PTE Speaking Tips 2026: How to Score 90 in Every Task

You could have near-perfect English and still score 50 in PTE Speaking.

That’s not an exaggeration – it’s something thousands of test takers discover the hard way. The PTE Speaking section isn’t assessed by a human examiner who appreciates your vocabulary or rewards your effort. It’s scored by Pearson’s Versant technology — a proprietary speech processing system that analyses and scores speech at the level of syllables, segments, and phrases, using statistical modelling trained on expert human ratings.

That system is looking for very specific signals in your speech. Task by task. Dimension by dimension.

Once you understand what those signals are, a score of 90 becomes a system — not a miracle.

This guide breaks down all seven PTE Academic Speaking tasks, explains exactly how Pearson’s scoring criteria evaluate each response, and gives you the precise strategies that move scores from 60 to 90. Every tip here is grounded in Pearson’s official scoring documentation.

How PTE Speaking Is Actually Scored

Before tactics, you need to understand the engine — and it’s more precise than most guides suggest.

PTE Speaking responses are assessed across three core dimensions:

• Content — does your response cover the required information accurately and completely?
• Oral Fluency — is your speech smooth, natural, and free of hesitations and false starts?
• Pronunciation — do your vowels, consonants, and stress patterns meet English language standards?

Not every task scores all three. And the weighting differs by task. Here’s what you need to know about each dimension.

Oral Fluency — The Official 0–5 Scale

Oral Fluency is not a general impression of how naturally you speak. It is scored on a precise 0–5 rubric:

Score	Label	What it means
5	Highly Proficient	Smooth rhythm and phrasing. No hesitations, repetitions, false starts, or phonological simplifications.
4	Advanced	Acceptable rhythm with appropriate phrasing. No more than one hesitation, one repetition, or one false start. No significant phonological simplifications.
3	Good	Acceptable speed but may be uneven. More than one hesitation allowed, but most words spoken in continuous phrases. No long pauses. Does not sound staccato.
2	Intermediate	May be uneven or staccato. At least one smooth three-word run. No more than two or three hesitations, repetitions, or false starts. Maximum one long pause.
1	Limited	Irregular phrasing or sentence rhythm. Multiple hesitations, repetitions, or false starts. Notably uneven or discontinuous.
0	Disfluent	Slow and laboured. Little discernible phrase grouping. Multiple hesitations, pauses, and false starts. Most words isolated.

The difference between a 4 and a 5 is striking: a single extra hesitation or false start drops you from Highly Proficient to Advanced. This is why eliminating hesitation habits through deliberate practice — not just speaking more — is the core fluency challenge.

Pronunciation — The Official 0–5 Scale

Pronunciation is equally precise:

Score	Label	What it means
5	Highly Proficient	All vowels and consonants easily understood. Appropriate assimilation and deletion. Stress correct in all words and sentences.
4	Advanced	Clear vowels and consonants. Minor distortions don’t affect intelligibility. Stress correct on all common words.
3	Good	Most vowels and consonants correct. Some consistent errors may make a few words unclear. Stress-dependent vowel reduction may occur on a few words.
2	Intermediate	Some consonants and vowels consistently mispronounced. At least two-thirds of speech intelligible. Listeners may need to adjust to the accent.
1	Intrusive	Many mispronunciations. Strong intrusive foreign accent. Listeners may struggle to understand around one-third of words.
0	Non-English	Pronunciation characteristic of another language. More than half of speech may be unintelligible.

Two things worth noting. First, Pearson’s system was trained on speakers from 158 countries speaking 126 first languages — it is not calibrated to a British or American standard. What it measures is intelligibility and phonemic accuracy, not accent. Second, word-level stress errors are explicitly scored — not just vowel and consonant production.

Content — Task-Specific Scoring

Content is scored differently for each task. For some tasks it is a detailed 0–6 rubric assessed by both AI and a human expert. For others it is a simpler correct/incorrect judgement. The task sections below cover the Content criteria for each.

The Seven PTE Academic Speaking Tasks

1. Read Aloud — Where the Most Marks Are Won or Lost

Read Aloud is the highest-frequency Speaking task and one of the highest-leverage tasks in the exam. It contributes to your Speaking score only, but because it appears 6–7 times per test, its cumulative impact on your Speaking score is enormous.

You’re shown a text of up to 60 words and given 30–40 seconds to prepare silently. Your recording window then opens.

What the AI is scoring:

• Content — each replacement, omission, or insertion of a word counts as one error. Maximum score depends on text length.
• Pronunciation — vowel, consonant, and stress accuracy (0–5)
• Oral Fluency — rhythm, phrasing, and absence of hesitations (0–5)

The strategies that work:

Use your preparation time to scan for difficulty. Identify any unfamiliar words before the microphone opens. Mark natural pause points at punctuation. The 30-second prep window is a scoring opportunity — use it fully.

Chunk the text into meaning groups. Native speakers don’t read word by word — they group words into phrases. “The researchers / found that daily exercise / significantly improved / cognitive function.” Chunking creates natural rhythm and prevents the robotic, word-by-word delivery that scores a 1 or 2 on Oral Fluency.

Never stop to self-correct. A stumbled word costs you one Content error. Stopping, pausing, and re-reading the word costs you a Content error and a fluency break. Keep moving. The net cost of stopping to correct is always higher than the cost of the original error.

Calibrate pace to text length. Longer texts require a slightly faster delivery to avoid running out of recording time. Shorter texts give you more room to be deliberate. Train this calibration across different text lengths.

Common mistake: Reading slowly because it feels more careful and precise. The Oral Fluency rubric is explicit — a score of 3 requires speech that does not sound staccato, with most words spoken in continuous phrases. Slow, halting delivery falls below this threshold. Read at a natural, conversational pace.

2. Repeat Sentence — An Integrated Listening and Speaking Task

Repeat Sentence contributes to both your Listening and Speaking scores — a fact many students don’t know and that changes how you should approach preparation for this task.

You hear a sentence once — 9 to 16 words — and repeat it as accurately as possible. No transcript. No second play. The microphone opens immediately after the sentence ends.

What the AI is scoring:

• Content — scored on a 0–3 scale: 
  • 3 = all words from the prompt in the correct sequence
  • 2 = at least 50% of words in the correct sequence
  • 1 = less than 50% of words in the correct sequence
  • 0 = almost nothing from the prompt reproduced
• Pronunciation (0–5)
• Oral Fluency (0–5)

Importantly, the official scoring guide confirms that hesitations, filled or unfilled pauses, and leading or trailing material are ignored in the scoring of Content for this task. Only replacements, omissions, and insertions count as errors. This means your delivery rhythm is assessed separately from your content accuracy — practise both, but understand they are scored independently.

The strategies that work:

Capture meaning chunks, not individual words. Trying to memorise 14 individual words under pressure collapses for most students. Train yourself to hear the sentence as 3–4 meaning groups and hold those groups, not a word list.

Speak immediately when the tone sounds. Any delay at the start is a fluency break before you’ve produced a single word.

Never go silent mid-response. Silence is scored more harshly than an approximated word. If you lose a section, keep the rhythm going — approximate the sounds or move to the next chunk you remember. A continuous, approximate response scores better than an accurate-but-interrupted one.

Train with volume and instant feedback. Repeat Sentence improves faster than almost any other task with high-volume, scored practice. You need to know your content accuracy percentage after each attempt — not just whether it felt right.

3. Describe Image — Content Is Human-Reviewed

Describe Image contributes to your Speaking score only and appears 5–6 times per test. It is one of two Speaking tasks where Content is reviewed by both AI and a human expert before the final score is confirmed. If the AI and human disagree, a second human makes the final judgement.

This matters. It means a nuanced, accurate description is genuinely evaluated by a human — not just pattern-matched by an algorithm.

You’re shown an image — a graph, chart, map, diagram, or photograph — and given 25 seconds to prepare. You then have 40 seconds to describe it.

Content is scored on a 0–6 rubric. A score of 6 requires a response that describes the image fully and accurately, expands on the relationships between features, uses varied vocabulary with ease and precision, and allows a listener to build a complete mental picture. A score of 3 — where many students sit — means mainly superficial descriptions with a narrow, repetitive vocabulary range.

What the AI and human are scoring:

• Content (0–6) — accuracy, coverage, vocabulary variety, relationships between features
• Pronunciation (0–5)
• Oral Fluency (0–5)

The 4-part template that works:

Part	What to say	Time allocation
Opening	Introduce what the image shows	5–7 seconds
Key feature	Identify the most significant trend or element	10–12 seconds
Supporting detail	Add one or two supporting observations	10–12 seconds
Conclusion	Brief closing statement or comparison	5–8 seconds

For a score of 6: Don’t just list features — identify relationships. “The sharp rise in X between 2010 and 2015 contrasts with the gradual decline in Y over the same period, suggesting that…” This kind of analytical observation is what separates a 5 from a 6 on Content.

4. Re-tell Lecture — Integrated Listening and Speaking, Human-Reviewed

Re-tell Lecture contributes to both your Listening and Speaking scores and — like Describe Image — Content is always reviewed by a human expert before the final score is confirmed.

You watch or listen to a lecture clip of up to 90 seconds. After a short preparation beep, you have 40 seconds to retell the key points in your own words.

Content is scored on a 0–6 rubric. A score of 6 requires a clear, accurate response that paraphrases the main ideas seamlessly in your own words, expands on important points with specificity, uses varied and precise vocabulary, and presents ideas in a well-connected, logical sequence. A score of 3 — the common plateau — means ideas are captured from the lecture but not accurately or selectively, with possible repetition of lecture language without reformulation.

What is scored:

• Content (0–6) — comprehension, paraphrasing, synthesis, organisation
• Pronunciation (0–5)
• Oral Fluency (0–5)

The note-taking system that works:

During the lecture, write down:

• Topic — one or two words: what is this about?
• Key point 1 — the main argument or finding
• Key point 2 — supporting detail or contrast
• Conclusion — outcome or implication

Four notes. Students who try to transcribe sentences miss the next point. Students who write nothing can’t reconstruct specific content. Four notes is the balance.

To score 6 on Content: Paraphrase in your own words — do not repeat lecture language. The rubric explicitly penalises responses that repeat language from the lecture without reformulation. Use your own vocabulary to express what you understood. A human reviewer will notice the difference.

5. Answer Short Question — Scores Listening, Not Speaking

This is one of the most misunderstood tasks in PTE Academic. Answer Short Question appears in the Speaking section of the exam — but it contributes only to your Listening score, not your Speaking score.

You hear a question. You answer in one or a few words. The task is scored on Vocabulary only:

• 1 = appropriate word choice
• 0 = inappropriate word choice

There is no Oral Fluency or Pronunciation scoring for this task. Your delivery doesn’t affect your score here — only whether your answer is correct.

The strategy: Answer immediately and confidently. If unsure, give your best guess — silence scores the same as a wrong answer (zero), and hesitation costs you time. Build vocabulary across the common topic areas: medical, scientific, geographical, occupational, and general academic terminology.

6. Summarise Group Discussion — Integrated Listening and Speaking

Summarise Group Discussion contributes to both your Listening and Speaking scores. You listen to a discussion between two or more speakers — typically 60 to 90 seconds — then have 70 seconds to summarise the key points made by each participant.

This task is distinct from Re-tell Lecture. You’re not following a single narrative thread — you’re capturing multiple perspectives and the relationships between them.

Content is scored on a 0–6 rubric. A score of 6 requires a response that paraphrases the main ideas of each speaker’s contribution seamlessly in your own words, explores the relationships between different points of view, synthesises perspectives effectively, uses varied and precise vocabulary, and presents ideas in a well-connected logical sequence.

A score of 3 — the common plateau — means some ideas from the discussion are captured but not fully accurately, with possible repetition of discussion language without reformulation, and only some accurate information while missing the substance and general direction of the full conversation.

What is scored:

• Content (0–6) — coverage of multiple perspectives, synthesis, vocabulary, organisation
• Pronunciation (0–5)
• Oral Fluency (0–5)

The note-taking approach that works:

Speaker	Key point
Speaker 1	Main argument or position
Speaker 2	Contrasting or supporting view
Overall	Agreement, disagreement, or conclusion

Keep it tight. 70 seconds is enough for a well-structured 3-part response — not enough for a meandering one. Aim to finish in 55–60 seconds, leaving a small buffer without rushing.

The score-6 differentiator: Don’t just report each speaker’s point separately. Identify the relationship between them — agreement, contrast, complementarity, progression. “While both speakers agreed that X was a concern, they differed significantly on whether Y or Z represented the better solution.” This synthesis is what the top rubric band explicitly requires.

7. Respond to a Situation — Authentic Communication Under Assessment

Respond to a Situation contributes to your Speaking score only. You’re given a written prompt describing a real-life situation — professional, social, or practical — and asked to respond to it verbally. You have 20 seconds to read and prepare, then 40 seconds to respond.

Content is scored on a 0–6 rubric. A score of 6 requires a response that deals with the situation effectively, accomplishes the primary communication goal with full consideration of the context, communicates with ease, flexibility, and precision, and expands beyond the prompt language to provide a persuasive, fully developed response.

A score of 3 means only the most basic aspect of the communication goal is partially accomplished, with limited consideration of context, functional but restricted expression, and possible repetition of language from the prompt.

What is scored:

• Content (0–6) — communication goal achieved, contextual appropriateness, development
• Pronunciation (0–5)
• Oral Fluency (0–5)

The strategy that works:

During your 20-second preparation, identify three things: Who are you speaking to? What do they need to know? What tone is appropriate for this relationship and context?

Structure your 40 seconds in three beats:

1. Acknowledge the situation — briefly reference the context (5–8 seconds)
2. Deliver the core message — the key information required (20–25 seconds)
3. Close naturally — a polite sign-off or suggested next step (5–8 seconds)

Example prompt: “You need to let your manager know you’ll be late to an important meeting due to a delayed train.”

Strong response: “Hi [name], I just wanted to let you know I’m going to be a few minutes late to the meeting this morning — my train has been delayed and I’m not sure of the exact arrival time yet. I’ll join as soon as I can and catch up on anything I miss. Sorry for the inconvenience.”

That response accomplishes the communication goal, is contextually appropriate, is natural in tone, and is well within 40 seconds.

Common mistake: Responses that repeat the prompt language almost verbatim. The rubric explicitly penalises this at the 2 and 3 score bands. To score 5 or 6, your response must go beyond the prompt — adding relevant detail, appropriate tone, and natural elaboration that a real person would include in that situation.

The Scoring Dimensions That Apply Across All Speaking Tasks

Two dimensions appear across every scored Speaking task: Oral Fluency and Pronunciation. Both use the official 0–5 rubric detailed at the start of this guide. Here’s how to train them deliberately.

Training Oral Fluency: Fluency is not a personality trait — it is a trainable skill. The rubric is explicit about what costs you points: hesitations, repetitions, false starts, and phonological simplifications. Train by recording yourself, identifying your specific pattern (do you hesitate before difficult words? Do you repeat filler phrases?), and drilling the specific behaviour — not just speaking more.

A score of 4 allows one hesitation, one repetition, or one false start per response. A score of 5 allows none. That’s a remarkably tight standard. Reaching it requires deliberate elimination of specific habits — not general fluency improvement.

Training Pronunciation: The rubric focuses on vowel and consonant accuracy, stress placement, and intelligibility. Common patterns among Indian and South Asian test takers that affect pronunciation scores: consonant cluster simplification, incorrect word stress on multi-syllable words, and vowel reduction errors. These are specific, identifiable, and fixable with targeted phoneme work — not general speaking practice.

The Four Habits of Consistent 90-Scorers

Students who consistently score 90 in PTE Speaking share four observable habits:

1. They practise with AI feedback, not just by speaking aloud. Recording yourself tells you nothing about what Pearson’s scoring engine detects. You need a platform that scores your responses against the same rubric the real exam uses — showing you Oral Fluency, Pronunciation, and Content scores per response, per task.
2. They know exactly which tasks contribute to which scores. Read Aloud and Describe Image: Speaking only. Repeat Sentence and Re-tell Lecture: Listening and Speaking. Summarise Group Discussion: Listening and Speaking. Answer Short Question: Listening only. Respond to a Situation: Speaking only. Strategic preparation means knowing where each task’s score lands.
3. They drill their weakest task asymmetrically. A student scoring 90 in Read Aloud but 55 in Repeat Sentence doesn’t practise Read Aloud. They run Repeat Sentence sets daily until the gap closes. Identify your lowest task score and treat it as your only job for two weeks.
4. They take full-length scored mock tests regularly. Speaking under real exam conditions — unfamiliar content, time pressure, no second chances — is different from comfortable practice. Regular full mock tests build the stamina and composure that keep your score stable on exam day.

Why Repeat Test Takers Stay Stuck — And How to Break Through

If you’ve taken PTE more than once and your Speaking score hasn’t moved, the problem is almost never your English.

The most common causes of a plateau:

Practising without feedback — you’re reinforcing current habits, not improving them. Without dimension-level scoring after every attempt, you can’t know whether the issue is Oral Fluency, Pronunciation, Content, or a combination.

Misunderstanding the rubric — students who think pronunciation is purely about accent, or that fluency means speaking fast, are optimising for the wrong things. The rubric is specific. The preparation needs to match it.

Test anxiety disrupting fluency — nerves create exactly the hesitation patterns that the Oral Fluency rubric penalises. Regular full mock tests in exam-like conditions are the fix — not more vocabulary lists.

Ignoring the integrated tasks — Repeat Sentence and Re-tell Lecture contribute to your Listening score. Students who deprioritise them are leaving Listening marks on the table in addition to Speaking marks.

Your Next Step: Stop Reading, Start Scoring

You now have the complete, Pearson-aligned picture for all seven PTE Academic Speaking tasks — the official scoring criteria, the task-by-task strategies, and the specific habits that separate students who score 90 from those who stay stuck.

The gap between knowing this and scoring 90 is scored practice. Specifically, practice on a platform that evaluates your responses the same way Pearson does — giving you Oral Fluency, Pronunciation, and Content scores after every attempt, for every task.

[Take a free PTE mock test →] See exactly where your Speaking score stands right now and find out which tasks to fix first.

[Join the Language Academy student portal →] Access scored Speaking practice, AI-powered feedback aligned to Pearson’s official rubric, and an exam-identical interface trusted by over 50,000 students preparing for PTE Academic.

Your 90 isn’t a lucky outcome. It’s a prepared one.

FAQ SECTION

Q: How is PTE Speaking scored?

A: PTE Speaking is scored by Pearson’s Versant technology — a proprietary speech processing system trained on expert human ratings. Responses are assessed across three dimensions: Content (whether your response covers the required information), Oral Fluency (the smoothness and rhythm of your speech, scored 0–5), and Pronunciation (vowel, consonant, and stress accuracy, scored 0–5). Not every task scores all three dimensions — Answer Short Question, for example, scores only Vocabulary and contributes to Listening, not Speaking. Understanding the scoring dimensions for each task is the foundation of effective preparation.

Q: How do I get 90 in PTE Speaking?

A: Scoring 90 in PTE Speaking requires consistent performance across all seven task types. The key is understanding exactly what Pearson’s scoring rubric looks for in each task — and practising with a platform that provides AI-scored feedback on Oral Fluency, Pronunciation, and Content after every attempt. Students who score 90 consistently practise with scored mock tests, identify and drill their weakest tasks, and train fluency and pronunciation as deliberate skills using specific rubric-aligned targets — not general speaking practice.

Q: What is the difference between Oral Fluency and Pronunciation in PTE?

A: They are two distinct scoring dimensions, each on a 0–5 scale. Oral Fluency assesses the rhythm and smoothness of your speech — a score of 5 requires no hesitations, repetitions, false starts, or phonological simplifications. A score of 4 allows no more than one of each. Pronunciation assesses vowel and consonant accuracy, word stress, and intelligibility — not accent. Pearson’s system is trained on speakers from 158 countries and 126 first languages. What it measures is whether your phoneme production and stress patterns are consistently intelligible, not whether you sound British or American.

Q: Does Answer Short Question affect my Speaking score?

A: No — despite appearing in the Speaking section of the exam, Answer Short Question contributes only to your Listening score. It is scored on Vocabulary only: correct word choice scores 1, incorrect word choice scores 0. There is no Oral Fluency or Pronunciation scoring for this task. This is one of the most commonly misunderstood facts about PTE Academic — and it matters for preparation, because time spent drilling Answer Short Question delivery is not improving your Speaking score.

Q: How many Speaking tasks are in PTE Academic?

A: PTE Academic includes seven Speaking task types: Read Aloud, Repeat Sentence, Describe Image, Re-tell Lecture, Answer Short Question, Summarise Group Discussion, and Respond to a Situation. Of these, Repeat Sentence, Re-tell Lecture, and Summarise Group Discussion are integrated tasks that contribute to both your Listening and Speaking scores. Answer Short Question contributes to Listening only. Read Aloud, Describe Image, and Respond to a Situation contribute to Speaking only. Understanding which tasks affect which scores is essential for strategic preparation.

Q: Why is my PTE Speaking score low despite good English?

A: This is one of the most common frustrations among PTE test takers. PTE Speaking is scored by Pearson’s Versant technology against a precise rubric — not by a human examiner who factors in your overall English ability. Students with strong English often unknowingly produce the specific patterns the rubric penalises: hesitations that drop Oral Fluency from 5 to 3, word stress errors that limit Pronunciation to a 2 or 3, or Describe Image responses that describe features without identifying relationships — scoring a 3 or 4 on Content rather than a 6. Practising with AI-scored feedback that shows you your dimension-level scores after every attempt is the fastest way to identify and fix exactly what’s holding you back.