7 ChatGPT Prompts for ESL Teachers

The seven ChatGPT prompts in this post cover tasks that consume a lot of lesson prep time for ESL and language teachers: generating level-appropriate reading texts, extracting vocabulary from authentic materials, creating professional role-play dialogs, building grammar exercises, writing comprehension questions, designing vocabulary review activities, and planning weekly practice schedules. Each prompt is copy-paste ready with CEFR level, learner context, and output format built in.

"Just use ChatGPT" or any other AI chatbot is advice that's easy to give and may be hard to act on. The gap between knowing the tool exists and knowing which prompts actually produce classroom-ready materials is wide. Most teachers who've tried it have stories about outputs that were technically fine but practically useless — wrong level, wrong register, generic examples that had nothing to do with the learner.

The good news is that the problem isn't the AI chatbot. It's often the prompt. A vague request produces vague output. A specific, well-structured prompt produces something you can use today. The following prompts cover tasks that eat prep time: reading texts, vocabulary lists, role-play dialogs, grammar exercises, comprehension questions, vocabulary review, and practice planning. I've tried to ensure each is fully written out and ready to copy-paste, with notes on what makes it work and what a real output looks like. The prompts should work equally well for ChatGPT, Claude, Gemini, or any large language model.

Why Does Prompt Structure Matter for Language Teachers?

Specificity is what separates a usable ChatGPT output from a generic one. The more concrete you are about the learner's CEFR level, profession, and target skill, the closer the result lands to something you can use in your next lesson. A simple way to be specific is to think in four parts every time you write a prompt: the role you want the AI to play, the task you're asking it to do, the context of your learner, and the format you want back. You don't need to memorize this as a formula — just check before hitting send that ChatGPT knows who the learner is, what level they're at, and exactly what you need.

This is already mainstream territory. A 2024 British Council survey of 1,348 English language teachers across 118 countries found 75% are already using AI tools — most commonly for creating learning materials (57%) and lesson plans (43%). The prompts below target exactly those use cases.

For a deeper look at how to approach AI prompting, see our guide to prompting fundamentals.

How Do You Generate a Level-Appropriate Reading Text?

Specify the CEFR level, the learner's profession, the topic, the word count, and the register in a single prompt, and ChatGPT will generate a targeted reading text you can use immediately. Without those inputs, "B1 reading on supply chains" usually returns something that drifts a level higher or lower and uses vocabulary you wouldn't have chosen — which leaves you rewriting it. Specifying all five up front turns ChatGPT into a draft generator for one of the slowest parts of prep: finding (or making) a text that's challenging but not overwhelming for this particular learner.

You are an ESL materials developer. Write a 250-word reading text for a B1-level adult learner who works in logistics and is preparing for business meetings in English. The text should be about supply chain disruptions, use a professional but accessible register, and avoid idioms or complex grammatical structures. Include 5-8 vocabulary items that are useful for this learner's context and emphasize them in the text.

Why it works: The CEFR level (B1), profession (logistics), topic (supply chains), length (250 words), register (professional), and format (vocabulary underlined) are all specified and drives ChatGPT towards a useful output. 

Customization: Swap in any CEFR level (A2 for beginners, C1 for advanced), any topic, and any profession to get a text tailored to your next learner. You can also try to add the learners interests for a more engaging text.

For a full walkthrough on creating reading materials, see also our post on how to create level-appropriate reading materials with AI.

How Do You Extract a Vocabulary List from Authentic Materials?

Paste your authentic text into the prompt along with the learner's L1, profession, and CEFR level — ChatGPT will extract relevant vocabulary with definitions, example sentences, and translations in one step.

Authentic materials are valuable, whether news articles, business emails, or industry reports. But extracting vocabulary, adding definitions, and writing example sentences takes time you don't have. This prompt handles all of it in one step.

You are an ESL tutor preparing materials for a Spanish-speaking B2 learner who works in marketing. Extract the 10 most useful vocabulary items from the text below and present them in a table with: the word or phrase, its definition in simple English, an example sentence relevant to the learner's marketing context, and the Spanish translation.

[Paste your text here]

Why it works: It extracts vocabulary from your chosen authentic material (not a generic list), personalizes definitions and examples to the learner's profession, and includes L1 translation for reference.

Customization: Change B2 and Spanish to any other level and language as needed. You can also experiment with changing "10 most useful" to "10 most difficult". You may even request specific word classes or words denoting certain things. Try to experiment and see how far you can make it work and where the result starts to fail. 

This prompt works well with news articles, company reports, or any text your learner has already encountered. The vocabulary feels immediately relevant because it came from something real.

How Do You Create a Role-Play Dialog for Professional Learners?

Define the learner's profession, a realistic scenario they'll face at work, the formality level, and the dialog length. The result mirrors something the learner will actually do, not a generic coursebook exchange. "Ordering coffee" is fine for low-stakes vocabulary, but it does nothing for a software sales rep preparing for a discovery call with a skeptical client, or a project manager handling a difficult stakeholder conversation. Asking for the role-play your learner will actually face turns ChatGPT into a near-bottomless source of relevant practice material.

You are an ESL materials developer. Write a role-play dialog for a B2-level learner who is a software sales executive preparing to conduct discovery calls in English with potential clients. The dialog should be between the sales executive and a skeptical potential client at a mid-sized tech company. The tone should be professional and moderately formal.

Length: 12-15 exchanges. Include a brief situation description at the top.

Why it works: The dialog reflects the learner's actual professional context. Compare this to a generic "making appointments" dialog from a coursebook. There is no comparison in engagement or usefulness.

Customization: Change the profession and scenario to match any learner: a lawyer preparing for client consultations, a doctor practicing patient conversations in a second language, a customer service manager handling complaints.

For more on how AI helps independent tutors, read how AI can empower freelance language teachers.

How Do You Build a Grammar Exercise for a Specific Structure?

Name the grammar structure, set the CEFR level, and contextualize the sentences to the learner's profession. You get a ready-to-share worksheet rather than a list of generic textbook examples. A nurse drilling present perfect on patient handovers retains the structure differently from the same learner doing the same exercise about travel. The grammar point isn't more or less learnable in either context, but examples that match the learner's daily working life are easier to remember and more useful when they next need the form for real.

You are an ESL tutor. Create 8 fill-in-the-blank sentences practicing the present perfect tense for a B1 learner who is a nurse. All sentences should relate to healthcare situations she encounters at work. Provide the answer key separately at the bottom.

Why it works: The grammar point is explicit (present perfect), the level is set (B1), the context is specified (healthcare / nursing), and the format is clear (fill-in-the-blank with answer key). The result is a worksheet you can share directly, not a list of generic examples about travelling or having breakfast.

Customization: Try swapping profession, situation description and grammar concept. You may also indicate more specific requirements on the grammar exercises, for instance "practicing the present perfect tense versus past simple tense".

For more complex grammar points, you can also ask ChatGPT to include a brief rule explanation at the top. Useful if you want to send the exercise as self-study homework.

How Do You Write Comprehension Questions for Any Text?

Ask ChatGPT for three question types in a single prompt — factual recall, inference, and opinion — to cover the full range from basic comprehension to discussion-ready conversation starters. The recall questions confirm the text was understood; the inference questions sharpen reading skill; the opinion questions turn the same reading into conversation fuel. Generating all three together means every text you bring into a session can do double duty without you drafting question sets one at a time.

You are an ESL tutor preparing a lesson. Based on the text below, write comprehension questions at three levels: 3 factual recall questions with short answers, 2 inference questions where the learner must read between the lines, and 2 opinion questions that invite the learner to relate the topic to their own experience. Format each section clearly.

[Paste your text here]

Why it works: It is again specific on the role of the AI, and what is needed. Factual questions check basic comprehension. Inference questions help develop critical reading skills. Opinion questions may turn reading into conversation. 

Customization: If you don't need the three types of questions just ask for "comprehension questions" and skip the rest of that sentence. You may also say "write yes/no comprehension questions" or "write comprehension questions with 4 options, one being the correct answer and the other distractors". Try experimenting to get what you need.

if the text does not fit what we are asking, e.g. does not have something to read between the lines, or the AI doesn't catch it, it may produce something odd. But try it out and review and just delete questions that are off.

How Do You Create a Personalized Vocabulary Review Exercise?

Feed previously taught vocabulary back into ChatGPT with a new professional context, and the result is a spaced-repetition exercise that feels like an extension rather than a re-run of last week's list. Vocabulary needs to reappear across sessions to stick, but learners notice when they're being shown the same five words a second time in the same format. Recycling the items inside a fresh sentence-completion or scenario task keeps the review productive without making the recycling visible.

You are an ESL tutor. Using the following vocabulary items from a previous lesson, create a sentence-completion exercise where each sentence describes a realistic work scenario relevant to a B2 learner who works in project management. Each sentence should have one blank where the target word fits naturally. The answer key should follow at the end.

Vocabulary: scope creep, stakeholder, milestone, deliverable, bandwidth

Why it works: It is again specific on the role of the AI, the level, and what is needed. It recycles vocabulary the learner already encountered (spaced repetition in practice). The work scenario framing keeps it relevant rather than abstract.

Customization: Change level, profession and vocabulary to match your learners. If your learners need more assistance, you might add "Add the blanked word in Spanish in parenthesis after the sentence" or whatever language matches your learner.

Use this after any lesson where vocabulary was introduced. It takes 30 seconds to write the prompt; ChatGPT handles the rest.

How Do You Plan a Weekly Practice Schedule for a Learner?

Give ChatGPT the learner's CEFR level, available daily time, target skill, and timeline, and it returns a structured self-study plan you can edit rather than build from scratch. This is the one prompt in the post that uses ChatGPT as a planner rather than a content generator. Instead of asking it to make materials, you're asking it to sequence practice across a week. The output won't be perfect, but having a five-day skeleton on the page in 30 seconds beats staring at a blank week and trying to design one from zero.

You are an experienced ESL learning coach. Create a realistic 5-day self-study plan for a B1-level adult learner who has 20 minutes per day on weekdays. Her goal is to improve spoken fluency for professional meetings in 8 weeks. She is a Spanish speaker. Each day's plan should include the activity type, the time allocation, and brief instructions she can follow independently. Focus on activities that don't require a live teacher.

Why this works: Again being specific helps; about level, time, learner and output. The plan may not be perfect, but it give you a starting point that you can more quickly adjust.

Customization: Change the specifics to your learners and their goals. You may also experiment with the output format and even get it to generate exercises if the goal is something else than speaking. 

It's a valuable tool when a learner asks "what should I practice between sessions?" You now have a concrete, personalized answer.

What Tips Help You Get Better ChatGPT Results?

Three principles make the biggest difference as you work with these prompts.

First, always specify the CEFR level explicitly (A1 through C2). "Intermediate" means something different to every teacher and even more different to ChatGPT. B1 or B2 leaves less ambiguity.

Second, include the learner's profession or area of interest whenever possible. A doctor and a software developer at the same CEFR level may need completely different vocabulary and example sentences. The specificity costs you a few words in the prompt and dramatically changes the output quality.

Third, review before you share. ChatGPT occasionally gets facts wrong or writes example sentences that are technically correct but semantically odd. A 60-second read-through catches those issues. You're still the one deciding whether the material is appropriate for your learner, the AI just reduces the time you spend writing it.

For more tip and tricks and a more comprehensive collection of prompts organized by teaching task, you can also explore our free AI Guide for Language Teachers.

The goal here isn't to hand over your lesson planning. It's to eliminate the mechanical parts of it so you can focus on what no tool can replace: knowing your learner, adapting in the moment, and building the kind of relationship that makes language learning stick.

Once you've generated the material, you still need to get it to your learners. If you're curious how that could work in practice, give Edumo a try.