Machine Translation Impact on Language Learning

Project Information

MTrill

Grant agreement ID: 843455

DOI

10.3030/843455

Project closed

EC signature date 21 March 2019

Start date 25 April 2019

End date 16 July 2021

Funded under

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Total cost

€ 184 590,72

EU contribution

€ 184 590,72

Coordinated by

DUBLIN CITY UNIVERSITY
Ireland

Periodic Reporting for period 1 - MTrill (Machine Translation Impact on Language Learning)

Reporting period: 2019-04-25 to 2021-04-24

Summary of the context and overall objectives of the project

The MTRILL project aimed at investigating the impact of popular web-based machine translation systems (MT) on the cognitive processing of English as a second language. The research was motivated by the observation that students of English as a second language are using MT as a tool to support their English learning, due to the easy access to the systems through applications on their phones which provide instant translations for the input entered either by voice, text or image. In the proposed research, the general question raised was: Would the interaction with the output of the MT result in changes in the cognitive processing of English as a second language, reflected by the learning of structures seen in the output of the MT?
To pursue the project’s goals, the researcher implemented two laboratory studies in which participants recruited were tested whether they would be primed by the MT output, i.e. whether the MT system would be capable of influencing the language behaviour of participants. The results of the studies implemented have shown that participants are influenced by the MT systems as they change their linguistic behaviour when speaking in English after exposure to the English syntactic structures seen on the MT output, even if there is an alternative preferred syntactic structure participants can use. The experiments have also revealed that participants trust the MT systems and that this trust can result in implicit (unconscious) learning of a challenging structure in English through the MT output. Furthermore, results of the experiments revealed that changes in linguistic behaviour triggered by the MT output can be elicited irrespective of the MT users’ English proficiency level. Thus, these experiments allowed the researcher to shed light on issues involving the cognitive impact of MT systems on second language processing with a focus on English as a second/foreign language (L2).
The results achieved with the research are important because they reveal the impact of the language technologies on language learning and on human language behaviour. Furthermore, they have the potential to be exploited by language teachers and tutors willing to understand the benefits of including MT systems in the workflow of their language classes. In addition to the pedagogical benefits, the results of this research can be exploited by system developers in the field of computer-assisted language learning (CALL) when designing language learning tools, as the results show which properties of the tools have the potential to positively impact language learning.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

Work was carried out through five (5) work packages which involved activities related to fellowship management, research, transfer of knowledge, training, outreach and dissemination. All work carried out during the reporting period resulted in invited talks and conference presentations, training courses completed, supervision and lectures, outreach activities and published work.
The researcher has published a total of seven works in the reporting period, either as first author or in collaboration with colleagues from the ADAPT Centre on topics related to Machine Translation. Two out of the seven works published were journal articles, reporting results of the MTrill project; two were conference papers reporting results of the MTrill project, and two conference papers in collaboration with colleagues from ADAPT CENTRE. The researcher has also published a book chapter for a non-expert audience in Brazil. Dissemination of the results was also carried out through participation in outreach activities. The researcher has participated in 10 outreach activities in diverse modalities throughout the fellowship, including activities targeting the expert technical audience, activities delivered in collaboration with the ADAPT Centre Education and Public Engagement team and activities involving the non-expert audience. All the works published as well as all the activities related to talks, outreach activities and conferences were communicated to the general public regularly through posts on social media mainly using the researcher’s personal account on Twitter (@_NatResende), Instagram (@natcarol2000) and Facebook (Natália Resende). Throughout the fellowship, dissemination using social networks was carried out through a total of 70 posts. In relation to training activities, the researcher has attended various courses during the fellowship, including topics related to Education and Public Engagement Activities and courses aimed at building on her programming skills in Python and Machine Learning. The fellow has also attended courses involving the improvement of communication skills.
As regards activities related to transfer of knowledge, the researcher was able to transfer her knowledge to people in the field of MT through the tutorial she organised in collaboration with Dr. James Hadley at the MT Summit 2021 conference on theory and practice for research in Post-editese and also supervising M.Phil students in the field of Literary translation. She has also transferred her knowledge through the talks given to non-expert audiences such as talks delivered at the R-ladies meetup Dublin or as a mentor in the Women in AI programme supervising groups of students working on data analysis projects using language R. Thus, the results of the fellowship impacted the society, research in the fields of Machine Translation and Language Learning as well as the fellow's skill set and CV.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

This MTrill project has moved MT cognitive process-oriented research to the next level. The fellowship brought to MT research a new perspective in which the central element of the investigation is the end user of the systems and the direct impact of these systems on users' daily lives, rather than on the performance of the system itself. The research shed light on the societal impact of the current technological context of MT systems influence on learning strategies and students’ linguistic behaviour by showing that language learners are using these tools for language learning and they are improving language skills through their use. This learning occurs because the MT output functions as a template for the use of syntactic structures that they have not yet mastered, triggering in the subsequent utterance a syntactic learning process of that structure. Therefore, contrary to the criticism of the adoption of translation and translation tools for language learning, , the MTrill research has shown a positive side to using MT systems for language learning, that students can learn from the MT output and such learning can be generalised by participants going from a task involving comprehension (reading the translation) to a task involving production (speaking). Furthermore, the methodology employed in the experiments has proven to be ecologically valid so that other researchers will be able to adopt this methodology to address other questions involving issues in MT-human interaction. The outreach activities, in which the fellow had the chance to participate, and the research outputs were crucial to disseminate the idea that new technologies are here to stay and they help raise language awareness. Therefore, MT tools should be integrated in the classroom of the 21st century as this integration will provide a more realistic scenario in which learners will have the opportunity to interact with the systems as in a real societal context.

Important moments of MTrill project