Plotting Poetry 5 - Popular Voices

Following from the manual typesetting interventions pioneered by Mallarmé in the nineteenth century, one of the key innovations of poetic modernism was to use the typewriter as a device to precisely score a written poem to the page. As Charles Olson put it: “For the first time the poet has the stave and the bar a musician has had. For the first time he can, without the convention of rime and meter, record the listening he has done to his own speech and by that one act indicate how he would want any reader, silently or otherwise, to voice his work” (“Projective Verse”). For Olson, the typewriter provides a spatial notation that allows the reader to reconstitute poetic sonic intention in the way a musician can work from sheet music. But even with such precision, the amount of sonic information that can be scored to a printed page is limited. The reader-listener could potentially glean something about phrasing or speech that resists convention through the use of dialect spellings. But in the end, the printed page is a work of visual art.

The adoption of audio recordings by poets, which reaches back to nearly the beginning of recorded sound, was a moment of rupture. Not only could the poet demonstrate the intonation, pitch changes, and tempos they imagined as aspects of the poem’s form, but the recordings captured what we might call surplus inscriptions, or dimensions of poetic performance not necessarily consciously intended to be an aspect of the poem: regional accent, the influence of literary scenes, an adherence to other sonic genres (e.g. political speeches or sermons), voicing shaped by the technologies of recording, and a panoply of other performances of individual and collective identity. And the presence of such voices, regardless of their level of conscious performativity or so-called authenticity, provides a rich space for the empirical analysis of literary audio and the potential to ask new questions about poetic performance and everyday speech.

This talk will explain the inextricable bond between literary audio and the “popular voices” of dialect recordings in the early twentieth century as an entrée to the concept of what I term “doing voices,” or the ways poets perform identity through voicing. We will tarry with the question of to what degree these voiced identities can be measured and compared computationally through deep learning methods like audio clustering, genre identification, and sound classification. At the heart of this conversation will be the concept of language itself as hermeneutic (cf. Walter Ong) and how the voices we do every day (re)construct a protean identity.

Charles Bukowski’s poetry was always aimed at being performed. In the 1960s and 1970s, his editor Ferlinghetti among others, organized public poetry readings, where Bukowski was the star. Naturally shy, he would get drunk before performing, and will continue drinking and reading poetry on stage in front of an electric audience. Some of his performances were recorded into compact disk and documentaries released after his death. Some other were retold in books by his friends or scholars. His voice was deeply melancholic, adding value to his words.

Our proposition of communication aims at discussing the transfer of Bukowski’s poetry performance skills online. Indeed, the readers of the authors have invested time and effort to remediate (Bolter, Grusin, 1999) Bukowski’s poetry, using a new medium to create new ways of performing poetry.

With a corpus of around 50 YouTube videos created by amateurs, using a variety of poems by Charles Bukowski, we will discuss the remediation of his poetry on a new medium and how it affects its reception. The voice is not the only tool to perform, sound and image are also taken into account online. This presentation will evoke the theories of Jenkins’ convergence culture (2006) and participatory culture (2015), in which the readers (or fans for Jenkins) are involved in the making of the art. We will also present a reflection on the reception this new form of poetry performance entails, as we can either read, listen to, watch poetry independently, or all of the above.

In his paper “Meter and Performance” (2010), Paul Kiparsky emphasized the significance of Roman Jakobson’s elaboration on the Russian Formalist dichotomy of meter and rhythm for the analysis of performed poetry. In “Linguistics and Poetics” (1960) Jakobson renamed this dichotomy a dichotomy of “verse design” and “verse instance” and, as Kiparsky reminds us, added a concept of “delivery instance” (text-setting, recitation).

Acceptable deliveries are pre-ordained by verse instances but do not coincide with them. It is only in the delivery instance that words proper appear: verse instance is the structure of syllables, stresses etc., but within this structure every word can in principle be replaced by another word or even by quasi-words like ta-TA-ta TA-ta-ta TA-ta.

To be sure, Jakobson complemented this triad with yet another element, “delivery design” (performance conventions that govern the delivery and determine the reciter to “cling to a scanning style or tend toward proselike prosody or freely oscillate between these two poles”). Therefore, “delivery instance” is influenced by verse instance, on the one hand, and by delivery design, on the other.

Jakobson was thinking about recitation of written poetry rather than songs. However, his model can be adapted to the description of the rhythm of sung poetry. In this case, “delivery design” can be defined more formally than just a style of performance—it can be defined as the musical rhythm of the poetical line, as it is determined by the musical phrase on which it is sung. (The musical terms “meter” and “rhythm” do not coincide with their poetic homonyms; therefore, delivery design can be equated to musical rhythm, such as ♩♪♪|♩. ♪, but not to musical meters, such as 2/4 or 3/4.)

In relation to a sung text, delivery design, or musical rhythm, such as ♩♪♪|♩♪♪, can be represented as a metrical scheme, such as SWWSWW. Obviously, verse design, or versification meter can also be represented as a metrical scheme, such as WSWWSW. Characteristically, the patterns of verse design and delivery design may not coincide. When they diverge, the performer will be tempted to structure a delivery instance in conformity with either of the two structures or, to use Jakobson’s words, to “freely oscillate between these two poles”.

In the proposed paper, two late Soviet songs will be used to illustrate this problem. One, “На Тихорецкую состав отправится” (Mikhail Lvovsky’s lyrics set to music by Mikael Tariverdiev), has different position of the caesura in the patterns that we have (re)defined as verse design and delivery design. Consequently, the performer has either to accept intraverbal enjambments such as “Про мое прошлое и на||стоящее” (this is how Alla Pugacheva performed this song) or to avoid them (as Vladimir Vysotsky did). The other, Bulat Okudzhava’s “Песенка о дураках” (“Антон Палыч Чехов однажды заметил...”), has a very complicated textual pattern that is set to a more simple musical pattern. In Okudzhava’s own performances, some lines are sometimes trans-accentuated (“Áнтон Палыч Чехов...” instead of “Антóн Палыч Чехов...”) or double-accentuated (“Д‎ýракóм быть выгодно, да очень не хочется”), thus revealing the impact of the delivery design.

There are several ways to adapt a poem to a new medium and create a song based on it (reference, continuation, parody or reflection). One of the most common methods is setting poetic texts to music with minimal interventions, such as transposing, repeating, or even omitting verses. It is commonly thought that this type of musical adaptation disseminates the poem and gives it a new life in another medium, but studies of the textual mechanisms which come into play and their rationale are scarce.

This work presents an analysis of almost a thousand musical adaptations of poems that elicits the most frequent types of modifications, teasing out the criteria for the addition, deletion, or repetition of verses or sequences.

Our corpus was provided by the PoeMAS project, which collects, in a public available database, a wide variety of Spanish lyrics (from 1975 onwards) based on poems. Each entry provides valuable metadata (publication date, performer or author, genre, etc.), and the texts for poem and song.

An automatic collation of each poem and its musicalisations was performed. The textual differences are then categorised by implementing a taxonomy that draws from the theories of intersemiotic translation (Jakobson, 1959; Romano, 1994, 1999) and intermediality (Rajewsky, 2005; Wolf, 1999).

This systematic classification of changes between poem text and song allows us to reach conclusions about the types of adaptations and to work on the criteria underlying these modifications in lyrics, explaining why some verses or sequences are repeated, omitted, added, or changed.

REFERENCES:
Jakobson, R. (1959). On linguistic aspects of translation. En R. A. Brower (Ed.), On translation (Vol. 3, pp. 30-39). Cambridge, Massachusetts: Harvard University Press.
PoeMAS UNED - Poesía para más gente. Relación entre poesía y música. (n.d.). Retrieved 29 April 2022, from PoeMAS website: https://poemas.uned.es/
Rajewsky, I. O. (2005). Intermediality, Intertextuality, and Remediation: A Literary Perspective on Intermediality. Intermédialités, (6), 43-64. https://doi.org/10.7202/1005505ar
Romano, M. (1994). La canción de «autor» en el cruce de escritura y oralidad. En L. Scarano, M. Romano, y M. Ferrari, La voz diseminada: Hacia una teoría del sujeto en la poesía española. Buenos Aires: Biblos.
Romano, M. (1999). Un diálogo selecto: Góngora y Paco Ibáñez. Letras, (52), 105-120.
Wolf, W. (1999). The Musicalization of Fiction: A Study in the Theory and History of Intermediality. Amsterdam: Rodopi.

There is a strong popular and poetic tradition for the reuse of known tunes, called “timbres”, for new songs, as well as for couplets within vaudeville plays. Because these “timbres” are widely familiar to contemporaries, their scores are rarely found, and the only information available is often their titles. Unfortunately, the same tune may have several different titles, making it difficult to track their reuse. Works such as La Clé du caveau provide compile famous tunes, listed according to various criteria (titles, score, adapted metrical scheme). Such valuable sources, although they have been converted into accessible electronic versions with text, have yet to be compiled into digital versions with smart tables, to facilitate their use. Moreover, they are not always complete, and might not adjust to specific corpora.

As part of our operationalization of two corpora (19th Century theatrical Vaudeville couplets and 19th-20th Centuries songs), we have built our own database of tunes. In our presentation we shall report on the successive stages involved in the construction of this database: preparing our corpora, XML encoding, organizing the database in Python/MySQL, and discuss the methodological difficulties we encountered, the purposes for which we intend to use the database, and finally our first results.

The Estonian national epic „Kalevipoeg” by Friedrich Reinhold Kreutzwald has been the subject of controversies and criticism since its publication in 1857. The author’s method of creating an epic transforming heroic legends into the form of runoverses, the alliterative tetrametric poetry, gave the biggest reason for criticism, causing certain preconceived and rejective attitudes among researchers of runosongs. The epic is estimated to contain approximately 12% of „authentic” verses from the runosong tradition, the rest are „forged” both using and misusing the rules of runosong poetic structure. The result is a distinct poetic idiom called Kalevipoeg-verse.

Regardeless those prejudices towards the Kalevipoeg-verse, it has been used to create literary works; also some correspondents dispatched records of songs from the „Kalevipoeg” epics to folklore collectors as traditional songs.

In April 2022, „Kalevipoeg” was added to FILTER.

FILTER is a project (https://blogs.helsinki.fi/filter-project/) which has brought together Estonian and Finnish voluminous data collections to the corpus of ca. 250 000 poetic texts. The Filter-corpus contains mostly traditional runo-songs collected in the 19th and 20th centuries from Karelia, Ingria, Estonia, and Finland, but also literary works: the Finnish epic “Kalevala” and “Kanteletar”. The addition of “Kalevipoeg” gives a great opportunity to answer or review a number of questions related to the Kalevipoeg-verse and the history of collecting Estonian folk songs.

‘textklang’ is an interdisciplinary research project based at the German Literary Archive in Marbach and the University of Stuttgart, including literary studies, digital humanities, computational linguistics, laboratory phonology and speech technology. In ‘textklang’, we develop a mixed-methods approach to investigate the interrelation between written lyric poetry and its sonic realisation. The project’s corpus is centered on poetry of Romanticism and is fed from holdings of the German Literature Archive.

In our contribution, we would like to illustrate a multi-perspective approach to one collection within the corpus entitled ‘The Boy’s Magic Horn’ (‘Des Knaben Wunderhorn’), edited in three volumes by A. v. Arnim and C. Brentano in 1806 and 1808 and including more than 700 poems. ‘The Boy’s Magic Horn’ is considered to be one of the most influential collections of lyrics in German literature because of its vivid reception both in ‘low’ folkloristic cultures and in ‘high’ cultures, especially in musical settings (G. Mahler, J. Brahms, F. Silcher). We would like to share some first outcomes of the ongoing research process, considering quantitative, textual, prosodic, and sonic aspects. More specifically, we will give an overview of this collection in numbers by using an interactive tool for data visualization and exploration, looking at both metadata and textual properties. Additionally, we will present our speech synthesis model trained on this subcorpus which results in better realization of poetic speech compared to synthesis models exclusively trained on prose data.

This paper regards early Soviet theories elaborated in the 1920s by two opposite figures in the literary scene of the time. The both theories focused on new kinds of meters spreading in poetry of the time and borrowed their key terms from Aleksey Kubarev, 19th-century’s scholar who strived to apply musical theory to analyze Russian metrics. These attempts were inspired by Isaac Vossius and Richard Bentley who’s works on Greek and Latin meters were well-known and acknowledged at the turn of the 18th and 19th century. The first theory regarding in the presentation was elaborated by Aleksey Kviatkovskiy, the second one—by Mikhail Malishevskiy, Valeriy Briusov’s student who published his book Metrotonica in 1925. Kviatkovskiy’s theory was rather fortunate: although his opus magnum Rhythmologia was not published during his life, many principal elements of the theory were transferred through his Poetic Dictionary which regularly re-issued for nowadays. Malishevskiy’s views, in contrast, were almost forgotten although Mikhail Gasparov mentioned him in his Modern Russian Verse (1974) as the most principle predecessor of Kviatkovskiy’s theory which he was skeptical about. Therefore, both figures obtained a marginal status in Russian theory of verse although their ideas frequently foresaw those of the later metrical typology which steadily developed after the WWII and after Roman Jacobson’s emigration in USA. In this presentation, I will trace the similarities and differences between Kviatkovskiy’ and Malishevskiy’s theory and contemporary metrical typology in order to show how conceptions taken as marginal and out-of-fashioned through decades can be actualize anew, in the new phase of versification theory.

The strong association between meter and semantics or "semantic halo" is known as a feature that is retained in specific meters over time. While most of the studies focus on tracing the origins and transmission of the halo, this paper aims to examine the processes of thematic innovation and expansion in a meter that (allegedly) lead to the loss of relations between form and semantics. As an example of such scenario we will use the data of the 19th century Russian poetry where the semantic changes in iambic tetrameter (Iamb-4) has been thought to be a development from a narrow genre to its expansion to various themes and genres. This paper aims to scrutinise these processes formally.

To present each poem as a comparable set of semantic features we will be using topic modelling. Our hypothesis is that we would be able to find a short period of thematic “innovation” in Iamb-4 via calculating the Kulback-Leibler divergence and then trace the form’s expansion and downturn with networks (taking topics as nodes and its appearances within a meter as edges). Using networks, we expect to find not a random “expansion”, but semantic clusters (sets of topics) intrinsic to other metrical forms to be added to Iamb-4 semantic structure but then reallocated to other metrical forms. We think these exploratory methods will be helpful in studying the form and meter relations and mechanisms of their change (or stability).

Three-ictic dolnik (Dk3) is the most common form of non-classical Russian meters and its rhythmical features have been analyzed more than once. The most famous study was realized by Mikhail Gasparov (1963, 1968, 1974) who examined Russian Dk3 of the 1890–1950s, identified its rhythmical variations (forms) and proposed a typology of the main rhythmical types of Dk3 (i.e. combinations of the most frequent rhythmical forms): “Yesenin’s type”, “Tsvetaeva’s type” and “Gumilev’s type”. In the last decade, the analysis of the metrical structure of Russian dolnik has yielded new results, in particular, it has been shown that the typology of Dk3 requires further elaboration (see Плунгян 2008, 2010; Ляпин 2011; Левашов — Ляпин 2012; Ляпин — Пильщиков 2014, Liapin — Pilshchikov 2015, etc.).

Dolnik verse arose in Russian poetry of the 19th century from translations and imitations of German tonic versification, but the structure of the dolnik in translated poetic texts is less studied than the structure of original Russian dolnik. There are two exceptions: the earliest Russian examples of this meter (translations of German romantic poetry by Zhukovsky, Lermontov, Tiutchev, Fet, Apollon Grigoriev, and Mikhail Mikhailov) and translations from Heine. In the 1930s, the rhythmical features of German and Russian dolnik were discussed by Russian scholars such as Osip Brik and Boris Yarkho in connection with the contemporary translations of Heine’s works into Russian (Брик и др. 1934/2012; Полилова 2014; Liapin — Pilshchikov 2015). For example, Brik argued for equirhythmical (rather than only equimetrical) translations of Heine’s dolniks, that is, for reducing the number of disyllabic unstressed intervals within the line in accordance with the rhythmical structure of German dolnik verse. Not only scholars, but also translators themselves recognized the importance of reproducing the original structure. For example, as James Bailey showed in the article “Blok and Heine: An Episode from the History of Russian dol'niki” (1969), the rhythm of Alexander Blok’s translations from Heine is significantly different from the rhythm of Blok’s original dolnik poems. In Blok’s translations the proportion of Form IV [(0/1/2) × 1 × 1 × (0/1/2/3)], which is not typical for Russian Dk3, but is popular in German dolnik verse, is much higher.

Russian translators used Dk3 not only for equimetrical translations of German poetry, but also as an equirhythmical analogue of Spanish Romancero verse. In Russian translations, Dk3 replaced trochaic tetrameter, which was traditionally used to render this Spanish form in the 19th century (Polilova 2018). In the 1930-1940s, Valentin Parnakh began to use isosyllabic Dk3 (8 syllables in the lines with a feminine clausula and 7 syllables in the lines with a masculine clausula) in translations of Federico García Lorca’s romances. Soon, other translators also mastered this form (although their Dk3 was not always isosyllabic), and began to use it not only in the translations of Spanish 20th-century octosílabos, but also in the translations of the old Spanish romances. The rhythm of Dk3 in the Russian translations from Spanish is studied fragmentarily (Goncharenko 1980; Kamelina 2007) and it has never been compared with the rhythm of original Russian Dk3. I analyzed the translations of Lorca’s Romancero gitano and old romances (60 texts, translators: V. Parnakh, I. Tynianova, M. Zenkevich, A. Geleskul, O. Savich, P. Grushko and others) and compared the rhythmical structure of Spanish original texts with the rhythmical structure of its Russian versions. I discovered the main feature of the rhythm of Russian Dk3 used in translations from Spanish: it is characterized by an almost complete absence of lines with trochaic cadence (whose proportion in the Spanish originals varies from 26% to 56%).

In this paper, I will present my observations on the rhythmical specificity of Russian Dk3 used in the translations of Spanish romances. Its rhythm will be compared to the rhythm of original Russian dolniks and the dolniks from the translations of Heine.

The research presented in this abstract results from the project based at the Institute for World Culture of Lomonosov Moscow State University (MSU) and supported by Russian Science Foundation grant 19-78-10132.

REFERENCES:
Bailey J. 1969: “Blok and Heine: An Episode from the History of Russian Dol'niki”, The Slavic and East European Journal, vol. 13, № 1, pp. 1–22.
Liapin S., Pilshchikov I. 2015: “Ein Fichtenbaum steht einsam” and the typology of the Russian dolnik (following Osip Brik’s, Boris Jarcho’s and Andrei Fedorov’s remarks on the Russian translations from Heine), Studia Metrica et Poetica, vol. 2.1. pp. 58–80.
Plungian V. 2011: “Two requiems, or the English dolnik on the Russian soil”, Formal Methods in Poetics: a Collection of Scholarly Works Dedicated to the Memory of Professor M. A. Krasnoperova, Lüdenscheid, pp. 173–181.
Polilova V. 2018: “Spanish Romancero in Russian and the semantization of verse form”, Studia Metrica et Poetica, vol. 5.2, pp. 77–108.
Брик О.М. и др. 1934/2012: “Доклад О.М. Брика о новых переводах «Германии» Гейне и его обсуждение на секции переводчиков союза писателей (1934)” / Подготовка текста, публикация и примечания Т.Ф. Нешумовой, Philologica, т. 9, № 21/23, с. 280—333.
Гаспаров M. Л. 1963: “Статистическое обследование русского трехударного дольника”, Теория вероятности и ее применение, т. 8, № 1, с. 102–108.
Гаспаров M. Л. 1974. Современный русский стих: Метрика и ритмика. Москва, с. 220—244.
Гаспаров М. Л. 1968: “Русский 3-ударный дольник XX в.”, Теория стиха, Ленинград, 1968, с. 59–106.
Гончаренко С. Ф. 1980. “Контрастивная лингвистика и поэтический перевод (испанский восьмисложник в зеркале русского стиха)”, Сборник научных трудов МГПИИЯ, вып. 166, 1980.
Камелина А. В. 2007:“Стиховые структуры «Цыганского романсеро» Федерико Гарсиа Лорки в оригинале и русских переводах: Диссертация на соискание ученой степени канд. филол. наук”, Москва: МГЛУ.
Левашов А. М., Ляпин С. Е. 2012: “Ритмико-синтаксическая структура «Прощальной оды»: к гексаметрической концепции шестииктного дольника Бродского”, Иосиф Бродский: проблемы поэтики: Сборник научных трудов и материалов, Москва, с. 139–151.
Ляпин С. Е. 2011: “«Сегментный» дольник: к описанию метрических новаций Иосифа Бродского”. Вестник Московского университета. Серия 9: Филология, вып. 6, с. 36–46.
Ляпин С. Е., Пильщиков И. А. 2014: ““Ein Fichtenbaum steht einsam” и типология русского дольника”, Методология и практика русского формализма: Бриковский сборник II. Москва, с. 146–157.
Плунгян В. А. 2008: “Об одном «незамеченном» типе русского стиха: логаэдический пеонический дольник”, Динамические модели: слово, предложение, текст: Сборник статей в честь Е. В. Падучевой, Москва, с. 646–663.
Плунгян В. А. 2010: “Тонический стих Вячеслава Иванова: к постановке проблемы”, Вячеслав Иванов: Исследования и материалы, Санкт-Петербург, вып. 1, с. 291–309.
Полилова В. С. 2014: “Осип Брик и стихотворный перевод (к проблеме эквиритмического перевода)”, Методология и практика русского формализма: Бриковский сборник II. Москва, с. 158–169.

Quantitative drama analysis has delivered new insights on major European dramatic traditions, e.g. German or French theater. However, “smaller” traditions have been less fortunate. Take the case of Alsatian dialect theater, with its rich production in which comedy predominates, without excluding more serious forms. Alsatian (Germanic varieties from Alsace, Eastern France) was the main oral communication language in the area historically, until its decline in favour of French late in the late 20th century. The Alsatian dramatic tradition is surrounded by two hegemonic literatures (German and French), and their respective influence is underexplored, less so using computational means. Macroanalytically inspired analyses of Alsatian theater were so far impossible given lack of an appropriate corpus. The MeThAL project is building the first large-scale TEI-encoded corpus for Alsatian theater, focusing mostly on the 1870-1941 period. So far, a corpus of 300,000 tokens (25 plays) has been published; 50 further plays are underway. I will present our corpus selection criteria and encoding automation workflow, paying attention to what might be generalizable to other low-resource literatures. I will also present our corpus annotations: bibliographic metadata and a manual annotation of social variables for the corpus' 2,386 characters (profession, social class, sex, age). These annotations allowed us to obtain generalizations about the evolution of social groups in the period, which complement existing knowledge of the tradition. Finally, I present the corpus exploration interface.

Humour studies have developed description protocols to reliably annotate jokes. The linguistic and logical mechanisms used for this purpose efficiently describe the careful overlapping of incompatible meanings that produce and resolve incongruity, and thus jokes.

Jokes can also be found in poetic corpora, and obviously there is a large body of humorous and sometimes outrageous poetry. Yet more broadly, poetic texts – probably more than other genres – rely on a heavy use of ambiguities, shortcuts, overlaps, that seem to function somewhat like jokes, but without involving humour necessarily.

We explore the use of joke-like occurrences, with or without a humorous intent, in a poetic corpus. For this, we are adapting the existing description protocols to our own purposes, and manually tagging joke-like patterns within a poetic corpus.

The data obtained combines the rich subjectivity of human observation with a volume large enough for statistical analyses. The functioning of non-humorous joke-like patterns informs the poetics of ambiguity as much as that of humour, and opens up a discussion on the complex seriousness to silliness spectrum.

Our paper is dedicated to rhyme in Estonian rap culture, being the main device of instrumentation in its poetic structure. For this, we have sampled the lyrics from the rap songs by two artists, Genka and Metsakutsu, and compared their rhyming models to the ones found in Estonian literary poetry. We will show that while in the beginning of the 21st century the idea began to spread that the possibilities of Estonian rhyme have been exhausted, rap practices have opened up new ways of rhyming and made use of completely new rhyming strategies, which results in the richest rhyming repertoire in the Estonian poetic culture. At the same time, it is not a counter-reactive creation with regard to some earlier standard or prescription, but rather an adherence to Western models, on which the aesthetic of Estonian rap rhyme heavily relies.

The collections Suomen Kansan Vanhat Runot (Old Poems of the Finnish People) and Eesti Regilaulude Andmebaas (Estonian Database of Runosongs) contain each around 100,000 songs in several closely related Finnic languages collected from oral tradition mainly in the 19th and early 20th century. They feature a lot of similar texts and passages, but the high linguistic variation, as well as the sheer size of the material, make it difficult to spot the full range of relationships between the texts. In this talk, I present computational methods for recognizing text similarity, which have been successfully applied to those collections. The results enable large-scale quantitative studies of similar texts, addressing questions such as the relationship between oral tradition and printed works or the transfer of texts across places and times, as well as the recognition of interesting individual cases via outlier analysis. In the second part of the talk, I show some initial insights from such studies.

I am seeking for a scientific solution to the methodological question of effective text choice – for research and scholarly editing – for certain genres of folk poetry where texts are relatively short, but highly variable and at times have no clear boundaries. My focus is on postmedieval þulur (ca. 15th–20th c.; hereafter PMÞ): versified, but not stanzaic lists of names, sequences of short motifs and/or longer narrative episodes in very free poetic form. PMÞ are relatively short (25 lines at average), fragmentary and intersecting.

The central challenge is that of producing a manageable, but representative selection that reflects the specificity of the genre in question. A specific feature of PMÞ is their loose and highly variable two-level structure based on the listing principle. The lists at the first level consist mainly of names and motifs which combine in various ways to produce relatively stable second-level units that I have designated as “blocks”; PMÞ-texts can be made of units from both levels. My hypothesis is that the selection which most adequately represents a useful overview of highly variable texts such as the PMÞ, while also reflecting their specificity, should be based on the most typically encountered variant(s) of most common structural units of PMÞ – primarily, of PMÞ-blocks (which have more clear boundaries than PMÞ-texts). I would like to discuss the advantages and drawbacks of using quantitative method to identify the most common structural units and their most typical variants.

This paper tries to examine deeply the correspondence between the oral performance and the degree of elaboration of the versification in Hungarian epic poems of the 16th century. All these poems were performed with a musical accompaniment, and in many cases the melodical lines are conserved.

A previous pilot study performed on a much smaller corpus suggests that the few freely, orally composed poems of the corpus, which survived only accidentally in a written form, have a somewhat looser versification that the poems composed with more care, on paper, keeping several patterns of their essentially written character (like the acrostics).

This time the entire corpus of Hungarian epic poems of the period (c. 180 poems, appr. 400000 tokens) will be analysed to confirm or to disprove this hypothesis. Some additional questions will be answered in the paper: Is the versification of the poems composed to an original melody more precise than that of contrafacta? Is the popularity of a poem (or of its melody) related to the quality of its versification? Can we identify with certainty personal styles in the versification despite the popular style and the generic patterns of the epic poetry (like the formulae)?

Metrical scansion is a subtle endeavour, in which the smallest details are of great importance. Moreover, one must be aware that a proposed scansion of a verse line not necessarily falls into the categories of ‘right’ or ‘wrong’. This is especially true for medieval Dutch poetry. This vernacular poetic tradition is characterised by a variable length of the verse line, which mostly contains four - but frequently also less or more - stressed syllables, positioned in an unregular fashion. This great poetic freedom poses present-day researchers for a challenge, as it is not an easy task to determine objectively which verse line scansions are to be preferred and which ones are less favourable.

In this paper, we focus on the challenging task of evaluating scansions of medieval rhymed texts. To this end, an essential bridge is established between an automatic scansion machine for medieval Dutch poetry and the assessment of its output. It will become clear that building such a bridge is not without difficulty. We will seek to overcome several obstacles by exploring creative workarounds. One of them is found in the development of an annotation task, in which twenty-five experts in the field of medieval literature participated. From these different experts, scansion proposals were collected for various medieval Dutch poems. This provides us with a more solid basis to gauge the preferences of the automatic scansion machine. Finally, our search will result in a meaningful method that allows us to judge the quality of computer-generated scansion for historical languages.

Diminution as a linguistic feature is clearly more common to the poetic register of the Estonian runosong (regilaul, Kalevala-metric song) language compared to contemporary Estonian. However, the exact functions and factors determining the variation of diminutive markers in the runosongs are not clearly determined and no corpus studies focusing on the diminutives of the Estonian runosongs have been carried out. In my presentation I will discuss the first results of my quantitative analysis performed with the aim of determining some general tendencies involving the prevalent diminutive suffix’ (-kene) variation in regilaul, starting with the distribution of this suffix in relation to geographical variance. The datasets that I used for this analysis have been extracted from the corpus of the Estonian runosongs with the help of our hackathon workgroup of the Estonian Folklore Archives. The dataset underlying my initial analysis contains the relative frequencies in regions of Estonia of all verse lines that contain the diminutive suffix -kene in nominative form, complemented with the geographical coordinates associated with the regions. I will present some figures and diagrams to visualize this data in my presentation. The results show that the relative frequency of -kene in runosong texts is incrementally increasing in direction from West to East. I will also present some initial results of analysing other datasets involving the diminutive suffix -kene in runosongs.