The Linguist

12 The Linguist Vol/64 No/2 ciol.org.uk/thelinguist FEATURES linguistic footholds. Some of these footholds are in the form of grammar notes or vocab boxes; others are less obvious. Do you have to work to the Common European Framework of Reference for Languages (CEFR)? Yes. The CEFR provides standardised guidelines on language achievement. We deliberately choose which language structures to use and weave them artfully into the text. One of the most common stylistic guidelines in CLIL projects is to avoid the passive voice in English. Ok, fine, so how to do that? Some shorter sentences can be flipped around but others might need a complete rewrite to create something coherent and natural in the target language. Another common guideline is to use shorter sentences. Again, this is not always simple. First we identify which parts of a long-winded sentence contain key information and which parts can be safely cut. We might have to completely restructure what is left. It's not just a case of chopping up the source sentence into smaller pieces. Translators with previous ELT experience in the same language combination tend to be even better at this work. Their first-hand knowledge of why pupils struggle with certain aspects of the language feeds into the translation process, helping pitch the finished product to pupils' expected level. Familiarity with how CLIL materials are actually used in the classroom is also a big help. Many subject teachers are not language teachers. The English ability across teaching staff may vary, so teachers need considerable language support too, in the form of linguistically accessible teaching notes. This is sounding a lot like transcreation… We'd argue that all translation is really transcreation but that's a topic for an entire article! What about AI and machine translation? CLIL translation is not a good candidate for machine translation (MT). We need to consider this from the perspective of both translators and CLIL translation buyers. Translators hoping that AI or MT will save them time are likely to be disappointed. We're talking here about the work of CLIL translation, not translation-adjacent tasks where AI may prove useful. In our experience, the level of adaptation required renders MT engines and GenAI not just unhelpful but a hindrance. CLIL translation buyers duped by the false promise of cheaper, faster translation must consider the potential impact on their end users. Teachers and pupils assume the English content in the CLIL lessons is pitched to their level as well as being technically accurate. What are the risks when this is not the case? Pupils struggle with the content and lose motivation, or pick up mistakes that have to be unlearnt later on. Young learners may think they are the problem when the real problem is the way the content has been presented to them. What does the future hold for CLIL translation? It's common knowledge that children who are exposed to language immersion may have improved cognitive skills, perform better academically and enjoy better job prospects in our global economy. Governments know this so there is a demand for CLIL translation, but CLIL translators need to remain flexible. The subjects on offer change, and the demand for new materials is cyclical and very much linked to legislative changes that dictate the overall structure of the education system in a country or region. For pupils to reap the benefits of a bilingual education, teachers need support to deliver the programme and the CLIL material must be fit for purpose. That means putting skilled translators in the driving seat who know how to optimise the user experience. EASE OF USE CLIL materials need to work for school students and for teachers of non- language subjects, who may not be fluent in the language of instruction IMAGES © PEXELS

• Cover