SoTEL Symposium 15-17 February 2023 7 Should Machine Translation have a role in language classrooms or not? Neil Cowie Okayama University, Japan Keiko Sakui Kobe Shoin Women’s University, Japan Presentation URL: Keywords: Machine Translation, language teaching, teacher guidelines, student use Abstract Machine translation (MT) of languages has been around nearly 30 years but the importance of its role in language learning has grown exponentially in recent years. This paper summarizes recent research on teacher and learner attitudes to MT, and suggests ways that MT can be used in language classrooms. Studies in the 2010s (Pym, 2013) suggest that teachers were against the use of MT because of its poor quality. However, the level of MT dramatically improved from 2016 when Google Translate adopted a neural-network system. As a result, teachers’ attitudes shifted to more acceptance of MT. Even so, teacher views about MT tend to fall into two camps: those who feel it is a form of cheating (Carré et al., 2022) and those who see it as an appropriate teaching tool. The former take the general approach of “detect, react and prevent”, whilst the latter wish to “integrate and educate” (Jolley & Maimone, 2022). Research has shown that students use MT in different ways according to their level. More advanced students tend to check words and phrases rather than translating a whole report. They understand the limits of MT but at the same time they believe it can help learn a language (Godwin-Jones, 2022; Jolley & Maimone, 2022). Research suggests that training in the use of MT can increase chances for such students to reflect on their language learning (Pellet & Myers, 2022) and that they can become aware of and correct MT errors (Zhang & Torres-Hostench, 2022). On the other hand, lower level students use MT differently as they may lack confidence in their language abilities (Organ, 2019). There are studies that claim lower level students can be linguistically overwhelmed in trying to notice and compare their own translations with MT; therefore, they do not correct the output of MT and submit it as their own work (Lee, 2022: Niño, 2020). In general, the accuracy of MT has improved so quickly that many teachers who previously dismissed MT as poor can no longer ascertain whether their students have actually used it or not (Jolley & Maimone, 2022). This creates doubt in how to assess student work fairly. Furthermore, as teachers vary in their attitudes towards the use of MT for learning, students can be very confused as to whether they are allowed to use MT in different teachers’ classes; and, if they are allowed, in what ways can they do so appropriately. In order to overcome this uncertainty and confusion, it is suggested that, after Reinders (2022), institutions, students and teachers become partners in exploring MT to find the best way to use it for learning. This will vary according to each educational context, particularly concerning student level, but it is vital to create commonly accepted guidelines, approaches and practices so that MT can be best used for language learning and not just as a tool to complete tasks with little or no educational meaning. References Carré, A., Kenny, D., Rossi, C., Sánchez-Gijón, P. & Torres-Hostench, O. (2022). Machine translation for language learners. In D. Kenny (Ed.), Machine translation for everyone: Empowering users in the age of artificial intelligence (pp. 187–207). Language Science Press. Doi: 10.5281/zenodo.6760024 Godwin-Jones, R. (2022). Partnering with AI: Intelligent writing assistance and instructed language learning.