Cambridge B2 First Speaking: Four Parts, Two Candidates, One Plan
Vlad Podoliako
Founder & CEO, LinguaLive
Vlad Podoliako is the founder of LinguaLive, an AI-powered language learning platform focused on making useful speaking practice available on demand.
Follow on LinkedInCambridge B2 First Speaking has four parts that move from a short examiner conversation to an individual photo-based turn, a collaborative task and a wider discussion. It is normally taken with another candidate, so a complete plan must train both independent speaking and responsive collaboration.
The current Cambridge candidate information describes a four-part Speaking paper lasting about 14 minutes for a pair and says candidates may be tested in pairs or, at the end of a session, in a group of three (Cambridge B2 First information for candidates). The official handbook specifies about 14 minutes for pairs and 20 minutes for groups of three (Cambridge B2 First handbook).
Give each part one clear job
| Part | Main interaction | Practical focus |
|---|---|---|
| 1 | Brief conversation with the interlocutor | Direct personal answers with relevant development |
| 2 | Individual long turn using photographs, plus a short response to the other candidate’s turn | Compare, answer the task focus and organise a one-minute contribution |
| 3 | Candidate-to-candidate discussion and decision | Invite, respond, develop and negotiate |
| 4 | Examiner-led discussion related to Part 3 | Explain and support broader opinions |
This table paraphrases the current official candidate guide and handbook; it is not a replacement for their full instructions. Part 2 gives each candidate about one minute for their main turn and about 30 seconds to respond after the other candidate. Part 3 includes a discussion followed by a decision phase, and Part 4 develops the same broad subject further (Cambridge B2 First handbook).
An older official Cambridge classroom overview—the source anchor for this page—also presents the four-part, paired format (Cambridge B2 First Speaking overview). Because that PDF is a legacy training resource, use the current candidate guide and handbook for present-day logistics and instructions.
Part 1: answer, develop, release
Part 1 is an exchange, not a personal presentation. Use a compact route:
- Answer immediately.
- Add a reason, example or contrast.
- Stop when the idea is complete.
Here is an original practice response, not an official or predicted Cambridge question:
I usually study in the public library because my flat can be noisy in the evening. The library also has large tables, so I can organise my notes more easily.
The response is personal, developed and proportionate. Avoid memorised biographies that ignore the examiner’s angle. Cambridge’s candidate guidance explicitly advises practice across different topics rather than learning a speech that may not fit (Cambridge B2 First information for candidates).
To train, ask a partner to change one word in a familiar-topic turn: “weekdays” becomes “weekends,” or “alone” becomes “with others.” Your answer must adapt. If the same paragraph still appears, you are recalling a script rather than interacting.
Part 2: compare through the task
Do not describe each photograph from top to bottom. Select visible details that support the comparison and the task focus.
Use this four-stage map:
- identify the shared situation;
- choose one meaningful similarity;
- choose one meaningful difference;
- connect those observations to the task focus.
An original practice fragment might sound like:
Both situations involve people learning together, but the level of interaction seems different. In the first, the participants appear to be solving something as a group, whereas the second person is working independently. Group work might provide faster feedback, while working alone could make concentration easier.
This is generic language practice, not a response to an official photograph set. Notice that visual claims are cautious: “seems,” “appear” and “might” distinguish observation from inference.
Create a two-column note after practice:
| Observation | Interpretation linked to the task |
|---|---|
| People are facing one another | They may be exchanging ideas |
| One person has headphones | The setting may support concentration |
If your transcript has many observations but no task-linked interpretation, narrow the description. If it has opinions unsupported by anything visible, return to evidence.
When responding briefly to the other candidate’s material, answer the examiner’s follow-up rather than extending your own earlier turn.
Part 3: collaborate visibly
Part 3 is not two separate monologues. The interaction should show that you can introduce an idea, react to a partner and work toward the required decision.
Use a collaboration loop:
- Offer one option with a reason.
- Invite the partner’s view.
- Respond to the substance of that view.
- Compare it with another option.
- Signal when you are ready to decide.
Useful language has a clear function:
- invite: “Which of these do you think would make the biggest difference?”
- build: “That point about cost also affects…”
- qualify: “I agree in general, although…”
- refocus: “We have considered access; should we compare the time involved?”
- decide: “We seem to agree that this is the strongest option because…”
Do not perform agreement. A respectful difference creates interaction when you explain it and remain open to the partner’s evidence. Do not dominate the full discussion, but do not become so polite that you contribute no position.
Part 4: widen the reasoning
Part 4 moves into a broader discussion related to Part 3. Prepare to explain causes, consequences, different groups’ perspectives and conditions.
Turn a basic opinion into a developed one:
Online resources are useful.
Then add scope and limitation:
Online resources are useful when learners need flexible access, especially if they cannot attend a fixed class. However, they may be less effective for someone who needs immediate interaction, so a mixed approach can work better.
This original example shows a B2-relevant skill: supporting a position while recognising a reasonable condition. Do not add a counterargument mechanically; use it when it sharpens the meaning.
The comparison with Cambridge C1 Advanced Speaking can help you see how similar interaction types demand more nuanced development at a higher qualification. For a framework view rather than an exam format, read CEFR spoken interaction levels.
Run one complete practice cycle
Once a week:
- Practise Part 1 with changing follow-ups.
- Record a one-minute visual comparison using licensed or self-created images.
- Complete a collaborative decision task with a partner.
- Discuss two wider implications of the same topic.
- Review one individual target and one interaction target.
- Repeat only the weakest part with new content.
Use the AI English tutor when a human partner is unavailable, but return to real partner practice before the exam. An automated exchange cannot fully reproduce another candidate’s timing, uncertainty or negotiation style.
Limitations and official scoring
LinguaLive is not affiliated with or endorsed by Cambridge University Press & Assessment or Cambridge English. It cannot provide official confidential materials, certify a B2 level, award a Cambridge result or predict a score. Confirm current format, timing and test-day rules in official Cambridge materials and with your authorised exam centre.
Sources and editorial review
This guide was checked against its primary official or academic reference on 29 July 2026. Language usage can vary by region, relationship, and situation. Review the primary source.
Related Topics
Share this article
Ready to Start Learning?
Try LinguaLive's AI-powered conversation practice free. 10 minutes a day can transform your fluency.
Start Free - 10 Min DailyMore Articles
30-Day Speaking Practice Plan: Build a Daily Language Habit That Transfers
This 30-day speaking plan uses 15 to 25 minutes a day, one weekly scenario, and a record–review–repeat loop. You will not become universally fluent in a month.…
Accessibility Checklist for Voice Language Apps
An accessible voice language app must provide a workable path when a learner cannot hear, speak, see, touch, read, process, or respond on the product’s default…