This is a stylometric study of human and machine literary translations. The research seeks linguistic measures that might help distinguish human- and machine-translated literary novels. The corpus consists of 2600 texts (originals, human and machine translations) translated from Italian, Spanish, Serbian, Croatian, Greek, Japanese, Danish, Norwegian, Swedish, Czech, Hebrew, Hungarian, Dutch, Chinese, Korean, Finnish, and German into Polish, Lithuanian, English, and French. The study examines differences in machine translation of analytical and inflectional languages and considers variables such as sentence and word length, lexical density, and the use of different punctuation marks.

We describe a new feature for authorship attribution and present the results of some preliminary studies of how practical it is. In our presentation we will further explore some advantages of this over more typical features such as character clusters.

This study investigates whether large language models (LLMs) recognize authorial style across English-to-Japanese translation and aims to distinguish memorization-based from abstraction-based explanations of model performance. In an early experiment conducted shortly after the release of ChatGPT, the model attributed a passage from Cormac McCarthy's The Road, presented in Japanese translation, to Haruki Murakami (Author 2023). Since then, frontier models have dramatically improved, yet freely accessible AI tools still produce confident misattributions. This persistent gap raises a broader question: does authorial style survive translation, and when models succeed, are they recognizing style or recalling memorized associations?

Focusing on nineteenth-century American narrative prose, the study tests both frontier commercial models and freely accessible AI tools on a corpus of six works by Irving, Poe, Hawthorne, Douglass, Melville, and Alcott. Each excerpt is prepared in three conditions: the English original (EN), a published Japanese translation (JP-HT), and a machine translation (JP-MT). Because the English originals are in the public domain and widely available online while their Japanese translations largely are not, the three conditions create a scale of likely data exposure.

The experimental design combines LLM-based attribution under controlled prompting, a masked condition replacing culturally specific markers with neutral placeholders, and a memorization test using n-gram overlap metrics. Multilingual Sentence-BERT provides a complementary embedding-based perspective. The study places particular emphasis on the qualitative analysis of model errors, examining misattribution patterns through the lens of literary history and reception.

This paper examines the identity of Eva Bloch, credited as the translator of L’Immortalité (1990), the French version of Nesmrtelnost by Milan Kundera. Beyond this single translation, no trace of Bloch can be found in bibliographic or archival records. Given Kundera’s well-known tendency to closely supervise and revise translations of his works, many scholars have suggested that Eva Bloch may have been a pseudonym and that L’Immortalité represents Kundera’s first attempt at a self-translation into French.

Because archival sources remain inaccessible, the paper approaches this question through stylometric analysis. While authorship attribution of original texts has proven methodologically robust, identifying a translator presents additional challenges, particularly in cases where authorial intervention in the translation process was substantial. Building on recent research in translator authorship attribution, the study employs the stylo package in R to compare L’Immortalité with Kundera’s acknowledged French-language novels and essays, as well as with French translations of his Czech works by other translators.

The analysis combines cluster- and distance-based methods based on the distribution of the most frequent words with selected measures of stylistic complexity, including length-controlled lexical diversity, average sentence length, and bi-gram entropy. The aim is to determine whether L’Immortalité aligns stylistically with Kundera’s later French works, with known translators, or suggests a more collaborative authorship. The case thus serves both as a test of the pseudonym hypothesis and as an experiment in applying stylometry to the complex process of literary translation.

The 20th-century history of Latvia is a history of forced displacement, with approximately 175,000 Latvians forming global communities from the USA to Australia (Plakans 2021, 64). This "Latvia outside Latvia" created a robust media landscape comprising journals and magazines that served as a surrogate public sphere for those severed from their homeland. These were European voices echoing far beyond Europe’s borders, preserving their heritage through the press. To analyze these discourses, we employed BERTopic for topic modeling. While the topic modeling approach has long been criticized for low reproducibility and interpretability, frameworks like BERTopic are based on enhanced semantic coherence and are potentially better. In this study we check whether high coherence scores actually indicate meaningful topics. Analyzing about 7,000 English-language articles from 27 Latvian diaspora newspapers (1940s–2000s), we compare four n-gram configurations (unigram, bigram, trigram, and combined n-grams) using two coherence metrics (C_V and C_NPMI). Our findings reveal a paradox: bigram models achieving near-perfect coherence scores in many cases captured just formulaic content – telephone directories, contact information, and administrative listings – rather than historical themes. The combined n-grams model achieved optimal balance, while trigram models showed weakest characteristics. Our findings indicate that topic modeling captures sign patterns and statistical co-occurrences rather than semantic content directly. High coherence can signal structural regularity without guaranteeing historical meaningfulness.

The study proposes a large-scale stylometric comparison between human-authored English novels and machine-generated texts produced from schematic summaries of those novels. Its main hypothesis is that both human authors and large language models (LLMs) exhibit distinct, identifiable stylistic profiles, and that these profiles show systematic patterns of similarity and divergence. The research also examines whether author gender influences stylometric clustering.

Using a corpus of about 1,000 nineteenth- and twentieth-century novels, the study generates abstracts and then produces new texts from these using four different LLMs. Stylometric analysis is conducted with TF-IDF features, k-nearest neighbors, UMAP visualization, and Burrows’s Delta to compare human and machine texts across dimensions of authorship, origin (human vs. machine), and gender. The design controls for narrative content in order to isolate stylistic differences.

The expected outcome is evidence that LLMs develop their own recognizable stylistic tendencies, which may align with or diverge from human authors’ styles and potentially form independent stylistic clusters. The study aims to shift stylometry beyond plagiarism detection toward understanding machines as possible stylistic agents, contributing to debates in digital humanities and NLP about authorship, stylistic identity, and the nature of style in artificial language production.