Speaking more, speaking better

Introduction – ‘What if they could speak more?’

In most English language classrooms, speaking practice is a precious resource. Even in well-planned lessons, a class of 20 might give each student under two minutes of speaking time. Add performance anxiety, large groups and time pressure, and fluency often stalls. But what if students could speak more, not just in class, but before, after and even outside it? What if they had access to conversation partners, pronunciation feedback and structured practice without us needing to be there every minute? This is where artificial intelligence (AI) is beginning to make a difference.

In this article, I share AI-powered tools I’ve used in my own teaching. These aren’t futuristic experiments – they’re real, often free and they help students build fluency, improve pronunciation and stay engaged.

More importantly, I offer a practical framework that shows when and how to use AI to support speaking across four stages:

• pre-class
• in-class
• post-class
• autonomous.

Every tool comes from real classroom use, not theory. These are things I’ve assigned, tested and adapted based on what works with real learners. This isn’t about replacing the teacher – it’s about expanding what’s possible.

A teacher’s framework: four types of AI-supported speaking

When AI enters the language classroom, we often talk about automation or chatbots. But if we focus on teaching, a more useful question emerges:

When can AI help students practise speaking?

Over time, I found that tools naturally fell into four roles depending on classroom rhythm.

• Pre-class – preparing, diagnosing or idea-generating
  e.g. students brainstorm or take a diagnostic test.
• In-class – practising, modelling or extending talk time
  e.g. students speak with a grammar bot or AI character.
• Post-class – reviewing, reflecting or completing assignments
  e.g. learners record a role play or complete pronunciation drills.
• Autonomous – practicing independently over time
  e.g. a structured 24-week fluency or pronunciation plan.

These categories aren’t rigid. Tools like SmallTalk2Me work well both before and after class. But this structure helps me decide when to use each tool, especially when time is tight.

Tool highlights:
from theory to practice

Below are the tools I’ve used at each stage of this framework. These aren’t tech reviews, they’re real examples of how I integrated each one into class, what it achieved and why it worked.

Pre-class activities

Before lessons, AI tools can help students activate background knowledge, generate ideas or assess readiness. These tasks align with weekly speaking goals.

Brainstorming and idea generation

Students use tools like ChatGPT Voice Mode, Pi or Character AI to talk about the week’s topic (e.g. fast fashion, public transport). They can speak in their home language first or go straight to English – the goal is to build comfort with the theme.

Diagnostic assessment

SmallTalk2Me provides a CEFR-level report based on a short speaking task. It analyses fluency, grammar, pronunciation, vocabulary and interaction.

How I used it

At the start of the term, I asked students to complete the free diagnostic. The feedback helped me group learners, set individual goals and compare progress later. It’s a great way to identify who needs what before a speaking task begins.

In-class activities

For many teachers, in-class AI integration remains the most unfamiliar and under-explored territory. This is exactly where I focused my most creative efforts, designing activities where students could speak, receive feedback and stay engaged using AI.

Brainstorming with CoPilot

To activate students’ ideas at the start of class, I used Microsoft CoPilot (built into Microsoft Edge) as a live brainstorming tool. I projected a voice chat and asked CoPilot questions like ‘What are some negative effects of globalisation?’ or ‘Why do people choose fast fashion?’. Students listened, responded and debated the AI’s suggestions – instant energy, zero prep.

Individual voice practice with Pi or Character AI

To increase individual speaking time, I set up a rotation: students wore headphones and talked to an AI assistant before partner work. Some used Pi, others used Character AI. They scanned a QR code to start a voice call.

• With Pi, students received flexible conversation support and vocabulary help.
• With Character AI, they could roleplay with themed characters (travel agent, angry customer) or even clone my voice for added immersion.

Grammar practice with Mizou

To reinforce grammar in context, I used Mizou.com to create custom bots. For example, after a lesson on past tenses, I created a bot that asked questions like ‘What did you do last weekend?’. Students had to use the correct tense and got immediate feedback if they didn’t.

These tools transformed my classroom. AI allowed students to speak more, get real-time feedback and enter communicative tasks better prepared. Teachers don’t need advanced tech, just a QR code, a headset and a simple task with a purpose.

Post-class and autonomous assignments

In many cases, the tools students use independently are also excellent for post-class assignments. The distinction is less about the tool itself and more about who initiates the use – the student or the teacher. When teachers assign tools with intention, AI becomes a powerful extension of classwork.

Here are some tools that can be given as structured post-class tasks.

FlowSpeak – structured, diagnostic speaking practice

FlowSpeak offers a complete speaking solution that combines diagnostic assessment, CEFR-aligned courses and phoneme-level pronunciation feedback. Unlike many AI tools, all audio models are human-recorded, providing authentic listening and speaking targets.

Learners begin with a diagnostic task that pinpoints their speaking level and pronunciation gaps. From there, they follow structured courses – ranging from beginner to advanced, as well as scenario-based tracks like Smart Shopper, On Vacation and IELTS Booster.

Each mini-lesson takes just two–three minutes but provides granular feedback on each word and even specific sounds. Students can re-record, compare with L1 models and track progress over time.

I used FlowSpeak to assign weekly pronunciation practice. Students received pinpoint feedback and could retry until they improved. For me, it added structure and visibility to something that’s usually hard to track.
– Learner A

FlowSpeak stands out for its balance of structure, feedback and learner autonomy – a rare combination that makes it ideal for post-class or independent fluency work.

Character AI

Assign students to interact with a specific character (job interviewer, tourist, hotel receptionist) and submit their conversation audio. This is especially effective for reinforcing situational language.

GetPronounce.com

Ask students to complete pronunciation-focused tasks (read aloud three–four sentences, complete a vocabulary chat) and reflect on their feedback. It’s great for helping learners recognise phoneme-level issues.

ELSA Speak

Assign a specific pronunciation goal or role-play scenario from ELSA’s lesson bank. Teachers can ask for screenshots of scores or audio recordings as evidence of completion.

For learners who want to improve pronunciation and structured fluency on their own, FlowSpeak is the most robust tool I’ve used. Its CEFR-aligned courses are broken into short daily lessons that provide feedback on every word – and even individual sounds.

How I used it

I assigned modules from the Advanced and IELTS Booster tracks as out-of-class practice. Students would complete two- to three-minute lessons daily and I could monitor their progress through the platform. The speech feedback is so detailed that it allows students to hear, see and fix specific pronunciation issues independently.

Why this approach works

AI tools are not magic – and they certainly don’t replace a thoughtful teacher. But when used with intention, they extend and enhance what teachers already do well: provide opportunities, build confidence and offer personalised support.

Here’s why this framework has worked in my classroom:

1. It fills the speaking gaps
   In traditional classrooms, most learners get minimal speaking time. AI tools allow students to engage in speaking tasks before, during and after class, maximising output without increasing teacher workload.
2. It provides personalised feedback
   Many of the tools I’ve used – like FlowSpeak, SmallTalk2Me and Pronounce – offer instant, granular feedback on grammar, vocabulary or pronunciation. This helps learners notice and correct issues in the moment rather than waiting for teacher correction.
3. It supports autonomy
   Students can use these tools outside the classroom to reinforce what they’ve learnt, develop fluency or explore topics in creative ways. The tools reward consistency and build routine, making speaking practice part of daily life.
4. It gives teachers visibility
   With transcripts, scoring and downloadable audio, I can monitor student progress over time. I can assign tools strategically, track which students are engaging and adjust support as needed.

In short, this approach doesn’t ask teachers to do more – it asks us to work smarter, using technology that brings value, variety and voice to our learners.

Conclusion: teachers still lead

AI can give learners more time, more chances and more freedom to speak – but it’s still teachers who make the difference. The tools I’ve shared here are not meant to replace you. They are here to extend your reach, your voice and your creativity.

The moment we stop seeing AI as just another ‘tool’ and start seeing it as a partner in lesson design, student practice and feedback, our classrooms transform. We create more space for our learners to grow, especially in the area of speaking, where practice and feedback are essential.

Let this be an invitation: experiment with one or two tools. Create space for student voice. And let AI help you help your learners speak more and speak better.

https://pavilionelt.com/wp-content/uploads/p78-80_MET_34.5_SepOct_2025.pdf