Intégrer GoogleTranslate comme support d’enseignement au niveau débutant : atelier pratique pour un public de réfugié·e·s ukrainien·ne·s
Le dispositif didactique analysé et décrit dans cet article suit un objectif double : amener les participant·e·s à développer leur apprentissage en autonomie et leur littéracie en traduction automatique neuronale (TAN). Ces outils sont massivement utilisés par les migrant·e·s, mais pas toujours à bon escient. Il est donc crucial de cadrer et de thématiser leur utilisation dans les classes de langues, même au niveau débutant. Plusieurs types d’exercices ont été créés : 1. Pour introduire ces outils à la fois comme de bons modèles et des bases pour la communication ; 2. Pour montrer que la TAN a des limites (ne représente pas toute la langue ; peut proposer un output fautif) ; et 3. Pour fixer un cadre et des situations où l’utilisation de cette technologie n’est pas utile. L’évaluation de ce dispositif a montré son succès relatif, même si une introduction théorique à la TAN aurait pu permettre un meilleur développement de la littéracie en TAN des participant·e·s.
Policy on the use of Machine Translation (MT): A good model for wider policies on Generative AI (GenAI)?
Since the advent of ChatGPT and other automatic text generators, educators from many disciplines, including language learning and teaching, have published numerous articles exploring this technology’s “pitfalls and potentials” (Barrot, 2023) and offering recommendations based on their own practice to teachers, users, and institutional decision-makers. But it is early days yet, and, while recognising the need to offer guidance, there is not enough scientific data to create evidence-based policies. Having been working on machine translation (MT) literacy (Bowker & Buitrago Ciro, 2019; Cotelli Kureth & Summers, 2023) for several years, we have developed guidelines for the use of machine translation (MT) tools in higher education, which have been implemented in a Swiss university. Given that MT tools share technical features with generative AI (GenAI) tools like ChatGPT, we believe that applied knowledge of the former could facilitate understanding of the latter. This article will draw on both our own experience and a thorough literature review of recommendations for the use of GenAI for higher education institutions (HEI) to map what guidelines on the use of GenAI should include and how they should be presented to teachers and users.
“I Looked It Up in DeepL”: Machine Translation and Digital Tools in the Language Classroom
This article looks at a widespread yet erroneous use of machine translation (MT): looking up single words, thus treating MT systems as online bilingual dictionaries (ODs). After a literature review of this trend in research about MT, we consider data from a large survey that we carried out in 2021 at all Swiss universities on MT use and users’ attitudes. When analysing users’ metalinguistic awareness, we discovered that nontranslators perceive the text to translate, mostly at word level, leading to the misuse of MT systems as ODs. Moreover, the survey results revealed confusion between the different digital tools for language learning, namely MT, online parallel corpora like Linguee and ODs. We therefore suggest broadening the scope of MT literacy to include training learners in general digital literacy to enable them to use such tools appropriately.
Fostering transparency: a critical introduction of generative AI in students’ assignments
This article analyses a specific strategy designed to include generative artificial intelligence (GenAI) tools in students’ written assignments. While we recognise that GenAI tools represent a challenge for teachers in terms of their classroom use and the development of digital literacy among students, we believe that banning them is not a viable option. In our view, students need to develop a sustainable, critical approach to these tools, informed by the need to be transparent. With this in mind, we have thus developed, tested and evaluated a protocol for language learners in two Swiss universities. In our experiment, students were allowed to use any online tools available for their written assignments, but they were required to clearly highlight in their texts any output derived from text generators (ChatGPT), machine translation tools (DeepL), online corpora and online dictionaries in their texts. They also had to report on their writing process in an additional, meta-analytical paragraph. After submitting their assignments, students were asked to answer a questionnaire investigating their use of, and attitude to, GenAI tools as well as their transparency in completing the task. The data gathered allowed us to gauge students’ trustworthiness as to their self-reported tool use and to determine whether our protocol could help teachers preserve the take-home written assignment in the GenAI era. Finally, the analysis yielded interesting insights into students’ use of GenAI in L2 writing and highlighted different ways in which teachers can foster more transparency. This innovative action-research study brings much-needed data and offers practical guidance to language teachers interested in GenAI.
Comment développer la littéracie digitale des enseignant-es et des apprenant-es?
Cet article présente les résultats du projet ‘Digital Literacy in University Contexts’ dont l’objectif est de développer la littéracie digitale du personnel enseignant et estudiantin des hautes écoles suisses. Ces conclusions illustrent un modèle de littéracie en IA (Cardon et al. 2023) et nous permettent les suggestions suivantes pour développer, pour soi-mêmes et pour ses étudiant-es, l’application, l’authenticité, l’agentivité et la responsabilité dans leurs interactions avec les outils d’IAGen. Il est important de : - discuter entre pairs et avec les étudiant-es au sujet de l’IA ; - connaitre et présenter des informations sur : les détails techniques du système qui ont une influence sur l’usage ; les faiblesses des relectrices et relecteurs humains ; - inclure des tâches avec l’IA en classe.