent transcription, or both. Very importantly, detailed analyses for all of the individual tabular entries are provided (as necessary) in the Commentary, to which I urge readers seeking further explanation to turn. Such analyses are signaled by the instruction “See note” in the third column.

NV’s textual resinging of HB’s actual performance consists of several main types of alterations. The least complicated case is what I have termed lapsus calami, the simple miswriting of a word or phrase. This category includes outright misspellings, as when a letter or word is unwittingly omitted from NV’s text (kandžkog for kanidžkog at line 258, e. g.), as well as conventional orthographic accommodations such as NV’s consistent substitution of mejdan for HB’s megdan.³ In kind and scope, such disparities are not so different from what we might expect in any scribal environment.

Another familiar type of discrepancy is NV’s irregular word-division. That is, in some instances he breaks whole, unitary words into increments as he transcribes them, while at other points he constructs composites from smaller units usually taken as wholes in themselves. Even attempting to document this practice of course begs the question of what a “word” really is, a more complicated matter in oral poetry than in poetry confined to the page, as well as the vexed problem of how to define a “standard” language or idiom.⁴ For the sake of brevity and convenience, I have resorted to conventional printed lexicons as the basis for comparative measurement (as well as for establishing HB’s performance-text), but remain quite aware of the complications that this policy overlooks.⁵ At any rate, it becomes clear that two procedures underlie NV’s variant word-division: separation by analysis into morphemic constituents and amalgamation by attachment of unstressed, linguistically dependent particles to a proximate word. For instance, he chooses to break HB’s Bećirbega to Bećir bega at 254, isolating two potentially free-standing morphs that can and do occur alone elsewhere in the poem or its tradition. Complementarily, NV occasionally merges a proclitic with the following word (HB’s ne preteže > nepreteže at line 34, e. g.) or an enclitic with the preceding word (HB’s Je li > Jeli at line 189, e. g.). At least as a general phenomenon, irregular worddivision is again a phenomenon not unknown within other scribal practices.⁶

Transposition from HB’s performance to NV’s text also involves the singers-cribe’s frequent modification in the palatalization of words. Of course, we may view this kind of difference as a general dialectal disparity, citing broad distinctions such as the ekavski versus jekavski versus ijekavski dialects (e. g., mleko, mljeko, and mlijeko, respectively), or we may choose to see personal, idiolectal peculiarities as the operative phenomenon. Given the tendency of the oral epic register to harbor different dialect forms alongside one another, as well as uncertainty about NV’s orthographical standards, however, it is difficult to pinpoint a particular source for each divergence. What we can observe is that the process can go either way: NV both palatalizes (Djerdjelez < HB’s Djerdelez in line 1, e. g.) and depalatalizes (Petericu < HB’s Pjetericu in line 39, e. g.) in the course of his transcription. Moreover, the singer HB is himself inconsistent in certain cases, as with the doublet devoj-/djevoj-, and at times NV shows a greater consistency in his interpretation than HB does in his original articulation.⁷

Another major category is that of performatives, excrescent consonants inserted to bridge hiatus and thus allow the guslar to sing through a sequence of contiguous vowels without inserting a glottal stop (e. g., the sound between the syllables of the common vernacular English interjection, “uh-oh”).⁸ The continuity such consonants provide is crucial to articulation in sung performance. For example, in line 4 of the ŽBM HB faces an articulatory problem caused by juxtaposition of the word zore (“dawn”) immediately followed by the conjunction i (“and”). Without an intervening consonant, a glottal stop will occur between words and voicing will be momentarily interrupted; the guslar’s vocalization will likewise be interrupted because he cannot sing through the sequence without a stoppage of voice. What HB does to avoid that aesthetic blemish (and very practical handicap) is to insert a [v] between zore and i, thus maintaining vocalization across the gap between vowels and easing the burden of sung performance. Thus it is that the performed line takes the acoustic shape of “Prije zore vi bijela dana” (“Before dawn and the white day”).

Performatives can and do occur anywhere that two successive vowels threaten to cause hiatus. Very often, they appear at junctures between formulaic phrases, which, because they have independent lives of their own in the poetic tradition at large, often produce an unwanted interface in particular combinations.⁹ In the eighteenth line of the epic HB furnishes a clear example of bridging a formulaic juncture when he links the first and second cola of the decasyllable (that is, syllables 1-4 and 5-6) via performative [j]: “Samo sebe ji svoga dorata” (“Just himself and his bay horse”). Such excrescent consonants—and HB employs [v], [j],¹⁰ [h], [m], [n], [l], [w], and very rarely [s], [d], and [nj] for the purpose—can also serve to bridge from one poetic line to the next and at a variety of intralinear loci, even, if more rarely, within a single word (e. g., Davulhana for Daulhana, “Drummers,” at line 154) or inflection (e. g., prečitavo, for prečitao, “finished reading,” at line 777).¹¹ Numerous examples of all of these deployments are found throughout the ŽBM and recorded in the table below.

Typically, however, NV transcribes virtually none of HB’s performatives.¹² As a singer fluent in the epic register, NV recognizes the excrescent sounds as part of the event of performance and apparently feels that they do not qualify for admission to the textual record. Because the added consonants are non-lexical, they are deleted without a trace, except on those rare occasions when NV misunderstands a word and assumes that the [v] or [j] or another bridging sound is a bona fide participant in the lexical unit.¹³ His implied distinction between performance and text on this score is extremely instructive: while performatives play an important role in the actual singing of an epic in this tradition, from the perspective of textual dynamics they are vestigial features. They serve no obvious function in the silent decoding of texts, and NV, with a foot in each world, seems to recognize that reality. Of course, our point of view is different: I have included the performatives in this edition of HB’s ŽBM so that silent readers (and listeners at www.oraltradition.org ) can better appreciate both the sounded reality and the compositional structure of the performance.

NV’s treatment of performatives leads naturally to the question of emendation — what he “corrects” versus what he lets stand. In addition to deleting excrescent consonants, he regularly repairs the relatively rare instances of lapsus linguae. These slips of the tongue can be as simple as reversing HB’s metatheses, as when NV restores turbeta from HB’s trubeta at line 738. Misspeakings can also be as complex and multifaceted as line 52: “Slabo svlači va dobro doblači” (“He doffed his poor garments and donned grand ones”). Here HB has added a [d] to the expected oblači (“donned”) under the alliterative influence of the proximate dobro (“grand”); the desirability of a consonantal bridge between dobro and oblači to avoid hiatus also enters the picture. Whatever the source(s) of the original deflection, NV restores the expected form, minus the excrescent [d].¹⁴

As against cases like these, NV sometimes lets HB’s incorrect or incomplete inflections stand. In line 427, for example, he mirrors HB’s vokreće (3^rd sing. historical present, “turned”), when the clearly plural subject is dvanajes’ bešlija (“twelve horsemen”).¹⁵ Whether as a manifestation of his own singer’s ear or in an effort to record as precisely as possible what was sung or for both reasons, NV does not often intervene in such instances of grammatical non-fit. Whatever its determining motivation, this kind of loyalty points up the significant distance between his transcription and an edition of the poem prepared according to customary philological principles. In fact, if we consider the performance on its own terms rather than impose an external frame of reference, we must come to the conclusion that the acoustic networking behind such “errors” is the lifeblood of traditional composition, and that to “correct” them — and in the process to fracture a functioning oral-aural pattern — is in some ways more wrong than right. Not seldom, as here with recurrent [e], an acoustic sequence actually encourages such a shift, leading the singer to misinflect for euphony’s sake, as it were. Complementarily, on most occasions NV simply echoes HB’s nonstandard inflections, never more consistently than with the genitive plural adjectival ending in [-ih], which HB enunciates without the terminal aspiration (reflecting a popular pronunciation in the language at large) and NV represents as -i. I have represented this usage as -i’ in the edited performance-text.

Another very frequent instance of lapsus auris, in which NV typically defaults to idiolectal compositional habits, is his frequent transcription of HB’s line-initial vA (“And, But” preceded by performative [v]) to Pa (“Then”), a substitution that occurs no fewer than 29 times over the 1030 lines of the epic.¹⁶ This small shift, which entails no major difference in meaning or function (both are introductory proclitics), occurs at the metrical site most vulnerable to change in the decasyllable: the opening of the first colon. Under the general rule of right-justification, which prescribes increasing length of words and increasing stability from left to right (beginning to end) of the verse,¹⁷ the line-initial position is the slot most likely to support idiolectal instability. Lapsus auris functions most readily at this site.

What is observably true of the first position in the decasyllable is theoretically applicable to and empirically verifiable for the entire first colon of the line: many more cases of lapsus auris occur over the first colon (syllables 1-4) than the second (syllables 5-10). I offer a few examples here, with the proviso that many more are available in the table below. Here are the two versions of line 138, with literal translations:

HB: Na vonu ga stranu prenosijo (“Bore him over to that shore”)

NV: Na drugu ga stranu prenosijo (“Bore him over to the other shore”)

NV has remade the first colon, using drugu (“the other”) in place of HB’s onu (“that”), but the two decasyllables are essentially equivalent in overall meaning.¹⁸ Sometimes lapsus auris motivates more substantial shifts, as in line 729:

HB: Saštevili, redom iskupili (“They assembled, gathered themselves in order”)

NV: Kad se bili redom iskupili (“When they had gathered themselves in order”)

Instead of the double-verb construction that HB actually sang, NV construes the line as a dependent clause with the common colonic opening of “Kad se bili” (“When they had”).

Most dramatic of all, even though no major change in meaning results, are the cases in which NV either trims or supplements a line to yield a regular ten-syllable increment.¹⁹ Line 12 offers an example of reduction (asterisks mark an extrametrical element):

HB: *hI* jednu, dvije sebi natočijo (“And one, two cups he poured himself”)

NV: Jednu, dvije sebi natočijo (“One, two cups he poured himself”)

The extrametrical I (“And”) plus performative [h] make for an eleven-syllable verse, but since hI is sung in advance of the musical and rhythmic onset of the line it does not result in a hypermetric unit. And just as NV crops “extra” syllables in what amounts to his resinging of the ŽBM, so he also adds them to eke out HB’s verses that seem to be missing one or more syllables. Line 745 provides an example (## indicates a vocal rest):

HB: ## Svanu i jogranu sunce (“[When] day broke and the sun rose”)

NV: A kad svanu i ogranu sunce (“And when day broke and the sun rose”)

Here NV dips into his own competence as a guslar and his particular idiolect to “fill in” what HB “left out” during his vocal rest at the beginning of the line. Again, HB’s verse, though only eight syllables in sung extent, is not hypometric: musically and rhythmically, he starts his vocal articulation with the third-syllable position.

So who is right — the singer who occasionally performs lines of 8, 9, or 11 syllables or the singer-scribe who remakes them all as tidy decasyllables? From one perspective, we must credit HB with having produced verse-units that — as they are actually sounded in performance — do not violate metrical rules. He has sung viable poetic lines. This is a fundamentally important distinction because it goes to the heart of what a verse line really is in performative terms.²⁰ Of course, as page-trained metrists we tend to see verses of more or fewer than ten syllables as flawed, and NV reveals a similar reaction as he either trims or augments every line that includes more or less than the canonical number of syllables. From a textual standpoint — one that is by definition irrelevant to the understanding of oral epic in performance — the amanuensis is “right” in resinging a (textually) more uniform version of the supposedly hypermetric or hypometric line. To give him due credit, NV was after all transcribing for readerly consumption, a reality that may make his manipulation of HB’s performance more understandable. But we nonetheless need to remember that the poetic line is at its core much more than an approved number of syllables; it is a multi-dimensional creature of performance, so that its music, rhythm, vocal rests, and extrametrical syllables are as much a part of its performative definition as the textualizable words we customarily privilege.²¹

For these and other reasons, NV’s “singer’s ear” — what I have termed lapsus auris above — emerges as a crucially significant dimension of his resinging of HB’s ŽBM. Because it influences the making of the song-text at every level, this is doubtless the single most powerful principle informing his transcription. Indeed, most of the changes discussed above are in some fashion manifestations of this recompositional dynamic: recognition and deletion of performatives, palatalization and depalatalization, emendation or failing to correct, substitution of cola, and trimming or augmenting lines all stem in one way or another from the interposition of NV’s singer’s ear.

Overall, then, what Nikola Vujnović’s Resinging offers is a stereoscopic view of a song that comes to us from two guslari at once. In transcribing the ŽBM for the Parry-Lord project, this singer-amanuensis filtered what he heard through the medium of his own compositional idiom. Via the process of lapsus auris, he does much more than simply reproduce HB’s words and syllables with what we like to characterize as verbatim accuracy; instead, he reconstrues the song, remakes it on his own terms. Since NV was set the task not of singing but of auditing and recording in writing what he heard, his path amounts to a negotiation between wholesale remaking of the ŽBM, which would of course involve much more divergence from HB’s performance, and mere mirroring. The result is a double fidelity to both HB’s language and his own idiolect, to both the individual performance and the collective tradition.²²