You've set up a great partner activity. The prompt is clear, the pairs are ready, and then you hear it: awkward silence followed by nervous laughter, maybe some whispering in the home language, or one student carrying the entire conversation while the other nods along.
The problem often isn't the activity itself. It's that students are being asked to perform before they've had a chance to rehearse. Live partner work puts two learners on the spot simultaneously, and for many students, that's where anxiety wins.
One way to shift this pattern is to let students warm up with low-stakes AI practice before you put them together with a classmate.
Why rehearsal matters before live speaking
Think about how professional speakers prepare. They don't walk onto a stage cold. They practice their material, often out loud, until they feel some control over the words.
Students benefit from the same approach. When they've already said something similar once or twice, they arrive at partner work with phrases loaded and ready. They've worked through a few rough spots. They know where their gaps are. That confidence, even if it's modest, makes the live interaction less overwhelming.
The rehearsal doesn't need to be long. Five minutes of structured AI conversation practice can give students enough of a foundation that the partner activity actually works.
A simple before-and-after structure
Here's a routine you can adapt to most speaking activities.
Step 1: Individual AI warm-up (5 minutes)
Give students a prompt that mirrors what they'll do in partner work. If today's pair activity is discussing weekend plans, the AI warm-up prompt might be: "Ask me about my weekend plans and tell me about yours."
Students work individually with headphones or in a quiet corner if you're rotating through a speaking station. The goal is to hear themselves say the target language out loud, get feedback on their responses, and try a second version if something felt clunky.
If you're using speaky.space, you can set up a conversation scenario that matches your lesson topic. Students get real-time feedback and can see vocabulary they used, which helps them remember useful phrases for the next step.
Step 2: Partner work (7 to 10 minutes)
Now pair students as you normally would. Because they've already rehearsed, you'll notice fewer false starts. Students have phrases ready. They're also more willing to take risks because they've already made mistakes in private.
Step 3: Brief share-out (3 minutes)
Ask a few pairs to share one thing they learned about their partner or one phrase that helped them. This brings the room back together and gives you quick formative data on what language stuck.
Differentiation options
This structure works well for mixed-proficiency classes because students can customize their rehearsal.
For emerging speakers: Simplify the AI prompt. Instead of "Discuss your weekend plans in detail," try "Tell me one thing you will do this weekend." Shorter turns build confidence without overwhelming them.
For stronger speakers: Add a twist. "Tell me about your weekend plans, but you can only spend 5 euros total. Explain your choices." This pushes them to think more flexibly.
For anxious speakers: Let them do two rounds of AI practice before partner work. The extra repetition helps words feel familiar in their mouths.
You can also pair students strategically after the warm-up. If you circulate during AI practice, you'll notice who seems comfortable and who still looks uncertain. That information helps you make pairs that will actually talk.
What to do when devices are limited
Not every classroom has one-to-one devices. Here are a few workarounds.
Station rotation: Set up an AI speaking station with two or three devices. While one group practices with AI, other groups do a vocabulary sort, a listening task, or a written preparation activity. Rotate every five to seven minutes.
Homework rehearsal: Assign the AI warm-up as homework the night before. In class, you can skip directly to partner work because students have already practiced. This approach works well if your students have reliable home access. For more on this, see the first week of AI speaking practice in a classroom.
Pair sharing: If you have limited devices, two students can share one. They take turns speaking to the AI while the other listens. This creates a mini audience, which is still less pressure than a full class but more than practicing alone.
A sample mini-lesson plan
Here's a 25-minute segment you can drop into a lesson on making suggestions.
| Time | Activity | Notes | |------|----------|-------| | 0:00 | Introduce target phrases (Let's..., How about..., Why don't we...) | Board or slide | | 3:00 | AI warm-up: "Suggest three activities for a rainy Saturday. I'll respond to each one." | Individual, headphones | | 8:00 | Partner work: Plan a rainy Saturday together using at least two suggestion phrases | Mixed pairs | | 18:00 | Share-out: Each pair shares their favorite plan | Whole class | | 23:00 | Quick check: Students write one suggestion phrase they used | Exit slip |
You can adjust timing based on your class. If students are newer to AI practice, give them a minute longer on the warm-up.
Addressing common concerns
"Won't students just read from a script during partner work?"
Some might try. But because AI conversation is interactive (the AI responds unpredictably), students can't prepare a fixed script. They're practicing flexible use of language, not memorizing lines.
"What if the AI gives feedback that contradicts what I taught?"
This happens occasionally. Use it as a teaching moment. "The AI suggested a different phrase. Let's look at both options." Students learn that language has variation, which is valuable.
"Is this just adding more screen time?"
The AI warm-up is short (five minutes) and purposeful. It replaces the silent thinking time students might otherwise spend staring at a prompt, and it produces actual speaking practice. For quieter students who struggle with participation, this private rehearsal often makes the difference between speaking up and staying silent.
Checking that it's working
After a few rounds of this routine, you should notice:
- Faster starts to partner conversations
- More even turn-taking (both students contributing)
- Students using target vocabulary or structures without heavy prompting
- Fewer requests to switch to the home language
If you want harder data, try a simple tally. During partner work, note how many pairs are on-task and speaking in the target language at the two-minute mark. Compare this to a similar activity without the AI warm-up.
You can also check saved vocabulary in speaky.space after class to see which words students actually used during practice. This gives you insight into what's sticking and what needs more work. For ideas on vocabulary follow-up, see this post on reviewing words after speaking practice.
One thing to try this week
Pick a partner activity you already have planned. Before class, write a matching AI prompt that covers the same topic in a simpler way. Give students five minutes to warm up with AI practice, then run the partner work as usual.
Notice the difference in how students start the conversation. That's your signal that rehearsal is working.