IELTS Speaking’s Bare Tests: Practising Hard Without Model Answers

Problem: model answers become crutches — candidates read fifty samples yet freeze on the fifty-first question because they have never produced under exam pressure. Answer in brief: train on bare question sets with a cold-attempt protocol: predict Part 3 from Part 2, answer aloud with no preparation beyond exam timing, then build your own answer key by recording and transcription. Method, models and 12 tasks below.

Learning outcomes

By the end you will be able to: attempt unseen Speaking tests cold under exact exam timing; predict likely Part 3 questions from any Part 2 card before hearing them; run partner practice with examiner scripts and solo practice with voice-prompt modes; and convert recordings into self-made answer keys via transcription and shape-labelling. All guidance follows the source book, 50 IELTS Speaking Bundle for 9 Band Practice (50-ielts-speaking-bundle-for-9-band-practice) (Jobin Thomas), whose format/timing/preparation chapters and fifty complete three-part tests (bare question sets, no model answers) were verified by full-structure scan with test sampling.

Why answer-free practice works

Reading a model feels like learning; producing an answer is learning. Bare tests enforce four production realities models hide: idea-finding under time (the one-minute Part 2 prep with no sample to lean on); unscripted Part 3 argument (no model to echo, so the opinion–reason–example–balance shape must be genuinely owned); recovery from blanks (stalling mid-answer with only your own resources); and honest calibration (a recording of your cold answer cannot flatter you). Use models to study architecture, then prove ownership on bare tests — the sequence matters: models first for shapes, bare tests second for proof, never bare tests as mere reading material.

The cold-attempt protocol

Each bare test runs exactly as the exam does: read the Part 1 questions aloud (or have them read to you), answer each in 2–4 sentences; take exactly one minute on the Part 2 card with keyword notes; speak 1–2 minutes; answer Part 3 at discussion length. Record everything. Then build your own key in three passes: (1) transcribe the Part 2 and two Part 3 answers verbatim; (2) label shapes (extension shapes, brainstorm coverage, discussion moves); (3) prescribe exactly two fixes (the two highest-value gaps, not ten). File the transcription with its fixes; re-attempt the same test after two weeks and compare transcripts — visible shape improvement is the progress metric no band estimator can fake.

Predicting Part 3 from Part 2

Part 3 grows from the Part 2 topic, so prediction is trainable and strategically gold: thirty seconds of anticipation buys organised answers. The prediction map: person cards → qualities, role models, influence questions (“Can ordinary people be role models? How has the concept changed?”); journey/place cards → travel benefits, perspective change, tourism impact (“How has travel changed? What impact does tourism have?”); object/habit cards → lifestyle, technology, change-over-time questions. Practise by covering each test’s Part 3, writing three predicted questions from its Part 2 card, then uncovering and comparing — when predictions routinely match, you are thinking topically like the test, and no Part 3 opener will ever surprise you.

Partner scripts and solo examiner modes

Bare tests are examiner scripts waiting for a reader. Partner mode: the partner reads questions verbatim (including instructions), times strictly, never rephrases, and follows up twice per Part 3 answer with “why?” or “can you give an example?” — scripted pressure, fair pressure. Solo mode: text-to-speech reads each question with a five-second gap you must fill; or cover-and-reveal with a timer. Rotate modes weekly: partner for listening pressure and follow-up robustness, solo for volume. Log per-test scores on four crude axes (extension held? bullets covered? discussion moves present? recovery clean?) — trends across ten tests diagnose more honestly than any single mock band.

Worked models: one bare test, answered live

Labelling: the demonstration below cold-attempts the book’s Test 1 pattern (journey card with travel Part 3) with newly written answers produced as if live — including one scripted stumble and recovery. The book’s questions are paraphrased, not reproduced verbatim; no model answers exist in the book to copy. Suggested models — other valid answers exist.

Model 1 — Part 1: free time and hometown, cold (newly written)

Q: What do you do in your free time? “Mostly walk — long evening loops with a podcast (answer). It unwinds the desk out of me better than scrolling does (reason). Last week I finally mapped the canal path end to end, four evenings (detail).” Q: Can you describe your hometown? “A river town of about eighty thousand, famous locally for its boat-building past (what). The yards are mostly museums now, although the festival still launches one new boat a year (contrast + although). I miss its unhurriedness whenever capital traffic tests me (feeling).” Live note: shapes held without preparation — proof that shape drilling transfers to unseen questions.

Model 2 — Part 2: journey card, one-minute plan shown (newly written)

Card (paraphrased pattern): a memorable journey — where, how travelled, what happened, why memorable.

Minute’s notes: night bus — coast road — breakdown, fishermen’s tea — kindness remembered.

Talk (suggested model, ~105 seconds): “The journey was an overnight bus down the coast to my grandmother’s village — twelve hours scheduled, nineteen delivered (where + hook). Halfway, the engine died in a fishing hamlet at 2 a.m., and forty passengers stood blinking under tube lights (episode with time-marker). What happened next redeemed everything: fishermen’s families brought out tea and mats, and strangers shared biscuits and phone chargers till dawn (episode 2, warmth). We reached at sunrise to a whole village laughing at our soot-smudged faces (arrival scene). What made it memorable wasn’t the delay but the unplanned hospitality — timetables break, people mend (reflection layer). It has set my travel philosophy ever since: plan lightly, receive gratefully (close).” Live note: keyword notes only; episodes and layers from the architecture method, story from the core.

Model 3 — Part 3: predicted questions, answered (newly written)

Predicted from the journey card before revealing: why people travel / benefits / perspective change / tourism impact. Q: Can travel change a person’s perspective? “Genuinely, yes (opinion). Routine hardens assumptions — the same streets, same conversations (reason). Travel breaks the loop: a 2 a.m. breakdown teaches reliance on strangers faster than any lecture on community (evidence from the talk, legitimately reused once). People return with widened tolerance because they have practised being the outsider (mechanism). The effect fades without reflection, though — souvenirs without thought teach nothing (balance).” Q: What impact does tourism have on a country? “Foreign exchange and jobs first (answer). Although fragile sites erode under footfall, entry revenues fund the rangers and repairs (concession + verdict). Managed well — caps, local hiring, behaviour codes — tourism preserves what it showcases; unmanaged, it consumes it (conditional close).” Live note: predictions matched; one talk detail reused once (allowed), never re-narrated.

Model 4 — Scripted stumble and recovery (newly written)

Mid-talk blank: “…and then we — sorry, I’ve lost the thread (acknowledge, don’t freeze). What I was saying is the waiting mattered more than the arrival (re-anchor on the final bullet — always legal). Stranded passengers… (continue).” Live note: acknowledgement plus re-anchor costs five seconds and zero fluency marks; silent panic costs both. Rehearse the recovery sentence until it is automatic — bare tests are where it gets match practice.

Second live demo: person card with role-model Part 3 (newly written)

Predictions from the card (“a person you admire”): leader qualities / ordinary role models / parents and children / celebrities / concept change over time.

Minute’s notes: school librarian — Mrs Iyer — stayed late, found books — attention as care.

Talk (suggested model, ~100 seconds): “The person is Mrs Iyer, my school librarian for six years (who + how known). She did one extraordinary ordinary thing: she noticed what each child reached for and reserved it before they asked (what done). For me it was atlases — she once kept a new atlas behind her desk for a fortnight until my exams ended, then handed it over like contraband (episode with detail). What I admired wasn’t the books but the attention: in a school of two thousand, being specifically seen (reflection layer). That attention has shaped how I lead small teams ever since — notice first, assign second (close with transfer).”

Part 3 cold answers: “Can ordinary people be role models? More effectively than celebrities (opinion) — proximity beats publicity, because children copy what they can observe daily (reason). A librarian’s consistency teaches more than a star’s highlight reel (evidence by contrast). Although fame amplifies reach, it dilutes verifiability — followers cannot check the performance (concession). How has the concept changed? It used to mean public achievers (past); now micro-communities elect their own — coaches, nurses, moderators (present). The internet didn’t kill role models; it localised them (evaluation).” Live note: five predictions, two asked, both answered from the map — prediction converts surprise into routine.

Practice bank: 12 tasks with answers

Attempt cold and recorded before opening each answer. All items are newly written; the book’s tests are paraphrased, never reproduced.

Common errors and corrections

• Reading models as practice: fifty samples read, zero produced. Correction: models study architecture; bare tests prove ownership.
• Untimed comfort attempts: leisurely answers without pressure. Correction: exam-exact timing or the proof is void.
• Restart habit: re-taking stumbles in practice. Correction: recovery sentences rehearsed; silence never.
• Prediction neglect: Part 3 always a surprise. Correction: three written predictions per test before revealing.
• Ten-fix reviews: overwhelming post-mortems. Correction: two highest-value fixes per test.
• Retelling Part 2 in Part 3: repetition mistaken for development. Correction: reuse once as evidence; argue anew.

Independent task and self-check

Task (newly written): cold-attempt three book tests this week (two solo, one partner), transcribe and label each, prescribe two fixes per test, and re-attempt the weakest after a fortnight. Chart the four log axes across all attempts.

Self-check: were all attempts exam-timed and recorded? Were Part 3 predictions written before revealing? Does each test have exactly two prescribed fixes? Did the re-attempt transcript improve on both? Four yeses = production training working.

Study sequence with the book

1. Read the format/timing/preparation chapters; fix exam conditions and the record-yourself habit.
2. Study shapes from the companion articles (frameworks, brainstorm, discussion moves) before touching the tests.
3. Cold-attempt Tests 1–5 with full self-made keys; calibrate prediction accuracy.
4. Run the 10-week rotation (new tests plus re-attempts), partner alternating with solo.
5. Chart log axes every five tests; drill the flattest axis for a week.
6. Final fortnight: partner-only simulations; last full key 48 hours before the exam.

FAQs

Aren’t model answers necessary to learn? They teach architecture, not production. Learn shapes from models, then prove them on bare tests — both stages are necessary, neither sufficient.

What if my transcriptions embarrass me? Good — embarrassment with labels is the fastest teacher. Every strong speaker has a drawer of bad transcripts.

How accurate are my Part 3 predictions likely to be? High within weeks: Part 3 grows lawfully from Part 2 topics, and the prediction map in this article covers the common families.

Partner or teacher for examiner mode? Any consistent reader works — strictness matters more than expertise. Save expert feedback for monthly reviews of your keys.

How does this book differ from the 50-Sets simulation volume? That volume gives complete tests with model responses plus delivery systems; this bundle gives fifty bare tests — pure production material. Simulate with models there; prove yourself here.

The matching book

50 IELTS Speaking Bundle for 9 Band Practice contains fifty complete three-part tests as bare question sets — Part 1 interviews, Part 2 cards and Part 3 discussions — with format, timing and preparation guidance. If cold production training helped here, the book is where you repeat it fifty times. Product page link to be added when the authenticated staging catalogue mapping is available — no URL is invented here.

Continue your practice with 50 IELTS Speaking Bundle for 9 Band Practice

Want more practice? This article introduces the key principles, but 50 IELTS Speaking Bundle for 9 Band Practice provides additional practice, worked examples and structured preparation. View the book →

Sources and editorial note

Teaching follows the verified source EPUB (format/timing/preparation chapters; 50 complete bare tests, confirmed by structural scan and sampling including Tests 1–2 and 6). The book’s questions are paraphrased in this article, never reproduced verbatim; all answers are newly written suggested models. Test-format summaries are editorial paraphrase; current exam rules should be confirmed against the official IELTS documentation before booking.