Researchers Call for Clear Policies around Machine Translation Use in Higher Education – slator.com


In a paper published in June 2023, researchers Jasper Roe of James Cook University, Willy A. Renandya of Singapore’s Nanyang Technological University, and George M. Jacobs of Universiti Malaya examine the use of digital tools in academic writing, including machine translation (MT), and their implications for educators and students.

The researchers refer to MT tools as “machine translators” and mention Google Translate and DeepL as examples of two of the most popular tools, noting also their increasing accuracy and ease of access. They claim MT augments the potential for “translation plagiarism,” a term that refers to those cases when an author takes credit for a translation in academic materials. 

The researchers remark on the difficulty in detecting this type of plagiarism, despite the existence of cross-language detection tools. Furthermore, translation plagiarism also occurs with back-translation, which implies that text is machine-translated multiple times to alter the language enough to prevent tools like Turnitin from detecting plagiarism. 

Plagiarized works are harder to detect when combined with a student’s original work. In either forward or back translation, common practice is for students to edit the output, at times using other digital tools to correct and refine the language, and adding their own writing.

Original vs. Translation

Banning MT in academic work could solve translation plagiarism, but besides not being a practical solution, there are benefits for non-native speakers of English who use it, argue the researchers. They add that “if a student from a non-English-speaking background needs to submit an assignment in English but is not sure how to say what they want to say in English, then using [an] MT is a helpful pedagogical tool. If a student wishes to clarify their understanding of a text in English, then translating it back to the first language may equally be helpful.”

In their review of existing research, the paper authors cite work by Mundt and Groves (2016). Specifically, the notion that technology assistance in academic work is the norm and MT is therefore an acceptable tool.

Considering MT both an educational tool and a study aid, Mundt and Groves also propose for institutions to deal with MT through clear policies about its use in academic work and research. According to them, MT use “may lead to greater equality and social justice outside of the language classroom, given their ability to ensure that all students will be assessed solely on content, rather than language use (in non-language based assessments).”

“If a student from a non-English-speaking background needs to submit an assignment in English but is not sure how to say what they want to say in English, then using [an] MT is a helpful pedagogical tool. If a student wishes to clarify their understanding of a text in English, then translating it back to the first language may equally be helpful.” — Roe et al.

The researchers conclude that instructors should become familiar with the tools students might use, including ChatGPT and large language models (LLMs). The focus should be on the value of digital tools for students while formulating clear guidelines for their use as plagiarism detection tools improve.

The paper was published a few weeks ahead of the European Conference on Ethics and Integrity in Academia (held at the University of Derby in England in July 2023), which centered around discussions on AI tools and plagiarism.



Source link