PTE template · Framework

PTE Summarise Group Discussion Template: 3 speakers, 2 minutes, one clear summary

Summarise Group Discussion (SGD) is one of two new tasks Pearson added on 7 August 2025. You hear a group discussion of up to 3 minutes involving 3 speakers, you get 10 seconds to prepare, and you have 2 minutes to speak a summary. There are 2 to 3 SGD items per PTE Academic test. Pearson scores Content (AI + human review), Oral Fluency (AI only) and Pronunciation (AI only). This is one of the 7 hybrid-scored task types where a human examiner reviews Content, so real fidelity to the actual discussion matters more here than on AI-only tasks.

Take a free scored mock →All PTE templates

Quick answer

The Summarise Group Discussion template is a 3-part spoken structure filled from a 3-column note grid: name the topic and each speaker's position (30 to 40 seconds), summarise 2 to 3 supporting arguments raised across the discussion (60 to 75 seconds), and close with the point of agreement, disagreement or conclusion (20 to 30 seconds). Total: roughly 2 minutes at natural pace. Draw the 3-column speaker grid before the audio starts.

Read this first

SGD Content is scored by AI plus a human examiner. A memorised template that ignores the actual discussion earns 0 on Content and zeroes the whole item. The frame below is a skeleton, not a script. Every speaker attribution, every position and every supporting point must reference the real discussion. Never leave a stock 'The three speakers discussed important issues' sentence in.

The framework

How the framework works

Read the sections in order. Each one is a step of the framework, with adaptable sentence starters you fill from the actual prompt.

The 3-column note grid (draw before audio starts)

On the erasable whiteboard, draw three columns labelled Speaker 1, Speaker 2, Speaker 3. Add a fourth strip below for AGREEMENT / DISAGREEMENT / CONCLUSION. As the audio plays, capture in each column: - POSITION: the speaker's stance in 2 to 4 words ("supports remote work", "opposes", "neutral / mixed view"). - REASON: their main supporting argument in 3 to 6 words. Speakers are usually introduced by name or role. Do not try to catch every word: catch the position and the strongest reason each speaker gives. Example: | Speaker 1 (Aisha) | Speaker 2 (Ben) | Speaker 3 (Chen) | | supports 4-day week | opposes 4-day week | mixed / conditional | | productivity data + wellbeing | wage cost + fairness to shift workers | depends on industry sector | Agreement: all three agree the current 5-day norm needs review.

The 3-part spoken structure (2 minutes)

Part 1 - Topic + speaker positions (30 to 40 seconds): "The discussion focused on [topic]. Speaker 1 [position], while Speaker 2 [contrasting position]. Speaker 3 [third position, often mixed or conditional]." Part 2 - Supporting arguments (60 to 75 seconds): "Speaker 1 supported this view by referring to [reason 1]. In response, Speaker 2 argued that [reason 2], particularly noting [specific detail from the discussion]. Speaker 3 pointed out that [reason 3], which added a further dimension to the debate." Part 3 - Agreement, disagreement or conclusion (20 to 30 seconds): "Ultimately, the three speakers [agreed / disagreed / partially agreed] on [specific outcome]. The discussion concluded with [any resolution or open question the recording ends on]." Total: roughly 110 to 145 seconds at 130 to 145 wpm.

Openers and connectors that carry a group discussion

Positional verbs (choose one that fits the actual speaker): - 'argues', 'contends', 'maintains', 'insists', 'supports' - 'opposes', 'questions', 'challenges', 'is sceptical of' - 'takes a mixed view', 'sees both sides', 'is conditional on' Connectors between speakers: - 'In response, Speaker 2 argues...' - 'Speaker 3 offers a different angle...' - 'The second speaker builds on this by...' - 'By contrast, the third participant emphasises...' Agreement / disagreement closers: - 'The three speakers agreed that...' - 'The discussion ended without a full consensus, though all three accepted that...' - 'The panel concluded that further evidence is needed before...' These connectors let you show relationships between positions without inventing content the speakers did not actually say.

What to do in the 10 second preparation window

1. Read your grid top to bottom. Confirm the three positions and the agreement strip. 2. Silently rehearse the Part 1 sentence: 'The discussion focused on [topic]. Speaker 1 [position], while Speaker 2 [contrasting position].' 3. Decide which speaker's argument you will cover most fully in Part 2. If you have strong notes on Speaker 1, lead with them. 4. Take one deep breath. You are about to speak for close to 2 minutes; you need pace management, not adrenaline. Do NOT try to add extra notes in the 10 second window. Your job now is to speak, not write.

What Pearson scores on Summarise Group Discussion

Three traits, per the July 2025 Score Guide: - Content: did you capture the topic, each speaker's position, key supporting arguments, and any agreement/disagreement? Scored by AI PLUS a human examiner (SGD is one of 7 hybrid task types). - Oral Fluency: smooth, connected 2 minute delivery. Long pauses, restarts and repeated filler ('um', 'er') cost Fluency marks fast on a 2 minute recording. - Pronunciation: clear stress placement and articulation. Speaker attributions and names should be pronounced distinctly so the AI can transcribe them. SGD is Listening + Speaking scored, so a strong summary lifts two skills at once. If Content = 0, the whole item scores 0 with no further scoring.

The most common SGD mistakes

- Trying to remember exact wording from the discussion. You cannot. Capture positions and reasons, not sentences. - Merging speakers into 'they'. Naming each speaker (or at least numbering them) is required for Content. A merged summary that says 'the speakers discussed the pros and cons' scores much lower than one that says 'Speaker 1 supported X, Speaker 2 opposed it'. - Running out of content before 2 minutes are up. If you finish early, loop back with the agreement/disagreement summary in different words rather than sitting silent. - Adding your own opinion. SGD is a summary, not an argument. The words 'I think', 'in my view' or 'personally' cost Content marks because they are not in the source. - Skipping the agreement/disagreement strip. Every group discussion ends somewhere: consensus, split, or open question. Missing this misses a Content chunk.

Worked examples

The framework applied

Same framework, different prompts. Each answer is filled with real content, not a memorised script.

Example 1Speaker 1 (Dr Aisha Rahman, industry engagement director): supports the requirement, cites Business Council data on the employability gap and graduate outcome improvements in existing Work Integrated Learning cohorts. Speaker 2 (Ben Torres, student union representative): opposes, argues it disadvantages students who need paid work outside their field to fund their studies. Speaker 3 (Dr Chen, curriculum design academic): mixed view, supports the principle but says universities lack the industry-partner network to make it work at scale in the next 3 years. Agreement: all three accept the employability gap is real; disagreement is about the mechanism.

Group discussion topic: whether Australian universities should require a semester of paid internship for all undergraduate degrees. 3 speakers, 2 minute 40 second audio.

Framework-filled answer

The discussion focused on whether Australian universities should require a semester of paid internship for every undergraduate degree. Doctor Aisha Rahman, an industry engagement director, argued in support, citing Business Council data on the graduate employability gap and improved outcomes among existing Work Integrated Learning cohorts. In contrast, Ben Torres, representing the student union, opposed the requirement, pointing out that many students rely on paid work outside their field to fund their studies and that a compulsory internship would disadvantage them financially. Doctor Chen, a curriculum design academic, took a mixed view. He accepted the principle behind the proposal but questioned whether universities currently have the industry-partner network to place every undergraduate within the next three years. Ultimately, the three speakers agreed that the employability gap is real and worth addressing, but they disagreed on the mechanism. Doctor Rahman favoured a mandate, Ben Torres favoured expanded voluntary schemes with better financial support, and Doctor Chen favoured a phased rollout. The panel concluded without full consensus, though all three accepted that the current model leaves too many graduates underprepared for employment.

Why this scores: Roughly 190 words. At 130 wpm this delivers in about 88 seconds; at 145 wpm about 78 seconds. Adding brief speaker attributions and one specific detail per speaker (Business Council data, financial disadvantage argument, industry-partner network) is what earns Content marks. Every position is attributed to the correct speaker; the agreement/disagreement is captured at the close.

Example 2Speaker 1 (Professor Kim, computer science): supports controlled use, argues that AI is a workplace tool students must learn to use responsibly. Speaker 2 (Doctor Adeyemi, humanities dean): opposes, argues that undergraduate writing exists to develop original thinking, which is undermined if AI drafts the ideas. Speaker 3 (Rachel, third year student): mixed, says students already use AI regardless of policy, so universities should focus on transparent citation rather than prohibition. Agreement: all three agree the current policy vacuum is unworkable.

Group discussion topic: whether AI-generated writing should be permitted in university coursework. 3 speakers, 2 minute 55 second audio.

Framework-filled answer

The discussion focused on whether AI-generated writing should be permitted in university coursework. Professor Kim, from computer science, argued in favour of controlled use. She contended that generative AI is now a standard workplace tool and that universities have a responsibility to teach students to use it critically rather than pretending it does not exist. Doctor Adeyemi, dean of humanities, took the opposing view. He argued that undergraduate writing exists precisely to develop the student's own thinking, and that outsourcing the drafting stage to AI hollows out the educational purpose of the essay itself. Rachel, a third year student, offered a mixed position. She acknowledged the risks Doctor Adeyemi described but pointed out that students already use AI regardless of institutional policy, so the practical question is how to require transparent disclosure rather than whether to ban it. Ultimately, the three speakers agreed on one point: the current policy vacuum, where every faculty applies a different rule, is unworkable. They disagreed on the direction of reform. Professor Kim favoured formal integration into the curriculum, Doctor Adeyemi favoured a stronger prohibition on generative use in assessed work, and Rachel favoured a disclosure-first framework similar to citation of secondary sources.

Why this scores: Roughly 210 words. At 135 wpm this delivers in about 93 seconds; the remaining time can be used for a brief reflection sentence or simply to close naturally. Notice how each speaker gets a named attribution, a distinct verb ('argued', 'contended', 'took the opposing view', 'offered a mixed position') and one specific detail. The agreement/disagreement close mirrors the fourth strip of the note grid.

Frequently asked questions

What is the best PTE Summarise Group Discussion strategy?

A 3-column note grid (one column per speaker, plus an agreement/disagreement strip) filled during the audio, delivered in a 3-part spoken structure: topic and positions in 30 to 40 seconds, supporting arguments in 60 to 75 seconds, agreement or conclusion in 20 to 30 seconds. Total: roughly 2 minutes at natural pace.

How many Summarise Group Discussion items are on the PTE Academic test?

2 to 3 items per test per Pearson's July 2025 Score Guide. SGD is one of two new task types added on 7 August 2025 (the other is Respond to a Situation). It appears in Part 1 (Speaking and Writing) and contributes to both Listening and Speaking scores.

How long is the audio for Summarise Group Discussion?

Up to 3 minutes, featuring 3 distinct speakers. The audio plays exactly once (no replay control), followed by a 10 second preparation window and a 2 minute recording window. This is the longest audio and the longest recording window of any speaking task in PTE Academic.

Is SGD scored by AI or by a human?

Both. SGD is one of 7 PTE Academic task types where Content is scored by AI plus a human expert reviewer (the other 6 are Describe Image, Retell Lecture, Respond to a Situation, Summarize Written Text, Write Essay, and Summarize Spoken Text). Oral Fluency and Pronunciation are AI-only, always.

Do I need to name each speaker in the summary?

Yes, or at least distinguish them clearly. Naming the speakers ('Doctor Rahman argued', 'Ben Torres opposed') is stronger than numbering ('Speaker 1', 'Speaker 2') if you caught the names, because it demonstrates fidelity to the source. If names are unclear, use role or number: the key is that each position is attributed to a specific speaker, not merged into a generic 'they'.

What if I run out of content before 2 minutes are up?

Loop back with the agreement/disagreement summary in different words, or add a one-sentence reflection on how the three positions relate. Sitting silent for more than 3 seconds can auto-stop the recording and always costs Oral Fluency marks on a 2 minute delivery.

Should I add my own opinion on the SGD topic?

No. SGD is a summary of what the three speakers said, not an argument for what you think. Phrases like 'I think', 'in my view' or 'personally I agree with' cost Content marks because they contain no source information. Stick to what the speakers actually said.