Skip to main content

Staff Guide to the Use of Machine Translation Tools

frosted clear translucent sculptures on a black table
Oreoluwa Mary Adekoya, 2023, BA Graphic And Media Design, London College of Communication, UAL | Photograph Oreoluwa Mary Adekoya

What do we mean by Machine Translation?

Machine Translation (MT) is when text in one language is translated into another, using computer software via a personal computer or smart phone (Wang and Ke, 2022). So, an MT tool takes the text inputted in the source language and automatically translates it into a version in the target language.

Although these tools are driven by AI, they are different from generative AI tools, such as ChatGPT, which create content.

Guidance

Guidance for staff

MT tools, like any tool, can be used in ways that are helpful, and ways that are not - see sections on ‘How do MT tools help?’ and ‘Concerns about the use of MT tools’ below. Below are some considerations to support the appropriate use of MT tools by students at UAL:

  • Encourage open conversations with your students about the use of translation tools and learn about how they are using the tools. If students feel they have to hide their use, it will be harder to talk about using the tools to best effect. Questions might include:
    • Which tools do they use and what do they use them for – reading, writing, listening?
    • Which ones seem to work best and how do they help?
    • Are there any difficulties around using them?
    • Could you as the tutor do anything different to reduce their use of the tools?
  • Be aware of the huge cognitive demands on students who are studying in an additional language and focus on creating inclusive learning environments. It is important we allow enough time for reading, processing, translating, thinking and speaking. We need to demonstrate our empathy and patience. If students feel they can ask their tutor questions, if they understand what others are saying and have more time for reading, they may feel less compelled to use MT tools.
  • Think through when it is appropriate to use MT tools and when it is not. Share this clearly with your studentsso they understand your expectations. It might be useful to think about your desired learning outcomes. If English competency is not one of them but demonstrating an understanding of a principle is, does it matter that MT has been used to support the students in communicating that understanding?
  • Think about your assessments – do you ‘design in’ or ‘design out’ the use of MT? (Mundt and Grove, 2023). If we are assessing students’ ability to talk about something in English, we may decide to design out the use of MT tools (by using a handwritten or oral exam where there is no access to MT, for example). If the focus of the assessment does not necessitate the assessment of English proficiency but a technical skill, then you may feel it is acceptable for students to be assessed with MT support. (Staff should, however, be aware of the OfS guidance on Assessment Practices in English Higher Education Providers guidance). If the assessment requires gathering information from numerous global perspectives, and that involves accessing ideas from work in a range of languages, marks could be allocated for the digital literacy demonstrated in the effective and critical use of MT tools.
  • Encourage students to use complementary ways to improve their English language and academic communication skills, in addition to the use of MT toolseg. by attending sessions run by Language Development and Academic Support tutors.
  • Try out some MT tools yourself (if you don’t currently use them), so you have a better idea of how they work.
  • MT tools can also be a nice way of introducing languages other than English into our classes. Do be careful about copyright though if you are translating other people’s work as you may not have permission to use their work in this way.
Key messages for our students

Have a look at key messages to students in our Student Guide

Case studies

Case study A

A student who is struggling to express themselves in the last section of their essay, writes it in their L1 (first language), then gets an MT tool to translate it into English and then submits it without reading it through or doing a final edit.

Reflections:

  • We would not see this as good practice as the student has not checked the accuracy of the MT. Although the technology powering MT tools has improved dramatically in recent years, there could still be many errors in the MT output. The tone used by the MT tools may be inappropriate for an academic piece of work. The student is not taking responsibility for what they are submitting.

    In addition to this, the student is not using the tool to support their English language development because they are not actively engaged in reviewing and evaluating the outputs.
  • The student is also not using their own voice nor developing the ability to write in their own voice.
  • Better practice would be for the student to use comparative analysis of two different MT tool outputs, or a comparative analysis of an MT output and their attempt in English and then choose the best version or weave the two together.

Case study B

An English as an Additional Language (EAL) student uses MT to help decide which articles (written in English) are most relevant for them to read in detail for the following week’s seminar.

Reflections:

  • This use for gist seems a sensible use of MT. We know it can take EAL students much longer to read and process work written in English and so can save the student valuable time in arriving at the best articles to read.
  • Individual use of translated work is unlikely to be problematic; however, sharing translations might infringe copyright.

Case study C

A Fashion Journalism student uses one MT tool to understand the words used in a lecture and another MT tool to understand the text on a PowerPoint slide, simultaneously.

Reflections:

  • Using an MT tool to access a lecture in an additional language can reduce the cognitive load for the student (Shadiev and Huang, 2019). However, taking in two inputs simultaneously might be more cognitively challenging.
  • MT tools can be better at translating written text rather than spoken communications. If we notice a student using speech to text translation, it might be a good opportunity to remind students that mistranslations may occur and to ask if anything is unclear.
  • If it is just one student doing this (as far as you know), it might be useful to gently talk to the student to see if you can help them feel more confident in class. It may be hard for the student to tell you but are you speaking too fast? Are you using English that they don’t know? Do they feel they can ask you for clarification if they can’t catch something etc? Can you sign post them to Language Development tutorials and/or Language Development self-access resources?

Context for the guidance

Examples of MT tools

The accuracy of MT tools has rapidly improved in recent years, and this has fuelled their use.  Google Translate, DeepL and Microsoft Translator are some of the best-known examples of MT tools. Each tool has slightly different functionality and may perform better at different tasks. Tools may also vary in the quality of translation depending on the language you are translating from and to (Alhaisoni and Alhaysony, 2017).

For this reason, the Language Centre does not recommend particular software, but recommends a critical evaluation of any tools used.

How MT tools are used

We know that MT tools are widely used in the world, in universities in the UK and at UAL (Alhaisoni and Alhaysony, 2017; UAL research, 2023).

Students may use MT tools in many different ways: to look up a word or phrase they do not know in the target language, to translate several abstracts to see which articles might be most relevant to read, to understand a lecturer or to translate written academic work into the target language, for example.

It is important that tutors and students think about the specific way the tools are being used and weigh up the helpful and less helpful aspects of this use.

How can MT tools help

MT tools can:

Reduce anxiety

People operating in an additional language can experience significant levels of anxiety; this is often known as ‘Foreign Language Anxiety’ and these feelings can have a significant impact on a student’s well-being (Dovchin, 2022; UAL research, 2023).  International students speaking in an additional language, new to the UK education system and paying international fees may feel they are under huge pressure to understand everything they hear and read (UAL research, 2023). MT tools can provide a low threat means of decoding this new environment (Wang and Ke, 2022).

Provide feedback 24/7

Tutors are not always available for assistance (Zhao et al, 2023)

Save students’ time

MT tools can help students use their time effectively by giving them the gist of several articles so they can focus in on the most relevant one. This may be particularly useful for postgraduate students who have a very short time to read large amounts, and a short time to enhance their English language skills. Trenkic’s (2018) research with international students suggests that EAL (English as an Additional Language) students’ vocabulary and reading speeds can be half of that of a control group of home students. MT tools may offer significant support for these students.

Assist students with language acquisition

MT tools can help students learn new words andphrases. Students can also compare their original draft in English and the MT version and use this to detect errors and to see where their writing could have been improved, for example, in terms of grammar or phrasing. Students might compare outputs from two different MT apps to find the best way to express their ideas. Noticing the differences and choosing between options is an active process and will aid language acquisition (Wang and Ke, 2022).

Help students develop their first language skills

As students discover that the quality of their inputs into an MT tool impacts heavily on the outcome, students will learn the importance of precision in their L1 inputs (Wang and Ke, 2022).

Help facilitate connections with other students

MT tools can assistwhen there are blocks over unknown vocabulary (both in class and social communications) (UAL research, 2023).

Allow students to focus on the academic content of their course rather than the medium of instruction

As Mundt and Groves (2021) put it “this technology [MT tools] could be seen as a levelling mechanism…which could potentially go some way to remove the extra burden on those who are studying challenging degrees in an additional language”. Note that in 22/23 the attainment gap between Home and International students at UAL was 8% points; c90% of International students at UAL are speakers of English as an additional language.

Enhance students’ knowledge of language/s, and develop their translation and digital literacies

Using MT tools can build students’ knowledge of language/s. For example, the tools may demonstrate to students that sometimes there are no exact equivalent phrases between the source and target language. We know that MT tool use is likely to increase in the future as the tools become even more accurate. Learning about the advantages and limitations of MT tools is an example of the digital skills that many students will be needing in their futures (Zhao, 2023).

Concerns about the use of MT tools

The use of MT tools may:

Be unhelpful for language development, if overused

There are some concerns that overdependence on MT tools may undermine some other methods of language learning (UAL research, 2023). For example:

  1. Looking up a specific word on an MT tool may not give as much contextual information as someone explaining it. Translating ‘frock’ into Spanish on an MT tool may not give you the difference in usage between ‘frock’ and ‘dress’, for example.
  2. Translating chunks of text without being active in this process may mean that new language does not ‘stick’ (Alhaisoni and Alhaysony, 2017).
  3. If you are reading a simultaneous translation of an oral presentation, you will not be practising your listening skills.
  4. Overreliance on MT tools may mean students read or write less in English (Alhaisoni and Alhaysony, 2017).

Not produce extremely high-quality translations in all instances

Although MT tools are improving rapidly, the translations they produce still need to be checked for inaccuracies. For example, the word ‘sculpture’ may be ‘heard’ by a tool as ‘culture’ and translated as the latter. Research suggests that while students are aware that MT tools may sometimes produce errors, they may sometimes overestimate their accuracy (Alhaisoni and Alhaysony, 2017). Students may also feel that AI will produce fewer errors than they will (de Vries and Groves, 2019).

Not produce the type of language suitable for an academic piece of writing

The outputs may not be appropriate in language accuracy or tone (Alhaisoni and Alhaysony, 2017).

Encourage complacency in others

Some students using MT tools may allow staff and other students who feel very comfortable in communicating in English to become complacent about the need to accommodate their language. This potentially puts the burden of communication on EAL students, and reduces everyone’s opportunity to develop their own intercultural communication skills.

Raise issues of equity in access

Some students will be able to afford better versions of MT tools that are paid for. Unless universities can support equal access to the apps, this raises issues of equity.

Facilitate “Translation plagiarism”

This is where work has been translated between languages one or more times and either intentionally or unintentionally disguises that it is plagiarised work (Roe et al, 2023).

Raise ethical concerns

The training data used for AI may be gathered by unethical and exploitative means. There may be concerns about the privacy of data entered into the app. Students should be encouraged to research the companies that run these apps.

Further reading, workshops and resources

Further reading

References

Updates and Feedback

Machine Translation is a rapidly developing area. This guide is a living document, and we plan to update it in December 2024 initially and after that every summer.

If anyone would like to talk to us about the guide, please get in touch - j.bloxham@arts.ac.uk or h.mcallister@arts.ac.uk.