Stylometrics

This article continues a historical review on the emergence and development of stylometrics which was published in the previous volume of “Structural and Applied Linguistics”. Being founded by Wilhelm Dittenberger, stylometrics expanded its geography in the Prst half of 20th century and enriched its methodological framework and interdisciplinary contacts. In Russia stylometrics appeared due to endeavours by N. A. Morozov. Later, the works by A. A. Chuprov, A. A. Markov, Andrei Bely, as well as the successes of Russian formal school and mathematical achievements of English biometrical school signiPcantly expanded the range of philological problems solved by means of mathematical methods. Later, there emerged new metric disciplines. In particular, artmetrics appeared Prst in the process of searching for culmination in verbal and musical text composition with help of the golden section. And then it was further developed in the search for criteria for measuring text aesthetics.
Данная работа является продолжением статьи, опубликованной в предыдущем выпуске сборника «Структурная и прикладная лингвистика». В первой половине XX века расширилась география стилеметрии, основанной В. Диттенбергером, обогатилась и ее методическая база и междисциплинарные контакты. Стилеметрия пришла в Россию благодаря усилиям Н. А. Морозова. В работах А. А. Чупрова, А. А. Маркова, Андрея Белого, в успехах русской формальной школы, в математических достижениях английской биометрической школы существенно расширился круг филологических проблем, решаемых с помощью математических методов. Появились новые метрические дисциплины, в частности возникла искусствометрия — сначала при поиске кульминации в композиции вербального и музыкального текста с помощью золотого сечения, а затем при поиске критериев измерения эстетичности текста.

The aim of this study is to perform an evaluation of ninety-nine Spanish Golden Age theatre plays of undisputed authorship using the R package Stylo, the stylometric analysis tool developed by Eder, Rybicki, and Kestemont (2016). In this paper I will determine which algorithms obtain best results on authorial classification (statistic method, MFW, culling, and word n-grams). I will also evaluate the text length at which stylometry begins to be an effective diagnostic tool for authorship attribution in our corpus. This cross-validation evaluation can serve future analysis of similar corpora and will show the possibilities of applying stylometry to Spanish Golden Age theatre, which presents many cases of dubious authorship.

An advantage of corpus-based language studies is that global descriptions of linguistic texts can be obtained by examining a broad range of linguistic features. However, multivariate statistical techniques are required to analyze the multiple linguistic features found in a number of texts. This study compared the strengths and weaknesses of several multivariate statistical techniques, thereby demonstrating the effectiveness of using heat map with hierarchical clustering as a powerful method for visualizing multivariate data. Explanations are also provided for how these techniques can be used in the R programming language as well as indicating how the results obtained can be interpreted.* 1. Introduction An advantage of corpus-based language studies is that global descriptions of linguistic texts can be obtained by examining a broad range of linguistic features. However, multivariate statistical techniques are required to analyze the multiple linguistic features found in a number of texts. These techniques can identify the complex interrelationships among linguistic features and texts, as well as the associated patterns between linguistic features and texts. The appropriate-ness of these techniques for analyzing language has been confirmed in the past few decades. For example, Burrows (1987) employed principal components analysis to analyze the idiolects of the major characters in Jane Austen's novels. Biber (1988) applied factor analysis to describe the linguistic characteristics of speech and writing. Nakamura and Sinclair (1995) used Hayashi's Quantification Method Type III, a method that is mathematically identical to correspondence analysis, to compare the frequencies of the collocates of the word woman in four components of the Bank of English, a representative subset of the COBUILD Corpus. Moreover, Hoover (2003) used cluster analysis to distinguish texts written by different authors using commonly occurring words in these texts. In the present study, I compared the strengths and weaknesses of several multivariate statistical techniques, thereby demonstrating the effectiveness of using heat map with hierarchical clustering as a powerful method for visualizing multivariate data. I also explain how these techniques are used with R (Ihaka and Gentleman 1996) and how the results can be interpreted.

This article analyses the use of film style in Rashomon (1950) to determine if the different accounts of the rape and murder provided by the bandit, the wife, the husband, and the woodcutter are formally distinct by comparing shot length data and using multiple correspondence analysis to look for relationships between shot scale, camera movement, camera angle, and the use of point-of-view shots, reverse-angle cuts, and axial cuts. The results show that the four accounts of the rape and the murder in Rashomon differ not only in their content but also in the way they are narrated. The editing pace varies so that although the action of the film is repeated the presentation of events to the viewer is different each time. There is a distinction between presentational (shot scale and camera movement) and perspectival (shot types) aspects of style depending on their function within the film, while other elements (camera angle) fulfil both these functions. Different types of shot are used to create the narrative perspectives of the bandit, the wife, and the husband that marks them out as either active or passive narrators reflecting their level of narrative agency within the film, while the woodcutter’s account exhibits both active and passive aspects to create an ambiguous mode of narration. Rashomon is a deliberately and precisely constructed artwork in which form and content work together to create an epistemological puzzle for the viewer.

Монография посвящена актуальному направлению изучения проблем стилистики методами математической лингвистики. В книге анализируются роль и место стилеметрии в филологических исследованиях, обсуждаются ее познавательные принципы и задачи, а также ее взаимодействие с другими измеряющими дисциплинами (наукометрией, искусствометрией, психометрикой и т. п.). Книга рассчитана на широкий круг специалистов в области текстологии, источниковедения, документалистики, стилистики, квантитативной лингвистики и автоматической обработки текста.

This article (co-authors: David J. Wrisley and Liliane Dulac) attempts to identify the different styles employed by Christine de Pizan in all of her works using empirically based stylometric analysis. The results, at once heuristic but also statistically significant, suggest as well that besides Christine de Pizan, other medieval French authors such as Jean de Meun, Froissart and Gerson, have at least two, if not more, stylometric footprints. The methods employed and the ramifications of the results are discussed briefly as well. Résumé Il s'agit d'un essai d'identifier les différents styles employés par Christine de Pizan dans l'ensemble de sa production littéraire à partir d'une analyse stylométrique empirique. Les résultats, à la fois heuristiques mais aussi statistiquement signi-ficatifs, suggèrent aussi qu'en plus de Christine de Pizan, d'autres auteurs fran-çais médiévaux comme Jean de Meun, Froissart et Gerson, possèdent au moins deux « empreintes » stylométrique, sinon encore plus. Les méthodes employées et les ramifications des résultats obtenus sont brièvement discutés.

This article presents an event probability versus event frequency notation for the linguistic power law, lifting several imperfections of the commonly used Zipf-Mandelbrot expression. The improved notation fulfills the conditions to normalize a power law by data transformation, allowing i.a. authorship attribution problems to be solved by Gaussian statistics.

Seit der „Rückkehr des Autors“ gewinnen Fragen literarischer und wissenschaftlicher Autorschaft weiter an Bedeutung. Der vorliegende Band bereitet die seitdem erschienene Forschungsliteratur anschaulich auf und erschließt in historischen und systematischen Fallstudien Theorien und Praktiken der Autorschaft zwischen Hermeneutik und Poststrukturalismus, Narratologie und (Auto-)Fiktionstheorie, Gattungswissen und Inszenierungsstrategien.

Η παρούσα μελέτη ασχολείται με τις υφολογικές εκφάνσεις των μηνυμάτων του Έλληνα Πρωθυπουργού, υπό τη μορφή διαγγελμάτων, κατά τη διάρκεια της καθολικής απαγόρευσης κυκλοφορίας στην ελληνική επικράτεια την περίοδο της πανδημίας του κορονοϊού. Μεθοδολογικά υιοθετήθηκε κυρίως ο όρος σωματοκειμενικά καθοδηγούμενη ποσοτικοποίηση των δεδομένων (αλλά χρησιμοποιήθηκε μερικώς και ο όρος κειμενικά καθοδηγούμενη ποσοτικοποίηση), αναδεικνύοντας ενδιαφέροντα στοιχεία ύφους και γλώσσας, φωτίζοντας ιδιαίτερες πτυχές του πολιτικού λόγου στις πρωθυπουργικές δηλώσεις προς τους Έλληνες πολίτες σε μία ιδιαιτέρως κρίσιμη κατάσταση, που εξελισσόταν ραγδαία παγκοσμίως. Η ποσοτικοποίηση των αποτελεσμάτων και η ανάλυσή τους, σε μία πρώτη προσέγγιση σύμφωνα με τις αρχές της υπολογιστικής υφομετρίας, επικεντρώθηκε κυρίως στη μελέτη συχνότητας εμφάνισης λέξεων στο σώμα κειμένων. [CITATION: Μαθιουδάκης Ν. 2021. «“Μένουμε σπίτι”: Υφογλωσσικές προσεγγίσεις των διαγγελμάτων του Πρωθυπουργού». Στο Ν. Τσιτσανούδη-Μαλλίδη & Δ. Βαρδαβάς (επιμ.), Γλώσσα, ηθική και ιδεολογία στα ΜΜΕ, σσ. 135-180. Αθήνα: Εκδόσεις Πεδίο.]

Данная работа является первой частью более крупной двухчастной статьи под заявленным названием. В предлагаемой части рассматривается эпоха первичного становления стилеметрии. Рассматривается широкий междисциплинарный контекст, сформированный другими измеряющими общественных дисциплинами: антропометрией, биометрией, психометрикой, искусствометрией и др.

In 1833, the year prior to his death, Coleridge stated that ‘I never put pen to paper as translator of Faust’, but Burwick and McKusick believe that he did just that.11 The editors’ account of Coleridge's alleged translation of Faust has divided critics of the Clarendon book. Susanne Schmid, for example, reports ‘the editors offer overwhelming evidence’ that convinces her that the ‘author is extremely likely to have been Coleridge’ (Schmid, 1–2), while Paulin et al., refuting the ‘discovery’ of a work by Coleridge, conclude that ‘this volume is not what it appears to be’ (Gentleman, 31). I am sympathetic with both positions, but I agree mostly with Paulin et al.: while the claim that Coleridge authored Faustus is enticing, and is backed by a lengthy critical introduction, scrutiny of the evidence presented compromises the editors’ hypothesis.

Μαθιουδάκης, Ν. (2015). Η γλωσσική ποιητική ταυτότητα του Καβάφη και του Καζαντζάκη: μια συγκριτική μελέτη. Στο Κ. Α. Δημάδης (επιμ.), Πρακτικά του Ε΄ Ευρωπαϊκού Συνεδρίου Νεοελληνικών Σπουδών/ Proceedings of the 5th European Congress of Modern Greek Studies. Θεσσαλονίκη, 2-5 Οκτωβρίου 2014. τομ. Γ΄, σσ. 319-341. Αθήνα: Ευρωπαϊκή Εταιρεία Νεοελληνικών Σπουδών (ΕΕΝΣ).

PDF AVAILABLE HERE: https://users.utu.fi/lmputk/

In this dissertation, I study indirect translation. Indirect translation is a translation made from a translation, and it may include compilative and/or collaborative practices, that is, many source texts may be used, or the translator may collaborate with someone.

The case study in this dissertation consists of 22 novels translated from Modern Greek into Finnish between 1952 and 2004. Indirect translation is studied from the perspectives of status (what translations are claimed to be), origin (what the genesis of a translation was like), and features (what kind of linguistic features translations have), and the findings are contextualized by examining the culture-bound norms governing the production of translations. The framework, proposed by Delabastita (2008), stems from descriptive translation studies (Toury 1995/2012).

In this study, mixed methods are used. One the one hand, the materials are mixed, comprising paratexts, translator bios, the 22 novels, and a corpus of translated and non-translated Finnish novels. On the other hand, both quantitative and qualitative methods are employed, including paratextual analysis, methods of genetic translation criticism and textual criticism, and corpus research tools.

In one article of this study, the analysis shows that, although bibliographical metadata offers information on the status of translations, this information is not always in line with whether the translations were done directly or indirectly. In another article, the origins of a compilative translation are studied to uncover how translators work when using several source texts. In the third article, the study of features suggests that the linguistic profile of indirect translations is different from those of direct translations and non-translated Finnish texts. In the fourth article, the study of norms shows that, in Finland, attitudes towards indirect translation are negative, and translators may use compilative and collaborative strategies to respond to criticisms.

The mixed-methods approach allows us to gain a holistic picture of indirect translation. In addition, the present study shows that indirect translation challenges the idea of the source text–target text relationship as exclusive, binary, and unidirectional.

There is a long history of stylometric approaches to questions of authorship of the Íslendingasögur, most particularly the question of whether Snorri Sturluson wrote Egils saga. Wieselgren (1927) used the frequencies of a series of syntactic features to substantiate the argument that Egils saga shared an author with Snorri’s Edda. Although the value of the particular features Wieselgren used was undermined (Nordal 1933), later researchers—van den Toorn (1959) and particularly Hallberg (1962) and West (1980)—have demonstrated conclusively that Egils saga is closer to Snorri’s known works than the other Íslendingasögur are, and closer than to Snorri’s known works than it is to the other Íslendingasögur. Outside the Old Norse-Icelandic sphere, such methods, particularly in the form proposed by Mosteller & Wallace (1964), have had many successes, settling the case of disputed authorship of the Federalist Papers, the Junius letters (Ellegård 1962), proving that Robert Galbraith is really J.K. Rowling (Juola 2013), and confirming that Shakespeare authored Double Falsehood (Boyd & Pennebaker 2015). In this paper, rather than attempt to answer a single question of shared or distinct authorship for some pair of texts, I use statistical methods to map the stylistic geography of the Íslendingasögur. Are there any other clusters of texts as similar as Egils saga and Snorri’s works? Can we see stylistic change over time, or stylistic differences between different types of narrative?

The aim of this article is to study the complete poetry of Luis de Góngora using a contrastive analysis. To this end, the poems, which do not pose any attribution problem, published in digital format by the Observatoire de la vie littéraire (OBVIL) have been selected (with the exception of theatre) and divided into three parts on the basis of a chronological criteria. Next, the tool Stylo has been used to obtain three lists of preferred and avoided words in the poems dated between 1580 and 1610 (sub-corpus A), 1611 and 1618 (sub-corpus B) and 1619 and 1625 (sub-corpus C). The results obtained allow us to characterize the contrast between these three sub-corpus, suggest an evolution of Góngora’s aesthetics and confirm a change produced in the poems written between 1611 and 1618. However, while Stylo may be useful for detecting a trend or dominant aesthetics, we believe that the data generated is unsatisfactory for analyzing the style at the scale of the sentence.

This article focuses on some of Robert E. Howard's stylistic tendencies and qualities. It also offers a general introduction to my theory of "Subjective Stylometry"—done by the enlightened "tagging" of a text through close reading and observing genre, rhetorical, and formalist attributes. This is what I call the texts "DNA" -- "Distinctive Narrative Attributes."

Popular songs can be regarded as a fine representation of modern society and culture. In particular, the lyrics of popular songs are the most important aspect for understanding the sense of values and linguistic sensitivity in a given generation and community. The purpose of the present study is to investigate the chronological variation of popular Japanese songs using stylometric techniques. This study draws on the lyrics of 858 songs, which appeared on the Oricon annual top 20 single hit chart between 1976 and 2015. The linguistic features investigated in this study include five different types of lexical indices, namely (a) number of words, (b) parts-of-speech, (c) word types, (d) character types, and (e) vocabulary level. Multiple regression analysis was conducted to explore the chronological change in the frequencies of lexical indices. The results showed that the frequencies of word types and character types dramatically changed before and after 1990. Moreover, the usages of auxiliary verbs as well as lower level vocabulary became more prominent, whereas the frequencies of adjectival nouns and conjunctions decreased. The findings suggest that a turning point in cultural trends corresponds with the historically significant political and economic events, such as the end of the Showa era and the burst of the bubble economy.

On the Nominal Style of Pre-Election Statuses The paper forms part of a broader research aimed at the linguistic and communicative specific features of the pre-election social-media communication. By means of orientation upon the so-called nominal style and within the context of the determinants of the social-media environment to try to make transparent the characteristics of status as a specific textual type, and point out some of the features of the particular communication register. As to its methodology, the paper draws on the quantitative-linguistic methods that enable to base the final interpretations on exact data. The database of the material is formed by nearly 2 000 statuses of the selected political parties that were published on the social network Facebook in the year 2020 in the pre-election period.

The quantitative analysis of writing style (stylometry) is becoming an increasingly common research instrument in philology. When it comes to medieval texts, such a methodology might be able to help us disentangle the multiple authorial strata that can often be discerned in them (issuer, dictator, scribe, etc.). To deliver a proof of concept in 'distant diplomatics,' we have turned to a corpus of twelfth-century Latin charters from the Cambrai episcopal chancery. We subjected this collection to an (unsupervised) stylometric modelling procedure, based on lexical frequency extraction and dimension reduction. In the absence of a sizable 'ground truth' for this material, we zoomed in on a specific case study, namely the oeuvre of the previously identified dictator-scribe known as 'RogF/JeanE.' Our results offer additional support for the attribution of a diplomatic oeuvre to this individual and even allow us to enlarge it with additional documents. Our analysis moreover yielded the serendipitous discovery of another, previously unnoticed, oeuvre, which we tentatively attribute to a scribe-dictator 'JeanB.' We conclude that the large-scale stylometric analysis is a promising methodology for digital diplomatics. More efforts, however, will have to be invested in establishing gold standards for this method to realize its full potential.

**NB: This paper is taken from the draft of my article 'Quantitative analysis of sound in a short horror film' above.

Audio segmentation comprises a set of techniques for analysing the features of audio signals, including motion picture soundtracks. Parsing the structure of film audio allows us to identify scenes and change points, features in the audio envelope (attack, decay, sustain, release), the distribution of sound energy, and the presence of affective events in a soundtrack. Despite the availability of a wide range of software capable of audio analysis (e.g. Python, R, Sonic Visualiser, Audacity, Speech Filing System, Raven Lite, etc.) statistical analyses of cinematic style have yet to apply these methods to the formal analysis of film soundtracks systematically. In this paper I demonstrate the analysis of film sound using the time contour plot of generated by the R package seewave. The time contour plot is a graphical method for visualising the temporal structure of sound energy based on the short-time Fourier transform of a signal and allows analysts to identify interesting features in a film’s soundtrack that warrant further examination and to communicate is a straightforward manner that is easy to comprehend. Applying this method to the soundtrack of Behold the Noose (Jamie Brooks, 2014), a short horror film, I show that the sound energy in the film increases non-linearly in creating of a state of heightened anxiety in the viewer.

Old English poems are difficult to date, largely anonymous, and written in a formulaic idiom, which makes it difficult to establish the authorship and intertextual relationships among the various poems. This essay employs computational stylometry—specifically, n-gram analysis—and network analysis to address this problem. It proposes methods for screening out chance n-grams and for employing n-grams of various lengths within a single analysis. This analysis shows that in Old English poetry formulaic diction exists at the level of idiolect, sociolect, and language. The signed poems of Cynewulf evince idiolectal characteristics; along with Andreas, Christ III, The Phoenix, and Guthlac A and B, they also form a poetic sub-tradition (sociolect).

Cet article a pour objectif de quantifier les spécificités syntaxiques gongorines par rapport à l'usage poétique de son temps, par extraction de séquences grammaticales récurrentes (sgr, ou n-grams de pos-tags) et comparaison avec un corpus-témoin. Après une clarification terminologique autour de la notion de motif, nous avons entamé une ca-ractérisation stylistique de l'oeuvre gongorine, grâce à l'observation de la séquence la plus surreprésentée chez notre auteur. L'analyse qualitative nous a conduit à identifier certains effets de sens récurrents qui feraient de cette séquence un motif à proprement parler, celui du rapprochement expressif entre entités hétérogènes, et, dans les cas où la séquence in-tègre un complément antéposé, celui de la recréation poétique du processus d'intellection du monde, allant du particulier au général. Zusammenfassung : Die Besonderheiten der gongorinischen Syntax werden quantita-tiv im Vergleich zum poetischen Sprachgebrauch seiner Zeit untersucht. Dazu werden wie-derkehrende grammatikalische Sequenzen (sgr, n-grams, pos-tags) mit einem ähnlichen zeitgenössischen Korpus verglichen. Nach einer theoretischen Betrachtung des Motivbe-griffs, wird versucht, das Werk Góngoras stilistisch zu beschreiben, indem eine bestimmte überrepräsentierte Sequenz genauer betrachtet wird. Die zeitgleiche qualitative Analyse ermöglicht es, bestimmte wiederkehrende Sinneffekte zu erkennen, die diese Sequenz als Motiv gelten lassen. Es geht dabei um ein ausgesprochenes Annähern von heterogenen Entitäten, und um eine poetische Reflexion über die Intellektualisierung der Welt (in den Fällen, in denen die Sequenz ein vorangestelltes Komplement integriert), immer vom Be-sonderen auf das Allgemeine schließend.

Variation among human translations is usually invisible, little understood, and under-valued. Previous statistical research finds that translations vary most where the source items are most semantically significant or express most 'attitude' (affect, evaluation, ideology). Understanding how and why translations vary is important for translator training and translation quality assessment, for cultural research, and for machine translation development. Our experimental project began with the intuition that quantitative variation in a corpus of historical retranslations might be used to project quasi-qualitative annotations onto the translated text. We present a web-based system which enables users to create parallel, segment-aligned multi-version corpora, and provides visual interfaces for exploring multiple translations, with their variation projected onto a base text. The system can support any corpus of variant versions. We report experiments using our tools (and stylometric analysis) to investigate a corpus of forty German versions of a work by Shakespeare. Initial findings lead to more questions
than answers.