Memorised language is penalised, on purpose
This surprises people, because a large part of the preparation industry sells scripts. But the speaking descriptors are explicit that memorised or formulaic language does not demonstrate ability — and in IELTS, an answer identified as memorised can be discounted.
The logic is straightforward once you look at what the criteria measure. They ask whether you can use language flexibly and appropriately: whether you can select the right words for this question, connect ideas as they arise, and adapt when the examiner pushes back.
A memorised paragraph demonstrates none of that. It demonstrates that you memorised a paragraph. It is evidence about your memory, and the test is not measuring your memory.
How examiners can tell
This is the part that is rarely explained, and it is worth understanding because it also explains why the damage spreads.
1. The fluency changes gear
A rehearsed chunk comes out fast, smooth and above the candidate's level — no hesitation, no self-correction, complex structures delivered perfectly. Then the question moves somewhere unscripted, and the delivery collapses: pauses appear, the grammar simplifies, the vocabulary shrinks.
The examiner hears a step change, and it is unmistakable, because real speech does not work that way. Genuine fluency is roughly consistent within a person. It does not switch between two levels at a paragraph boundary.
2. It does not quite answer the question
A memorised answer is built for a topic, but examiners ask specific questions. The script is about technology; the question was whether older people are disadvantaged by technology. The script gets deployed anyway, because it is what the candidate has, and it slides past the actual question.
That is scored directly — the criteria ask whether you develop the topic appropriately and answer what was asked. Meanwhile the examiner has heard it slide, which is its own signal.
3. The register is wrong for speech
Memorised material is almost always written English delivered aloud. In contemporary society, there exists a plethora of factors which contribute to this phenomenon is not something a person says. Nobody talks like that, in any language, and it is instantly audible.
4. It survives the follow-up question, or it does not
Part 3 is a discussion, and examiners probe. You said older people struggle — but is that changing? A script has no answer to a question it did not anticipate, and the candidate either restarts the same script or stalls. This is the moment the whole thing becomes obvious, and it is why Part 3 is where memorisation is most reliably caught.
What to learn instead of a script
"Just be spontaneous" is useless advice. Nobody speaks a second language spontaneously by being told to. What actually works is to memorise the machinery rather than the content — the connective frames that let you say anything, rather than the sentences that let you say one thing.
| A script (do not) | A frame (do) |
|---|---|
| Technology has revolutionised our lives in myriad ways… | It depends on … / I suppose it varies, but… |
| There are three main reasons for this phenomenon… | The main reason, I think, is … and on top of that… |
| In conclusion, I firmly believe that… | So on balance I'd say … / If I had to pick… |
| This is a double-edged sword… | Although X, the trouble is that Y. |
The right-hand column is memorisable, and it is not detectable, because it is what fluent speakers actually use. These phrases carry no content — they carry relationships, so they attach to any topic and they never slide past a question.
The features of actual spoken English
Natural speech is not polished writing. It has properties that candidates often try to suppress because they believe they sound uneducated — and suppressing them is what makes speech sound recited.
- Contractions. I'd, it's, doesn't, there's. Speaking in full forms (I would, it is, does not) sounds like a formal announcement.
- Hedges and vagueness. Sort of, kind of, I guess, more or less, something like that. Fluent speakers hedge constantly, and hedging is real vocabulary.
- Discourse markers. Well, actually, I mean, you know, anyway. These organise speech and are used by every native speaker in every conversation.
- Self-correction. It was in 2019 — no, sorry, 2018. Repairing yourself is a normal feature of speech, and doing it smoothly is evidence of control.
- Occasional hesitation while you think. The descriptors distinguish content-related hesitation (deciding what to say) from language-related hesitation (hunting for a word). The first is normal and is not what caps you.
None of these are errors. Candidates avoid them because they are trying to sound like an essay, and an essay read aloud is exactly what a memorised answer sounds like.
How to actually get there
- Learn ten frames, not ten answers. The connective machinery above. Drill them until they arrive without thought — that is memorisation used correctly, on the part that generalises.
- Practise on questions you have not seen. A script only survives in the presence of a predicted question. Deliberately answer prompts you have not prepared for.
- Practise the follow-up. Say something, then have someone (or something) challenge it, and defend or revise. This is Part 3, and it is where rehearsal breaks.
- Record yourself and listen for the gear change. If one part of your answer is noticeably more fluent than the rest, an examiner will hear it too — and the fluent part is the part you should be worried about.
- Talk about real things. Concrete memory produces natural language automatically; abstract opinion invites the script back in.
For the test, and for real English
For the test: memorise frames, never content. Answer the question that was asked, in the register of speech rather than of writing, and let the small imperfections of natural speech stay in — they are not costing you what you think they are.
For real English: this is one of the rare places where the test is asking for exactly the right thing. Nobody in a meeting, an interview or a conversation wants a recited paragraph, and the ability to build an answer while speaking — hedging, self-correcting, adapting to what the other person just said — is the whole of being able to talk to people.
The frames are what you keep. They are not exam technique; they are how English speakers assemble a sentence in real time, and they will still be useful long after the certificate stops mattering.
Frequently asked questions
Are memorised answers allowed in IELTS speaking?
They are not rewarded, and they can be actively discounted. The speaking descriptors measure whether you can use language flexibly and appropriately — selecting words for this question, connecting ideas as they arise, adapting when the examiner pushes back. A memorised paragraph demonstrates none of that; it demonstrates that you memorised a paragraph, and the test is not measuring your memory.
How do examiners know if an answer is memorised?
Mainly by a change of gear in the fluency. A rehearsed chunk comes out fast, smooth and above the candidate's actual level, and then the delivery collapses the moment the question moves somewhere unscripted. Real fluency is roughly consistent within a person and does not switch levels at a paragraph boundary. Memorised answers also tend to answer a topic rather than the specific question, use written rather than spoken register, and fall apart under a follow-up question.
Why is using a template worse than not using one?
Because the contrast does not just expose the rehearsed part — it casts doubt on the rest of your speech. Once an examiner has heard a fluent stretch that does not match the rest, they have reason to treat it as unrepresentative and to weight your unscripted speech more heavily. The template buys you a beautiful thirty seconds and invites the examiner to disregard it.
What should I memorise instead of answers?
Frames rather than content — the connective machinery that attaches to any topic. 'It depends on whether...', 'the main reason, I think, is...', 'although X, the trouble is that Y', 'so on balance I'd say...'. These are undetectable because fluent speakers genuinely use them, they never slide past the question because they carry no content of their own, and they produce conditionals, subordinate clauses and cohesion for free.
Do contractions and fillers make me sound less educated?
No — avoiding them is what makes speech sound recited. Contractions, hedges like 'sort of' and 'I guess', discourse markers like 'well' and 'actually', and smooth self-correction are all features of fluent natural speech, and every native speaker uses them constantly. Candidates suppress them because they are trying to sound like an essay, and an essay read aloud is exactly what a memorised answer sounds like.