Effective use of AI to support EAL learners

TOM BEAKES, TRAINER, THE BELL FOUNDATION, UK

This article explores where generative artificial intelligence (AI) might be used to help to enhance the learning of students who use English as an additional language (EAL) in integrated classrooms. Learners with EAL are defined by the Department for Education (DfE, 2020) as any child who is exposed to a language other than English at home or in the community. EAL learners account for more than one in five students in state-funded schools in England (1.77 million), and multilingual classrooms are increasingly the norm across the country.

EAL learners are a highly diverse group, and research shows that proficiency in English, rather than EAL status itself, explains four to six times as much variation in achievement as gender, free school meals and ethnicity combined (Strand and Hessel, 2018). This highlights the importance of schools accurately assessing English proficiency in order to identify needs and provide targeted support, enabling learners with EAL to develop English alongside ensuring full access to the curriculum.

Learners who are new to English, particularly recent arrivals, may face additional barriers, including unfamiliar education systems and assessments that assess knowledge that they may have but can’t yet express in English. Meanwhile, more proficient users of EAL, who may have grown up in the UK, may still require support with developing specialist academic language. These differing needs require a strong focus on adaptive teaching.

While not without limitations, advances in AI present potential opportunities to support more adaptive and responsive teaching in everyday classrooms. This article therefore explores how generative AI might be used to address the diverse and evolving needs of EAL learners, drawing on research-informed principles of language development to support both language development and access to the curriculum.

In this article, ‘AI’ refers to freely available large language models (e.g. ChatGPT, Copilot, Gemini) that generate responses to text or speech prompts. These tools can produce text, audio or visuals, offering potential ways in which to support EAL learners’ access to both language and curriculum content. They offer a range of practical applications for supporting both access to the curriculum and language development, including:

• live translation (including text to speech or speech to speech)
• translation of key information into multiple languages
• modification of texts to suit the English proficiency of the learner, without losing content
• bespoke materials, such as visuals to accompany texts, language-focused tasks to address specific areas of vocabulary or grammar, or scaffolds such as glossaries and speaking and writing frames.

While these tools offer practical ways in which to support curriculum access and language development, their impact depends on how they are used (Guest, 2025). Overreliance on translation may limit opportunities for developing language, while dependence on AI-generated scaffolds may reduce teachers’ insight into learners’ needs. Given the limited evidence on the impact of AI use on language learning in the school setting, this article takes a cautious approach, exploring how AI might complement established, evidence-informed EAL strategies.

Integrating language and AI policies

A school-wide language policy provides an essential foundation for coherent and consistent provision for EAL learners. Schneider et al. (2016) highlight the importance of schools developing a whole-school language policy that sets clear expectations around the use of languages in school and classroom practice, including approaches to home languages.

As AI tools become more common in education, their use should sit within a clear language policy and wider teaching approach. Starting with pedagogy and whole-school language principles helps to make sure that AI supports and extends practice, rather than shaping or narrowing teaching and learning.

When reviewing or developing policy in this area, leaders might consider:

• whether AI tools used in school can effectively detect and filter prohibited content across all languages used, in line with the Department for Education’s 2026 product safety standards
• how the school’s policy on home languages is defined and what role AI-based translation or communication tools play within it
• whether the accuracy and reliability of AI translation tools have been evaluated across the full range of languages used in the school and in different contexts (e.g. pastoral care, curriculum content, social interaction)
• which types of school communication are appropriate for AI translation and when human interpretation or a professional interpreter is more suitable.

AI applications for supporting EAL learners in the classroom

The following sections explore how AI can enhance evidence-informed strategies for supporting EAL learners.

Using comprehensible English

Modifying language to suit the proficiency level of the learner is a crucial element in the language acquisition process (Lightbown and Spada, 2021), and AI can support this by:

• simplifying or grading classroom texts
• generating glossaries of key vocabulary
• breaking down complex instructions
• providing summaries in accessible language.

Table 1 shows an example of a prompt and the simplified text that it produced, alongside the original. This example shows how AI can restructure complex sentences into shorter, more accessible units, without reducing content. However, teachers would still need to critically engage with the text to check for accuracy and omissions of key language or information. As learners progress in their language development, such simplifications may not provide a good model for the type of language that we expect learners to produce, so it is important to provide models that reflect their developing proficiency in English.

Table 1: Example of an AI prompt and the generated result

Prompt: 

I am a Key Stage 3 science teacher, teaching a class about types of enzymes in the digestive system. I have some EAL learners in my class who are at the early acquisition stages of learning English. Can you simplify the following text to make it more accessible to them while retaining the key information?

Use of home languages

Bringing home languages into the classroom can have a range of benefits for learners of English (Cummins, 2007; Tian and Li, 2024). AI can be used to translate texts from other languages into English and vice versa and for quickly generating multiple versions of materials such as instructions, glossaries, rubrics and content in written or audio form. While AI models are improving their translation abilities, it is important not to view translation as a panacea for EAL support. Learners also need to develop their English alongside learning the curriculum, so its use should be targeted, particularly for new arrivals or those new to English, who otherwise have limited access to the language of learning. A key consideration is whether carefully graded English supported by visuals may be more effective than potentially inaccurate translation in a given context.

Here are some tips and considerations when using AI for translation:

• Be cautious when translating into less common languages or language varieties. AI works best when it has been trained on a lot of good examples of language, so it tends to be more accurate with languages that have a large written or online presence. There is evidence that AI is better at translating into English in comparison to out of English (Briakou et al., 2023).
• Modify/simplify the English text before translating it into the target language.
• Check AI translations into other languages with a proficient speaker or interpreter where available, and always in cases of critical or sensitive information.
• Be aware that AI may generate convincing but incorrect language that is not easily recognised as an error.
• Do not use students to check or translate sensitive information.

Using visuals

Visuals are a powerful tool for helping EAL learners to access the language used in the curriculum. The ability of AI to produce customised visuals means that it can be used to:

• create bespoke word mats or illustrated lists and glossaries specific to the lesson or topic
• create visual ‘social stories’ to explain school routines and expected behaviours (especially useful for new arrivals coming from different cultures and education systems)
• visualise scenes from stories or other texts to support understanding.

Tips for practitioners:

• Be selective about which words or concepts that you illustrate using visuals. Not all vocabulary items can be easily represented visually (e.g. abstract concepts such as ‘curiosity’ or ‘hope’), and too many visuals on a text can be overwhelming.
• Check AI imagery for biased/stereotyped representations and for specific cultural references or assumptions that may not be understood by EAL learners.
• Check AI-generated images for misspelled, inaccurate or even incomprehensible text.

Exploiting emergent language

Reacting to language development opportunities as they spontaneously occur in the classroom (e.g. providing learners with the phrase or word they need for a certain task) allows learners to acquire English that is directly connected to their curricular learning and communication needs (Chinn and Norrington-Davies, 2023). To facilitate this, AI could be used to:

• provide lists of synonyms or alternatives for a word or phrase to broaden EAL learners’ range of vocabulary
• provide immediate language-focused feedback on written or spoken texts (e.g. highlighting grammar mistakes or offering suggestions for improvement)
• explain the differences in usage between similar words or concepts (e.g. ‘interested’ vs ‘interesting’).

The accuracy of AI responses cannot be taken for granted, and students and teachers should be made aware of this and taught to use AI responses as the basis for critical discussion rather than accepting them as the final authority.

Assessment of language proficiency and feedback on language use

Evidence on AI for assessing language proficiency is mixed, with concerns about accuracy and validity. However, AI can support formative assessment when used cautiously to create:

• personalised assessment rubrics or success criteria focused on specific language demands of tasks
• curriculum-linked grammar or vocabulary assessments (e.g. gap-fills or multiple-choice quizzes)
• targeted feedback on student writing, highlighting specific areas such as tense use, sentence structure or cohesion
• model texts (e.g. exam responses) that exemplify the language features towards which learners are working.

These are complex processes, and AI may require detailed and multi-step prompts that outline the specific language areas where your learners need support. Guides such as the Cambridge Generative AI Idea Pack (2024) can provide tips and examples. The need to tailor AI prompts specifically to address language learning needs means that accurate and regular assessment of language proficiency, using a robust and well-established EAL assessment framework and setting targets for language development, is crucial for EAL learners.

Integrating language and content in lesson planning

By integrating language learning targets and content into lesson planning, teachers can potentially use AI to aid language-aware planning by asking AI to:

• adapt or suggest changes to lesson plans or schemes of work to make them more accessible to specific groups of EAL learners
• suggest adaptations to individual tasks or activities for learners at different proficiency levels
• provide a list of useful vocabulary or language structures that could be integrated into the teaching of a curricular scheme of work or lesson sequence.

As with using AI for feedback and formative assessment, these strategies rely on robust EAL assessment systems so that teachers are aware of and can refer to the proficiency levels and language needs of their learners, ensuring that suggestions are specifically targeted and useful rather than vague and generic. Experimenting with these processes alongside a colleague who has expertise in EAL or language learning will help to ensure that AI-generated suggestions are effective, and could also lead to useful discussions and opportunities for professional learning.

Conclusion

Learning a language requires sustained cognitive effort, and overreliance on AI to reduce this effort, such as through constant translation or automatic correction, may limit language development. Effective use of AI should therefore focus on easing the cognitive load associated with the dual challenge that EAL learners face: learning curricular content while developing English. Used thoughtfully, AI can support access without removing the productive effort needed for language acquisition. Ultimately, AI should be used in ways that support – not undermine – multilingual learners’ ability to develop both their English and their subject knowledge.

The examples of AI use and specific tools in this article are for context only. They do not imply endorsement or recommendation of any particular tool or approach by the Department for Education or the Chartered College of Teaching and any views stated are those of the individual. Any use of AI also needs to be carefully planned, and what is appropriate in one setting may not be elsewhere. You should always follow the DfE’s Generative AI In Education policy position and product safety standards in addition to aligning any AI use with the DfE’s latest Keeping Children Safe in Education guidance. You can also find teacher and leader toolkits on gov.uk.