Preliminary proceedings now available

Programme

09:00 – 09:05 Opening notes - Sheila Castilho

09:05 – 10:00 Keynote Speech: Dr Marzena Karpinska

Can readers enjoy LLM-based machine translation?

A successful literary translation is not only faithful to the original but also conveys the author's voice, style, and, perhaps crucially, entertains the reader. As large language model (LLM)-translated literature leaks into the trade and self-publishing industries and reaches the real readership, we need to understand how it shapes readers' perception of the work, and how well LLMs can carry the style and unique voice of the author into another language. In this talk, I'll argue that while LLM-translated literature may be "fine," it is often characterized by stylistic flatness and a lack of creativity.

I will first situate the problem within recent work on diversity and style in open-ended generation before turning back to a reader-centered study of literary machine translation. I will discuss how avid readers perceive LLM-based literary translation compared to a professionally published human translation. I will share what surprised us, including how acceptable readers found the machine output and where it nonetheless fell short. I will close with a reflection on these findings and what they mean for the future development of machine translation systems and the evaluation of style in machine-translated content.

10:00 – 10:30 Oral Presentations: Style in GenAI-Generated Literary Translation

Toying with Style: Can GenAI Mimic a Literary Translator’s

Voice? – Beniamin Sopot and Dorothy Kenny

10:30 – 11:00 Coffee break

11:00 – 12:30 Oral Presentations: Style in GenAI-Generated Literary Translation (cont.)

Emotion Profiling in LLM-Based Literary Translation:nSystematic Shifts Across MT and Post-Editing – Antonio Castaldo, Johanna Monti and Sheila Castilho

Retranslation at the Intersection of Style, Machines, and Plagiarism – Hüseyin Emir Akdağ, Yusuf Mert Aygün, Mehmet Sǎhin, Ena Hodzik, Sabri Gürses and Tunga Güngör

12:00 – 12:30 Oral Presentations: Comparative Studies on Style in NMT, LLMs and HT

Lexical and Syntactic Diversity: Still Lost in Machine Translation? - Lise Volkart and Pierrette Bouillon

12:30 – 13:30 Lunch break

13:30 – 14:30 Keynote Speech: Prof Carl Vogel - 

Theory of Style in Language

Contemplation of stylistic features of the products of large language models qua intelligent translators invites reflection on the mechanistic -- stochastic if not deterministic -- nature of style in artificial and natural language production.  Compositionality in style may be addressed with the same rigour as in the computation of sentence and discourse meaning in formal semantic theory of natural language.  Artificial systems driven by large language models are imbued with human stylistic preferences twice over -- firstly, in distributions of behaviours visible in the linguistic data on which the models are trained and secondly, in the design and implementation strategies that favour systematicity in output behaviours.  Features of systematicity are computationally discoverable.

14:30 – 15:00 Oral Presentations: Comparative Studies on Style in NMT, LLMs and HT (cont.)

Tracing Style in English–Arabic Translation: A Stylometric Comparison of Human and Machine Outputs – Nooredeen Awwad, Ebtihal Enfes and Kolawole John Adebayo

15:00 – 15:30 Coffee break

15:30 – 17:00 Oral Presentations: Comparative Studies on Style in NMT, LLMs and HT (cont.)

Style and Terminology in Commercial Flows: Mixing APE with Iterative Feedback - Marthe Lamote, Ewoenam Tokpo, Tom Vanallemeersch, Sara Szoc and Koen Van Winckel

Lexical Variation in English–Italian News Translation: A Comparative Study of Google Translate and ChatGPT – Aurora Trapella, Lieve Macken and Alessandra Molino

The Style of Machines: A Stylometric Study of LLM Generation and Translation - Natalia Resende and Sheila Castilho

17:00 – 17:15 Closing remarks - Natalia Resende