PTE Retell Lecture: the content mark is a listening score in disguise

Search for Retell Lecture help and you get the same product every time: twenty-five lectures, twenty-five audio files, twenty-five model answers. It is a useful thing to have and a terrible thing to rely on, for a reason worth stating plainly — an example bank is the easiest asset in test preparation to copy. Anyone can build one in a weekend. Nobody's score moved because a bank existed.

The candidates who are stuck on this task are not stuck for want of examples. They have usually already done the obvious work. They have slowed down, stopped restarting sentences, learned an opening line, and their speaking score has not moved.

That is the situation this article is about, and it has a specific cause.

We build the engine that marks speaking and writing answers on Hilingo, so this is written from the marking side rather than the coaching side.

What the task is, exactly

	Retell Lecture
Lecture audio	Up to 90 seconds
Preparation after the audio ends	10 seconds
Recording window	40 seconds
Number of items on a test	2–3
Communicative skills it feeds	Listening and Speaking

Timings from Pearson's PTE Academic speaking and writing format page; item count and skills from the PTE Academic Test Taker Score Guide, page 21. Retell Lecture does not exist on PTE Core — if you are sitting Core for Canadian immigration, this task is not on your test, and the task-to-skill map shows what is.

Note the last row. This is an integrated item. It is not a speaking question that happens to have audio attached. Pearson counts it in your listening score as well, which means a bad answer here costs you in two columns, and a good one pays in two.

The three traits, and why one of them is not like the others

The Score Guide lists the traits scored on Retell Lecture on pages 21 to 24. There are three:

Trait	Scale	Scored by
Content	0–6	AI plus a human expert review, with a second human deciding if they disagree
Pronunciation	0–5	AI only
Oral Fluency	0–5	AI only

Two things fall out of that table, and both are published facts rather than inference.

Content is the largest single trait on the task. It carries a six-point scale where the other two carry five.

Content is scored first, and it is a gate. Pearson's guide sets out the order: the response is scored on Content, and if Content is zero, no score points are given for the response at all. Pronunciation and Oral Fluency are only reached if Content cleared zero. The same page lists, among the responses that receive nothing, one that fails to deal with the prompt because of "significant amounts of pre-prepared/memorized material" (Score Guide, page 8).

So the fluency work that so many candidates do is real work on a trait that a zero Content score prevents anyone from ever looking at. That is not a claim about how Pearson's software is written — Pearson does not publish its algorithm, and nobody outside it can tell you how the number is produced. It is simply the scoring order printed in the guide.

The part that reframes everything: read the Content rubric

Here is the thing the example banks skip. The Content descriptors for Retell Lecture are not a description of speaking. They are a description of comprehension.

Paraphrasing Pearson's own 0–6 scale (Score Guide, pages 21–23) into plainer wording, the ladder runs like this:

Content band	What the descriptor is actually testing
6	You understood the lecture fully, restated its main ideas in your own words, and added specific important detail
5	You got the main ideas and some real detail right, in your own words, with small slips that do not change the meaning
4	Some main ideas, some detail, a few inaccuracies, or attention paid to the less important material
3	Some ideas, not reliably accurate, main points not distinguished from detail — and language lifted from the lecture without reformulation
2	Mostly inaccurate or incomplete, leaning heavily on repeating the lecture's own wording
1	Isolated words and phrases from the lecture, on topic, communicating nothing
0	Related to the lecture but too limited to score

Read bands 2 and 3 again. Repeating the lecture's language without reformulating it is written into the rubric as a symptom of a weak answer. Not a neutral choice, not a safe fallback — a descriptor of the bottom half of the scale.

That single fact rules out the two strategies candidates most commonly bring to this task. Echoing back remembered phrases is explicitly a low band. And a memorised frame is worse, because it is not in the lecture at all, which puts it on the wrong side of the Content-zero gate described above. The full argument against scripts is in why PTE templates do not work; the point here is narrower and harder. On Retell Lecture, you are not asked to reproduce the lecture. You are asked to demonstrate that you understood it, by saying it differently.

You cannot say differently what you never encoded in the first place. Which is why this is a listening problem in a speaking costume.

Why fixing your fluency does not fix your score

Follow the chain and you get a diagnosis that is checkable rather than motivational.

1. Content is scored first and gates the other two traits.
2. Content is a comprehension judgement about this lecture.
3. Comprehension here is comprehension after one hearing, with no replay.
4. Nothing you do in the 40 seconds can add information you did not take in during the 90.

So a candidate who speaks smoothly, on topic, for the full window and still scores badly is almost always failing at step 3. They heard the lecture, followed it while it was playing, and had nothing specific left when the microphone opened. What comes out is fluent, generic, and — by the rubric above — a band 2 or 3.

This is also, incidentally, why the task lands in your listening score. Pearson is not being clever. It is scoring the thing the task actually measures.

What to write down, and what to deliberately not write down

Note-taking is the intervention, and most advice about it is one line long: "write keywords, not sentences." True, and not enough to act on. The useful version is a fixed page you draw before the audio starts, with slots that force the right kind of note.

Rule these four zones onto your erasable notepad in the seconds before the lecture plays, one under the other:

Zone	How much goes in it
TOPIC	Two or three words
CLAIM	Five or six words, in your wording
1) 2) 3)	One word each
SO	Five or six words

Four zones. Here is what each one is for and why.

TOPIC — the subject. Almost always available in the opening sentence. Write it and stop; the topic is the cheapest information in the lecture and you will not forget it.

CLAIM — the speaker's position. This is the zone that decides your band, and it is the one nobody fills in. A lecture is not a list of facts; it has a point of view. The speaker is arguing that something changed, or that a common belief is wrong, or that two things are connected. Capture that, in five or six words, in your own wording. If your CLAIM line is filled, you can produce a band 5 answer from almost nothing else.

1) 2) 3) — the supports. Nouns, names, numbers. Their job is not to be the answer. Their job is to be retrieval cues: one word that will pull a whole sentence back out of your memory thirty seconds later. Three cues is the right number for a 40-second window. Four is greed.

SO — the closing move. A consequence, a contrast, a caveat, a "which matters because". Lecturers signal it audibly and candidates almost never write it down, which is why so many answers stop rather than end.

Now the harder half, because what you refuse to write matters as much as what you write.

Do not write full sentences. Every sentence you form on paper is several seconds during which you are composing rather than listening, and the lecture does not pause for you. A candidate transcribing sentence one reliably misses sentence two.

Do not write the speaker's exact phrasing. This is the counterintuitive one, and it comes straight out of the rubric. If your notes are the lecture's words, your answer will be the lecture's words, and bands 2 and 3 name that explicitly. Reformulate as you write. Doing the paraphrase during the 90 seconds — when you have spare capacity — rather than during the 40 seconds, when you have none, is the single highest-value habit on this task.

Do not write connectives, adjectives or verbs. You will supply those fluently while speaking. Paper is for the things your memory will not hand back: names, numbers, the claim.

Stop taking notes once SO is filled. Late detail is the cheapest thing on the page and the most expensive thing to write, because it competes with the closing move you actually need. If the lecture keeps going after you have your four zones, just listen.

The 10 seconds, and the 40

The 10 seconds of preparation are not for writing and not for composing sentences. Spend them doing one thing: decide the order and fix your first four words. Candidates who pause at second one did not do this. Candidates who pause at second twenty ran out of notes, which is a 90-second problem, not a 10-second one.

Then a budget for the window. This is our guidance, not Pearson's — nothing about it is published, and it is simply what fits:

Seconds	What should be on the recording
0–5	Topic and claim, in one sentence
5–30	The three cues, one sentence each
30–38	The SO line
38–40	Silence, deliberately

Two notes on that table. Finishing at 38 and stopping is fine; being cut off mid-clause is not, because a half-finished clause is less language on the tape than a shorter clause you completed.

And notice what a memorised opening costs here. Time your usual "The speaker was discussing…" preamble, then subtract it from 40. Whatever is left is your real content window. On a 40-second task, a preamble does not merely fail to score — it takes a support point off the tape, and you cannot get it back.

What to check in your own recording

You cannot hear your own answer the way a machine hears it, which is the actual reason this task feels unfixable from the inside. Three checks, in order, on a recording of yourself:

1. Read the transcript and mark every word that could only have come from this lecture. Names, numbers, the specific claim. If you are marking three or four words, your Content is the problem and no amount of fluency work will touch it. If you are marking many but they are the lecture's own phrases in the lecture's own order, you are looking at a band 3 for a different reason.
2. Find where your pauses cluster. Pauses in the first five seconds mean you did not fix your opening during the 10. Pauses after twenty seconds mean your notes ran out — go back and look at whether you wrote three cues or one.
3. Check whether your last sentence finished. If the cut landed mid-clause, your budget is wrong, not your English.

The first two are hard to do by ear, which is the point. On Hilingo, Read Aloud answers come back with a transcript, a pause count, and pronunciation broken down to individual sounds in phonetic notation — so Read Aloud is where you can calibrate what your own hesitation actually looks like on paper, and then apply checks one to three to your Retell Lecture recordings knowing what you are listening for.

Practise it as the task it is

If the diagnosis above is right, the practice that helps is not more Retell Lecture examples. It is the note page, under real audio, repeatedly — and only then the speaking.

A drill that works: play a 90-second lecture, fill the four zones, then put the recording away and wait ten minutes before speaking. If you can still produce a coherent answer from four zones and no audio, your encoding is sound and the rest is delivery. If you cannot, you found the real problem, and it is upstream of your mouth.

Hilingo gives you one full scored mock free, with no card. Speaking and writing come back marked per criterion rather than as one number, so you can see whether it was content or delivery holding the task down. There is a single-question practice mode alongside the full mocks, which is what you want for a task like this one — you can drill Retell Lecture on its own rather than sitting two hours to reach three items. The PTE Academic library runs to 45 full mocks plus 45 section tests per section, objective questions come back with the reason the answer was wrong and the evidence sentence quoted, and any question can be shown in 50 languages if the English of the prompt is getting in the way of the English being tested.

Take a free scored mock

• PTE marks distribution — the full task-to-skill map, and why Retell Lecture pays into two columns
• Why PTE templates do not work — the Content-zero gate in full
• What is PTE Academic? — why nothing in the room can infer what you meant
• Official PTE practice vs third-party — which kind of practice this drill belongs to
• PTE score chart — 10 to 90 against IELTS bands and CEFR

Frequently asked questions

How is PTE Retell Lecture scored?

On three traits, listed in Pearson's Test Taker Score Guide at pages 21 to 24: Content on a 0 to 6 scale, Pronunciation on 0 to 5, and Oral Fluency on 0 to 5. Content is scored first and a zero there means no score points for the response at all, so the other two are only reached if Content cleared zero. Content on this task also receives a human expert review alongside the AI, with a second human deciding if the two disagree; Pronunciation and Oral Fluency are machine-scored only. You never see any of these traits on your report — Pearson removed the old enabling skills in November 2021, and what you receive now is one overall score plus four communicative skill scores.

How long do you get for PTE Retell Lecture?

The lecture runs up to 90 seconds, you then get 10 seconds to prepare, and the recording window is 40 seconds. There are normally 2 to 3 Retell Lecture items on a PTE Academic test. There are none on PTE Core, which does not include this question type.

What should I write in my Retell Lecture notes?

Four things, in fixed positions: the topic in two or three words, the speaker's claim in five or six words of your own, three single-word retrieval cues for the supports, and the closing move — the consequence or contrast the lecture ends on. Write the claim in your own wording rather than the speaker's, because Pearson's Content descriptors treat repeating the lecture's language without reformulation as a marker of the lower bands. Do not write full sentences; every sentence you write is time you are not listening.

Can I use a template for PTE Retell Lecture?

Not usefully. Pearson's Score Guide lists a response that fails to deal with the prompt because of significant pre-prepared or memorised material among the responses that receive no score points, and Content on this task is reviewed by a human as well as the AI. Beyond the detection question there is simple arithmetic: you have 40 seconds, and every second spent on words that are the same whatever the lecture said is a second not spent on the words that earn Content. A fixed order of what to mention is fine — a fixed set of sentences is not.

Why is my PTE speaking score low when I speak fluently?

Because on the integrated speaking tasks, fluency is not the first thing marked. Retell Lecture is scored on Content first, and Content is a judgement about how much of that specific lecture your answer accurately reproduced in your own words. A smooth, generic, on-topic answer with nothing specific in it sits low on that scale no matter how it sounds. If your fluency work has not moved your score, the next thing to test is not your delivery — it is how much of the lecture you still had in your head when the microphone opened.

Does Retell Lecture affect my listening score?

Yes. Pearson lists it as an integrated skills question scored for Listening and Speaking, so the score on the item contributes to both communicative skills. That cuts both ways: a weak answer costs you twice, and it also means listening practice and Retell Lecture practice are the same activity rather than competing ones.