How IELTS Speaking Part 3 exposes the limits of Part 2 rehearsal

IELTS Speaking is a single live interview that runs for roughly 11 to 14 minutes and is divided into three parts, each with its own question types, register, and scoring pressure. Most candidates arrive at the test centre over-prepared for Part 2 cue cards and under-prepared for the rest of the interview, and that imbalance is exactly what separates a Band 6 from a Band 7. This article walks through the question-type progression across IELTS Speaking Parts 1 to 3, names the examiner behaviours that should change the way you answer, and gives a tactical preparation plan centred on the productive skills syllabus at TestPrep İstanbul's IELTS Academic preparation programme.

Throughout, the focus stays on what the examiner is actually listening for in each part and how your answers must evolve across roughly twelve minutes of speaking. The aim is not to memorise scripts; it is to recognise the question types in real time, adjust register, and produce answers that are clearly audible, developed, and relevant under timed pressure.

The three-part architecture of the IELTS Speaking test

IELTS Speaking is one continuous interview with one examiner, recorded for marking and possible re-moderation. Part 1 lasts four to five minutes, Part 2 lasts three to four minutes including one minute of preparation, and Part 3 lasts four to five minutes. The total interview sits inside an 11 to 14-minute window. Knowing those numbers matters because pacing decisions are made against them: a candidate who speaks for 90 seconds in Part 1 has eaten into a budget that the examiner normally expects to span five distinct questions.

Part 1 is a factual, familiar-topic exchange: the examiner introduces themselves, confirms your identity, and then moves through a small set of short questions on everyday subjects such as your hometown, your studies, your work, food, weather, reading, or daily routine. The questions are direct. They expect short, grammatical, personally relevant answers of roughly two to four sentences. The examiner is listening for fluency at a comfortable speed, basic verb tense control, and the ability to extend an answer without losing the thread.

Part 2 is the long turn. The examiner hands you a cue card with a topic, a short prompt, and four bullet points telling you what to include. You get exactly one minute to prepare, then you must speak for one to two minutes without interruption. After you finish, the examiner may ask one or two short follow-up questions before moving on. This is the part most candidates rehearse, and the part where the marking is most generous to prepared language — provided the prepared language still sounds like spontaneous speech.

Part 3 is a two-way discussion. The examiner picks a theme related to the Part 2 topic and pushes you into abstract questions: reasons, comparisons, predictions, opinions, and evaluations. The questions are longer, the answers are expected to be longer, and the lexical and grammatical range is judged more strictly. For most candidates, this is where the band score is actually decided, because Part 1 alone is rarely enough to demonstrate upper-band ability and Part 2 can be lifted by rehearsal.

Architecturally, the test moves from concrete to abstract, from personal to general, and from short turns to extended turns. The examiner is testing whether you can ride that gradient without collapsing back into yes/no answers or rehearsed mini-monologues. Treat the three parts as three different micro-tests wearing the same interview.

Question types in Part 1: the familiar-topic exchange

Part 1 question types are tightly constrained. You will see present-simple questions about habits and routines, present-simple questions about preferences, past-simple questions about experience, and a small number of future forms about plans. Wh-questions dominate. Auxiliary do/does/did is the standard frame. The examiner rarely asks a question that needs more than three or four sentences to answer well.

Concrete examples help here. A typical sequence on the topic of reading might run: Do you like reading? What kind of books do you read? Did you read more when you were a child? Do you think reading is important? Each question is structurally simple, but the topic shifts and the verb tense shifts, and the examiner is watching whether you can keep up. A candidate who answers every question in present simple — even the past one — is signalling a ceiling around Band 5.5 on grammatical range.

The tactical question is how long to speak. Most candidates reading this default to one-sentence answers because the questions feel small. That is a mistake. The examiner wants two to four sentences, with a clear main answer, one piece of supporting detail, and a natural closing remark. A model answer for What do you usually do on weekends? might run: I usually spend Saturday morning at a café near my flat, then in the afternoon I meet a friend for a walk. Sundays are quieter — I cook, watch a film, and try to read for an hour before bed. Three sentences, three different verbs, three different time anchors. That is the texture Part 1 is built to reward.

Common pitfalls in Part 1 cluster around three habits. The first is the monosyllable answer — yes, no, sometimes, often — which gives the examiner nothing to score. The second is the over-rehearsed mini-monologue, in which a candidate treats Do you like cooking? as the prompt for a 30-second speech on the joys of Italian cuisine. The third is the tense slip, where a past question is answered in present simple because the candidate stopped listening after the first three words. Each of these is fixable with the same drill: listen for the auxiliary, answer the verb tense you heard, extend by one sentence with a concrete example, then stop.

Question types in Part 2: the cue card long turn

Part 2 introduces a new question type: a written prompt with structured bullet points and a one-minute preparation window. The prompt always contains a topic, a short framing sentence, and three to four bullet points telling you what to include. A representative example might read: Describe a place you often visit to relax. You should say: where it is, how often you go there, what you do there, and explain why you find it relaxing. The four bullets function as a checklist, not a script. You must cover all of them, but you do not have to cover them in order, and you can add anything relevant that is not on the list.

The two-minute speaking window is what catches candidates out. Two minutes is longer than it sounds. A native-speed speaker produces roughly 250 to 300 words in two minutes, which means your answer needs to be a small structured talk, not a paragraph. The examiner is listening for the ability to sustain relevant speech without trailing off, repeating, or running out of content at the one-minute mark.

The preparation minute is the most under-used minute in the test. Most candidates reading this probably waste it trying to write full sentences on the cue card. A more reliable approach is to spend the minute on three tasks: pick a clear example, decide the order you will cover the bullets, and underline two or three vocabulary items you want to use without forcing. If you cannot remember a word during the talk, paraphrase — do not freeze. Examiners do not penalise paraphrase; they penalise silence.

Lexical resource is judged more strictly in Part 2 than in Part 1. The examiner is listening for the ability to use less common vocabulary, to collocate naturally, and to paraphrase when a precise word does not come. A candidate who says The place is very nice and I go there a lot is producing fluent but flat language. A candidate who says It's a small café tucked behind the post office, and I tend to drift in there on Saturday mornings when the streets get noisy is showing range, precision, and the ability to handle a less common item like drift in. Both are grammatical. Only one is going to push the lexical resource descriptor upward.

A useful preparation drill is the four-bullet drill. Pick any past cue card, set a two-minute timer, and speak continuously until the timer ends. If you stop before the timer, you have not yet internalised the rhythm of two minutes. If you run past the timer, you are over-explaining. The aim is to land within ten seconds of the two-minute mark on the second or third attempt. Most candidates reading this will not pass that drill on their first try. That is the point.

Question types in Part 3: the two-way abstract discussion

Part 3 is the part the public conversation about IELTS Speaking under-represents, and it is the part that decides most band scores. The examiner chooses a theme that connects loosely to the Part 2 topic, then asks four to eight questions that move from concrete to abstract. The question types you will encounter include: reasons and explanations, comparisons between groups or time periods, predictions about the future, hypotheticals, opinions and preferences, and evaluations of advantages and disadvantages.

Concrete examples anchor the abstract grammar. On a Part 2 topic about a place you visit to relax, the Part 3 theme might be Leisure and public spaces. The questions could run: Why do some people prefer outdoor activities to indoor ones? How have public parks changed in your country over the past few decades? Do you think the government should be responsible for providing leisure facilities? What might happen to traditional forms of leisure as cities get busier? Notice the verbs: prefer, have changed, should be, might happen. Each question demands a different verb tense and a different rhetorical move.

The answer length expectation rises sharply. Where Part 1 wanted two to four sentences, Part 3 wants four to eight sentences, with a clear position, a reason, an example, and a consequence or contrast. A Band 6 answer is often a position plus a vague example: Yes, I think the government should provide parks because they are good for people. A Band 7 answer keeps the position but adds structure: I think local councils do have a role, mainly because not everyone can afford private gyms or weekend trips, but I would not say the government should build everything — they can partner with community groups to maintain smaller green spaces. The grammar is similar. The discourse structure is what changed.

Lexical resource is where Part 3 really bites. The examiner is listening for topic-appropriate vocabulary: words like funding, infrastructure, accessibility, well-being, urbanisation, demographic shift, work-life balance on a leisure theme, or curriculum, autonomy, rote learning, critical thinking, standardised assessment on an education theme. Candidates who rely on the same general-purpose adjectives — important, big, good, bad — through Part 3 are capping themselves at Band 6 on lexical resource regardless of how fluent they sound.

Examiner behaviours that change how you should answer

The examiner is not a passive interlocutor. They use a small repertoire of behaviours that should change what you do next. Five of them matter most.

The first is the topic shift signal. The examiner says something like Let's move on to talk more about… or Now I'd like to discuss some more general ideas related to… That phrase is the border between Part 1 and Part 2, or between Part 2 and Part 3. When you hear it, the register is about to change. In Part 1, register is casual and personal. In Part 3, register is academic and discursive. Failing to change register is one of the most common reasons a Part 1 answer sounds like a Part 3 answer, or vice versa.

The second is the follow-up probe. After you finish a Part 1 answer, the examiner may ask Why? or Why is that? This is not a new question; it is an invitation to extend. Candidates who treat it as a new question often contradict themselves. Candidates who hear Why? and produce one extra sentence of reason-keeping score higher on fluency and coherence.

The third is the paraphrase. The examiner will sometimes rephrase a question if you look lost. A candidate who does not notice the paraphrase and tries to answer the original wording will sound confused. A candidate who recognises the rephrase as a second chance is showing the listening skill the band descriptors reward.

The fourth is the silence. After you finish speaking, the examiner will often pause for one to two seconds before asking the next question. Candidates who fill that silence with and also, in my opinion, I think that… are showing they do not trust the structure of the test. Candidates who sit in the silence show the examiner they are confident enough to let an answer end.

The fifth is the non-verbal. The examiner is trained to keep a neutral face. A candidate who is reading facial cues for grade signals is reading the wrong data. The grade signals are in the audio you produce, not the expression the examiner wears.

Common pitfalls and how to avoid them

Five pitfalls account for most of the lost marks across Parts 1 to 3. Each is fixable with a specific rehearsal habit.

• Pitfall 1 — answers that are too short. Part 1 wants two to four sentences, Part 3 wants four to eight. Drill with a timer; count the seconds, not the sentences.
• Pitfall 2 — answers that are too long. Part 1 over-extension is just as costly as under-extension. If the examiner interrupts you in Part 1, your answer was too long. Shorten by removing the second example.
• Pitfall 3 — yes/no answers with no extension. Every opinion question in Part 3 needs a position, a reason, an example, and a consequence. Practice the four-move template until it becomes automatic.
• Pitfall 4 — grammatical range that collapses under pressure. Candidates who produce complex grammar in Part 2 sometimes flatten in Part 3 because the questions are abstract. Drill Part 3 with a friend who asks cold questions, not pre-rehearsed ones.
• Pitfall 5 — lexical resource stuck on general-purpose words. Build a topic vocabulary list of 15 to 20 items per theme, then use them in timed Part 3 drills until they are automatic.

Register and tone: the silent scoring axis

Register is the layer of language most candidates do not think about because it is not a grammar rule. It is the level of formality, the choice of vocabulary, and the sentence structure that signals whether you are speaking in a casual interview or an academic discussion. The IELTS Speaking test uses two registers. Part 1 is casual. Part 3 is academic. Part 2 sits between them, leaning casual because the topic is personal but tilting academic because the examiner wants to see range.

Casual register is marked by contractions, personal pronouns, simple connectives, and a relatively narrow vocabulary range. I usually go to the gym on Saturdays because I like staying fit is casual register. Academic register is marked by full verb forms, topic-specific vocabulary, more complex connectives, and a willingness to take a position and defend it. Regular exercise tends to be a habit people build in their twenties, and gyms often play a social role as well as a physical one is academic register. Both can be Band 7. The mistake is mixing them in the wrong part.

The reason register matters for scoring is that the examiner is calibrating the band descriptors against what is reasonable for the question. A casual-register answer to a Part 3 question about government policy sounds under-developed, even if the grammar is perfect. An academic-register answer to a Part 1 question about your favourite food sounds rehearsed, even if the vocabulary is impressive. The fix is to listen for the topic, not just the question word.

Scoring signals across the three parts

The four band descriptors — fluency and coherence, lexical resource, grammatical range and accuracy, and pronunciation — are applied to the interview as a whole, but each part stresses different descriptors. A simple comparative view helps.

The table makes a point that is easy to miss in a syllabus. The scoring pressure rises as the test progresses. A candidate who performs well in Part 1 and Part 2 but underperforms in Part 3 will be scored on the underperformance, because Part 3 is where the examiner has the most material to judge upper-band ability. Conversely, a candidate who struggles in Part 1 can still recover in Part 3 if the rest of the interview is strong, because the descriptors are applied holistically.

A six-week preparation plan for Parts 1 to 3

A realistic preparation plan runs about six weeks, with three sessions a week of 30 to 45 minutes each. Longer than that and the gains flatten. Shorter than that and the muscle memory does not build.

Week one focuses on Part 1 only. Record yourself answering eight Part 1 questions on three different topics. Listen back, count the seconds, count the verb tenses, and count the number of distinct vocabulary items per answer. The aim is to land at three to four sentences per answer, with a tense that matches the question and at least one piece of concrete detail.

Week two introduces Part 2 cue cards. Take one cue card per day, run the one-minute preparation drill, and speak for two minutes against a timer. Review the recording for lexical range and for the moment you ran out of content. The aim is to identify the third or fourth sentence where the answer typically collapses, and to build a transition that carries you past it.

Week three introduces Part 3. Pair with a study partner, choose a Part 2 topic, then move into Part 3 questions. The aim is to get used to the register shift. Record both parts and listen for whether your Part 3 answers sound like longer Part 1 answers or genuinely discursive answers. Most candidates find the second week of Part 3 practice is where the band score starts to move.

Week four is consolidation. Mix the three parts in one continuous mock interview, timed. The aim is to reproduce the real test format under fatigue.

Week five is pressure rehearsal. Do two full mock interviews, with the second one recorded and reviewed in full against the band descriptors. The aim is to surface any remaining register mistakes and any vocabulary items that collapse under time pressure.

Week six is rest and light review. One full mock on day one, then three days of light vocabulary review, then the test. Avoid new material in the final 48 hours; the gains have already been made.

How TestPrep İstanbul's productive-skills syllabus covers the three parts

The IELTS Academic preparation programme at TestPrep İstanbul treats Speaking as a single integrated module rather than three separate lessons. The productive-skills syllabus spends roughly equal time on Parts 1, 2, and 3, with two specific design choices that differ from a self-study plan. The first is paired mock interviews every fortnight, recorded and reviewed against the four band descriptors. The second is a structured topic-vocabulary list that is built once and then re-used across Part 2 and Part 3, so the language you rehearse in the cue card carries directly into the abstract discussion.

For most candidates, the largest gain from a structured course is in Part 3, where self-study tends to under-rehearse abstract question types. A diagnostic assessment at the start of the programme places a candidate on the right week of the plan, and the subsequent pacing is adjusted to the candidate's actual band ceiling rather than a calendar date.

Conclusion and next steps

IELTS Speaking is one interview in three movements. Each part has its own question types, its own register, and its own scoring pressure, and the candidate who treats the three parts as three separate tests is the candidate who walks out of the test room with a lower band than their real ability. The work that pays off is the work on Part 3 register, on Part 2 timing, and on Part 1 extension. None of that is memorisable. All of it is drillable.

A sensible next step is to book a diagnostic Speaking mock so the band ceiling is known before the six-week plan begins; TestPrep İstanbul's diagnostic Speaking assessment is the natural starting point for candidates building a Part 1 to Part 3 preparation plan around the productive-skills syllabus.

Frequently asked questions