Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv* Introduction Plotting Poetry 5 was the fifth meeting of an international group of schol- ars with a common interest in the machine-assisted exploration of poetry and poetics. The conference was hosted by the Estonian Literary Museum in Tartu between July 4 and 6, 2022, organized by Mari Sarv (Estonian Literary Museum), Maria-Kristiina Lotman (University of Tartu), Susanna Mett (Estonian Literary Museum & University of Tartu), Anne-Sophie Bories (University of Basel), Pablo Ruiz Fabo (University of Strasbourg) and Petr Plechač (Czech Academy of Sciences). The conference comprised 16 presen- tations by scholars from 11 different countries. This year’s conference theme was popular forms of poetry, such as folkloric poetry (traditional or novel), besides song lyrics (e.g. in folk, pop, or rap) and the connections between text and performance, as well as poetry delivered through new media channels like the social web.1 Besides this main theme, work on other topics was also presented, including on genres other than poetry; this has been traditional for the Plotting Poetry conference series as long as the research is quantitatively * Authors’ addresses: Nils Couturier, Philosophisch-Historische Fakultät, Departement Sprach- und Literaturwissenschaften, Fachbereich Französische Sprach- und Literaturwis- senschaft, Maiengasse 51, 4056 Basel, Schweiz. E-mail: nils.couturier@unibas.ch; Antonina Martynenko, Department of Slavistics, Lossi 3, 51003 Tartu, Estonia. E-mail: antonina.mar- tynenko@ut.ee; Lara Nugues, Philosophisch-Historische Fakultät, Departement Sprach- und Literaturwissenschaften, Fachbereich Französische Sprach- und Literaturwissenschaft, Maien- gasse 51, 4056 Basel, Schweiz. E-mail: lara.nugues@unibas.ch; Pablo Ruiz Fabo, Faculté des Langues, 22 rue Descartes, F-67084 Strasbourg Cedex. E-mail: ruizfabo@unistra.fr; Mari Sarv, Eesti Kirjandusmuuseum, Vanemuise 42, 51003, Tartu, Estonia. E-mail: mari@haldjas.folklore. ee. Author order is alphabetical; all authors contributed paper summaries. The report was coordinated by Pablo Ruiz and Mari Sarv. 1 The call for papers is at https://www.plottingpoetry.org/conference/2022tartu/cfp https://doi.org/10.12697/smp.2022.9.2.05 Studia Metrica et Poetica 9.2, 2022, 134–146 mailto:nils.couturier@unibas.ch mailto:antonina.martynenko@ut.ee mailto:antonina.martynenko@ut.ee mailto:lara.nugues@unibas.ch mailto:ruizfabo@unistra.fr mailto:mari@haldjas.folklore.ee mailto:mari@haldjas.folklore.ee https://www.plottingpoetry.org/conference/2022tartu/cfp https://doi.org/10.12697/smp.2022.9.2.05 135Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia or computationally inspired. The present conference report is structured as follows: First, we provide an overview of the main topics covered by the confer- ence papers. After that, individual paper summaries are presented, following the order of papers in the conference program chronologically.2 Overview of paper topics The majority of the papers were related to this year’s main theme. Mustazza’s keynote examined the relation between literary audio and popular voices in dialect recordings in the early 20th century. The interplay between poetry and 20th/21st-century popular song was the focus of several papers. Pilshchikov’s paper illustrates the tension between musical rhythm and metrical scheme based on two late Soviet songs, working with concepts “verse design”, “verse instance”, “delivery instance”, and “delivery design”, introduced by Jakobson (1960). Sabel et al. examined textual modifications in musical adaptations (since 1975) of poems in Spanish, via an automatic collation followed by manual labeling of modifications, in order to explain repetitions, omissions, additions or other changes to adapt to the new medium. Lotman & Lotman compared rhyming models by two Estonian rap artists with the ones found in Estonian literary poetry, finding rich novel rhyming strategies in Estonian rap as a result of contact with Western models. Still within 20th-century poetry, Macaud presented her study of the online remediation of Bukowski’s poetry performances based on a corpus of YouTube videos by amateur readers. Moving on to poetry-music relation in historical corpora, Bernhart & Koch presented a quantitative overview of textual, prosodic and sonic aspects in “The Boy’s Magic Horn” (“Des Knaben Wunderhorn”), an early 19th-century German poem collection appreciated in both folk culture and “high” culture settings, including via musicalizations by Mahler or Brahms. They also pre- sented speech synthesis experiments trained on poetry corpora. Couturier & Nugues presented a database of tune reuse in a corpus of 19th-century vaude- ville couplets and 19th–20th century songs. Plecháč et al. turned to Hungarian epic poems of the 16th century, examining the impact of oral performance (accompanied by music) vs. written composition in versification complexity. Traditional folklore poetry was also addressed by several papers. Saarlo et al. introduced their research on Kalevipoeg, the Estonian national epic, how its versification both uses and misuses runosong poetic structure. They also 2 https://www.plottingpoetry.org/conference/2022tartu/programme https://www.plottingpoetry.org/conference/2022tartu/programme 136 Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv presented a large-scale analysis of Kalevipoeg in the context of the FILTER project, along with Estonian and Finnish runosongs and epic poetry. FILTER’s 275,000 poem corpus was also discussed in Janicki’s keynote talk, where he presented methods for automatic similarity computation and alignment within this corpus, towards new findings about the relations among its texts. Veskis presented a further paper on Estonian runosong, focusing on the distribu- tion of diminutive suffixes and how they correlate with geographic variation. Helgadóttir discussed methods to find structural units and their variants in postmedieval Icelandic þulur, a folk poetry genre. Several papers dealt with topics beyond this year’s main theme. Bories presented work on non-humorous joke-like patterns (involving different types of ambiguity or “script oppositions” using Attardo & Raskin’s 1991 framework) in works by Apollinaire and Aimé Césaire. Martynenko & Šeļa examined semantic change in 19th-century Russian iambic tetrameter, using topic modeling to represent poem content and network analysis to assess topic distribution across different meters. Polilova analyzed the use of Russian meter Dk3 (three-ictic dolnik) in the translation of Spanish romances, paying atten- tion to how its use compares to original dolnik rhythm and dolnik used in translations of Heine. Finally, Ruiz Fabo presented work towards creating the first large electronic corpus of theatre in the Alsatian dialects, besides analyses of character distribution diachronically. Individual paper reports The conference opened with a keynote talk by Chris Mustazza (University of Pennsylvania), delivered online and entitled “The Voices We Do: Surplus Inscriptions in the Poetry Audio Archive”. His work focuses on audio record- ings of poetry. He is interested in what the recording can capture that the written document cannot, and how this audio data can be measured and compared computationally. The audio data he is working on comes from the PennSound online archive, and primarily from the collection of sound recordings he has edited known as The Speech Lab Recordings – the largest prewar collection of American poetry recordings, made in a dialect labora- tory founded by William Cabell Greet and George W. Hibbit in the 1930s. The pitch, duration, pauses and tempo of this audio data are analysed to see if it is possible to determine poets performing in relevant sonic genres such as sermons, vaudeville, political speeches and comic monologues using these parameters. In addition, special attention is paid to the voice, and how its 137Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia sound is shaped by context, the technology that records it, the influence of personal literary influences, the identity of the speaker, or the accent linked to a geographical location. In his presentation, Chris Mustazza examined in detail several recordings, William Butler Yeats – “The Lake Isle of Innisfree” (recorded 1936), Ezra Pound – “Sestina: Altaforte” (recorded 1939), Robert Creeley – “I Know a Man” (recorded 1965), William Carlos Williams – “The Red Wheelbarrow” (recorded 1942), and concluded with a study of Robert Frost – “The Code” (recorded 1933), which features a demotic North American regional dialect. The paper concluded that there is no singular voice in a poetry recording, but rather a heteroglossic amalgam of various influences, each of which is capable of being measured, defined, and located as it reverberates through an audio corpus and through literary history. After the first keynote, we listened to the paper “Charles Bukowski’s per- formance poetry, reperformed online”, by Amélie Macaud from Université Paris 1. Bukowski’s poetry was written to be performed, and indeed his editor (Lawrence Ferlinghetti) organized poetry readings by Bukowski in the 1960s and 70s, some of which were recorded. The paper discusses the online “reme- diation” (Bolter and Grusin 1999) of Bukowski’s poetry, based on a corpus of 30 YouTube videos created by amateurs, focusing on the poem “The Crunch”. The paper describes not only text selection in the videos, but also how they combine voice (Bukowski’s or other readers’, sometimes the video authors’ voice), music (and how its rhythm compares to poem rhythm) and image (often with different images bound to each stanza, generally moving images, sometimes still ones). Besides remediation forms, the paper also discusses the comments section for the videos, carrying out both a statistical analysis of their number and temporal distribution and a textometric analysis of their content, to see salient aspects pointed out by viewers in their mostly positive comments. The overall research goal is to examine how Internet users read, understand and re-perform poetry, thanks to systematically studying the vid- eos’ text, audio and image. Igor Pilshchikov (UCLA/Tallinn University) in “Jakobsonian ‘broad met- rics’: a model for musical verse (exemplified by two late Soviet Russophone ‘bard’ songs)” proposed a model for understanding and analysing performed poetry. The problem of verse recitation influenced both by verse and recitation features was earlier outlined by Roman Jakobson, who formulated it as a four- part structure: on the one hand, a text has poetic meter and rhythm (or verse design and verse instance respectively), on the other hand, it can also have a separate delivery design, i. e. recitation rules; all the three above influence a specific performance or a delivery instance. Pilshchikov expanded the defini- tion of performed poetry from recitation to singing and exemplified how the 138 Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv notion of delivery instance might be applied to “sung poetry”. In this case, the four-part scheme is supplemented with musical design (musical meter) and instance (musical rhythm), which do not always coincide with the verse fea- tures. Having two examples of Russian “bard” songs, Pilshchikov showed how an actual performance (delivery instance) depends on verse and musical struc- ture. The first song (“Na Tikhoretskuiu sostav otpravitsia...” [The train will go to Tikhoretskaya...]) was an illustration of different caesura patterns in music and verse that might result in two ways of performance: while the main Soviet star singer Alla Pugacheva embraced musical design with intraverbal enjamb- ments, another important Soviet singer and songwriter (Vladimir Vysotsky) avoided the enjambments and preferred verse design over the musical features. Another example was Bulat Okudzhava’s song (“Pesenka o durakakh” [A lit- tle song about fools]) that demonstrates how a very simple musical pattern influenced the complex verse design and led to reaccentuation. This theoretical framework thus allows to formally align verse and musical (or performative) structures in order to see how they differ and influence performer’s choices in verse or musical features. The paper by Helena Bermúdez Sabel (Université de Neuchâtel), Clara Martínez Cantón (UNED Madrid) and Pablo Ruiz Fabo (Université de Strasbourg) was entitled “From poetry to song: A corpus-based approach to textual variation”. They presented results from the PoeMAS project, carried out at Universidad Nacional de Educación a Distancia in Madrid, Spain. This project is collecting, in a publicly available database, a large variety of Spanish lyrics (from 1975 onwards) which bear some type of relation to a poem. The project targets both specialists and a general enthusiast public. The paper’s originality lies in its being the first corpus-based approach examining textual changes within musical adaptations of poems, in Spanish; the concept of adap- tation is based on works like Albrecht (2010), Hopkins & O’Leary (2005) and Ingham (2013). Concretely, over 800 songs which represent a similar or identi- cal adaptation to a poem were analyzed. It is proposed that similar adaptations and identical adaptations are preferred over other strategies as a way to adapt the poem to a different medium while trying to imitate the original cultural object (Badía Fumaz, 2020). An automated collation was carried out to identify types of change (transposition, repetition, omission or addition). Differences were manually annotated to determine the type of modified sequence (e.g. a line, a word or other fragment) and the level of analysis (phonic, semantic, structural). Most text changes were found to be motivated by requirements of the new (song) medium. The most frequent text change was repetition, which can be tied to its importance in pop song, where repetition is remembered as associated with a specific melody and rhythm (Abeillé 2013: 167). Structural 139Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia changes to the poem text in the song, also based on repetition, include the creation of refrains and choruses. The talk by Nils Couturier and Lara Nugues (University of Basel) was enti- tled “Constructing a TIMBRE Database: Handling Popular Tunes Reuses”. They presented their work on timbres, a French word that designates well- known tunes reused to set new texts to music. Two popular corpora were analysed. First, a corpus of vaudeville with couplets, a kind of play mixed with songs, chosen between 1830 and 1832. Second, a corpus of chanson sung on well-known tunes from the early 19th century (Pierre-Jean de Béranger) and from the end of the 19th century (Jules Jouy). Based on these two corpora, they built a database containing more than 2,000 “timbres”. The aim was to compare corpora and sub-corpora of a similar nature (popular corpora performed and sung on well-known tunes) at different times in the 19th century in order to measure the interplay of influences between the vaudeville with couplets genre and the chanson genre and to gain a better understanding of how timbres are used and reused (passing fad, importance of the composer’s reputation, life span and original genre of the tune). From the recovery of the plain text to the extraction of the data, including the tagging and standardisation phases, Nils Couturier and Lara Nugues described all the steps that enabled them to develop interoperable data sets and to draw interpretations from them that shed light on the history of the genres sung on well-known tunes. Liina Saarlo, Mari Sarv and Susanna Mett explored in their paper “Kalevipoeg in the FILTER-machine: New opportunities to study the links between literary and oral poetries” the similarities and differences between archival collections of Estonian runosongs and the Estonian national epic Kalevipoeg by Fr. R. Kreutzwald, which aims to use the same poetic form combining excerpts from the oral tradition with the writer’s own diction. The literary scholars and folklorists in general have agreed that Kreutzwald’s aim has not been fulfilled. He has either not succeeded in imitating or follow- ing the complex poetic system of oral poetry or has decided to modify it according to his own aims and taste. The most striking difference between the poetic form of oral tradition and Kalevipoeg lies in sentence structure: while in oral tradition usually there are simple sentences with a length of one line, Kreutzwald uses more complex sentences that may span over several lines. Kreutzwald also added a layer of his own formulaic language with a romantic tint, in addition to more traditional formulae. Relying on the verse and song similarity calculations provided in the FILTER database, the presenters aimed to detect the source region of more traditional parts of Kalevipoeg. Although a region in Central Estonia appeared to have closest parallels to Kalevipoeg, a closer observation showed that a majority of the texts available in the folklore 140 Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv collection were either direct copies from Kalevipoeg or literary creations in the romantic style of Kalevipoeg. Elements of Kalevipoeg style could be also detected in the songs by genuine folk singers. All this demonstrates that the national epic was not only relevant for scholars and learned people, but was known and esteemed widely in the society. In their paper “Blowing ‘The Boy’s Magic Horn’: Plotted and Synthesised Romanticism”, Julia Koch and Toni Bernhart began by presenting the interdis- ciplinary research project “textklang – Mixed-Methods Analysis of Poetry in Text and Tone”, which is based at the German Literary Archive in Marbach and the University of Stuttgart. This project aims to investigate the interrelation between written lyric poetry and its sonic realization in recitation and musical performance, in a corpus centred on poetry of Romanticism. Julia Koch and Toni Bernhart then illustrated this multi-perspective approach on one specific collection within the corpus, entitled “The Boy’s Magic Horn” (“Des Knaben Wunderhorn”), edited in 1806 and 1808 by Achim von Arnim and Clemens Brentano, in three volumes, and containing more than 700 lyric texts (versified and rhymed) and a few parts of text in prose. They shared their first outcomes by presenting on the one hand how to plot data and metadata by using the Keshif app, and on the other hand how to plot sounds using Text-to-Speech Synthesis (TTS). Keshif is a web-based visualization and analytical tool, that allows to explore a dataset quickly: the metadata of “The Boy’s Magic Horn” that were imported to Keshif concerns titles of the text, number of tokens per text, text genres, level of expressivity, authors of the sources and of the pre- texts, publication years of the sources and of the pre-texts. The TTS is trained on “The Boy’s Magic Horn” in order to propose a better realization of poetic speech compared to synthesis models exclusively trained on prose data. Julia Koch and Toni Bernhart presented the challenges raised by the use of this TTS, such as historical language and non-standard speech features, and showed how it can nevertheless capture some stylistic properties, for example speech rate and pauses. The synopsis of text, audio files and metadata finally allows us to study prosodic patterns in text and sonic realisation. Antonina Martynenko (University of Tartu) and Artjoms Šeļa (Institute of the Polish Language of the Polish Academy of Sciences and University of Tartu) in “The fall of genres that did not happen: formalising historical dynamics of Russian poetic meter semantics’’ introduced a computational approach for tracing changes in poetic meter semantics. Taking as an example the historical development of the “universal” Russian meter, iambic tetrameter, Martynenko and Šeļa questioned whether there is a formal way to model the decomposition of the relationships between meter and meaning (effect known as “semantic halo”) in a widely popular meter. To formalise the semantics of poems, authors 141Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia used topic modelling, which allowed them to represent the 19th century poetic corpus as probabilities of topics (sets of co-occurring words) and then look at the chronological dynamics of topics in different meters. Historical changes in topical composition per meter were explored using networks: each topic served as a node and two nodes were connected if two or more topics appeared in the same poem with high probability. Networks were used to chronologically compare the behaviour of a “univer- sal” meter with meters which were presumed to have specific genre domains. Indeed, only the most popular meter was occupying most of the network’s space for a short period of its “expansion” time. However, when the ratio of shared thematic connections between meters was compared to a random base- line, it became apparent that the majority of new connections was a function of a frequency of iambic tetrameter and not its “invasiveness” into the semantic domains of other forms. This paper also shows that changes in semantics of Russian meters were not rapid: most stable connections between topics (edges in the network) tended to persist across the whole period. This highlights the conservative role that poetic forms play in semantic retention. In her “3-ictic dolnik in Russian translations of Spanish and German verse: a comparative rhythmical analysis”, Vera Polilova (Moscow State University) reported on the recent findings in rhythmical features of translations made in Russian dolnik (accentual verse with mono- and disyllabic inter-ictic intervals). Although historically this meter was influenced by translations of German dolniks, in Russian tradition it is routed in ternary meters and avoids binary (trochaic) cadence. However, in the case of translations, these tendencies of Russian original dolnik could have been altered for the sake of reproduc- tion of the source texts’ rhythm. The classic examples are Alexander Blok’s equirhythmical translations from Heine, drastically different from the rhythm of Blok’s original dolnik in proportion of lines with binary rhythm. Polilova investigated the strategy that prevailed in translations made in dolnik during the 20th century: the equirhythmic method that presumes the usage of dolnik to reproduce the original rhythm, or the exorhythmic method that adapted the meter to its rhythmic norms in the target language. The further examination of translations from Heine proved that the equirhythmic tradition started by Blok in the early 20th century was quite unstable and in the second half of the century translators switched to the exorhythmic type, abandoning binary lines and showing strong preference to “pure” dolnik lines with irregular unstressed intervals. While earlier equirhythmic translations clearly show the possibil- ity to reproduce the rhythm of German dolnik in the Russian language, this switch to Russian dolnik’s rhythm in the later translations provides a strong evidence for the cultural reasons to avoid lines with binary and even ternary 142 Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv rhythm in this meter in Russian poetry. Polilova deepened this conclusion with an example of the same trend in dolnik translations of Spanish octosyl- labic verse into Russian (García Lorca’s romances). Though the initial idea of representing syllabic rhythm using rhythmically-free dolnik could have been pursued in early 20th century translations, later translations imbibed the rhythm of original Russian dolnik (with a limited variability of rhythmic patterns). Hence, both the German and Spanish cases confirmed that even in the free domain of translation, Russian dolnik was a stabilising force and functioned as if a new classical meter. The talk by Pablo Ruiz Fabo (University of Strasbourg) was entitled “MeThAL: Towards a macroanalysis of theatre in Alsatian”. Alsatian refers to Germanic varieties spoken in the Alsace region (Eastern France) and a rich dramatic production has existed in Alsatian for over two centuries, in spite of the oral nature of these language varieties. The goal of the project is to create the first large open access electronic corpus of Alsatian theatre; the targeted corpus volume is above 500,000 tokens and the corpus is being encoded according to the Text Encoding Initiative (TEI) Guidelines. The major traditions that have influenced Alsatian theatre most are German and French drama. Numerous computational projects have delivered insights into these traditions, but such analyses were so far impossible for Alsatian given lack of a corpus. The corpus under construction will help to perform the first macro- analytically inspired analyses of Alsatian texts and will facilitate a comparison with German and French theatre. Corpus selection criteria were discussed, besides automation methods in the TEI encoding workflow. The corpus anno- tations were also described: bibliographic metadata and a manual annotation of social variables for the corpus’ 2,386 characters (profession, social class, sex, age). Based on these annotations, generalizations about the evolution of social groups in the period were presented, which complement existing knowledge of the tradition. An interface for structured navigation of the corpus based on the annotations was also presented. Anne-Sophie Bories (University of Basel), in her presentation titled “Jokes without Humour”, started out from the protocol developed by Salvatore Attardo and Victor Raskin to annotate jokes, and proposed extending the use of their script-opposition system to describe incongruities in contexts that are not humorous. She works mainly on poetry, a genre largely concerned with the use of incongruity and ambiguity. Her study material is composed of three distinct corpora: two collections of poems by Guillaume Apollinaire: “Le Bestiaire, ou Cortège d’Orphée” (1911) and Alcools (1913), and Aimé Césaire’s lengthy poem “Cahier d’un retour au pays natal” (1939). Through statisti- cal analyses, she showed which script oppositions were most present in her 143Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia different corpora, using the following categorisations: [possible_impossible], [normal_abnormal], [actual_non-actual], [good_bad], [high_low-stature], [obscenity_non-obscenity], [money_no-money], [human_non-human], [life_death]. She also established which logical mechanisms (mapping, meta- phor, homophony, polysemy, garden path, personification, priming) and which narrative strategies (Q&A, narrative, description, address, list) were recur- rently deployed in her texts, with the aim of identifying trends specific to certain corpora and authors. This led her to find an overrepresentation of the [human_non-human] opposition in Aimé Césaire’s non humorous incongrui- ties, and to the surprising observation that, in a text primarily preoccupied with issues of racism and dehumanisation, this script opposition is mostly found in the depiction of things or concepts as humans, and not in the depic- tion of humans as objects. In their “Rhyming strategies in Estonian rap lyrics: a statistical view”, Maria-Kristiina Lotman and Rebekka Lotman (University of Tartu) demon- strated how rhyme returned to Estonian poetic space thanks to modern rap artists. While in Estonian literary poetry rhyming possibilities are very lim- ited by the language itself and rather seen as exhausted by the 19th and 20th century poets, in rap culture rhyming remains an important and evolving device. Using the corpus of four 19-20th century literary poets, the presenters showed that the selected Estonian canonic poets almost exclusively used full rhymes. By contrast, the samples of two contemporary rappers (Genka and Metsakutsu) showed a dramatically low proportion of full rhymes. In case of Genka’s rap verses, the full rhyme avoidance is compensated by the use of different types of imperfect rhymes, especially the ones with inexact syllable- end consonants. Metsakutsu’s verses showed an even more equal distribution over different types of inexact rhymes. The authors of the paper emphasised that this rich rhyming repertoire in rap verse might be caused not by the inner overcoming of rhyming limitations in Estonian but by the influence of rap versification in Western traditions. The keynote paper by Maciej Janicki, “Text similarity and alignment in the study of Finnic oral folk poetry”, introduced the methodology for detecting similar verse lines and song texts implemented in the exploratory research environment Runoregi. The multilingual database that aggregates large text collections (275,000 song texts) in various Finnic languages as well as the Runoregi environment have been created in the framework of the FILTER project (Formulaic intertextuality, thematic networks, and poetic variation across regional cultures of Finnic oral poetry). This project combines the fields of computer science and folkloristics and is funded by the Academy of Finland. The methodology for similarity detection developed by Janicki is based on two 144 Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv components. Verse similarity calculation is based on the cosine similarity of verses’ bigram vectors, which enables to also detect the similarity in case of slight orthographic and dialectal variation as well as in case of variants with different word order or slightly different word selection (e. g. addition of filler words). The methodology has been tested and proved to correspond quite well with manual similarity annotations. The second component of similar- ity detection is based on verse similarity and aims to detect the songs where similar verse lines occur in the same sequence. A weighted edit distance cal- culation implemented by Janicki was able to find the similarity between all the song pairs within only 64 hours, which is a short time taking into account a corpus volume of over 275,000 poems and 16 million pairs of similar poems. The results implemented in the FILTER database and Runoregi environment enable researchers to explore the types of relationships between similar texts in terms of song types, regions, individual singers etc. They also enable schol- ars to detect the exchange of texts between oral and literary traditions, and to build and assess the type indexes (or folkloristic classification of the texts), among other research activities. Yelena Helgadóttir (Árni Magnússon Institute for Icelandic Studies) pre- sented a paper entitled “Textual Variation and Representative Selection of Texts: The Case of Post-Medieval Icelandic þulur”. She discussed work towards developing a methodology for selecting representative examples of post-medi- eval Icelandic þulur, a folk poetry genre. Several challenges exist for effective text selection in this tradition in order to describe trends or for scholarly edit- ing, e. g. high variability of the texts in spite of their short length, and unclear boundaries between poems. These factors make it difficult to select examples that are generic enough while covering genre specificities appropriately. In their paper “Orality and versification in 16th century Hungarian epic poems: the epic formulas”, Petr Plecháč, Szilvia Maróthy, Levente Seláf, Margit Kiss and Villő Vigyikán examined the correspondence between oral perfor- mance and the degree of versification elaboration in Hungarian epic poems of the 16th century. Based on a previous pilot study, their work aims to assess a hypothesis suggesting that the few freely, orally composed poems, which have only accidentally survived in written form, have somewhat looser versification than the poems composed with more care, on paper. The enlarged corpus used in their present paper contains circa 1,530 pieces of Hungarian poetry, that are mostly strophic, sung, rhymed and syllabic. Their approach takes into account the uses of melodies related to each text thanks to annotating the tune, which allows them to propose graphical visualizations of these relations. To highlight the oral patterns within their corpus, their research seeks out the “epic formu- lae”, by using computational means such as word and lemma frequencies. A 145Plotting Poetry 5: Popular Voices, 4–6 July 2022, Tartu, Estonia typology of repetition can be outlined from these results (e. g. inner formu- lae, shared of external formulae, or formulaic schemes). Plecháč and al. also presented diachronic statistics regarding vala-rhymes (an archaic pattern of rhyming), proportion of unrhymed lines in dated and undated poems, and vocabulary richness: this latter aspect enables them to establish an evident growth of the lexical richness, a complexification of rhyming-technique, and some intertextual relationships during the first well-documented 60 years of the corpus genre. A comparison with a small corpus of Hungarian folk-ballads collected in the 19th century shows that lexical variety was much stronger in most epic poems of the 16th century, and also, that a stronger connection between some of the early epic poems and orality shall be proven by examin- ing lexical richness: the most repetition tends to indicate the oral character of the composition. In his presentation “Quantitative analysis of distribution of the nomina- tive form of the diminutive suffix -kene in Estonian runosongs”, Kaarel Veskis introduced his research results on the spread and use of diminutives in the Estonian song tradition. Diminutives are widely used in various languages of the world to express smallness, endearment, intimacy etc. Compared to contemporary colloquial language, runosongs include many more nouns in diminutive form; however, the same has been noted in other peoples’ oral poetry. The wide use of diminutives in runosongs is partly ascribed to the constraints of meter – it is an easy way to add additional syllables to a word. On the other hand, diminutives are systematically used when addressing someone or something – as if to contribute to a more positive acceptance by the inter- locutor. There is not much statistical information on the use of diminutives within different dialectal varieties. However, it has been noted to be a more prevailing way of expression in Southern-Estonian dialects, possibly as a result of the contacts with speakers of Russian, which is considered to use diminu- tives comparatively more than Estonian. The statistical analysis of diminutive word forms in Estonian runosongs demonstrated that the most widespread referents in diminutive word forms are girls, boys, other family members, as well as metaphoric figures denoting the same (e. g. little bird). The study also showed that nominative-case word forms with the most widespread disyllabic diminutive suffix were more frequent in the Eastern part of Estonia, diminish- ing towards the West, and that there were no significant differences between South and North. The alternative, monosyllabic ending and its variants were, however, impossible to retrieve with simple queries, thus the total spread of the use of diminutives still remains to be detected. As a whole, papers presented covered a variety of periods and geo- graphic areas (medieval Hungarian epic poetry, post-medieval Icelandic 146 Nils Couturier, Antonina Martynenko, Lara Nugues, Pablo Ruiz Fabo, Mari Sarv þulur, 19th-century poetry (Estonia, Finland, Germany, Russia) and popular drama (vaudeville, Alsatian theatre), 20th or 21st-century poetry and song in English, Spanish, Russian and French, besides Estonian rap. Beyond versifica- tion theory, theoretical frameworks for humour analysis and for remediation (adaptation to new media) underlay several of the papers. Methods ranged from descriptive statistics based on manual annotation to unsupervised machine learning, network analysis, speech analysis and synthesis. References Abeillé, Constanza 2013. Las formas musicales en la era digital: La crisis del álbum como principio organizador y la nueva función social de la canción. In: Signa: Revista de la Asociación Española de Semiótica 22, 185–204. https://doi.org/10.5944/signa.vol22.2013.6349 Attardo, Salvatore; Raskin, Victor 1991. Script theory revis(it)ed: Joke similarity and joke representation model. In: Humor: International Journal of Humor Research 4(3/4), 293–347. https://doi.org/10.1515/humr.1991.4.3-4.293 Albrecht, Roger 2010. Song of the Poet: Lost in Translation or Re-Discovered in a New Form? In: ETC: A Review of General Semantics 67(2), 177–190. https://www.jstor.org/stable/42579028 Badía Fumaz, Rocío 2020. Apropiaciones poéticas. Hacia un modelo de representación de la poesía en la música popular actual. https://canal.uned.es/ video/5fbf82dab6092302eb2352a4 (not accessible as of 30 December 2022). Bolter, Jay David; Grusin, Richard 1999. Remediation: Understanding New Media. Cambridge, MA: MIT Press. Hopkins, Gerald Manley; O’Leary, Sean F. 2005. The Alchemist: Gerard Manley Hopkins Poems in Musical Adaptations. Ingham, Mike 2013. “The true concord of well-tuned sounds”: Musical adaptations of Shakespeare’s sonnets. In: Shakespeare 9(2), 220–240. https://doi.org/10.1080/17450918.2012.705879 Jakobson, Roman 1960. Linguistics and Poetics. In: Sebeok, Thomas (ed.), Style in Language Cambrige: MIT Press, 350–377. https://doi.org/10.5944/signa.vol22.2013.6349 https://doi.org/10.1515/humr.1991.4.3-4.293 https://www.jstor.org/stable/42579028 https://canal.uned.es/video/5fbf82dab6092302eb2352a4 https://canal.uned.es/video/5fbf82dab6092302eb2352a4 https://doi.org/10.1080/17450918.2012.705879