Invited Speakers
We are pleased to announce that we will have the following plenary speakers confirmed for their participation at the UCCTS2025 conference in Hildesheim:
Elke Teich, Saarland University in Germany, with the talk entitled "Translation as Rational Communication"
Translation has been shown to exhibit specific linguistic traces left in the output by the process of cross-lingual transfer due to simplification, explicitation, normalization, shining-through etc., known as “translationese”. While translationese effects are very well covered at a descriptive level (languages, linguistic features) for both translation and interpreting, there are much fewer attempts at modeling their deeper underlying mechanisms.
In my talk, I propose to turn to rational communication for an explanatory framework of important aspects of translation and language variation more widely. According to rational communication, speakers’ overarching goal in linguistic interaction is communicative efficiency. The prerequisite for efficiency is predictability in context, also known as surprisal and defined formally in Shannon's information theory. Information theory provides a powerful approach to address translationese for two reasons. First, surprisal is proportional to cognitive effort, i.e. highly surprising items incur higher processing costs (indexed e.g. by longer reading times) and vice versa. Information theory may thus act as a linking theory between language use and cognitive explanations of linguistic choice. Second, information theory lends itself as an analytic tool for exploring phenomena of linguistic variation on the basis of computational language models. Illustrating these two uses of information-theoretic models, I will present selected studies on translation and interpreting in the language pair English-German conducted in project B7 of the Collaborative Research Center (CRC) 1102 “Information Density and Linguistic Encoding” (https://sfb1102.uni-saarland.de/). These studies are joint work with Mariia Kunilovskaia, Ekaterina Lapshinova-Koltunski, Christina Pollkläsener and Heike Przybyl.
Dylan Glynn, Université Paris 8, Vincennes - Saint Denis, with the talk entitled "Behavioural Profiles for Contrastive Linguistics: A context-dependent qualitative-quantitative approach to equivalence"
In both contrastive linguistics and translation studies, the notion of cross-language equivalence is fundamental. From cognitive science to language typology, the ability to compare symbolic associations within a language, across different languages (and even between individuals) is sine qua non for making descriptive generalisations. However, if one adopts a function-driven, usage-based model of language, the question of how form-meaning pairs can be compared across languages becomes empirically problematic.
This talk will present the Behavioural Profile Approach and demonstrate how it can be employed to identify context-dependent cross-language equivalences. It will focus on:
- Manual Annotation – heuristics proposed to improve reproducibility
- Data Analysis – statistical methods proposed to better model complexity
Three case studies will be outlined. The first examines equivalences in the lexical field of anxiety in Chinese and French; the second examines grammatical constructions of Futurity in Norwegian and English. The third extends the methodology to examine discursive equivalence in multimodal data, specifically boasting strategies on Instagram in French and Russian.
Photo taken from: 2023.aclweb.org/committees/program/
Christian Hardmeier, IT University of Copenhagen, with the talk entitled "Chasing discourse across languages".
Discourse-level linguistic phenomena such as pronominal anaphora and the use of discourse connectives are realised differently across languages. This is due to language-specific preferences and constraints or, in the case of translated texts, to the translation process. My talk is a synthesis of a series of studies of discourse phenomena across European languages and Arabic with a variety of methods on a continuum from fully data-driven corpus analysis to analysis of manually annotated corpora and studies with human participants. I will present some results from our studies and contrast the various methods used across studies with their strengths and limitations.
