Search

Die vorliegende Arbeit beschäftigt sich mit der KI-gestützten (maschinellen) Übersetzung von Audiodeskriptionsskripten (AD-Skripten) in der Sprachrichtung französisch-deutsch. Da nach wie vor die meisten AD-Skripten pro Sprache erstellt werden und keine (maschinelle) Übersetzung stattfindet, soll untersucht werden, welche Qualität die KI-gestützte (maschinelle) Übersetzung von AD-Skripten in der genannten Sprachkombi-nation erreicht. Zunächst wurde das AD-Skript der Sendung «Passe-moi les jumelles» mit Chat-GPT-4 aus dem Französischen ins Deutsche übersetzt. Dabei entstanden drei Versionen: eine ohne Bildinput, eine mit Bildinput und eine durch ChatGPT selbst post-editierte Version. Der Grossteil der gewonnenen Daten wurden quantitativ anhand eines Direct Assessments von professionellen Audiodeskribierenden bewertet. Die Ergebnisse zeigten keinen signifikanten Unterschied zwischen den drei übersetzten Versionen. Die Adequacy der Übersetzungen wurden von den Teilnehmenden als hoch, die Fluency mittelmässig und die Eignung für die Textsorte AD hingegen als tief eingestuft. Einen Teil der Daten wurden an-schliessend qualitativ anhand der DQF-MQM-Methode analysiert, um die Qualität der übersetzten AD-Segmente zu eruieren. Es stellte sich heraus, dass insbesondere Probleme bei der Fluency und dem Stil auftraten. Die Auswertung steht in Abhängigkeit von verschiedenen Faktoren wie der Sprachkombination, des gewählten Übersetzungstools und des Sendungsinhalt, daher sind weitere Forschungen in diesem Bereich erforderlich., The present thesis examines AI-assisted (machine) translation of audio description scripts (AD scripts) from French to German. Since most AD scripts are still created for individual languages without (machine) translation, this study aims to assess the quality of AI-assisted (machine) translation for this specific language pair. Initially, the AD script for the program "Passe-moi les jumelles" was translated from French to German using ChatGPT-4. This process produced three versions: one without visual input, one with visual input, and one post-edited version by ChatGPT itself. Most of the obtained data were quantitatively evaluated through direct assessment by professional audio describers. The results showed no significant difference among the three translated versions. Participants rated the adequacy of the translations as high, the fluency as moderate, and the suitability for the AD text type as low. Subsequently, a portion of the data was quali-tatively analysed using the DQF-MQM method to determine the quality of the translated AD segments. It was found that issues primarily arose in fluency and style. The evaluation depends on various factors such as the language pair, the chosen translation tool, and the content of the program, indicating the need for further research in this area.

...

In their capacity as language experts, interpreters are sometimes expected to deliver target texts that are better than their underlying source text, especially when the latter was produced by a speaker in a language that is not their L1. The spread of global English has given rise to ever more occasions when interpreters encounter non-L1 speakers of English as a lingua franca (ELF). The question as to whether or not interpreters try to optimise those speakers’ input is addressed by applying Relevance Theory (RT) as a conceptual and methodological framework that helps to understand interpreters’ needs or readiness to augment relevance for their audience. The paper builds on data from the larger project CLINT (Cognitive Load in Interpreting and Translation). The 56 renditions by all 28 professional interpreters participating in the project’s interpreting part of two original ELF speaker texts and their edited versions are analysed with a view to the enrichment processes undertaken by the interpreters. A comparison of the renditions of the original versus edited versions of the two texts shows that interpreters do engage in such processes considerably more when rendering ELF texts, especially if they are technical in nature. Determining whether or not these interventions lead to actual cognitive ffects in terms of information gains on the part of the audiences or to increased cognitive effort on the part of the interpreters requires additional comprehension testing and triangulation with other indicators of cognitive effort.

The recent shift from on-site to video conference interpreting affects interpreters in many ways, one of these is related to space: When working on-site, interpreters are surrounded by the conference setting; during a videoconference, this is squeezed onto a screen together with other relevant visual input in numerous windows (documents, online dictionaries, etc.)(Saeed et al 2022). These windows often overlap and switching between them may require additional effort. Many interpreters have already taken to using several devices or screens during video conference assignments (Chmiel & Spinolo 2022, Saeed et al. 2022). Another option that might increase the available screen space, offer user-friendly interaction with intuitive gestures and hence, improve the ergonomics of video conference interpreting is the use of augmented reality technology (AR) (Ziegler & Gigliobianco, 2018). In order to explore the affordances of AR technology in interpreting, I am in the process of conducting a study comparing a videoconference setting on screen with one on AR glasses where I plan to collect data on participants’ interactions with AR objects, interpreting performance, physiological measures, and self-reported effort. The use of AR technology in research, however, presents some methodological challenges because the interactions with AR objects are not visible to the researcher and typical physiological measures (e.g. pupil dilation, skin conductance) are often not compatible with AR technology: relevant body parts are covered or cables restrain the body movements necessary for gesture recognition. In my presentation, I will address these methodological challenges and the solutions developed for the present project.

Speech Translation models designed to convert spoken Swiss-German to written German have been in existence for some time. While these models generally perform well, their performance in various scenarios remains poorly understood. In this thesis, we explore the influence of audio length on the performance of Swiss-German speech translation models and identify the necessary factors for achieving better performance on longer audio segments. To achieve this, we examined four speech translation models from different institutions. A model from the Zurich University of Applied Sciences (ZHAW), one from the University of Applied Sciences Northwestern Switzerland (FHNW), a model from Microsoft, as well as a model from OpenAI called Whisper. We conducted eight different experiments using a Swiss-German corpus collected by the ZHAW and FHNW. In the experiments, the audio length was augmented in various ways. From there, we found that while the ZHAW, FHNW and Microsoft models showed a tendency to perform worse on longer duration, extending the duration by adding silence did not influence on the performance. Changing the playback speed has a negative influence on the ZHAW, Microsoft and Whisper models, both when speeding segments up or slowing them down. The FHNW model exhibited extraordinary robustness to changes in playback speed, as the results when accelerated by a factor of 1.25 were nearly identical to the results when the playback speed was not altered. The biggest influence on performance was when adding more than one sentence to a segment. Without a segmentation of the input audio the ZHAW, FHNW and Microsoft models performed badly, indicating that segmentation should be introduced as soon as more than one sentence appears in an audio recording. Training a model specifically on multi-sentence segments showed promising results, on single sentence segments and multi-sentence segments as well as in scenarios where sentences are split while segmenting the audio recordings.

References: - Albl-Mikasa, M. (2022). Conference interpreting and English as a lingua franca. In M. Albl-Mikasa, and E. Tiselius (eds). The Routledge Handbook of Conference Interpreting, (546-563). London: Routledge. - Ehrensberger-Dow, M., Albl-Mikasa, M., Andermatt, K., Heeb, A. H., & Lehr, C. (2020). Cognitive load in processing ELF: Translators, interpreters, and other multilinguals. Journal of English as a Lingua Franca, 9(2), 217–238. https://doi.org/10.1515/jelf-2020-2039, In the wake of the growing impact of English as a lingua franca (ELF) on multilingual communication, it has been assumed and partially shown that deviations from conventional linguistic standards may affect interpreting and translation (T&I), especially in monological settings (Albl-Mikasa 2022). This raises the question as to how T&I professionals cope with these deviations in their effort to produce high-quality target texts. As part of the CLINT (Cognitive Load in Interpreting and Translation) project, various product- and process-based methods of the established repertoire in Cognitive Interpreting and Translation Studies have been applied in order to measure the (additional) cognitive demands involved and gauge the coping strategies targeted at managing those demands (Ehrensberger-Dow et al. 2020). To complete the picture, an international survey was conducted among professional conference and community interpreters as well as translators, designed to determine how often they actually deal with ELF texts, the characteristics of the texts that cause them the most difficulties and what coping strategies they use. Based on the responses of almost 900 professionals, this presentation highlights the self-reported results from conference and community interpreters as well as translators focusing, in particular, on their coping strategies. In addition, it will demonstrate how the addition of self-report data from members of the communities of professional practice and their experiences with ELF texts at their usual workplaces allow a richer description of ELF-related multilingual language processing than data collection from the simulated workplace setting and usability laboratory alone.

Talk within the symposium "Promises and pitfalls – how to handle and to use digital technology in academic writing. Three perspectives on digital writing from a Swiss project".

How is accessible content provided in different languages in Switzerland, an officially multilingual country? Are there common guidelines or does each language region have its own guidelines? To what extent do the guidelines differ for purely linguistic reasons? Using the example of SDH (subtitles for the deaf and hard of hearing) on Swiss public television, the guidelines of three of the language regions are categorized and then compared. The guidelines differ in format, scope and level of detail. In terms of content, it becomes clear that, on the one hand, the different language regions have different emphases, which cannot always be justified by linguistic differences. On the other hand, there are certain types of rules that have a high priority in all three language regions and are implemented in a very similar way in the different languages.

With the global spread of English as a lingua franca (ELF), T&I professionals and other multilinguals are increasingly exposed to English produced by non-native speakers. So far, research has focused primarily on the use of ELF in dialogical settings whereas the effects on language processing in monological settings are still under-researched. In a large research project, we addressed this gap with a multi-method approach. In this presentation, we draw on performance data and self-report data collected from professional conference interpreters, professional translators and teachers of English as a foreign language at secondary level. The participants were asked to process two texts on different topics: One of them in its original version produced by an ELF speaker and the other one in an edited version that conformed to English conventions. The texts and versions were counterbalanced for order. Depending on their professional background, they performed one of the following tasks: simultaneous interpreting, translation or content recall, respectively. Subsequently, they rated the comprehensibility of the texts on a six-point Likert scale. Based on the assumption that non-native speakers' texts are less comprehensible, we expect to find a negative effect of ELF on task performance and self-reported comprehensibility in all three groups. In line with the CTIS focus on the triangulation of methods, our presentation of results will show how the combination of self-reported and performance data contributes to an understanding of the effects of non-native English input on language processing.

In their capacity as language experts, interpreters are sometimes expected to deliver target texts that are better than their underlying source text, especially when the latter was produced by a non-native speaker (Déjean Le Feal 1990: 155). In the course of the global spread of English as a lingua franca (ELF), interpreters are faced with non-native input on more and more occasions. Applying Relevance Theory (RT) as a conceptual and methodological framework enables us to analyze in detail how interpreters deal with non-native input. More specifically, an RT-based analysis helps us to demonstrate what of the source input is understood by the interpreter and how it is presented in the target rendition in an attempt to augment or maximize relevance for the audience. This, in turn, provides clues as to whether or not interpreters (feel the need to) optimize those ELF speakers’ input and to the potential additional cognitive effort that may be associated with it. Based on authentic data from the CLINT (Cognitive Load in Interpreting and Translation) project, the renditions of 28 professional interpreters of an original non-native English speaker text and its edited (to comply with native speaker conventions) version are analyzed with a view to the enrichment processes undertaken by the interpreters. A comparison of the renditions of the original versus edited versions shows that interpreters do engage in considerably more such processes when rendering an ELF speech. Determining whether or not these interventions lead to actual cognitive effects in terms of information gains on the part of the audience may require an additional qualitative analysis as well as comprehension testing among listeners.

Community and medical interpreting have been described as socially situated interactions. This paper introduces a situated cognition and functional-pragmatic perspective as part of efforts to (re)integrate the cognitive dimension. It highlights the significance of institutional, domain-specific and discourse-related knowledge on the part of interpreters and draws attention to the fact that interpreters enact institutional processes through shared (speech) action patterns realized by language-specific devices. Part one draws upon the Heidelberg School’s theoretical framework, a cognitive and pragmatic discourse model of interpreting, and updates it with insights on doctor-patient communication (DPC) from Functional Pragmatics (FP). Part two discusses first-order and second-order knowledge relating to discourse structures and speech action patterns in DPC. Shared knowledge of institutional speech action patterns is highlighted as facilitating cognitive processing in interpreter-mediated DPC (IDPC). Part three uses FP-based analysis to illustrate two aspects of IDPC where knowledge of both institutional processes and linguistic structures may determine interpreting success. This is in continuation of previous analyses of 19 video-recorded IDPCs from the Swiss Basel University Hospital/ZHAW corpus transcribed and analysed for sequences where knowledge of institutional processes and language-specific devices can be shown to play a role in interpreting. Conclusions are drawn for an integrative view of the interactive, social and cognitive dimensions of interpreting.

In many MA programmes, input about translation technology is provided in dedicated courses that run parallel to the practical courses. In a session in a revision course module, co-taught by the translation technology lecturer and a translation lecturer, students discovered the collaborative possibilities that CAT tools can offer.

With the spread of English as a lingua franca (ELF), English texts and speeches are increasingly being produced by none-L1 speakers of English. A recent investigation of the potential cognitive impacts of ELF on mediated communication involves interpreters, translators and non-T&I multilinguals. This presentation reports on the impact on the non-T&I multilinguals who may engage in natural translation and interpreting. A task-based experiment in a simulated workplace setting was designed to investigate how these multilinguals approach ELF material. The material was sourced from authentic conference talks and their corresponding abstracts. Of each talk and abstract, an edited version was created. The study participants listened to one talk and read its abstract in their ELF versions and listened to another talk and read its abstract in their edited versions. After each task, they were asked to rate the comprehensibility of the talk and the abstract that they had just processed on a scale from 1 to 6. They then performed a recall task of the content of each talk and abstract and addressed their handling tactics in a post-task interview. We will report on the perceived comprehensibility of non-native text material and the handling tactics applied by the participants. The results intend to stimulate the discussion of how non-professional interpreters and translators cope with increasingly frequent ELF input.

Nel settore della traduzione l’importanza di tradurre contenuti in modo rapido e qualitativo è sempre più marcata. La tesi ha l’obiettivo di identificare la modalità migliore di visualizzare i risultati della traduzione automatica (TA) nei CAT-Tool per aumentare la produttività, senza porre limiti alla qualità. A tal fine è stato condotto un esperimento con sette partecipanti (traduttori professionisti e studenti) e tre modalità a confronto: TA direttamente nel segmento (post-editing convenzionale), TA all’esterno del segmento come suggerimento e TA interattiva. Ogni partecipante ha tradotto tre brevi testi, uno per ogni modalità. Il processo di traduzione è stato registrato attraverso il keystroke logging e la registrazione dello schermo. Al termine dell’esperimento i partecipanti hanno completato un questionario on-line sulle loro preferenze e opinioni riguardo alle tre modalità. Infine, tre esperti hanno va-lutato la qualità dei testi finali. I risultati indicano che il post-editing convenzionale ha richiesto lo sforzo temporale, tecnico e cognitivo minore e ha permesso di ottenere una qualità finale leggermente maggiore rispetto alle altre due modalità. La maggior parte dei partecipanti ha preferito tradurre con il post-editing convenzionale. Si può concludere che il post-editing convenzionale permette di aumentare la produttività senza compromettere la qualità dei testi di arrivo., Finding productive ways to translate while maintaining high quality levels is becoming increasingly important. This thesis aims to identify the best way to display machine translation (MT) results in the CAT-Tool to boost productivity without compromising quality: directly in the segment (conventional post-editing), as a suggestion outside of the segment or interactively. An experiment with seven participants (professional translators and students) has been conducted for this purpose. Each participant translated three short texts, one for each mode. The translation process was registered with screen recording and keystroke logging. After the experiment, the participants completed an online questionnaire with questions about their preferences and opinions regarding the three modes. The quality of the final texts was then assessed by three experts. The results show that conventional post-editing required the least temporal, technical, and cognitive effort and produced target texts with a slightly higher quality. Most participants preferred to translate with conventional post-editing. The main conclusion that is drawn is that conventional post-editing increases productivity without compromising the quality of the target texts.

Das Ziel dieser Masterarbeit ist eine vertiefte Auseinandersetzung mit Metaphern im Dolmetschbereich. Metaphern mit ihren komplexen Interpretationsmöglichkeiten haben als Forschungsfeld nicht an Bedeutung verloren. Hier soll zusätzlich die Ebene der Mehrsprachigkeit durch eine Verdolmetschung in Betracht gezogen werden. Dazu wird folgende Forschungsfrage gestellt: Wie werden Metaphern zum GREEN DEAL am EU-Parlament verdolmetscht? Ausgang der Forschungsarbeit ist ein Korpus aus Metaphern, das anhand der EU-Datenbank erstellt wurde. Die Metaphern des Korpus werden anhand bestehender Theorien und dem Open Coding in fünf unabhängigen Durchgängen geordnet. Für die Analyse wird eine mixed-method Strategie angewendet. Einerseits wird quantitativ untersucht, wie viele Metaphern direkt mit derselben Metapher verdolmetscht, und wie viele umformuliert oder weg-gelassen werden. Andererseits wird qualitativ betrachtet, welche Metaphern verwendet worden sind, um anschliessend die verwendeten Strukturmetaphern nach ihrem Konzept-system zu untersuchen. Die Auswertung der qualitativen Analyse zeigt auf, dass Konzeptsysteme der verwendeten Metaphern zwischen Ausgangs- und Zielsprache selten übereinstimmen. Die quantitative Analyse zeigt, dass die untersuchten Metaphern mit einer deutlichen Mehrheit durch eine äquivalente, oder eine alternative Metapher verdolmetscht werden., This Master’s thesis aims to undertake an in-depth examination of metaphors in the field of interpreting. Metaphors, which offer rich opportunities for interpretation, have not gone unnoticed as a research focus. In this Master’s thesis, the topic of multilingualism is examined through the prism of interpretation. For this purpose, the following research question will be investigated: How are metaphors concerning the Green Deal in the EU Parliament interpreted? The starting point for this study is a corpus of metaphors, which was compiled using the publicly accessible EU online terminology database. The metaphors from the corpus are ranked using existing theories and open coding in five independent rounds. A mixed method strategy is used for the analysis. Firstly, a quantitative analysis examines the number of metaphors that are directly translated with the same metaphor, and how many are rephrased or omitted. Secondly, qualitative analysis investigates the particular metaphors in use, in order to subsequently examine the conceptual metaphors according to their concept system. The evaluation of the qualitative analysis shows that the concept systems of the metaphors used rarely match between source and target language. The quantitative analysis shows that the metaphors used are with a clear majority interpreted through an equivalent or an alternative metaphor.

The cognitive demands associated with performing a task involve at least two dimensions: (1) the load dimension that is related to the assumed task difficulty and (2) the effort dimension that reflects the resources invested in a task. This study considers whether this distinction is actually relevant to translators and interpreters when they report load and effort and, if so, how the assumed psychological reality of these two dimensions is related to task performance. In this study, professional translators and interpreters performed naturalistic tasks with comparable stimuli, working from English into German. After each task, they were asked to rate their experienced load and effort as part of the NASA Task Load Index. Their performance was measured by analysing process and product indicators that correspond in interpreting and translation. Results indicate that while self-reported load and effort are highly correlated, their relationships to process or product measures appear to be more complex.

Das laufende Swissuniversities-Projekt «Digital Literacy im Hochschulkontext» hat gezeigt, dass maschinelle Übersetzung sowohl von Studierenden als auch von Dozierenden an Schweizer Hochschulen regelmässig für das Verfassen und die Übersetzung aller Arten von Texten verwendet wird. Allerdings geschieht diese Verwendung bisher fast immer ohne explizite Instruktion und Sensibilisierung für den Umgang mit digitalen Tools, die auf künstlicher Intelligenz beruhen. In dieser Präsentation geben wir einen kurzen Überblick darüber, was MTLiteracy, also der verantwortungsvolle und kompetente Umgang mit maschineller Übersetzung bedeutet. Wir zeigen auf, dass MT-Literacy verschiedene Komponenten beinhaltet, von Wissen über die Tools, ihre Potentiale und Risiken bis hin zu Soft Skills, wie eine selbstbewusste Haltung und realistische Einschätzung der eigenen Fähigkeiten. Wir skizzieren verschiedene Szenarien, wie maschinelle Übersetzung für mehrsprachige Schreibprozesse genutzt werden kann. In unsere Überlegungen beziehen wir auch die immer häufiger verwendete KI-basierte Textgenerierung ein und wie diese mit maschinellen Übersetzungstools gezielt kombiniert werden kann.

Depuis quelques années, la qualité accrue des nouveaux systèmes de traduction automatique neuronale et leur utilisation massive, surtout dans le contexte éducatif (Bourdais 2022 pour l’école secondaire par ex.) rendent ces outils indispensables pour les apprenant-es. Les apprenant-es les utilisent pour de nombreuses tâches (réception, expression) et même lorsqu’on leur interdit expressément leur utilisation (O’Neil 2019). Il est donc crucial pour les enseignant-es de prendre le taureau par les cornes et d’aborder le sujet en classe. La première partie de cet atelier proposera une série d’informations sur la traduction automa-tque (TA), son fonctionnement et les implications pour l’utilisatrice et l’utilisateur. Même si les systèmes de TA actuels présentent une très haute qualité syntaxique et orthographique, de nombreuses chercheuses et chercheurs ont montré que ces systèmes présentaient encore de nombreux problèmes (des biais de genre, par exemple Savoldi et al. 2021). De plus, de nombreuses utilisatrices et utilisateurs n’emploient pas ces systèmes de façon optimale et confondent souvent les différents types d’outils disponibles en ligne, comme en témoigne une grande enquête menée récemment dans les hautes écoles suisses. Les apprenant-es doivent donc développer une instruction en littéracie de la TA (Bowker & Buitrago Ciro 2019) et plus généralement des outils bilingues dans l’enseignement. Dans une deuxième partie, nous résumerons les conclusions auxquelles nous avons abouti après plusieurs essais plus ou moins fructueux de protocoles pour insérer une instruction en littéracie de la TA dans tous les cours de langue d’une université suisse : présentations de 20 minutes ; puis présentation et développement par les enseignant-es de tâches spécifiques pour développer la littéracie en TA de leurs étudiant-es. La dernière partie de l’atelier sera consacrée à des discussions en petits groupes où les participant- es questionneront leur pratique et réfléchiront à une faç

L’atelier proposé repose sur les résultats d’un large projet consacré à la littéracie de la traduction automatique (machine translation literacy, Bowker & Buitrago Ciro 2019) en contexte universitaire. Nous présenterons tout d’abord les résultats d’une enquête à large échelle (6400 répondant-e-s) sur les usages et perceptions de la traduction automatique en regard de l’apprentissage des langues étrangères. Cette enquête indique une utilisation buissonnière (Bour- dais & Guichon 2020) de la traduction automatique aussi bien par les apprenant-e-s que par les enseignant-e-s, et montre que cette technologie reste largement taboue dans les échanges didactiques. Plusieurs autres enquêtes aboutissent à des conclusions similaires (Bourdais & Guichon 2020, Dorst & al 2022, Groves & Mundt 2021, Niño 2020, Tsai 2022). En analysant les données de plus près, on observe rapidement une désinformation quant à la nature des traducteurs automatiques neuronaux et à leur mode de fonctionnement, très différent par exemple d’un dictionnaire en ligne. Il en ressort un amalgame fréquent entre les multiples ressources numériques, qui résulte en une utilisation sous-optimale voire contreproductive de l’ensemble des ressources numériques, algorithmiques ou non. Cet atelier vise à confronter les pratiques, attitudes et perceptions établies par notre enquête aux réalités concrètes de la traduction automatique neuronale telle qu’elle est offerte au grand public aujourd’hui. Nous présenterons donc non seulement les pratiques observées mais aussi un état des lieux actuel concernant son développement technologique, son potentiel et ses limites. Nous mentionnerons également les travaux pionniers lancés dans divers pays pour intégrer la traduction automatique à l’enseignement des langues étrangères (Chung & Ahn 2021, Delorme Benites & Lehr 2021, Hellmich 2021, Knowles 2022, Lee 2020, Pellet & Meyers 2022, Yamada 2020, Wei 2022, Xu 2021, Resende & al. 2020). Enfin, nous proposerons des axes de travail

Recent developments in machine translation (MT) might have led some people to believe that soon professional translation will not be needed, but most translator trainers are aware of the high demand for the quality that MT systems cannot deliver without human intervention. It is thus important that professional translators, trainers and their students appreciate when and how MT can best be deployed, even if they do not use it much themselves. This can be accomplished by enhancing their MT literacy, which encompasses an understanding of the basics, risks and benefits of the technology. Trainers can prepare their students to provide advice to clients who might be interested in using MT for their multilingual content but do not have the expertise to judge when it would be enough to meet their needs. Drawing on the example of knowledge dissemination in higher education, this article presents survey results that suggest MT is being used far more widely than previously assumed. We highlight some of the risks associated with uninformed use of this technology, discuss how they can be mitigated by translation professionals with consulting competence, and outline some training scenarios which could contribute to developing societal AI literacy in general.

University students, especially language learners, have increasingly been using machine translation (MT) systems in the last decade and for all kinds of texts, including homework, assignments and exams. This ubiquity does not translate into visibility as few teachers address the subject in class. Several researchers have shown that MT systems, while technically very easy to access and use, are not always employed in a critical manner. They have therefore suggested that users should develop MT literacy skills. As part of a larger Swiss project on digital literacy in university contexts (DigLit), an action research project at the University of Neuchâtel Language Centre (UniNE LC) seeks to investigate whether delivering a 20-min presentation about machine translation in all L2 classes (French, German, English) at the beginning of the semester was sufficient to foster minimal MT literacy in language learners. All LC students were surveyed at the end of the semester. These survey results were compared with those from a survey of Swiss university students carried out in spring 2021 as part of the DigLit project. These results have allowed us to monitor and enhance the teaching of MT literacy skills in our LC.

References: Angelone, E. (2023). Weaving adaptive expertise into translator training. In G. Massey, E. Huertas Barros, & D. Katan (Eds.), The human translator in the 2020s (pp. 60–73). Routledge. Bernadini, S., Ferraresi, A., & Petrovic, M. M. (2022, June 24). A curious mindset and love of experimentation: A corpus perspective on emerging language industry jobs. [Paper presentation]. 10th EST Congress: Advancing Translation Studies, Oslo, Norway. Cimatti, B. (2016). Definition, development, assessment of soft skills and their role for the quality of organizations and enterprises. International Journal for Quality Research, 10(1), 97-130. https://doi.org/10.18421/IJQR10.01-05 EMT Board and Competence Task-Force. (2022). European master’s in translation. Competence framework 2022. European Commission. https://commission.europa.eu/system/files/2022-11/emt_competence_fwk_2022_en.pdf Matteson, M. L., Anderson, L., & Boyden, C. (2016). “Soft skills”: A phrase in search of meaning. portal: Libraries and the Academy, 16(1), 71–88. https://doi.org/10.1353/pla.2016.0009. Nitzke, J., & Hansen-Schirra, S. (2021). A short guide to post-editing (Translation and Multilingual Natural Language Processing 16). Language Science Press. https://doi.org/10.5281/zenodo.5646896 Risku, H., Schlager, D., Milosevic, J., & Rogl, R. (2022, June 24). The social construction of translation expertise in the language industry. [Paper presentation]. 10th EST Congress: Advancing Translation Studies, Oslo, Norway. Guest lecture at Jagiellonian University in Krakow., A dynamic, diversifying language industry is quite evidently placing a rising premium on so-called “soft” transferable or transversal skills (Angelone, 2023; Bernardini Ferraresi, & Petrovic, 2022; Risku et al., 2022). Under umbrella terms like “service provision,” “interpersonal and personal,” “consulting” or “risk assessment” competence (EMT Board and Competence Task-Force, 2022; Nitzke & Hansen-Schirra, 2021), soft skills have come to occupy a visible position in competence modelling and curricular outlines – alongside those traditionally perceived as hard. But both inside and outside translation studies and translator education, “soft skills” seem to represent a “phrase in search of meaning” (Matteson, Anderson, & Boyden, 2016). Largely ill-defined and vaguely described, they remain notoriously hard to operationalise, track and evaluate (Cimatti, 2016). This talk explores some confusions, contradictions and caveats in how soft skill sets have been viewed until now. It seeks to explain the fuzziness in distinguishing the soft from the hard as the inevitable outcome of trying to model expert performance in the intrinsically embodied social enaction of language mediation. It then tentatively outlines a basic, practicable framework for identifying, mentoring and assessing how students develop the soft – or better, transferable – skills that add tangible, much-needed human value to the processes and practices of today’s language industry.

The papers compiled in this volume reflect the key theme of the most recent Duo Colloquium sessions – contextuality. The psychological notion of context has been central to translation research for decades, and it has evolved along with the development of translational thought, translation types and tools. The theme of contextuality can be understood at any level, from the geopolitical to the textual, and embraced by both academic and professional considerations of translational and interpreting phenomena. It is centred on context, contexts and/or decontextualisation in translation and interpreting theory and practice from a variety of disciplinary, interdisciplinary and transdisciplinary perspectives. Discussing the above-mentioned notions is the subject of the present volume.

This introduction presents an overview of the key concepts discussed in the subsequent chapters of this book. The book reflects on the present and future of Translation Studies following the incorporation of terms and perspectives from industrial technology discourses and practices into the discipline. It examines the current status of translation technologies training in translation and interpreting curricula in undergraduate degrees in Spain. The book provides a state-of-the-art review of translation competence, translation revision competence and post-editing competence models, analysing their similarities and differences in the light of recent research on revision and PE training. It proposes a pedagogical approach to future-proofing the translation profession in the 2020s.

Has the language industry of the 21st century been racing ahead of the translation profession and leaving translators behind? Or are translators adapting to new sociotechnical realities and societal demands, and if so, how? The chapters in this volume seek to shed light on the profiles and position of human translators in the current decade. This collection draws together the work of leading authors to reflect on the constantly evolving language industry. The eight chapters present new perspectives on, and concepts of, translation in a digital world. They highlight the shifts taking place in the sociotechnical environment of translation and the need to address changing buyer needs and market demands with new services, profiles and training. In doing so, they share a common focus on the added value that human translators can and do bring to bear as adaptive, creative, digitally literate experts. Addressing an international readership, this volume is of interest to advanced students and researchers in translation and interpreting studies, and professionals in the global language industry.

This chapter summarises the past and present of the Lodz-Maastricht and Lodz-ZHAW Duo Colloquia on Translation and Meaning through the lens of its latest iteration on context, contexts and contextuality in translation and interpreting. It considers the newest developments in the language industry and the impact that this is having on the multifarious professions in its scope. It serves to introduce the broad range of papers selected for the volume, from literary translation to speech-enabled PEMT, from NMT to game localization, from church interpreting practices to translation quality assessment, reflecting the rich diversity of the language industry and those working within it.

Simultaneous interpreters’ primary concern is providing a high-quality target language rendition that accurately reflects source language speech in content including subtle nuances, terminology or register. Managing visual input feeds into this cognitively demanding process with interpreters taking cues from speakers’ facial expressions and body language, interspeaker dynamics and audience response, as well as consulting any projected visuals, texts or presentations they have before them in the booth, and also glossaries. To date, interpreters have used on-screen, printed glossaries or specific applications (Jiang 2013), requiring them to interrupt their visual information processing to look up technical terms. This interruption may momentarily increase the interpreters' cognitive load. This is anecdotally evidenced by the fact that it is good practice for interpreters to help each other out in the booth by pointing to or writing down the term in question. A similar issue may also concern CAI-tools: first studies suggest that the question of how to display the output of a CAI-tool is not a trivial one (Defrancq and Fantinuoli 2021; Desmet, Vandierendonck, and Defrancq 2018). A solution to improve the interpreters' cognitive ergonomics when looking up terms may be augmented reality (AR) (see also Ziegler and Gigliobianco 2018). This technology allows a seamless integration of auditory and visual information by displaying the translations of technical terms on augmented reality glasses in direct view of the interpreters whenever they would usually consult their glossary. To address this question, we are in the process of conducting a pilot study with ten professional interpreters. They are asked to interpret two highly technical conference talks of 10 minutes each: the first one with a conventional glossary on a laptop and the second one with augmented reality glasses. In both conditions, interpreters receive a copy of the corresponding presentation slides that they c

In this paper, we pursue two main (interconnected) goals, relying on a comparable and parallel corpus of French and Italian cleft sentences. First, we assess the most typical formal and functional properties of clefts in these languages, while also considering variation due to the different types of data. Second, we verify the presence of interference in the translation of clefts from French into Italian. Translators’ choices mostly align with information-structural patterns found in original Italian texts, but word-for-word translation styles can increase the frequency of less-common cleft types in Italian, such as Broad Focus and Information Focus clefts. However, odd or infelicitous pragmatic effects arise only in very few cases.

Die vorliegende Arbeit ist das Ergebnis einer Analyse eines aktivistischen Romans, Las Malas, der von einer argentinischen Trans-Autorin geschrieben wurde, um das gesellschaftliche Bewusstsein für die Demütigung und Marginalisierung, in der die Trans-Gemeinschaft lebt, zu sensibilisieren, sowie von drei Übersetzungen des Romans. Um herauszufinden, inwieweit professionelle Übersetzer:innen ihre Handlungsfähigkeit ausüben und ihre Rolle als Ermöglicher des sozialen Wandels in ihrer jeweiligen Sprachkultur anerkennen, wurden einige kulturell und emotional geladene Auszüge aus dem Originaltext von einer Gruppe von zehn Leser:innen ausgewählt, die der Autorin nicht bekannt sind. Diese emotional geladenen Auszüge wurden anschliessend mit den Personen diskutiert, die Las Malas ins Englische, Deutsche und Französische übersetzt haben. Als konzeptueller Rahmen für die Textanalyse dienten die kontrastive Grammatik, die Diskursanalyse, des Qualitätsbewertungsmodells von Juliane House und der Normen von Gideon Toury. Die Untersuchung des Verhältnisses zwischen diesen Übersetzungen und den sozioökonomischen Bedingungen der Kultur des Ausgangstextes ergab, dass die hervorstechenden aktivistischen Inhalte meist so übersetzt wurden, dass sie die gleiche oder zumindest eine ähnliche Wirkung auf das Zielpublikum ausübten. Obwohl einige geografische, soziale, historische oder kulturelle Bezüge in den Zieltexten verloren gingen, beeinträchtigte dies weder das Verständnis der übersetzten Texte, noch schmälerte es das berufliche Selbstkonzept der Übersetzer:innen und ihre Selbstwahrnehmung als Vermittler zwischen den Sprachkulturen. Somit stützen die bisherigen Ergebnisse die Hypothese, dass Übersetzer:innen, bedingt durch die soziokulturellen Verhältnisse des LGBTQ+-Kollektivs in Argentinien, das Potenzial haben, eine entscheidende Rolle bei der gesellschaftlichen Anerkennung von travestis zu spielen, und ihnen zu helfen, eine Stimme zu finden und ihre weibliche Identität neu zu definie

In einem von Zeit- und Kostendruck geprägten Markt ist es für Übersetzende wichtig, effizient zu arbeiten und gleichzeitig ihre beste Leistung abzurufen. Dies können sie erreichen, wenn sie sich in einem kognitiven Zustand befinden, der als Flow bezeichnet wird. Flow wurde bis anhin in der Übersetzungswissenschaft noch kaum untersucht. In dieser Arbeit wurden deshalb prozessbasierte Indikatoren identifiziert, anhand derer Flow-Zustände im Übersetzungsprozess untersucht werden könnten. Davon ausgehend wurden flow-bezogene Vergleiche zwischen Übersetzenden mit unterschiedlichen Erfahrungsniveaus sowie zwischen unterschiedlichen Texten und zwischen unterschiedlichen Teilen der Entwurfsphase des Übersetzungsprozesses gezogen. Diese Vergleiche sollten den Fokus für kommende Untersuchungen schärfen. Verbale Daten und Beobachtungsdaten aus Übersetzungsprozessen wurden auf eine Auswahl von Indikatoren überprüft, die anhand der zur Verfügung stehenden Forschungsliteratur erarbeitet worden waren. Es zeigte sich, dass neben verbalen Daten auch Indikatoren in Keystroke-Logging-Daten Hinweise auf Flow liefern dürften, während die gewählten Indikatoren in Eye-Tracking-Daten wenig aussagekräftige Ergebnisse generierten. Ausserdem lassen die Resultate vermuten, dass professionelle Übersetzende häufiger Flow erleben als Masterstudierende. Zwischen den in dieser Arbeit betrachteten Texten und Teilen der Entwurfsphase waren die flow-bezogenen Unterschiede weniger deutlich. In a market characterised by time and cost constraints, translators must be able to work efficiently while at the same time performing at their best. Translators can achieve this balance if they enter into a cognitive state described as flow. In translation studies, flow has not yet been studied in detail. In this paper, process-based indicators were identified that could be used to investigate states of flow in the translation process. These indicators were used as the basis for drawing flow-related comparisons bet

This book provides a unique roadmap for the design of an effective emotional intelligence (EI) intervention adapted to the needs of translators. Building on the growing interest in understanding the impact of translators’ emotions on their work, this book is the first to outline the key components of an 18-hour training intervention based on a sound theoretical basis and aiming to improve translators’ emotional competencies. Following a detailed review of the relevant literature in both translation studies and psychology, the book demonstrates the importance of soft skills for translators’ wellbeing, psychological health, and work performance. A number of important elements that need to be considered when designing an evidence-based intervention are discussed, such as appropriate safeguards, theory-based activities, psychometric measures, individual difference variables, and suitable coaching techniques. The book concludes with a discussion of other relevant aspects, such as the role of culture in EI interventions and promising areas for further study.

The messy reality of non-comparable source texts, domain knowledge, experience, familiarity with tools, and emotional factors is difficult to reconcile with attempts to measure and describe cognitive processes involved in translation. The focus of Cognitive Translation and Interpreting Studies includes both the micro level of cognitive processes involved in translational decision-making and the macro perspective in translators operating in specific situations and embedded in organizations, discourses, and societal structures. Cognitive load, a construct that is most often associated with psychological research on learning and the effects of instruction, might also be useful to understand the effects of various factors on translators and their translation processes. We postulate that affective states as well as the processing of emotion-eliciting content require the allocation of a certain amount of cognitive resources and can add to cognitive load. We focus on this link and explore how translators might regulate emotion in order to cope with any resulting increases in cognitive load. This strand of research can contribute to understanding how translators cope with the additional challenges of emotional aspects of their work and provide insights into how competences such as emotion regulation might be included in training.

https://www.zhaw.ch/en/linguistics/institutes-centres/iued-institute-of-translation-and-interpreting/research/duo-colloquium, The Łódź-ZHAW Duo Colloquium on Translation and Meaning is a successor to the internationally acclaimed event with the same concept, organised in Maastricht and Łódź from 1990 to 2015. The theme of the Duo Colloquium 2020/2021 is Contextuality in Translation and Interpreting. The Łódź session of the Faculty of Philology of the University of Łódź held online from 3-5 December 2020 focused on theoretical aspects (http://www.duo.uni.lodz.pl). The ZHAW session approaches applied aspects of the theme and has a more industry-oriented focus on the theme Contextuality in Translation and Interpreting. Contextuality can be understood at any level, from the geopolitical to the textual, and embraces both academic and professional considerations of translational and interpreting phenomena. The sub-themes of the Duo Colloquium 2020/2021 concern the context(s) and/or decontextualisation in translation and interpreting theory and practice, embracing both academic and professional considerations of meaning in translation and interpreting from a variety of disciplinary, interdisciplinary and transdisciplinary perspectives. The sub-themes of the ZHAW session include the interface of translation and/or interpreting with: • agency and empowerment • corpus and/or text linguistics • domain specificity (e.g. economics, law, science, medicine, religion) • education or training • lexicography or terminology • metaphor and/or phraseology • modality and/or media • (new) professional roles • quality assessment • situated practice(s) • sociotechnical systems • technology

Invited speaker / guest lecture (online), The talk focuses on key strategic initiatives in teaching and research at the IUED Institute of Translation and Interpreting, together with their operationalisation and results, to indicate more general trends in the Translation Studies community, highlighting areas in which fruitful didactic and research approaches have been, can be and should be applied. It elaborates on the themes of technology use, translator roles and responsibilities, and the growing diversification taking place in the language industry. The scope of the lecture embraces teaching, research and transdisciplinary social and community initiatives in human-machine interaction and in interlingual and intralingual, accessible communication.

The advent of neural machine translation in 2016 catapulted machine translation (MT) into a world that was already digitalizing at an accelerated pace. The last decade has also witnessed the digitalization of most well-established dictionaries (Merriam-Webster, Robert, Pons, Duden, etc.), including free online access and the emergence of bilingual concordancers (Linguee). As a result, language learners and translation students can use numerous online resources with a simple click. However, the process behind many online dictionaries (OD) is still human lexicographic work, while MT provides output based on natural language processing, i.e. artificial intelligence (AI), without human validation. A survey carried out among Swiss universities in 2021 shows that, in the context of higher education, machine translation tends to be used as a surrogate for online dictionaries. This chimes with observations by Briggs (2018) and O'Neill (2019), who noted an almost identical use of MT and OD amongst students. In the present survey, a large proportion of the respondents indicate that they use online MT systems primarily to look up words or to find synonyms. This conflation of MT and OD works both ways: respondents cited bilingual dictionaries when asked what MT systems they use. Further, the physical distinction between the two types of linguistic resources is also fading, since many online MT tools now include dictionary functions and vice versa. In this situation, the authority of human lexicographers seems to be tacitly challenged by AI and demonstrates the need to elucidate how human knowledge and AI should interact for building, maintaining and using linguistic resources. This communication will present the data consolidating this affirmation and will discuss practical consequences and implications, especially with respect to training needs in the area of MT literacy (Bowker 2020) and the informed use of digital literacy tools.

Ce fascicule est destiné aux personnes qui doivent communiquer avec des réfugié·e·s d'Ukraine lorsque aucun·e interprète (professionel·le ou bénévole) n'est disponible. Il contient des conseils élaborés après avoir testé des outils de traduction automatique gratuits et courants au vu de leur qualité et de leur maniabilité.

Since the advent of neural machine translation, we have been witnessing a major change in written multilingual practices not only on websites and social media but also in academic contexts. Both translation of content and multilingual communication have suddenly become available at the click of a button instead of being entrusted to trained translators and linguists. In a country like Switzerland, with its long tradition of multilingual text production in the three official languages plus English, these developments are welcome but are also being viewed with some caution. A variety of new and creative practices have emerged amongst individuals, yet most of them remain undocumented and unguided. Our project explores current attitudes and practices regarding MT in Swiss universities. Initial results show how user profiles (e.g. students, teachers, researchers, academic support staff) call for differentiated approaches to fostering MT Literacy.

With the spread of English as a lingua franca (ELF), more and more English texts and speeches are being produced by none-L1 speakers of English. Only recently have researchers started to consider the potential cognitive impacts of ELF on mediated communication such as translation or conference interpreting (cf. Albl-Mikasa & Ehrensberger, 2019). However, it remains unclear whether language mediators have developed ways of processing ELF that differ from those of non-mediating language professionals. This paper explores whether professional translators, conference interpreters and L2 teachers of English actually recognise ELF when they are engaged in taxing language processing tasks. A task-based experiment in a simulated workplace setting was designed to investigate whether ELF has an effect on their performance and how these professionals reflect on it. The ELF material was sourced from authentic video recordings of conference talks and the corresponding abstracts. They differed with regard to levels of subject specificity and technicality. The talks were then respoken in order to remove any accent and be able to investigate ELF specificities that are present both in spoken and written language. The participants in the study, who were all L1 speakers of a language unrelated to the L1 of the ELF producers, processed at least one text in the ELF version and one in an edited version in a translation, interpreting or recall task, respectively. Afterwards, their reflections on textual characteristics and on processing tactics were prompted by a structured interview, with no information provided about ELF. The recorded interviews were coded and analysed in order to identify categories of source text characterisations and processing tactics by participant group. We will report on the results and discuss potential implications for the work of language professionals dealing with ELF on a regular basis.

Keynote address, This presentation is divided into two parts. The first part considers current aspects of translator education and teacher development, ending with a scalable model of co-emergent learning that accounts for learning from the individual to the organisational or institutional level. Part two looks at phronesis (“practical wisdom”) as Aristotle describes it, and as it is defined, described and applied by educators and researchers in various professional sister disciplines of today. It then explores the significance of phronesis for current models and approaches in translator education and briefly sketches out its implications for, and applicability to, translator teacher development.