Fluency or Accuracy?

The ELT Identity Crisis No One Can Ignore

By Emy Stamou Teacher, Academic Associate, Researcher, University of Thessaly, Greece

In 1985, two influential accounts of second-language acquisition placed different demands on the language classroom. Stephen Krashen argued that acquisition depended primarily on learners understanding language slightly beyond their current competence. Merrill Swain, drawing on research in Canadian immersion programmes, proposed that comprehension alone could leave important gaps in learners’ productive ability. The distinction still matters.

Walk into one English lesson and the teacher may be explaining verb forms, correcting errors and asking students to produce structurally accurate sentences. In another, students may be debating, negotiating meaning and speaking with considerable freedom. Mistakes pass without interruption because the immediate objective is communication.

Language is being used in both rooms. Whether the same kind of learning is taking place is harder to determine.

For many learners, fluency provides the most visible evidence of progress. Words arrive more quickly. Sentences become easier to form. Speaking no longer feels like a test of endurance. A learner who once answered with a few guarded words can sustain a conversation, explain an opinion and recover when communication breaks down.

That progress is real. Yet fluent speech can conceal persistent problems. The learner communicates successfully while continuing to rely on inaccurate grammar, limited vocabulary or pronunciation that occasionally obstructs meaning. An error repeated often enough may become difficult to dislodge.

Teachers then face a familiar decision: correct the language or protect the conversation.

Frequent interruption may make learners more accurate in the moment, but it can also produce cautious speakers who monitor every sentence before saying it. Allowing every error to pass creates a different risk. Learners may become comfortable with language that works well enough for immediate communication but never develops much further.

Krashen’s Input Hypothesis changed the terms of this discussion. In his account, acquisition occurs when learners understand meaningful language containing features beyond their present level. Speech develops from sufficient comprehensible input rather than from the systematic teaching and correction of one structure after another. His wider Monitor Model also distinguished subconscious acquisition from conscious knowledge of grammatical rules. Krashen’s published account gave teachers a reason to place comprehension, meaning and low-anxiety exposure closer to the centre of instruction.

The theory was highly influential, but its stronger claims remained contested. Learners can understand far more than they are able to produce, especially in settings where listening and reading greatly exceed opportunities for sustained speaking and writing. Comprehension may develop without equivalent grammatical control in production.

Swain encountered this problem among learners in Canadian French immersion programmes. Students had received extensive, meaningful input and could understand the language well, yet weaknesses remained in their spoken and written production. Her Output Hypothesis proposed that producing language performs functions that comprehension cannot perform alone.

When learners attempt to express a precise meaning, they may discover that their existing resources are inadequate. A missing word, an uncertain tense or an awkward sentence forces them to pause. They reformulate, test a possibility, ask for help or compare their version with somebody else’s. Swain described this as noticing a gap between what learners want to say and what they can currently say. Output can therefore become part of the learning process, rather than simply its final product. Her later work also identified hypothesis testing and reflection on language as possible functions of production. Research discussing these mechanisms examined how producing language might contribute to acquisition.

This leaves teachers with no neat contest in which input defeats output or fluency displaces accuracy. Learners need language they can understand, opportunities to use it and reasons to pay closer attention to form. The proportion changes according to the learner, the task and the stage of instruction.

Correction is equally dependent on context. An error that makes no difference during an energetic discussion may deserve attention afterwards. The same error may require immediate intervention during pronunciation practice, examination preparation or work on a recently taught structure. Timing matters as much as the correction itself.

Similar tensions appear elsewhere in ELT.

Pronunciation teaching, for example, once commonly treated British or American models as the destination. International communication has complicated that assumption. Many learners use English with speakers who have different first languages, accents and cultural references. Intelligibility may therefore be a more relevant goal than imitation of a particular national accent. Individual learners may still want such a model, particularly for professional, academic or personal reasons.

The use of learners’ first language has undergone comparable reconsideration. English-only policies were often presented as evidence of methodological seriousness. Teachers now make more selective judgements. A brief explanation, comparison or translation can sometimes prevent confusion and free time for meaningful work in English. Used without restraint, the first language may reduce exposure. Used purposefully, it can support learning.

These questions become daily classroom choices:

Should the teacher correct the error or allow the conversation to continue?

Is the planned activity still useful, or has the learners’ response created a better opportunity?

Does the class need simpler input, or have the students been protected from difficulty for too long?

The answers will not remain consistent across every lesson. A teacher may tolerate errors during one activity and insist on precision ten minutes later. Free discussion may follow tightly controlled practice. Learners who need reassurance may receive a different response from those who have become too comfortable with approximation.

From outside the classroom, this can look inconsistent. In practice, it may be evidence of informed judgement.

Learners also need help understanding these shifts. Otherwise, one teacher’s attention to grammar may appear incompatible with another’s emphasis on free speaking. Correction can feel like failure, while an absence of correction may be mistaken for proof that everything is accurate. Explaining the purpose of an activity makes these differences easier to interpret.

Mistakes are evidence that learners are attempting something their current language cannot yet manage. That does not make every mistake valuable or every correction helpful. Its value depends on what happens next: whether the learner notices it, receives useful feedback, tries again or encounters the form in a context that makes it memorable.

ELT has spent decades searching for methods capable of resolving such tensions. Classrooms keep resisting the search. Learners arrive with different histories, ambitions, anxieties and reasons for learning English. A practice that releases one learner’s language may expose another’s gaps.

Effective teaching depends less on allegiance to a single theory than on knowing what each theory helps the teacher notice. Input provides the language from which learners can build. Output tests what they can do with it. Fluency gives them room to communicate. Attention to accuracy helps that communication become more precise.

The work of teaching lies in deciding what the learner needs now.

How I Use ChatGPT to Teach IELTS to A2 and B1 Students
(A Teaching Method I Never Thought Would Have Worked)

By Eliza Marabyan

English language educator, blogger, and author

For years, I had one firm rule as an IELTS teacher: if a prospective student was around A2 level, I politely declined.

It wasn’t because I didn’t want to help them. Quite the opposite. I simply believed that IELTS preparation should begin only after students had reached at least a solid B1 level. Otherwise, every lesson would become a General English lesson rather than focused exam preparation.

And honestly, I still think there is some truth in that.

An A2 learner is usually not ready for authentic IELTS reading passages, academic listening recordings, complex writing tasks or fluent discussions. Throwing them straight into official IELTS materials is like asking someone who has just learned to ride a bicycle to compete in the Tour de France.

Then ChatGPT completely changed the way I teach.

Today, I no longer reject lower-level students automatically. Instead, I build what I like to call a Mini IELTS Pathway. Rather than exposing students to the full difficulty of the exam from day one, I recreate every section of the IELTS exam using language that matches their current level. Month by month, we gradually increase the complexity until the student is finally ready to tackle authentic IELTS materials.

The results have genuinely surprised me.

The Student Who Changed My Mind

Several months ago, a student contacted me asking for IELTS preparation. After assessing her level, I concluded that she was approximately A2.

A few years ago, I would have immediately explained that she first needed several months of General English before we could even think about IELTS. This time, however, I decided to experiment.

I explained one very important thing from the beginning.

“These are not real IELTS tasks. They are simplified IELTS-style activities designed specifically for your current level. Think of them as miniature versions of the real exam.”

Setting expectations was essential. Students need to understand that these activities are stepping stones rather than substitutes for official IELTS practice.

Once she understood this, we got to work.

Why ChatGPT Makes This Possible

The biggest obstacle to teaching IELTS to lower-level students has always been lesson preparation.

Finding or creating A2 versions of IELTS tasks used to take hours. A single reading lesson often meant rewriting an authentic article. A writing task had to be redesigned from scratch. Listening activities required searching for suitable recordings or creating my own.

Now I simply describe my student’s level and ask ChatGPT to generate an IELTS-style activity at exactly the right level. Within seconds, I have a lesson that would previously have taken an hour or more to prepare.

The secret is not asking ChatGPT to create “an IELTS lesson.” Instead, I ask it to imitate the structure of IELTS while adapting the language to my student’s current ability. That small difference changes everything.

Reading: Creating Mini IELTS Passages

Instead of giving my A2 learner difficult academic texts filled with unfamiliar vocabulary, I asked ChatGPT to create shorter passages using A2 grammar and vocabulary while preserving authentic IELTS question types.

One of my favourite prompts is:

2 Prompt

Create an IELTS-style Reading task for an A2 English learner.

Topic: Healthy Lifestyle

Length: 180-220 words.

Include:

True / False / Not Given

Matching headings

Multiple-choice questions

Use only A2 vocabulary but imitate the IELTS format. Present everything as a visually attractive worksheet.

After two months, I made just one small adjustment.

Instead of writing A2, I changed it to B1. Nothing else.

That tiny modification immediately increased the difficulty without overwhelming the student.

Other reading topics that worked particularly well included environmental issues, education, technology, travel, healthy eating, online shopping, famous inventions, festivals, wildlife and sports psychology. My student gradually became familiar with IELTS question types without struggling with impossible vocabulary.

Speaking: Building Confidence Before Complexity

Speaking is often the most intimidating section for lower-level students.

If you ask an A2 learner a genuine IELTS Part 3 question like:

“To what extent should governments regulate social media?”

the conversation is usually over before it even begins.

Instead, I asked ChatGPT to recreate the IELTS speaking format using language my student could actually understand.

2 Prompt

Create an IELTS Speaking lesson for an A2 learner.

Include:

IELTS Part 1 style questions

A picture description similar to Part 2

Simplified Part 3 discussion questions

Useful vocabulary before each section

Follow-up questions

Format everything as a colourful worksheet.

At first, I only expected one or two sentences. Later, I encouraged three. Then five.

Eventually, complete answers with reasons, examples and opinions.

The format stayed familiar while the language became progressively more sophisticated.

Writing: Teaching IELTS Essays Without Writing Essays

Many teachers postpone IELTS writing until students become intermediate.

I decided to try something different.

Instead of asking my student to produce a 250-word essay, I created simplified versions of IELTS tasks.

For example, instead of discussing both views, she initially wrote six sentences expressing a simple opinion.

Six sentences became ten. Ten became one paragraph.

One paragraph became two.

Eventually, she was producing structured responses with introductions, supporting ideas and conclusions.

One of my favourite prompts was:

Create an IELTS-inspired opinion writing task for an A2 learner. Maximum 100 words. Include useful vocabulary, sentence starters, planning ideas and a model answer.

Later, I switched to: Rewrite this IELTS Task 2 question for a B1 learner while preserving the original argument. Include a Band 5 model answer and explain why it deserves that score.

This approach allowed my student to master IELTS structure long before worrying about advanced grammar.

Listening Without Audio Files

Listening was probably my biggest breakthrough.

Authentic IELTS recordings were simply too difficult, so I found another solution.

I asked ChatGPT to generate IELTS-style listening scripts.

Then I became the speaker.

This gave me complete control over the speed, pronunciation and repetition. If necessary, I slowed down. Sometimes I repeated difficult expressions or emphasised key words. As my student’s listening improved, I gradually increased my reading speed until it sounded like natural conversation.

My prompt looked like this:

Create two IELTS-style listening exercises for an A2 learner.

Include:

A complete script,

Gap-fill questions,

Multiple-choice questions,

Short-answer questions,

Natural distractors,

Highlight the key information naturally without making it obvious.

This simple technique proved far more effective than struggling through recordings that were well above her level.

Turning Every Lesson into a Professional Worksheet

Another feature I use constantly is asking ChatGPT to create visually appealing materials.

Instead of receiving plain text, I finish many prompts with:

Present everything as a modern, colourful worksheet with clear headings, icons, tables, boxes and plenty of white space suitable for printing.

Students immediately respond more positively when lesson materials look polished and professional.

Increasing the Difficulty Month by Month

One mistake many teachers make is increasing the difficulty too quickly.

For the first two months, almost everything remained at A2 level.

Once my student felt comfortable, I switched nearly every prompt from A2 to B1. Later, I introduced longer reading passages, more academic vocabulary, increasingly complex writing tasks and more challenging speaking questions.

Only after building confidence did we begin working with authentic IELTS materials.

The transition felt natural rather than overwhelming.

My Favourite ChatGPT Prompts for Lower-Level IELTS Students

Here are some prompts that have become permanent parts of my teaching workflow.

Vocabulary Builder

Create 25 IELTS-related vocabulary words on the topic of education for a B1 learner. Include pronunciation, definitions, collocations, example sentences and a gap-fill exercise.

Grammar Focus

Design an IELTS grammar lesson on relative clauses for B1 students. Include explanation, controlled practice, communicative activities and homework.

Reading Upgrade

Rewrite this authentic IELTS reading passage at B1 level while keeping the same question types and overall structure.

Speaking Extension

Create 20 follow-up questions based on my student’s answers. Increase the level of difficulty gradually and encourage longer responses.

Pronunciation Practice

Design an IELTS pronunciation lesson focusing on word stress, sentence stress and connected speech using common IELTS vocabulary.

Listening Practice

Write an IELTS-style dialogue about university life suitable for B1 learners. Include distractors similar to those used in the real exam.

Essay Expansion

Expand these six simple sentences into a Band 6 IELTS Task 2 essay. Explain every improvement step by step.

Error Correction

Correct my student’s IELTS essay. Explain every grammar mistake, suggest stronger vocabulary, estimate the band score and provide an improved version.

Paraphrasing Practice

Generate ten paraphrasing exercises using common IELTS Task 2 vocabulary. Start at B1 level and gradually increase the complexity.

Vocabulary Recycling

Create ten speaking questions that naturally recycle the following IELTS vocabulary without making the questions sound artificial.

More Creative ChatGPT Activities for A2 and B1 IELTS Learners

Beyond the four exam skills, ChatGPT has helped me create engaging activities that simply didn’t exist before.

Picture-to-Essay Challenge

Upload a picture and ask:

Create an IELTS-style writing lesson based on this image for a B1 learner. Include vocabulary, brainstorming questions, useful phrases and a model answer.

Band Upgrade Race

Provide a Band 4 essay and ask ChatGPT to improve it to Band 6, explaining every change. Students quickly notice patterns in vocabulary, grammar and organisation.

IELTS Detective

Ask ChatGPT to hide ten grammar mistakes inside a Band 5 essay. Students work together to identify and correct them.

Question Generator

Paste a reading passage and ask ChatGPT to generate additional True/False/Not Given, Matching Headings and Multiple Choice questions.

Vocabulary Ladder

Ask ChatGPT to replace every A2 word with a B1 alternative, then every B1 word with a B2 alternative. Students instantly see how vocabulary becomes more academic.

Role Play Interviews

Create examiner-candidate role plays based on familiar IELTS topics such as travel, hobbies, education or technology. Students practise fluency without feeling intimidated.

What Happened After Two Months?

The improvement was impossible to ignore.

My student no longer panicked when she saw reading passages. She recognised IELTS question types, understood how writing tasks were organised and produced much longer speaking answers.

Most importantly, she stopped seeing IELTS as something impossible. That psychological transformation was just as valuable as the language improvement itself.

The Biggest Lesson I’ve Learned

ChatGPT did not replace my expertise. It amplified it.

I still assess my students, decide what comes next, explain grammar, correct mistakes, teach strategies and provide feedback.

What ChatGPT has changed is the amount of time I spend creating personalised materials. Instead of manually adapting every worksheet, I can now generate level-appropriate lessons in minutes. That has allowed me to say “yes” to students I would once have turned away.

Would I recommend preparing a complete beginner for IELTS? Probably not. Students still need realistic expectations, and there is no shortcut to reaching the language level required by the exam.

However, if you adapt the format carefully, gradually increase the complexity and focus on building confidence first, lower-level learners can begin their IELTS journey much earlier than many teachers believe.

Bottom Line

One of the biggest misconceptions about AI is that it simply produces lesson plans more quickly.

In reality, its greatest strength is personalisation.

Every learner can receive materials designed specifically for their current level instead of struggling through resources intended for someone far more advanced.

That is exactly what happened with my A2 student. Instead of waiting months before introducing IELTS, we built a bridge towards it. Two months later, she was confidently working with B1-level IELTS-style tasks, and each lesson brought her one step closer to authentic exam materials.

As teachers, we often say that we should meet students where they are. With ChatGPT, doing exactly that has never been easier.

AI Won’t Replace Teachers, But It Will Replace Boring Lesson.

By Eirini Louizou

Owner and Director of a Foreign Languages Centre in Athens

“Will AI replace English teachers?”

It’s probably the question every educator has heard at least once this year. Teachers discuss it in staff rooms and at conferences. Every new AI tool seems to bring a new wave of excitement—and concern.

But perhaps we’re asking the wrong question.

AI is unlikely to replace great teachers. What it will replace are repetitive tasks, uninspiring lessons and hours of preparation that leave teachers exhausted before they even step into the classroom.

Teaching Has Never Been About Delivering Information

For many years, teachers were seen as the main source of knowledge. We explained grammar, introduced vocabulary, corrected exercises and prepared students for exams.

Today, AI can do many of those things in seconds. It can generate worksheets, write reading texts, create quizzes, suggest lesson plans and even adapt activities to different language levels.That sounds impressive—and, for some, a little intimidating.

The internet has provided instant access to information for decades, yet students still need teachers.

Why? Because learning has never been just about information. It has always been about connection.

Students Remember People, Not Worksheets

Think back to your own school days.

Most of us don’t remember every grammar rule we learned or every worksheet we completed.

What we remember are the teachers who encouraged us, believed in us and made learning enjoyable.

Students need to feel safe enough to make mistakes, ask questions and try again. That confidence grows through positive relationships—not through technology.

No AI tool can recognise the nervous smile of a student speaking English for the first time. No chatbot can notice when someone has had a difficult day and simply needs a little encouragement.

AI Can Save Time—If We Let It

Let’s be honest.

Many teaching tasks are repetitive. Creating another vocabulary worksheet. Writing comprehension questions. Looking for a suitable reading text. Preparing extra grammar practice for one student who needs additional support.

Instead of spending an hour creating materials, teachers can produce a first draft in minutes and then adapt it to suit their learners.

The real value of AI isn’t that it does our job. It’s that it gives us something we never seem to have enough of: Time.

The Best Lessons Still Need a Teacher

Imagine asking AI to create a role-play about ordering food in a restaurant.

Within seconds, the dialogue is ready.

But the magic doesn’t happen when the worksheet is printed.

It happens when the teacher transforms that activity into a memorable lesson—adding humour, encouraging reluctant learners, asking follow-up questions, introducing real-life situations and creating opportunities for meaningful communication.

The worksheet is only the beginning. The teacher creates the experience. That’s a difference no technology can replace.

The Skills That Matter Most

Recent reports from UNESCO and the OECD emphasise that AI should support teachers rather than replace them.

They need creativity, critical thinking, communication, collaboration.

These are exactly the skills that develop through discussion, teamwork and authentic interaction—areas where teachers continue to play an irreplaceable role. Technology can generate content.

A Smarter Way to Use AI

The most successful teachers won’t ignore AI, but they won’t depend on it either.

Instead, they’ll use it wisely.

In other words, AI becomes an assistant rather than the person leading the lesson.

The Future of ELT Is Still Human

Every major technological change has raised fears about the future of education.

The internet didn’t replace schools. Interactive whiteboards didn’t replace classrooms. AI won’t replace English teachers either.

What it will do is encourage us to rethink how we spend our time and where we can make the greatest difference.

The teachers who thrive won’t necessarily be those who know the most about technology.

They’ll be the ones who know how to combine technology with empathy, creativity and genuine human connection.

Final Thoughts

Artificial Intelligence is changing education, and there is no reason to fear that change.

Used wisely, it can reduce routine work, generate fresh ideas and help us personalise learning more efficiently than ever before.

But when students look back on their education years from now, they won’t remember the AI tool that generated a worksheet. They’ll remember the teacher who inspired them to speak, encouraged them after a mistake, celebrated their progress and made them believe they could succeed.

Perhaps that’s the real future of education.