# The Complete IELTS Academic Reference

- **Category:** `metodo`

---

ALL FOUR MODULES · ONE REFERENCE
The Complete IELTS Academic MasterDocument
A complete reference for all four modules: structures, lexical banks,
question mechanics, exam-day protocol, and a ranked account of the
errors most frequently reported in real IELTS scripts. Written for candidates
working from band 6 towards band 7.5 and above.
All four modules Computer-delivered Grounded in published error data
Print / PDF ready
CONTENTS
Part 0 — How to use this document
Part I — The exam map
Part II — The errors that are actually reported
Part III — Writing Task 1, the report
Part IV — Writing Task 2, the essay
Part V — Speaking, all three parts
Part VI — Listening
Part VII — Reading
Part VIII — Grammar priorities
Part IX — Lexis banks
Part X — The one-page card
Part XI — Exam day protocol
Part XII — After the exam
Part XIII — Drill appendix
1. 
2. 
3. 
4. 
5. 
6. 
7. 
8. 
9. 
10. 
11. 
12. 
13. 
14. 
1

PART 0
How to use this document
A single reference covering all four IELTS Academic modules: technique, structure,
lexis, and the errors that most reliably cost marks. It is written for candidates already
operating at band 6 or above who need to reach 7 or higher, which is a different
problem from learning English.
At that level the obstacle is rarely knowledge. Most candidates who plateau at 6.5 can state the grammar rule
they just broke. The failure happens under production pressure, and it repeats. That is why this document is
organised around three layers rather than around a syllabus.
LAYER WHAT IT CONTAINS HOW TO WORK WITH IT
The
mechanism
layer
Parts II and
VIII
What the published error data
actually shows, why those errors
happen, and the interventions that
shift them
Read closely, once. Return whenever a
score plateaus. This is where band 6.5
becomes band 7.5.
The
architecture
layer
Parts III–VII
Structures, formulas, question
mechanics, model answers
Learn the shapes until they are automatic.
Automatic structure frees working memory
for content and accuracy.
The retrieval
layer
Parts IX and
XIII
Lexical banks, collocations,
production drills
Never read passively. Produce aloud or in
writing. Reading a phrase bank builds
recognition, and recognition is not what the
exam measures.
THE GOVERNING PRINCIPLE
At the upper bands the discriminator is not vocabulary sophistication. It is error density — how many
small grammatical markers survive under time pressure. The official band descriptors make this explicit:
the step from band 6 to band 7 is described in terms of producing frequent error-free sentences, and
the step to band 8 in terms of most sentences being error-free. Nothing in those descriptors rewards
rare words for their own sake.
A short, precise, error-free answer outscores an ornate one carrying three dropped inflections.
Compression is not a weakness. Every section here is written on that assumption.
2

Notation
RED BLOCK · WHAT FAILS
The evidence suggest a clear pattern.
GREEN BLOCK · WHAT TO PRODUCE
The evidence suggests a clear pattern.
Blue boxes carry rules. Amber boxes carry warnings and traps. Green boxes carry protocols to execute.
Purple boxes are spoken language. Teal boxes carry the highest-leverage items in the document.
Where the data in this document comes from
The error frequencies in Part II are drawn from published error-analysis research on real IELTS scripts,
principally Müller and Han's 2022 study for the IELTS Partners, which tagged grammatical errors across 500
Writing Task 2 essays at half-band intervals from 5.5 to 7.5. Where a figure is quoted, it comes from that
dataset unless stated otherwise. Where a claim is a teaching judgement rather than a measured result, it is
written as one.
3

PART I
The exam map
Two hours forty-five minutes of continuous testing for the three written modules;
Speaking is scheduled separately or on the same day.
MODULE SHAPE TIME WHERE THE MARKS ARE ACTUALLY LOST
Listening4 parts, 40
questions
~30 min +
short check
window
Connected speech and distractor design, not
vocabulary. Also instruction compliance on word
limits.
Reading 3 passages, 40
questions
60 min, no
extra transfer
time
Question mechanics — paraphrase traps, True/
False/Not Given, matching headings. Rarely
comprehension.
Writing Task 1 (≥150 w) +
Task 2 (≥250 w)
60 min total; 20
/ 40 split
Under-length responses, unanswered parts of
the prompt, and grammatical accuracy under
time pressure.
SpeakingParts 1, 2, 3 11–14 min Morphological accuracy under load. Fluency
and lexis are usually already at target.
Computer-delivered specifics
There is no ten-minute answer-transfer window in Listening. Answers are typed as you go, so
transcription error between booklet and answer sheet is not a risk category. The trade-off is that there is
no protected review block either — use the short check time at the end deliberately.
The word counter is on screen at all times in Writing. Never estimate a word count. Under-length is a
hard penalty applied by the examiner and it is entirely avoidable.
For the administrative side — booking, ID requirements, cancellation windows, results timeline, One
Skill Retake, and Enquiry on Results — see [IELTS Logistics](ieltsdip-notes.html), which covers that
ground in full and is not duplicated here.
Highlighting and note-taking tools exist in Reading. Use highlighting on the question stem keywords, not
on the passage — the passage is not where the difficulty lives.
Speaking is face-to-face with a human examiner (or video-delivered in some centres). It is not automated.
How the bands are calculated
Listening and Reading are marked out of 40 and converted to a band. The conversion is fixed per test
version and not published in advance; the figures below are the widely reported approximations, not a
guarantee.
Writing is scored on four equally weighted criteria: Task Achievement/Response, Coherence and
Cohesion, Lexical Resource, Grammatical Range and Accuracy. Task 2 counts twice as much as Task
1 in the Writing band.
Speaking is scored on four equally weighted criteria: Fluency and Coherence, Lexical Resource,
Grammatical Range and Accuracy, Pronunciation.
• 
• 
• 
• 
• 
• 
• 
4

The overall band is the mean of the four module bands, rounded: .25 rounds up to the next half band,
.75 rounds up to the next whole band. Anything below .25 rounds down.
APPROXIMATE RAW SCORE LISTENING BAND ACADEMIC READING BAND
39–40 9.0 9.0
37–38 8.5 8.5
35–36 8.0 8.0
32–34 7.5 7.5
30–31 7.0 7.0
26–29 6.5 6.5
23–25 6.0 6.0
ARITHMETIC WORTH INTERNALISING
Where a requirement specifies a minimum in each module, the average cannot rescue a weak one. A
6.5 in Writing alongside 8s elsewhere still fails a "no band below 7" condition, no matter how strong the
overall figure looks. Where the requirement is an overall figure only, the opposite holds and a strong
module can carry a weak one. Establish which shape applies to you before planning, because it
decides whether the marginal study hour belongs to your weakest module or your most improvable one.
• 
5

PART II
The errors that are actually reported
Not a list of everything that can go wrong. A ranked account of what candidates at
bands 5.5 to 7.5 measurably do, drawn from tagged corpora of real scripts — followed
by why those specific errors survive so long, and what displaces them.
1. The overall picture
Across 500 tagged Writing Task 2 essays totalling roughly 145,000 words, about 12,300 words carried a
grammatical error. That is an average error rate near 8.5%, or close to one error per twelve words, pooled
across all bands from 5.5 to 7.5. The authors note their counting method was deliberately conservative: where
one point in a sentence could have been counted as two errors, only one was recorded, and punctuation and
nonsensical word choice were excluded. The real rate is therefore higher than the published figure.
ERROR RATE BY BAND
WRITING BAND MEAN GRAMMATICAL ERROR RATE IN PRACTICAL TERMS
5.5 14.8% About three errors per twenty words
6.0 10.1% About one error per ten words
6.5 8.3%
7.0 6.0%
7.5 4.9% About one error per twenty words
THE MOST USEFUL SINGLE FINDING
Improvement is not linear. The largest drop in error rate happens between 5.5 and 6.0 — roughly a 4.7
percentage-point fall. Between 7.0 and 7.5 the average drop is about 1.5 points. Each successive half-
band costs disproportionately more work than the last.
More striking: between 6.5 and 7.0 the data shows genuine turbulence — several error categories get
worse before improving again. Nouns, adjectives and conjunctions all show increased error rates at that
step. This is consistent with restructuring: as candidates start attempting more complex syntax,
accuracy in categories they had already mastered temporarily degrades. If your accuracy feels worse
while your writing feels more ambitious, that is a documented stage, not a regression in ability.
6

2. The ranked error table
Ranked by how often each grammatical category is used incorrectly, relative to how often it appears at all.
The second column matters, because a category that constitutes a quarter of the text produces far more
absolute errors than one that constitutes 4%.
RANK CATEGORY ERROR
RATE
SHARE
OF TEXT WHAT THIS ACTUALLY MEANS
1 Determiners (a, an,
the, this, that,
possessives)
12.5% 13.2% The single most frequent error type by a
clear margin — roughly two determiner
errors per twenty words written
2 Verbs (tense,
agreement, form)
8.8% 22.0% High rate and the second most common
word class, so verbs generate the
largest raw number of visible errors
3 Pronouns 8.7% 4.4% High rate but low frequency, so fewer
absolute errors
4 Prepositions 8.3% 15.2% Common and consistently wrong — a
major contributor to the impression of
non-native writing
5 Nouns (number,
countability,
possessive)
8.1% 25.6% The most frequent word class in the
text, so even a moderate rate produces
many errors
6 Conjunctions 7.5% 3.7% Infrequent, but each error can distort
the logical relationship between clauses
7 Adverbs 6.7% 6.3%
8 Adjectives 5.2% 9.6% The most reliably controlled category
READ THIS TABLE CORRECTLY
Determiners rank first, but that does not mean articles should absorb most of your study time.
Determiner errors are frequent and highly visible, yet they rarely destroy meaning. Verb errors are
slightly less frequent but far more consequential — a wrong tense changes when something happened,
and a dropped agreement marker is the error examiners are trained to notice because native speakers
do not make it. Prioritise by cost, not only by frequency.
7

3. What survives all the way to band 7.5
This is the most practically useful part of the data. Four error subtypes disappear entirely by band 7.0, while
others persist stubbornly at the top of the measured range. If you are already scoring near 7, these are your
remaining targets — the categories that keep firing when everything else has been cleaned up.
SUBTYPE
ERROR RATE STILL
PRESENT AT BAND
7.5
TYPICAL FORM OF THE ERROR
Possessive endings
('s / s')
18.6% the government's policies · students' results ·
misplaced or missing apostrophe, or an
apostrophe used to form a plural
Proper singular
nouns
12.2% capitalisation and article use with names of
countries, institutions and languages
Personal wh-
pronouns (who,
whom)
10.7% relative clause construction — who / which / that
selection and case
Determiners (general) 8.9% article selection, especially with abstract and
generic nouns
Third-person singular
present verbs
8.6% the dropped -s — still failing in roughly one use in
twelve at band 7.5
Adverb particles
(phrasal verbs)
7.4% wrong particle, or particle omitted
Subordinating
conjunctions
7.5% although / despite / whereas confusion, and
clause-type mismatches
THE HEADLINE FOR ADVANCED CANDIDATES
Third-person singular agreement does not go away. It is still failing roughly 9% of the time at band
7.5 — one of the highest-persistence errors in the entire dataset, and the one most visible to an
examiner, because it is a single letter that native speakers never omit. It is also the cheapest error in
the language to fix, which makes it the highest return on effort available to almost any candidate.
By contrast, past tense verb errors improve the most across the range, falling from about 38% at
band 5.5 to about 6% at 7.5. Past tense responds to instruction. Agreement resists it.
4. First-language interference: the effect is real and measurable
Holding the IELTS band constant, candidates from different first-language backgrounds produce significantly
different error rates. In the corpus data, with Italian speakers as the reference group, the modelled incidence
8

of errors was roughly 1.5 times higher for Chinese speakers, 1.7 times higher for Russian speakers, and 1.75
times higher for Arabic speakers — all statistically significant. Two candidates holding the same band score
can have quite different error profiles, because other criteria in the rubric compensate.
L1
BACKGROUND
MEAN ERROR RATE IN
THE STUDY CATEGORIES WITH THE SHARPEST DIFFICULTY
Italian 5.5% Consistently lowest across every category —
typological proximity to English
Chinese 7.7% Determiners (12.6%), verbs, subordinating
conjunctions
Russian 9.4% Determiners (18.4%) — by far the largest single
difficulty in the dataset
Arabic 10.4% Pronouns (13.8%), adverbs (11.1%), verbs (10.9%)
WHY THESE PATTERNS ARISE — AND WHAT GENERALISES
The mechanism is structural, not a matter of effort. Where the first language lacks a category, English forces
a decision that has no equivalent to fall back on.
9

IF YOUR FIRST
LANGUAGE…
EXPECT THE
DIFFICULTY IN… BECAUSE
Has no articles
(Russian, Chinese,
Japanese, Korean,
Turkish, Hindi, Polish,
Czech, Thai)
Determiners —
a/an/the/zero
Definiteness is carried by context or word order, not
by a separate word. There is no habit to transfer, so
every article is a fresh decision.
Does not inflect verbs
for person
(Chinese, Japanese,
Korean, Vietnamese,
Thai)
Verb agreement
and tense
Time and person are carried by adverbs and context.
The English inflection carries no meaning the
speaker feels, so it drops first under load.
Permits subject
omission
(Spanish, Italian,
Portuguese, Japanese,
Chinese, Arabic)
Dummy
subjects — it is,
there are
The verb ending already identifies the subject, so the
pronoun feels redundant and gets dropped in English
where it is obligatory.
Uses different verb–
preposition pairings
(all Romance
languages, Arabic,
Slavic languages)
Prepositions
and phrasal
verbs
Collocations do not map. depend on, different from, 
married to, arrive to are transfer, not ignorance.
Marks number
differently
(Arabic, Chinese,
Japanese, Korean)
Noun plurals
and countability
English divides nouns into countable and
uncountable in ways few languages replicate. 
Informations, advices, furnitures follow from this.
Shares Latin roots with
English
(Spanish, Italian,
Portuguese, French)
False friends
and register
The advantage in grammar is offset by lexical traps: 
actually / actualmente, eventually / eventualmente, 
assist / asistir.
Uses a different
rhetorical structure
(Arabic, Chinese,
Russian and others)
Coherence and
Cohesion
Written argument conventions differ across traditions.
English academic writing expects the claim first and
the elaboration after; several traditions build toward
the claim instead. This is scored, and it is invisible to
the writer.
10

HOW TO USE YOUR OWN PROFILE
Find your language family in the table above, then spend your grammar time on those two or three
categories rather than on general review. Error analysis of your own writing beats any generic syllabus:
collect fifteen of your own marked sentences, tag the error type in each, and count. The distribution will
be narrow. Most candidates are producing four or five error types repeatedly, not thirty.
5. Why these errors survive: the two mechanisms
MECHANISM 1 — PRODUCTION PRESSURE, NOT MISSING KNOWLEDGE
The acquisition literature is consistent on this. Untimed grammatical knowledge is typically intact at
intermediate-to-advanced level; the failure is specifically in real-time production, when attention is divided
between generating content and monitoring form. What shifts it is explicit instruction combined with active
correction and repeated production — not passive rule review, which candidates have usually already
done several times.
This also explains why the same errors appear in Speaking and Writing. They are not two problems. They are
one problem observed under two levels of time pressure.
MECHANISM 2 — FOSSILISATION AND SKILL REGRESSION
Fossilisation is not a global stop in learning; it is specific to particular structures. At advanced levels, the more
relevant phenomenon is skill regression: forms previously produced correctly start failing again when the
learner takes on more complex syntax. The functional categories — conjunctions, determiners, pronouns —
are the most vulnerable, and this is exactly what the band 6.5-to-7.0 turbulence in the data shows.
The practical consequence: when you begin using more ambitious sentence structures, expect a temporary
rise in basic errors. Do not respond by retreating to simple sentences, and do not respond by adding more
complex structures. Respond by adding a separated revision pass, described below.
6. The four interventions that actually move the number
INTERVENTION 1 · THE HALF-SECOND PAUSE BEFORE THIRD-PERSON VERBS
For Speaking. Not a pause to think about content — a pause placed at the subject, before the verb
arrives. Content-related hesitation is scored as normal; it costs nothing in fluency terms. What it does is
interrupt the automatic fast-production sequence in which the -s is dropped. Given that third-person
agreement is still failing at 8.6% in band 7.5 writing — where candidates have time to revise — the
spoken rate is certainly higher.
11

INTERVENTION 2 · REVERSE-ORDER PROOFREADING
Errors persist despite full awareness of the rule because reading your own text forward activates the
meaning you intended, and comprehension papers over surface form. The eye confirms the sentence
you meant to write.
Finish the full text. No revision during drafting.
Read the last sentence first, then the second-to-last, backwards to the opening.
Out of narrative order, each sentence is processed as a grammatical object rather than as a step in
an argument. Agreement and tense errors become visible where they were invisible forward.
Then run the six-step check in Part VIII, in the order given.
INTERVENTION 3 · SEPARATE PRODUCTION FROM REVISION
Editing while producing splits the working-memory budget between generating the next clause and
auditing the last one. Under that split, the audit takes the resources and the generation gets what is left
— which is precisely when inflections drop and tenses drift. This single habit explains errors that look
unrelated: hesitation in Speaking, and fragments plus comma splices in Writing.
Writing: produce the full paragraph without stopping, then revise in a structurally distinct pass.
Speaking: finish the sentence you started, even if it is not the sentence you wanted. A slightly
awkward completed clause is less visible to an examiner than a restart.
The silent reset: if a word is misjudged, take half a second of silence, reset, and deliver the
corrected phrase cleanly. Silence reads as thinking. Verbal stumbling reads as a language gap.
INTERVENTION 4 · SELF-TRANSCRIPTION
Post-task self-transcription outperforms simple repetition. Record an answer, transcribe it yourself — by
hand or typing, never with automatic transcription — and mark your own errors before consulting any
correction. Transcription forces the monitor onto surface form, which is exactly the channel that fails
under load. Three cycles of this beat thirty minutes of unrecorded speaking.
The same logic applies to Writing: type up your own handwritten essay and the errors surface, because
retyping is a second pass through the surface form.
7. A working error hierarchy
If the ranked table is what the data shows, this is what to do with it. Ordered by cost, not by frequency.
1. 
2. 
3. 
4. 
• 
• 
• 
12

LEVEL PATTERN WHY IT SITS HERE
Level 0
Architectural
Revising while producing The root habit. It generates hesitation in
Speaking and fragments in Writing
simultaneously. Fixing it improves two
modules at once.
Level 1
Expensive and
patterned
Verb agreement (especially third-
person -s and abstract singular
subjects), and tense inconsistency
under load
High persistence into the upper bands,
single-morpheme errors, highly visible to
examiners, and structurally fixable rather
than requiring more knowledge.
Level 2
Isolated,
checklist-
fixable
Determiners, prepositions,
possessive endings, relative
pronouns, comma splices,
fragments, countability, parallelism,
register
Frequent but not driven by a single
mechanism. They respond to a systematic
revision pass, which is why the six-step
check exists.
Level 3
Cosmetic
Spelling of high-frequency academic
words, minor collocation drift
Real but low-yield. Never spend planning
time here.
8. Verb agreement in detail — the highest-persistence error
Because it survives to the top of the measured range, it gets its own treatment. The failures cluster in three
predictable places.
A. ABSTRACT AND UNCOUNTABLE SINGULAR SUBJECTS
These feel plural because they refer to something large or diffuse. They are singular.
THE DECEPTIVE SET
evidence, research, data (in careful singular register), information, knowledge, advice,
feedback, furniture, equipment
trust, growth, exposure, migration, scarcity, availability, perception, effort, interaction
politics, economics, statistics, mathematics, news — -s ending, singular verb, when naming the
field
everyone, everybody, nobody, each, either, neither, none (formally)
the number of… is singular; a number of… is plural
13

FAILS UNDER LOAD TARGET
The evidence suggest otherwise. The evidence suggests otherwise.
Trust take years to build. Trust takes years to build.
Politics affect healthcare more than people
admit.
Politics affects healthcare more than people
admit.
Exposure to screens have increased.Exposure to screens has increased.
The number of applicants have risen.The number of applicants has risen.
She gave me many informations. She gave me a great deal of information.
B. THE INTERVENING-NOUN TRAP
A plural noun between the subject and the verb pulls the agreement the wrong way. The head noun governs,
regardless of what stands between.
WRONG
The findings of this research supports the
hypothesis.
RIGHT
The findings … support. Head noun is plural.
WRONG
The impact of rising costs on families are
severe.
RIGHT
The impact … is severe. Head noun is
singular.
C. AFTER INVERSION
When a sentence opens with a negative adverbial, the auxiliary carries the agreement and the main verb
stays bare. This fails often precisely because the structure is being used to demonstrate range.
WRONG
Not only does this policy reduces costs…
RIGHT
Not only does this policy reduce costs…
9. Tense inconsistency — the second Level 1 error
Present tense infiltrates a sustained past narrative, most reliably at the emotionally engaging or cognitively
demanding point. It is not random. One line of research frames tense shifting in L2 narrative as functioning
almost like private speech — a coping move that reaches for a simpler structure exactly when the content
gets harder. The drift is a load signal, not a discipline failure.
14

Three structural defences, in order of usefulness:
Let the structure carry the tense, not the memory. In Task 1 the chart decides: a dated axis means
past simple, undated static data means present simple, a process means present passive, a map means
past passive for the earlier state. In a Speaking Part 2 narrative, the cue card decides, and the decision is
made in the preparation minute, before the first word.
Pre-authorise the shift. Most collapses happen at the transition from narrating to evaluating. Decide in
advance that the closing sentence moves to the present, and the shift stops being a leak and becomes a
device.
The gerund escape route. When the correct past form will not arrive mid-sentence, replace the
conjugated verb with a gerund and the contradiction is defused. "I went home that day and I feel
exhausted" → "I went home that day feeling exhausted."
RECOVERY PROTOCOL
When a slip is noticed mid-utterance: return to the last correct verb, repair once, briefly, aloud — "Sorry,
took, not take" — and move forward. Never repair the repair. A single clean self-correction is neutral to
slightly positive at band 7–8. A cascade of corrections is a fluency penalty.
10. What the research says about study design
Progress slows at the top. The rule of thumb that ten weeks of intensive study yields half a band is not
supported at the upper levels. Gains from 6.5 to 7.0 take substantially longer than gains below 5.0, and
they are more easily lost.
Living in an English-speaking country is not a study plan. Longitudinal studies of students during
degree programmes found average gains well under half a band over one to three years, with a
meaningful proportion of students scoring the same or lower after their degree. Exposure without
dedicated language work does not reliably raise the score.
Spaced retrieval with interleaving beats blocked massed practice for long-term retention — with one
caveat: close to a test date, interleaving many drill types in a single session adds fatigue faster than it
adds benefit. Near the exam, block by error type.
Error analysis of your own output is the highest-yield diagnostic available, and it is free.
1. 
2. 
3. 
• 
• 
• 
• 
15

PART III
Writing Task 1 — the report
Twenty minutes. Minimum 150 words; target 170–190. One third of the Writing band.
The task is factual reporting, not analysis — the single most reliable way to lose marks
is to explain why something happened.
The unbreakable four-paragraph structure
PARAGRAPH JOB LENGTH NUMBERS?
1 ·
Introduction
The scope — what the graphic represents,
paraphrased
1–2
sentences
Timeframe
only
2 · Overview The diagnosis — 2 or 3 most prominent
features
2–3
sentences
None
3 · Body A Grouped evidence, first group 3–4
sentences
Yes, exact
4 · Body B Grouped evidence, second group 3–4
sentences
Yes, exact
NON-NEGOTIABLE
A report without a clearly identifiable overview paragraph is capped below band 7 on Task Achievement
regardless of how good the rest is. If time runs out, the overview is the paragraph you protect, not the
second body paragraph.
Paragraph 1 — the introduction
Formula: [Graphic type] + [reporting verb] + [what, paraphrased] + [where] + [when].
PILLAR A — UPGRADE THE VERB
LINES, BARS, PIES
illustrates · delineates · depicts · portrays · charts
COMPARISONS AND TABLES
compares [X] in terms of [Y] · sets out · presents data on
16

PROCESSES AND MAPS
explains · details the sequential stages of · outlines · traces the development of
PILLAR B — THE "HOW" TRANSFORMATION
Turn noun phrases into full clauses. This demonstrates grammatical control, not just synonym substitution.
shows the number of
people…
illustrates how many people…
shows the spending on…depicts how much was spent on…
shows the process of… explains how [X] is produced.
shows the changes in the
town
traces how the town was redeveloped
PILLAR C — THE STRUCTURAL FLIP
PROMPT WORDING
The table shows metro systems in six cities in
2020.
FLIPPED
Data regarding underground networks across
six cities in 2020 is set out in the table
provided.
PILLAR D — TIME AND GEOGRAPHY
Between 2005 and 2015 over a decade-long period commencing in 2005
From 1990 to 2020 across three decades · over a thirty-year span
In the UK and USA across two distinct Western nations
In six countries across half a dozen national contexts
TWO SPECIAL CASES
Multiple graphics: join them with while, whereas, or and. "The pie chart portrays [X], while the
accompanying table illustrates [Y]."
Overloaded prompts: never list every item. Condense to an umbrella term. TVs, radios and laptops → 
various types of household electronics. Hamburgers, pizza and fish and chips in the UK, France and
Germany, 1990–2010 → fast-food expenditure across three European nations over a two-decade period.
• 
• 
17

Paragraph 2 — the overview
Two or three most prominent features. No numbers, no percentages, no specific dates — the timeframe itself
is the only permitted temporal reference.
OPENINGS THAT WORK
Overall,
Broadly speaking,
The most prominent trend to emerge is…
A notable feature of the data is…
What stands out across the period is…
OPENINGS TO AVOID
It is immediately apparent that…
Looking from an overall perspective, it is readily apparent that…
An overarching perspective reveals…
18

OVERVIEW FORMULAS BY CHART TYPE
CHART TYPE FORMULA
Dynamic — up vs
down
[Opening] while [A] experienced an upward trajectory, [B] witnessed a marked
decline.
Static —
dominant vs
marginal
[Opening] [A] stood out as the dominant [subject]; conversely, [B] constituted
the smallest minority.
Process [Opening] this is a [multi-staged / continuous cyclical] process comprising [X]
distinct phases, commencing with [step 1] and culminating in [final step].
Map [Opening] the area underwent [a comprehensive modernisation / extensive
residential development], characterised by [the replacement of open land with
built infrastructure].
Large table —
across the board
[Opening] while [A] dominated across the board, [B] consistently registered the
lowest figures on almost every metric.
Massive gaps [Opening] the data highlights a striking disparity between [group A] and [group
B].
Erratic lines [Opening] the period was characterised by significant volatility across all
categories, though [A] ultimately emerged as the highest.
Paragraphs 3 and 4 — the body
GROUPING STRATEGY — NEVER LIST MECHANICALLY
CHART TYPE BODY A BODY B
Dynamic Categories that rose Categories that fell or fluctuated
Static The highest figures The lowest figures
Maps The past layout The modern additions
Processes First half of the sequence Second half to completion
19

THE COMMONEST TABLE TRAP
When every category moves in the same direction, direction cannot divide your paragraphs. Group by 
magnitude instead — the large figures together, the small figures together — or by rate of change: the
sharp movers versus the near-static. A report that says "all categories rose" and then lists them in table
order scores poorly on Coherence even when every number is correct.
PREPOSITIONS OF DATA — THE GRAMMAR TRAP
AT — a starting point or a
peak
The percentage began at exactly 20%. · It peaked at 60%.
TO — the final destinationThe figure climbed steadily, rising to 50%.
BY — the margin of changeUsage dropped by 15%. (from 40% to 25%)
FROM… TO… — full span It fell from 40% to 25%.
WITH — appending a figureGermany led the sector, with a rate of 60%.
OF — after a noun of
change
an increase of 15 percentage points
IN — after a noun, naming
the thing
a sharp rise in unemployment
THE DEGREE-OF-CHANGE MATRIX
DEGREE ADJECTIVE + NOUN VERB + ADVERB
Sharp / large a steep / pronounced drop fell steeply / markedly
Clear / substantiala significant / substantial riseincreased significantly / considerably
Slow / constant a steady / incremental growthgrew steadily / incrementally
Minimal a marginal / modest declinedeclined marginally / slightly
Alternate the two columns across your body paragraphs. Using only the verb form for eight sentences is a
Grammatical Range ceiling.
FLUCTUATION AND STABILITY
Fluctuation: significant volatility · fluctuated markedly · oscillated between X and Y
Plateau after movement: levelled off · reached a plateau · stabilised
Flat from the outset: held steady · remained unchanged · maintained extended stability
• 
• 
• 
20

STATIC CHARTS — COMPOSITION VERBS, NOT TREND VERBS
No time axis means no growth. Trend verbs are ungrammatical against the data.
accounted for — Coal accounted for the largest proportion.
comprised — Female students comprised exactly half the cohort.
made up — These two sectors made up 80% of total revenue.
represented — Rail travel represented a negligible share.
constituted — Manufacturing constituted the smallest category.
COMPARISON IDIOMS
Massive difference X exceeded Y by a significant margin.
Near-identical There was a negligible difference between X and Y.
Avoiding repetition of the
noun
The figure for X was double that of Y.
Absolute contrast In stark contrast to the steady growth of solar, coal declined
sharply.
Matching subjects to
numbers
France and Spain reported the highest figures, at 30% and 40% 
respectively.
Two items, second mentionBoth fell, but the former dropped to 15%, whereas the latter fell
to 5%.
FRACTIONS AND PROPORTIONS
Never use amount, number, or parts for percentages. Use proportion, percentage, share, rate.
74–76% just under / just over three-quarters
66% two-thirds
49–51% roughly half · almost exactly half
33% precisely one-third
10% a tenth · one in every ten
4% a negligible minority · a mere fraction
21

TIME EXPRESSIONS
START
At the onset of the period, · In the initial year under review,
MIDDLE
Over the subsequent years, · During the following decade, · Midway through the period,
END
By the close of the period, · In the final year, · By the end of the timeframe,
CONTINUOUS
Throughout the entire timeframe, · Consistently across the period,
Maps and processes — the voice rule
TASK VOICE AND TENSE VERB BANK
Maps, past
→ present
Past passive for the original
state; present perfect passive
for the change
erected · constructed · demolished · expanded ·
repurposed · relocated · pedestrianised · reforested
· converted into
Maps,
decline
Past passive / past simplefell into disrepair · underwent de-industrialisation ·
was left undeveloped
Man-made
process
Present passive — humans
operate the machines
the raw materials are extracted · the mixture is
heated · the product is packaged · is transferred
to · undergoes a filtration stage
Natural
process
Present active — nature does
the work
the caterpillar emerges · hatches · matures · water
evaporates and later condenses
PROCESS SEQUENCERS
Initially, · At the first stage, · Subsequently, · Following this, · Concurrently, · Once this has occurred,
· In the final stage, · culminating in
The boardroom rule — strict red flags
If it sounds like a disaster film or a politician, it is wrong. It must sound like an engineer reading a factual
report.
22

TRAP PROBLEM REPLACEMENT
dramatic, striking,
radical, unmistakable
Emotional and theatrical.
Data does not feel.
sharp, steep, marked, prominent, notable,
extensive
collapse, surge,
overhaul, plummet
(overused)
Violent and catastrophicsteep drop, rapid upward trend,
comprehensive modernisation, virtual
elimination
the data witnessed · the
chart experienced
Personification — paper
is not alive
the data reveals / indicates. Note: years
and places can witness — "the decade
witnessed a shift."
making electricity, using
energy, old people
Register too low power generation, energy consumption,
senior citizens
The zero-speculation rule
Task 1 never explains causation. If you must acknowledge an exception, state it as a mathematical fact.
SPECULATION — PENALISED
Egypt increased, possibly because of new
government investment in tourism.
FACTUAL EXCEPTION
With the notable exception of Egypt, the
decline was universal.
The inverted conditional is available for the same purpose and demonstrates range: "Had it not been for
Egypt's anomalous increase, the decline would have been entirely universal." Use it once at most.
Macro and micro lexicon
Macro terms belong in the introduction and overview; micro terms belong in the body.
MONEY, ECONOMY, WORK — MACRO
financial expenditure · revenue · the general
populace · the workforce
MICRO
allocated budget · affluent demographics ·
joblessness rates · white-collar professions ·
disposable income
ABSTRACT ACTION AND TIME — MACRO
utilisation · production output · engagement
levels · footfall
MICRO
the sheer volume of · the aggregate of · on an
annual basis · globally
23

ENVIRONMENT, MAPS, AGRICULTURE —
MACRO
environmental impact · structural
modifications · agricultural cultivation ·
commuting patterns
MICRO
carbon footprint · waste disposal ·
demolished / erected · livestock rearing ·
vehicular traffic · mass transit systems
SUPERLATIVES AND RELATIONSHIPS
reached its zenith · bottomed out · hit a trough
· the preeminent category
a direct correlation · an inverse relationship ·
a mirrored trend · to intersect and
subsequently surpass
Model report — line graph, dynamic data
PROMPT: PERCENTAGE OF HOUSEHOLDS WITH INTERNET ACCESS IN THREE COUNTRIES, 2000–
2020
The line graph illustrates how the proportion of households with internet access
changed in three countries over a two-decade period commencing in 2000.
Overall, all three nations recorded substantial growth in connectivity, although the
pace and the point of departure differed considerably. The country that began with the
lowest level of access ultimately narrowed the gap, while the initial leader saw its
advantage erode as its figures approached saturation.
At the onset of the period, Country A stood well ahead, at approximately 45%,
compared with 20% in Country B. Access in Country A then climbed steadily, reaching
80% by 2010, before levelling off and finishing the period at just over 90%. Growth in
this case was front-loaded, with the steepest gains concentrated in the first decade.
Country B followed a markedly different trajectory. Having started from a considerably
lower base, it rose incrementally until 2010 and then increased by more than 40
percentage points over the following decade, closing at around 85%. Country C, by
contrast, remained the least connected throughout, and despite consistent expansion its
final figure of roughly 60% left it well behind the other two.
24

TASK 1 FINAL CHECKLIST — 90 SECONDS
Is there a separate overview paragraph with no numbers in it?
Does the introduction paraphrase rather than copy?
Are the body paragraphs grouped, not listed in source order?
Prepositions: at a point, to a destination, by a margin.
Any speculation about causes? Delete it.
Any personification of the chart? Replace with reveals / indicates.
Word count on screen ≥ 160.
1. 
2. 
3. 
4. 
5. 
6. 
7. 
25

PART IV
Writing Task 2 — the essay
Forty minutes. Minimum 250 words; real target 270–290. Two thirds of the Writing
band. Task 2 is lost by silence on part of the prompt and by falling short of the word
count — almost never by vocabulary.
The eight-minute planning rule
Planning is not lost writing time; it is the mechanism that prevents mid-sentence decision-making. Writing
without a map produces the draft–pause–panic–delete cycle, and that cycle is where the Level 0 error lives.
Five to eight minutes of planning, then thirty minutes of uninterrupted production, then the reverse-order pass.
PLANNING PRINCIPLE
Choose the easiest defensible position, not your genuine view. The examiner marks the clarity of the
English, not the sophistication of the ethics. A nuanced position you half-believe produces hedged,
tangled sentences. A simple position you can defend cleanly produces controlled ones.
Technique 1 — the 5W + H extraction
This replaces open brainstorming. It is a mechanism for fixing the subject and the tense of each sentence
before writing it, which is precisely where agreement and tense errors are generated.
QUESTION WHAT IT PRODUCES
Who The agent or the affected group — fixes the grammatical subject and therefore the
agreement
What The specific action, policy or phenomenon, stated concretely
Why The mechanism or motivation — the best source of causal language (because, given
that, insofar as, since)
How The means by which the effect occurs
When /
Where
Scopes a claim or anchors a concrete example
THE PAIRING RULE — TWO OR THREE PER SENTENCE, NEVER ALL SIX
Topic sentence: usually Who + Why or What + Why.
Example sentence: usually Where + When — this is what satisfies the Task Response demand for
specific evidence.
• 
• 
26

Answering all six turns a sentence into an unreadable list. Picking two forces a choice about which
relationship carries the argumentative weight.
Technique 2 — vocabulary and syntax first
Pre-load the lexical chunks and the grammatical targets into the outline, so the mechanics are locked before
drafting begins.
Target collocations: plan the exact noun–verb or adjective–noun pairs. This is the single most effective
guard against literal translation from Spanish mid-sentence, and it locks the prepositions in advance (the
hope of creating, deal with, different from).
Pre-assigned connectors: ban the weak set (On top of that, Evenmore, Besides that). Assign the formal
set in the plan: Furthermore, Consequently, In addition, By contrast, Admittedly.
Advanced syntax markers: plan exactly one or two high-level structures per essay. More than two and
the essay reads as constructed rather than controlled, and each one is an additional agreement risk.
• 
• 
• 
27

The seven structural frames
TYPE EXTRACTION THESIS SHAPE TARGET
SYNTAX
BODY
ARCHITECTURE
1 · Opinion —
full agreement
What + Why Flat stance, no
hedging
Negative
inversion
Reason 1 (Why +
example), Reason 2
(Why + example)
2 · Opinion —
"to what
extent"
What + Why,
plus one
reservation
Scaled stance: 
largely, to a
considerable extent
Concession
then assertion
Main supporting
argument; then
concede the
counterpoint and
rebut it
3 · Discussion
— both views
Who + Why,
per view
Both views named;
verdict withheld
until the conclusion
Parallel
concession,
no evaluative
adjectives
View A (Who + Why),
View B (Who + Why),
in separate
paragraphs of equal
grammatical weight
4 · Advantages /
disadvantages
What + How,
per side
Neutral framing of
the phenomenon
Participle
clause
Mechanism of benefit;
mechanism of harm
5 · Problem /
solution —
causes
Why + how it
manifests
States the problem
plainly, no dramatic
framing
Absent +
noun phrase
Cause explained as a
mechanism, then a
concrete example
6 · Problem /
solution —
solutions
How + who
implements
Institutional
recommendation
Advanced
passive with
modal
Measure proposed,
then its expected
logical effect
7 · Two-part /
direct
questions
What + How,
per question
Names both
questions in one
fluid sentence
Cleft sentenceParagraph 1 answers
Q1, paragraph 2
answers Q2 — no
adversative
connectors between
them
STRUCTURAL TRAP
In a discussion essay, fusing both views into one body paragraph costs marks under Coherence and
Cohesion even when the content is perfectly balanced. Each view gets its own paragraph. There is
also an eighth prompt shape that appears periodically: a three-part prompt combining a reason with
advantages and disadvantages. Treat it as three body paragraphs, not two, and shorten the introduction
accordingly.
28

TARGET SYNTAX, WORKED
NEGATIVE INVERSION — OPINION ESSAYS
Not only does mandatory physical education foster lifelong healthy habits, but it also
significantly reduces the long-term burden on public healthcare systems.
ADVANCED CONCESSION — DISCUSSION AND PARTIAL-AGREEMENT ESSAYS
While proponents of rapid urbanisation argue that cities drive economic innovation, it
is equally evident that this expansion exacts a severe toll on local ecosystems.
THIRD CONDITIONAL INVERSION — PROBLEM/SOLUTION
Had stricter environmental regulations been enforced earlier, the current crisis of urban
air toxicity might have been largely averted.
PARTICIPLE CLAUSE — ADVANTAGE/DISADVANTAGE
The shift towards remote work decentralised urban populations, precipitating a decline
in commercial property values across city centres.
ABSENT + NOUN PHRASE — CAUSES
Absent a comprehensive public transit network, commuters default to private vehicles,
and congestion becomes structurally inevitable.
CLEFT SENTENCE — TWO-PART QUESTIONS
It is the prohibitive cost of housing that forces young couples to delay parenthood, not
a change in their underlying preferences.
29

FALLBACK RULE
If the target structure does not arrive cleanly within a few seconds, use a plain sentence and move on. 
A clean simple sentence outscores a broken complex one. Grammatical Range is assessed
alongside Accuracy, not instead of it — a broken inversion loses on both criteria simultaneously.
The four-paragraph skeleton
PARAGRAPH 1 — INTRODUCTION
Paraphrase the prompt. It is often argued that… / There is an ongoing debate over whether…
Thesis. State the position directly and unhedged (or scaled, for "to what extent").
Outline. Name the two ideas. This essay will examine X before turning to Y.
PARAGRAPHS 2 AND 3 — THE "SO WHAT?" METHOD
Topic sentence — the claim, built from your 5W+H pair. It goes first. Not context, not background.
Explanation — why or how the mechanism works. Assume the reader knows nothing. This means
that… / As a result…
Example — specific and generalised, not a personal anecdote.
Link — one short sentence tying it back to the thesis.
WEAK PARAGRAPH OPENING — CONTEXT FIRST
Nowadays, many people around the world use
social media in their daily lives.
STRONG — CLAIM FIRST
The central damage caused by social media is
not the time it consumes but the quality of
attention it destroys.
WEAK
There are many opinions about whether
governments should fund the arts.
STRONG
Public funding for the arts is justified precisely
because the market systematically underprices
what has no immediate commercial return.
PARAGRAPH 4 — CONCLUSION
Paraphrase the thesis and the two ideas. No new information. No maybe. In conclusion, although some may
argue [opposing view], the evidence points clearly towards [thesis], for the reasons of [idea 1] and [idea 2].
WEAK CONCLUSION
So maybe our future is not decided yet.
DECISIVE CONCLUSION
The evidence points in one direction. Unless
governments treat housing as infrastructure
rather than as investment, the crisis will
deepen.
1. 
2. 
3. 
1. 
2. 
3. 
4. 
30

The AEEC paragraph engine
A more granular version of the same shape, useful when a paragraph feels thin.
MOVE FUNCTION FRAMES
Assertion Establish the stance
immediately
From my perspective, I firmly believe that… · It is
undeniably evident that…
ExplanationUnpack the why and the
how
The primary rationale behind this is… · This is
fundamentally driven by the fact that…
Example Ground it in a specific
instance
A compelling illustration of this can be observed in… · For
instance, in the case of…
Conclusion Synthesise and link backConsequently, … · This is precisely why …
Calibrated hedging — the band 8 register
Absolute claims read as unexamined. Calibrated claims read as controlled.
This tends to occur when…
There is a strong tendency to…
In most developed economies,…
The available evidence suggests…
It is likely that…
This arguably reflects…
To a large extent,…
ONE CALIBRATED SENTENCE, FULLY BUILT
While technology has undoubtedly widened access to information, it is far less clear
that it has improved the depth with which that information is processed.
Examples: the rule that decides Task Response
Task Response improves measurably when placeholder examples are replaced with named, specific cases.
But specificity must be real. An invented statistic or a fabricated institution is worse than a general example,
because it is a factual error inside an argument.
31

NEVER REUSE THESE
There is no "University of Australia."
The commonly cited figure for emissions attributable to animal agriculture is roughly 14–15%, not
35%.
Omegle was an anonymous chat platform, not a dating platform, and it no longer operates.
The claim that the vast majority of care-home residents have no children is false.
Write this essay argues, never this essay believes.
Safe substitutes: a named country and a named policy area (Finland's teacher-training model,
Singapore's congestion charging, the Netherlands' cycling infrastructure), or a described mechanism
with no number attached at all. A qualitative claim you can defend beats a quantitative claim you cannot
source.
• 
• 
• 
• 
• 
32

Critical mechanics checklist
FAILS TARGET
people believes people believe
is hard to picture It is hard to picture — never drop the dummy
subject
without worry about it without worrying about it — gerund after every
preposition
in the hope of create in the hope of creating
the platform's are growing the platforms are growing — an apostrophe never
makes a plural
life's are changing lives are changing
Students should to attend. Students should attend — bare infinitive after a
modal
This are the main problems. These are the main problems.
Governments must act, they have the
resources.
Governments must act; they have the resources.
(comma splice)
Because the costs are high. Attach it: The policy has stalled because the
costs are high. (fragment)
don't, can't, it's do not, cannot, it is — no contractions in Task 1
or Task 2
more cheaper cheaper · more expensive — never both markers
THE SPELLING SET THAT RECURS
Environment · excessive · consequences · government · necessary · occurred · beneficial · which · because ·
society · believe · planet · careers · college · furniture · whether · absence.
33

Model essay — discussion type
PROMPT: SOME BELIEVE UNIVERSITY EDUCATION SHOULD BE FREE FOR ALL; OTHERS BELIEVE
STUDENTS SHOULD PAY. DISCUSS BOTH VIEWS AND GIVE YOUR OWN OPINION.
Whether higher education should be funded by the state or by the individuals who
receive it remains one of the more durable disagreements in public policy. Advocates of
free provision treat it as social infrastructure, while their opponents regard it as a
private investment with private returns. This essay examines both positions before
arguing that a means-tested model resolves most of the tension.
Those who defend fully subsidised universities argue primarily from access. Where
tuition is charged, the cost operates as a filter that removes capable students from low-
income households before their ability is ever assessed, and the loss is borne by the
wider economy rather than by the individual alone. Countries that have maintained
free or near-free provision, such as Germany, typically point to broader participation
across income bands as the return on that expenditure. On this account, education 
generates benefits that no single graduate captures.
The opposing camp begins from the distribution of the benefit. Graduates earn
measurably more across a working life, and funding their training from general taxation
asks non-graduates, who earn less on average, to subsidise a group that will out-earn
them. Not only does this arrangement strain public budgets, but it also transfers
resources upwards. Critics further note that where places are free but limited,
competition for them tends to favour applicants who arrive with the strongest
preparation, which is rarely the poorest.
In my view, the disagreement dissolves once the funding model is separated from the
access question. A system in which fees scale with family income, combined with
repayment contingent on later earnings, preserves the public interest in wide
participation without asking a non-graduate population to underwrite it. Absent that
calibration, both models fail: one on cost, the other on fairness.
34

TASK 2 FINAL CHECKLIST — 3 MINUTES
Word counter ≥ 270. Check the screen, never estimate.
Has every part of the prompt been answered? A discussion essay that omits your own opinion caps
at band 6 on Task Response.
Reverse-order pass: last sentence first, backwards.
Agreement — check every abstract singular subject and every intervening plural.
Tense — the spine is present simple. Past simple only for a single dated fact; present perfect only
for a completed change that still matters.
Comma splices and fragments.
Contractions, apostrophes, this/these, bare infinitive after modals.
1. 
2. 
3. 
4. 
5. 
6. 
7. 
35

PART V
Speaking — all three parts
Eleven to fourteen minutes, face to face. For most candidates who have reached band
6.5, fluency and lexical range are not what holds the score down — grammatical
accuracy under real-time pressure is the ceiling. Every technique below exists to
protect accuracy without costing fluency.
PART TIME WHAT IS BEING TESTED CALIBRATION
1 · Interview4–5
min
Baseline fluency on
familiar topics
Answer, reason, one detail. Roughly 27–39
words. Two to three sentences.
2 · Long
turn
3–4
min
Sustained monologue
without prompting
1 minute preparation, then speak for 1–2
minutes continuously.
3 ·
Discussion
4–5
min
Abstract and analytical
reasoning
Expansive, impersonal, comparative. Four to
six sentences.
THE CALIBRATION PROBLEM, NAMED
Two opposite habits appear in recordings: clipping — stopping after one sentence, which starves
Lexical Resource of anything to score — and drifting — continuing past the natural end of the answer,
which is precisely where tense collapse enters. The target is not elaborate. It is a short, specific,
factually grounded answer extended slightly further than feels necessary, then stopped deliberately.
Part 1 — the echo technique
Match the auxiliary or modal in the examiner's question in your first clause. This forces immediate tense
alignment and buys the half-second needed to build the rest of the sentence.
36

QUESTION FORM ECHO RESPONSE WORKED EXAMPLE
"Have you ever…?" —
present perfect
Actually, I have… / I
haven't, but…
Actually, I have. I've been trying to get back
into it over the last few months.
"Would you…?" —
second conditional
I certainly would,
provided that…
I certainly would, especially if it meant more
time outdoors.
"Do you usually…?" —
present simple
I typically do,
especially when…
I typically do. I usually spend the evening
reading rather than watching anything.
"Were you…?" — past
simple
I really was, mainly
because…
I really was, primarily because the project
required far more preparation than expected.
"Did you…?" — past
simple
I did, actually,
though…
I did, actually, though not as often as I would
have liked.
"Are you…?" — present
continuous / state
I am, to a degree…I am, to a degree. It depends on how
demanding the week has been.
HIGHEST-RISK STEM
"Would" is the stem that most reliably produces an error, because Spanish permits will-type forms
where English demands would. I will prefer → I would prefer. I will love to learn Nahuatl properly → I
would love to. Drill this stem separately from the others.
OPENERS — NEVER BEGIN WITH A BARE YES OR NO
HONEST, NUANCED OPINION
To be perfectly candid,… · Truth be told,… · I'd have to say…
PREFERENCE
I'm more of the type of person who… · I'd definitely say that…
CONDITIONAL
It largely depends on the day, but generally… · That varies quite a bit, though on the whole…
THINKING TIME — COVERS A PAUSE OVER TWO SECONDS
That's an interesting question, actually. · I haven't thought about that in a while. · Let me think about
that for a second.
37

PART 1 CALIBRATION, DEMONSTRATED
CLIPPED — TOO SHORT TO SCORE
"Do you enjoy cooking?" — Not really, no.
TARGET LENGTH
Not especially, to be honest. I tend to cook the
same three or four things on rotation, mostly
because I get home late and I'd rather spend
the energy elsewhere. Weekends are the
exception — that's when I'll actually try
something new.
DRIFTING — STOP MARKED
…that's when I'll actually try something new. 
[stop here] And my mother used to cook a lot
when I was young, she makes this dish that
takes all day, and I remember…
WHY IT COSTS MARKS
The drift introduces a past narrative with no
time anchor, and makes should be made.
Tense collapse enters through over-extension,
not through the original answer.
Part 2 — the long turn
THE PREPARATION MINUTE — FOUR FRAGMENTS, NOT SENTENCES
Fix the tense first. Most cue cards are past narratives. Write the time anchor at the top of the note paper:
2019 · residency. That single anchor is the defence against collapse.
Four keyword fragments only, one per beat below. Never full sentences — a written sentence invites
reading aloud, which flattens intonation and reads as rehearsed.
Decide the closing line in advance. It moves to the present tense deliberately.
THE FOUR-BEAT STRUCTURE
BEAT JOB APPROX. TIME
Set the scene When, where, who — establishes the past frame20 s
The story What happened, in sequence 50 s
The feeling Reaction, in the past 20 s
The close Why it still matters — authorised shift to present15 s
If a beat finishes early, add one more detail rather than rushing to the next. If time runs over, cut the close —
being stopped at two minutes is not penalised.
1. 
2. 
3. 
38

PAST-ANCHORED EMOTIONAL LANGUAGE
This is where tense collapses, so the phrases themselves are pre-anchored in the past.
I was absolutely thrilled.
I was completely taken aback.
I felt a huge sense of relief.
I was on the verge of giving up.
It was nerve-wracking, honestly.
I could not believe my eyes.
It meant the world to me.
I was torn between two options.
A wave of relief washed over me.
It caught me off guard.
THE LANGUAGE OF SIGNIFICANCE — FOR THE CLOSE
It left a lasting impression on me.
It shaped the way I work to this day.
It taught me something no textbook ever could.
It put things into perspective.
If that had not happened, I doubt I would be where I am now. (mixed conditional — high value, drill it)
DESCRIBING PEOPLE AND PLACES
PEOPLE
She was the kind of person who…
What set him apart was…
She had a gift for putting people at ease.
He was demanding but fair.
I owe her a great deal.
PLACES
It's a bit off the beaten track.
The place has a charm of its own.
It was breathtaking, in the most literal sense.
What struck me first was the silence.
It felt like stepping into another century.
39

We hit it off straight away.
She always went the extra mile.
I could picture myself living there.
PART 2 TRAPS
Forced idioms. One natural idiom raises the score; three forced ones lower it. If you have never
said it out loud before, do not debut it in the exam.
"In a nutshell" used as an opener. It closes; it does not open.
Filler inflation.You know, like, I mean more than twice a minute reads as a fluency gap dressed as
style.
Reading the notes. Glance, don't read. Notes are anchors, not a script.
Part 3 — the discussion
The register shifts entirely. Personal anecdote is replaced by societal, institutional and comparative reasoning.
Professional or specialist knowledge is a real asset here, but with the first person removed: not "at my
workplace I see…" but "in most organisations the pattern tends to be…". The knowledge is deployed; the
pronoun is not. This single move is what turns a Part 1-style answer into a Part 3-level one.
THE FOUR-MOVE SPINE — ONE FRAMEWORK, USED EVERY TIME
Position — a calibrated stance, not a flat yes.
Mechanism — why or how, stated as a causal process.
Illustration — a named, general, real case, or a described institutional pattern.
Qualification — the limit of the claim. This is the move that separates band 7 from band 8, because it
demonstrates conditional and concessive grammar under pressure.
LANGUAGE BY FUNCTION
STATING A CALIBRATED POSITION
I would argue that…
My sense is that…
Broadly speaking, I would say…
I'm somewhat sceptical of the idea that…
There's a strong case for saying that…
I'm in two minds about this, actually.
CAUSES
This largely comes down to…
One of the main drivers is…
It stems from…
The root of the problem lies in…
This is partly a consequence of…
Several factors feed into this.
• 
• 
• 
• 
1. 
2. 
3. 
4. 
40

EFFECTS
This has far-reaching consequences.
The knock-on effect is that…
In the long run, this leads to…
The ripple effects are felt across the whole
system.
One unintended consequence is that…
COMPARISON BETWEEN GROUPS
Older generations tend to…, whereas
younger people are more inclined to…
There's a growing divide between…
The gap between the two has widened
considerably.
What used to be exceptional has become
routine.
QUESTION ARCHETYPES AND THE MOVE THAT ANSWERS THEM
ARCHETYPE EXAMPLE STEM OPENING MOVE
Cause Why do you think…? This largely comes down to… + mechanism
Effect /
prediction
How will this change in the
future?
It's difficult to say with any confidence, but the
trajectory suggests…
Comparison Is it different from the
past?
Markedly so, in one specific respect: …
Evaluation Is that a positive
development?
On balance yes, though with one significant
caveat…
Responsibility Who should be
responsible?
Responsibility is probably shared, but the weight
falls on… because…
Hypothetical What would happen if…?If X were to happen, Y would probably follow —
second conditional, drilled
41

EMERGENCY PROTOCOLS
WHEN THE QUESTION IS STRANGE, ABSTRACT OR UNCLEAR
Ask for a rephrase — this is permitted and costs nothing: "Sorry, could you rephrase that?" /
"Do you mean in terms of X, or more generally?" Misreading a Part 3 question is far more expensive
than asking.
Bridge to solid ground: "That's not something I've considered directly, but if I think about the
related question of…"
Narrow the scope openly: "I can only really speak to the [education / healthcare] side of that, but
there…"
Admit the limit, then continue: "I genuinely don't know, though my instinct would be that…" — an
honest hedge in fluent English scores; a fabricated fact in halting English does not.
Advanced structures that demonstrate range
STRUCTURE STANDARD ADVANCED
Inversion after a
negative adverbial
I had rarely seen such a well-
designed venue.
Rarely had I seen such a well-
designed venue.
Inversion with only
when
I understood the complexity only
when I started.
Only when I started did I
understand the complexity.
Third conditional
inversion
If I had known the budget was
limited, I would have planned
differently.
Had I known the budget was
limited, I would have planned
differently.
Present participle
clause
Because she was exhausted, she
left early.
Feeling exhausted, she left early.
Past participle clauseThe proposal, which was rejected by
the committee, needs revision.
Rejected by the committee, the
proposal needs revision.
Perfect participle After I had finished the report, I
reviewed the data.
Having finished the report, I
reviewed the data.
Fluency mechanics
The silent reset. Eliminate mid-sentence restarts. Half a second of silence, reset, deliver the corrected
phrase cleanly. Silence reads as thought; stumbling reads as a gap.
Never start a standalone sentence with "Because." Link it: primarily because, owing to, since, given
that.
• 
• 
• 
• 
• 
• 
42

Enunciate word endings, particularly regular -ed suffixes, plurals, and the third-person -s. The -s is the
single most visible marker to an examiner: one letter, disproportionate weight.
Momentum over correction. If a minor error escapes, do not derail the narrative to fix it. Push forward.
Measured pace. Rushing generates both grammatical errors and consonant blurring. Pause naturally to
gather thought.
SELF-RECORDING PROTOCOL — THE HIGHEST-YIELD PRACTICE METHOD
Record a full answer without stopping.
Transcribe it yourself, by hand or by typing. Do not use automatic transcription.
Mark your own errors on the transcript before looking at any correction.
Re-record the same answer once.
Self-transcription outperforms plain repetition because it forces attention onto surface form — exactly
the channel that fails under load. Three cycles of this beats thirty minutes of unrecorded speaking.
• 
• 
• 
1. 
2. 
3. 
4. 
43

PART VI
Listening
Comprehension is not the bottleneck. The bottleneck is connected speech and
deliberate distractor design — the words are known, but they never sound the way they
do in isolation.
PART CONTEXT SPEAKERS TYPICAL QUESTION TYPES
1 Everyday transactional —
booking, enquiry
Two Form completion, note completion
2 Everyday monologue — a tour,
a briefing
One Map or plan labelling, matching, multiple
choice
3 Academic discussion — tutorial,
project
Two to
four
Multiple choice, matching, flow-chart
completion
4 Academic monologue — a
lecture
One Note completion, summary completion,
sentence completion
Difficulty rises across the four parts. Part 4 has no mid-section break, so the attention budget must be
managed deliberately.
Connected speech — the real difficulty
Native speech collapses word boundaries. Recognising these four patterns converts "I didn't hear it" into "I
heard it and now I know what it was."
PATTERN WHAT HAPPENS HEARD AS
Linking A final consonant attaches to a
following vowel
an apple → "a napple" · pick it up → "pi ki tup"
Elision A consonant between two others
disappears
next day → "nexday" · most common → "mos
common"
AssimilationA sound changes to match its
neighbour
ten minutes → "tem minutes" · did you →
"dijoo"
Weak formsFunction words reduce to a schwaof, to, and, for, can, was, from → "uv, tuh, un,
fuh, kn, wuz, frm"
44

DRILL FOR THIS SPECIFICALLY
Blind dictation. Take twenty seconds of natural-speed audio, transcribe it word for word without the
transcript, then compare. Focus only on rhythm and boundaries — where one word ends and the next
begins. This is the drill that moves Listening; more listening practice without transcription does not.
The distractor taxonomy
DISTRACTOR HOW IT SOUNDS DEFENCE
Speaker self-
correction
"It's at three — no, sorry, four
o'clock."
The second figure is the answer. Listen for 
actually, sorry, I mean, no wait, in fact — these
words are answer markers.
Negated option"We used to offer that on
Tuesdays, but not any more."
The mentioned option is the wrong one.
Negation carries the answer.
All options
mentioned
Every multiple-choice option
appears in the audio
Presence is not evidence. Listen for the
evaluative language attached to each.
Paraphrase The question says expensive;
the audio says beyond most
students' budget
Predict paraphrases during the preparation
window, not during the audio.
Spelling out Names and addresses are
spelled letter by letter
Write letters as you hear them. Double letters
are said as "double L," not "L L."
Mechanics that cost marks unnecessarily
Word-limit compliance. "NO MORE THAN TWO WORDS" means two. Three words is marked wrong
even when the meaning is perfect. A hyphenated word counts as one; a number counts as one word.
Spelling and grammar are marked. A correctly heard answer spelled wrong scores zero. Plural -s is part
of the answer: if the sentence requires a plural, write it.
Never leave a blank. There is no penalty for a wrong answer. Guess with something plausible in the right
word class.
Read ahead during the preparation windows. Underline the question-stem keywords and predict the
word class of each gap — noun, number, verb, adjective. Knowing you need a number before it arrives is
most of the work.
Never chase a missed answer. Losing one question while hunting for it costs the next two. Mark it, move
on, return in the check window.
• 
• 
• 
• 
• 
45

COMPUTER-DELIVERED NOTE
There is no ten-minute transfer window. Type each answer as it arrives and use the short review time at
the end for the questions you flagged — not for a full re-read. Capitalisation is not penalised, so type in
whichever case is fastest.
46

PART VII
Reading
Three passages, forty questions, sixty minutes, no extra transfer time. Where reading
speed and comprehension are already strong, the entire remaining risk sits in question
mechanics. The trap is in the question, not the passage.
The approach
Ninety seconds of orientation. Title, first sentence of each paragraph, any subheadings. Do not read the
passage in full first.
Go to the questions. Underline proper nouns, dates, numbers, and unusual or technical words — these
are the scannable anchors.
Scan for the anchors, then read the two sentences surrounding each one carefully. The answer is almost
always within that window.
Questions generally follow passage order, with the notable exceptions of matching headings, matching
features, and global summary tasks. Use that ordering to bound your search.
True / False / Not Given — the forensic label system
Rather than deciding intuitively, name the reason. Assigning an explicit label converts a feeling into a check,
and it is the single most effective intervention available for this question type.
LABEL WHAT IT NAMES VERDICT IT
PRODUCES
Absoluto The statement inserts an absolute — always, never, all, only, every
— where the passage was moderate (often, tends to, in most
cases)
FALSE
Invertido Causation or direction is reversed — the passage says A causes B,
the statement says B causes A
FALSE
Parcial Half the statement matches the passage and half is fabricated or
altered
FALSE
InferenciaThe statement is a reasonable conclusion the passage never
actually states
NOT GIVEN
Ausente The passage simply does not address the topic at all NOT GIVEN
1. 
2. 
3. 
4. 
47

THE SINGLE MOST EXPENSIVE READING ERROR
Marking FALSE when the answer is NOT GIVEN. If the passage does not address the point, it is Not
Given — even when the statement seems obviously untrue in the world, and even when your own
knowledge contradicts it. FALSE requires a locatable contradiction in the text. If you cannot point at the
contradicting words, it is not FALSE.
The mirror error also exists: marking NOT GIVEN when the contradiction is present but paraphrased.
Before settling on Not Given, spend one more scan looking for a synonym of the key term.
Matching headings
The wrong heading often contains a word copied verbatim from the paragraph. That is the bait. The
correct heading paraphrases the main idea and usually shares no distinctive vocabulary with the
paragraph at all.
Read the first and last sentences of the paragraph, then ask: what is this paragraph doing — describing a
problem, giving an example, presenting a counterargument, reporting a result?
Do the easy paragraphs first and cross off used headings. The list is deliberately longer than the number
of paragraphs.
A heading that fits one sentence of the paragraph is wrong. It must fit the whole.
Other question types
TYPE GOVERNING RULE
Sentence / summary
completion
Words must come from the passage exactly as written. Predict the word
class before scanning. Respect the word limit absolutely.
Multiple choice Eliminate rather than select. Three options are wrong for a locatable
reason — too extreme, partially true, or true but not stated here.
Matching features
(names to statements)
Not in passage order. Scan for the proper nouns first and mark them all,
then work through the statements.
Yes / No / Not Given Same mechanics as TFNG, but the object is the writer's opinion or
claim, not a fact. Look for evaluative language, modals, and reporting
verbs.
Diagram / flow-chart
labelling
Usually drawn from a single dense section of the passage. Find that
section first, then fill everything from it.
Timing and mechanics
Roughly twenty minutes per passage, but Passage 3 is hardest — take eighteen, twenty, twenty-two.
• 
• 
• 
• 
• 
48

Never leave a blank. No penalty for a wrong answer.
Word limits are absolute: three words where two were permitted scores zero regardless of meaning.
If comprehension is fast, the spare minutes belong to the TFNG items, not to re-reading passages you
already understood.
• 
• 
• 
49

PART VIII
Grammar priorities
Only the items that actually fail under load. Everything else is already secure and does
not need review.
The tense system, reduced to what is needed
CONTEXT SPINE TENSE PERMITTED DEPARTURES
Task 2 essay Present simple Past simple for a single dated historical fact. Present
perfect for a completed change that still matters. Nothing
else.
Task 1, dated
chart
Past simple Past perfect only to mark one event as prior to another.
Present simple for the sentence describing the chart itself.
Task 1, undated
static chart
Present simple None.
Task 1, processPresent passive Present active for natural processes.
Task 1, map Past passive for
the earlier state
Present perfect passive for the change; present simple for
the current state.
Speaking Part 1Present simple Present perfect for experience; past simple with a time
anchor.
Speaking Part 2Past simple Past continuous for background; one authorised shift to
present in the closing line.
Speaking Part 3Present simple Second conditional for hypotheticals; will / is likely to for
predictions.
ONE LEGITIMATE EXCEPTION WORTH KNOWING
A permanent trait inside a past narrative stays in the present: "I saw him yesterday, and he is always so
stubborn." This is correct, not a collapse. Knowing which shifts are legal prevents over-correction, which
is its own fluency cost.
50

VERB LADDERS — THE FORMS THAT COLLAPSE
go goes · went · gone · going
take takes · took · taken · taking
begin begins · began · begun · beginning
choose chooses · chose · chosen · choosing
lead leads · led · led · leading
arise arises · arose · arisen · arising
seek seeks · sought · sought · seeking
bear (a cost) bears · bore · borne · bearing
strive strives · strove · striven · striving
undergo undergoes · underwent · undergone · undergoing
False regularisation is the specific failure here: choosed, leaded, bringed, teached, catched, thinked. Say
the correct past form aloud; the eye will not catch it.
51

Prepositions — Spanish transfer
TRANSFER ERROR TARGET
different from what we expected different from what we expected
I arrived at the clinic I arrived at the clinic (a building) · in the city (a
place)
married to a colleague married to a colleague
depend on public funding depend on public funding
responsible for the department responsible for the department
discuss about the issue discuss the issue — no preposition
consist in three parts consist of three parts
based in the evidence based on the evidence
interested on the topic interested in the topic
an increase from 5% an increase of 5% (magnitude) · from 5% (starting
point)
in the last years in recent years · over the past few years
according to my opinion in my opinion · to my mind
Would versus will
English uses would for hypothetical preference and polite distance where Spanish permits a future form.
FAILS TARGET
I will prefer to move abroad. I would prefer to move abroad.
I will love to learn the language properly.I would love to learn the language properly.
I will recommend that if things were different.I would recommend that if things were different.
If I would have more time, I would travel.If I had more time, I would travel. — would never
appears in the if-clause
52

HABITUAL PAST — THREE FORMS, DISTINCT MEANINGS
used to + infinitive — a past state or habit, now ended: I used to work night shifts.
would + infinitive — a repeated past action (not for states): Every winter we would drive north.
Never I would be a student.
be used to + -ing — accustomed to, any tense: I am used to working under pressure.
False friends
actually ≠ actualmente → currently
eventually ≠ eventualmente → possibly / occasionally
realise ≠ realizar → carry out / perform
assist ≠ asistir → attend
career ≠ carrera (degree) → degree / course
success ≠ suceso → event
discuss ≠ discutir (to quarrel) → argue
sensible = sensato; for the emotional sense → sensitive
large ≠ largo → long
contest ≠ contestar → answer / reply
idiom ≠ idioma → language
support ≠ soportar (to endure) → tolerate / put up with
pretend ≠ pretender → intend / aim to
constipated ≠ constipado → to have a cold
53

Punctuation and sentence boundaries
FAILS TARGET
Governments must act, they have the
resources.
Governments must act; they have the resources. ·
…act, and they have…
However. The situation is more complex.However, the situation is more complex.
The policy failed; because funding was cut.The policy failed because funding was cut. — a
semicolon needs an independent clause after it
The report, shows a clear trend. The report shows a clear trend. — never one
comma between subject and verb
Which is why the system collapsed. Attach it: Funding was withdrawn, which is why
the system collapsed.
Parallelism
BROKEN
The policy aims to reduce costs, improving
access, and it will raise standards.
PARALLEL
The policy aims to reduce costs, improve
access, and raise standards.
Articles and quantifiers
a / an follows the sound, not the letter: a university, an hour, an MRI.
The for a shared or previously specified referent; no article for a plural generalisation. Students benefit
from feedback (in general) vs the students in this cohort.
amount for uncountable, number for countable. The amount of information · the number of students.
fewer for countable, less for uncountable.
this / these — a persistent slip. This is singular, these is plural, and there is no exception.
• 
• 
• 
• 
• 
54

THE THREE-MINUTE WRITTEN CHECK — RUN IT IN THIS EXACT ORDER
Agreement — every abstract singular subject (evidence, research, data, trust, the number of) and
every intervening plural.
Tense — did it move without the prompt or the chart moving it?
The small set — this/these · bare infinitive after a modal · a/an · amount vs number · comparatives
(more expensive, never more cheaper).
Prepositions — different from · arrive at/in · based on · depend on · consist of · discuss (no about).
False friends.
Fixed collocations.
The order matters. It runs from the highest-frequency, highest-visibility errors down to the cosmetic. If
time runs out at step three, the expensive errors are already caught.
1. 
2. 
3. 
4. 
5. 
6. 
55

PART IX
Lexis banks
Organised by topic domain, because retrieval is cued by topic under exam conditions,
not by alphabetical or grammatical category. Produce these aloud or in sentences;
reading them silently builds recognition, which is not what the exam tests.
1 · EDUCATION
rote learning · critical enquiry · a well-rounded
curriculum
vocational training vs academic pathways
widening participation · attainment gap
tuition fees · student debt burden
to instil discipline · to foster independent
thought
league tables distort what schools prioritise
2 · HEALTH AND MEDICINE
preventive care · primary care · a public
health burden
sedentary lifestyles · dietary intake
waiting times · resource allocation · triage
life expectancy · comorbidity · early
intervention
to place strain on health services
the social determinants of health
3 · TECHNOLOGY
pervasive connectivity · digital literacy · the
digital divide
automation displacing routine roles
data privacy · algorithmic curation
screen time · attention fragmentation
to streamline a process · to render a skill
obsolete
4 · ENVIRONMENT
carbon footprint · emissions targets ·
renewable capacity
biodiversity loss · habitat fragmentation
waste disposal · a circular economy
to exact a severe toll on ecosystems
mitigation and adaptation are not the same
problem
5 · WORK AND ECONOMY
job security · the gig economy · remote
arrangements
work–life balance · burnout · staff retention
6 · URBAN LIFE AND TRANSPORT
urban sprawl · population density · mass
transit
congestion charging · pedestrianised zones
56

disposable income · cost of living · wage
stagnation
to enter the labour market · transferable skills
affordable housing · gentrification
commuting patterns · vehicular traffic
7 · FAMILY AND SOCIETY
an ageing population · the dependency ratio
intergenerational contact · extended family
networks
shifting social norms · community cohesion
to delay parenthood · caregiving
responsibilities
8 · CULTURE AND THE ARTS
cultural heritage · intangible heritage ·
preservation
public subsidy · commercial viability
homogenisation of global culture
to broaden one's horizons · a sense of shared
identity
9 · CRIME AND GOVERNANCE
deterrence · rehabilitation · recidivism
custodial sentences vs community service
regulatory frameworks · enforcement capacity
to impose strict penalties · to allocate
resources
10 · TRAVEL AND GLOBALISATION
cultural exchange · mass tourism ·
overtourism
seasonal employment · local displacement
cross-border mobility · remittances
to erode the character of a place
57

Connectors, ranked by register
FUNCTION USE THESE AVOID THESE
Adding Furthermore · In addition · Moreover · What is
more
On top of that · Besides ·
Evenmore
Contrasting By contrast · Conversely · Nevertheless ·
Admittedly
But then again · On the other
side
Causing Consequently · As a result · This in turn · Owing toSo · That's why
Conceding While it is true that · Granted · Admittedly ·
Notwithstanding
Even so I think
ExemplifyingFor instance · A case in point is · As illustrated byLike for example
Concluding On balance · Taken together · The weight of the
evidence suggests
To sum up briefly · In a nutshell
(in writing)
Fixed collocations that resist translation
to place a burden on · to bear the cost · to meet a demand
to raise awareness · to address a problem · to tackle an issue
to draw a distinction · to reach a consensus · to strike a balance
to pose a risk · to pursue a career · to take measures
a vested interest · a compelling case · a pressing concern
in the hope of creating · with a view to improving · for the sake of clarity
deal with · account for · result in vs result from · catch up on
Idioms — the one-per-answer rule (Speaking only, never Writing Task 2)
Natural idiomatic language is explicitly named in the Speaking Lexical Resource descriptors, but the scoring is
asymmetric: one well-placed idiom helps, three forced ones actively hurt. Use only phrases already in active
use. In Writing Task 2, idioms read as too informal for the register the task expects — keep this list for
Speaking Part 2/3 only.
a blessing in disguise · off the beaten track · to go the extra mile
a double-edged sword · to be in two minds · to hit a wall
58

it's not set in stone · the jury is still out on that · to take its toll
a drop in the ocean · to pay off in the long run
59

PART X
The one-page card
Everything above, compressed to what fits in working memory on the morning of the
exam. If there is time for only one page, this is the page.
THE THREE THAT DECIDE THE OUTCOME
Half a second before every third-person verb in Speaking. One letter, disproportionate weight.
The tense is decided by the structure, not by memory. Chart type, cue-card frame, paragraph
function.
Finish the unit, then revise. Never edit mid-sentence, spoken or written.
MODULE THE ONE THING THE ONE TRAP
Listening Read ahead and predict the word class of each
gap
The word limit. Two words means
two.
Reading Name the trap type before choosing an answerFALSE where the answer is NOT
GIVEN
Task 1 A separate overview paragraph with no numbersExplaining why something
happened
Task 2 Word counter ≥ 270, and every part of the prompt
answered
Silence on one part of a two-part
prompt
Speaking
1
Echo the auxiliary; answer, reason, one detailClipping to one sentence
Speaking
2
Write the time anchor before anything elseTense collapse at the emotional
peak
Speaking
3
Position, mechanism, illustration, qualificationSlipping back into personal
anecdote
The eight sentences to have automatic
1. 
2. 
3. 
The evidence suggests otherwise, though the picture is more complicated than it appears.01
The findings of this research support a different conclusion.02
Not only does this approach reduce costs, but it also improves access.03
While there is undeniable merit in that view, I would argue that the opposite holds in most cases.04
60

Banned on sight
It is immediately apparent that · Looking from an overall perspective · dramatic · striking · radical · the data
witnessed · the chart experienced · making electricity · old people · this essay believes · maybe · don't, can't,
it's · In a nutshell (as an opener) · Because at the start of a standalone sentence.
Had the policy been introduced earlier, the outcome would have been substantially different.05
If that were to happen, the consequences would be felt across the whole system.06
I would prefer to approach it differently, primarily because the current model is unsustainable.07
Having worked in that setting for several years, I came to see the problem quite differently.08
61

PART XI
Exam day protocol
The taper — 48 hours out
Stop introducing new material. Anything learned in the last forty-eight hours is not retrievable under
pressure; it only competes with what is.
No full mock the night before. A bad score creates anxiety with no time to act on it; a good score creates
complacency. Neither is useful.
Light review only: the one-page card, the eight automatic sentences, twenty minutes of speaking aloud.
Seven or more hours of sleep. Two alarms. This is not a wellness suggestion — test-day stress
measurably narrows working memory, and reduced working memory is the exact condition under which
the -s drops.
The morning
Eat something. A three-hour written block on an empty stomach degrades performance in the final
passage of Reading, which is the hardest one.
Activate English before arriving. Twenty minutes of listening on the way in, or speaking aloud to
yourself. Arriving cold means the first five minutes of Listening are spent warming up, and Part 1 is where
the easy marks are.
Documents, water, arrive early enough that the queue is not part of the stress budget.
• 
• 
• 
• 
• 
• 
• 
62

In the room
MOMENT ACTION
Listening, before
each section
Read ahead. Underline stem keywords. Predict word class.
Listening, if an
answer is missed
Flag and move. Never chase. One missed answer costs one mark; chasing it
costs three.
Reading, start Ninety seconds of orientation, then straight to the questions.
Reading, TFNG
items
Name the trap type before answering.
Writing, first 8
minutes
Plan Task 2 fully. Subject, tense, connectors, one target structure.
Writing, order Task 2 first if that reduces anxiety about the word count; Task 1 first if the
timing discipline is stronger that way. Decide in advance, not in the room.
Writing, last 3
minutes
Reverse-order pass, then the six-step check.
Speaking, first thirty
seconds
Slow down deliberately. The first answer sets the examiner's baseline
impression, and nerves push the pace up exactly when accuracy is most
fragile.
Anxiety, treated as a technical problem
Stress reduces available working memory, and every high-cost error in this document is a working-memory
failure. "Rest" is not a strategy. These are:
Slow exhale breathing before each module. Four counts in, six or more out, for sixty seconds. The
longer exhale is the active component.
Externalise the load. Anything written down — the time anchor, the two body-paragraph ideas, the
predicted word classes — is memory you no longer have to hold.
Pre-decide the small choices. Which task to write first, which structure to attempt, how long each
passage gets. Decisions made in the room consume the same budget as the language.
A recovery line for Speaking. Have one rehearsed sentence for the moment a question lands badly: 
"That's an interesting question — let me think about that for a second." Rehearsed, it buys three seconds
without registering as hesitation.
• 
• 
• 
• 
63

PART XII
After the exam
Results and what they mean
Computer-delivered results are typically released within a few days; paper-based results took around thirteen
days, and paper-based testing has been progressively withdrawn. The Test Report Form shows a band for
each module, the overall band, and a CEFR level. Test results are normally treated as current for two years
by receiving institutions, though the institution sets that rule, not IELTS.
CHECK THE SHAPE OF YOUR REQUIREMENT, NOT JUST THE NUMBER
Requirements come in two shapes and they are not equivalent. An overall requirement can be met by
averaging — a strong Reading band can carry a weak Writing band. A requirement with a minimum in
each module cannot: one band below the floor fails the whole result regardless of the overall figure.
Many professional registration bodies use the second shape, and several also require all four scores to
come from a single sitting. Confirm which applies to you before booking anything, because it
determines whether your marginal study hour belongs to your weakest module or your strongest.
If one module falls short — the two instruments
ENQUIRY ON RESULTS (EOR) ONE SKILL RETAKE (OSR)
What it is A remark of your existing performance by a
senior examiner
A fresh sitting of a single module, with
the other three scores carried over
Use it
when
You believe the score does not reflect the
performance — typically a borderline half-
band in Writing or Speaking
You believe the performance itself fell
short and you can reliably do better
Risk The score can only stay the same or rise.
The fee is normally refunded if a band
changes.
The new score replaces the old one
even if it is lower. A 7.0 retaken and
scored 6.5 becomes 6.5.
Timing A limited application window from the
results date — check the deadline printed
on the TRF
Must be sat within 60 days of the
original test date, once per original test
TurnaroundSeveral weeks Typically three to five days
64

ONE SKILL RETAKE — THE CONDITIONS
Available for computer-delivered IELTS. Paper-based sittings have historically not qualified, and the
rule requires the retake to use the same delivery mode as the original test.
The retake must be sat within 60 days of the original test date — not sixty days from when results
arrived — at the same centre or another enabled centre in the same country.
It can be used once per original test, on exactly one of the four modules.
You receive a new Test Report Form combining the retake score with your three original scores.
Both TRFs remain valid, and you choose which to submit.
Availability is not universal: it is offered in a large and growing number of countries but not all, and it
has not been available for tests taken in the United States. Some visa and immigration routes do not
accept it.
Acceptance is decided by each receiving organisation. Many universities and professional
bodies accept it; some do not. Confirm with your specific institution before booking, not after.
These rules are set by the IELTS Partners and change periodically. Verify current terms with your own
test centre rather than relying on any secondary source, including this one.
The decision tree
If a retake is needed, where the time goes
The instinct is to do more of the module that failed. That is usually the wrong shape of work. In each case the
higher-yield activity is diagnostic rather than volumetric.
• 
• 
• 
• 
• 
• 
All modules at or above requirement. Nothing further. Submit within the validity window your
institution applies.
01
One module short by 0.5, and the performance felt better than the score. Consider an EOR first. It
cannot lower the score, and a successful remark saves the retake fee entirely. Note that the EOR
deadline is shorter than the OSR window, so this decision comes first chronologically.
02
One module short by 0.5, and the performance was genuinely below target. One Skill Retake,
booked early inside the sixty days to leave preparation time. Retake only when you can reliably match or
beat the original — the downside is real, and a retake that scores lower has cost you the original.
03
Two or more modules short. OSR does not apply. A full retake is the route, and preparation should be
targeted rather than general.
04
Short by a full band or more in one module. A gap that size rarely closes in a single retake cycle.
Treat it as a preparation problem first and a booking problem second. There is no limit on how many
times the full test may be taken.
05
65

MODULE LOW-YIELD HIGH-YIELD
Writing Writing more essays
and reading model
answers
Three essays under exact timed conditions, then typed up and
self-marked against the four criteria — checking specifically
whether every part of the prompt was answered and whether
the word count cleared the target
SpeakingMore unrecorded
conversation practice
Record, self-transcribe, tag your own errors, re-record. Blocked
by error type, not interleaved, near a test date
Reading Reading more
articles
Question-mechanics drilling with the trap-label system in Part
VII. Most lost marks are mechanics, not comprehension
ListeningMore listening
practice
Blind dictation of short natural-speed clips, focusing on word
boundaries and reductions
A REALISTIC EXPECTATION
The published data shows that the average error-rate improvement between band 7.0 and 7.5 is
roughly a third of the improvement seen between 5.5 and 6.0. A half-band at the top of the scale is not
the same unit of work as a half-band in the middle. Plan the retake window accordingly: two weeks of
targeted, diagnostic work on one module is a realistic project; two weeks of general English is not.
66

PART XIII
Drill appendix
Production drills, not reading material. Each set targets one error type. Say them aloud
or write them out; recognition is not the skill being tested.
Drill A — subject–verb agreement, abstract singulars
Say each correct sentence three times, fast, with no content thinking. Pure motor pattern.
Drill B — the intervening-plural trap
Identify the head noun first, then produce the verb. Say the head noun aloud before the sentence.
The evidence suggests a consistent pattern.01
Research shows the opposite in most settings.02
The data indicates a gradual decline.03
Exposure to screens has increased sharply.04
The cost of housing shapes where people settle.05
Each of those decisions matters more than it seems.06
Trust takes years to build and moments to lose.07
Migration from rural areas continues to accelerate.08
The availability of specialists varies considerably by region.09
Politics affects healthcare more than people admit.10
The number of applicants has fallen for three consecutive years.11
A growing body of work points in the same direction.12
Scarcity drives prices upward regardless of policy.13
Interaction between the two systems produces unexpected effects.14
What this points to is a structural failure.15
The impact of rising costs on families is severe.01
The findings of this study support the hypothesis.02
The range of options available to students has narrowed.03
The effects of prolonged isolation are well documented.04
A combination of factors explains the discrepancy.05
The quality of the arguments presented varies considerably.06
The proportion of households with access continues to grow.07
Increases in energy prices have affected every sector.08
67

Drill C — tense anchoring in narrative
Each pair: the collapse, then the anchored version. Produce only the second.
COLLAPSE ANCHORED
When I was seven, my grandmother gives
me an old radio and it has two speakers.
When I was seven, my grandmother gave me an
old radio and it had two speakers.
I went home that day and I feel exhausted
but hopeful.
I went home that day feeling exhausted but
hopeful.
We arrive at the hospital and nobody knows
what to do.
We arrived at the hospital and nobody knew what
to do.
The case has create a lot of discussion at
the time.
The case created a great deal of discussion at the
time.
He tells me to wait, so I waited outside for an
hour.
He told me to wait, so I waited outside for an
hour.
It was the first time I present in English and I
am terrified.
It was the first time I had presented in English,
and I was terrified.
Drill D — the authorised shift
Narrative in the past; final sentence deliberately in the present. Produce the whole pair as one unit so the shift
becomes a device rather than a leak.
…and by the end of that rotation I had learned more from a single difficult case than
from an entire year of reading. It still shapes the way I approach a first interview.
…so I rewrote the whole thing the night before and handed it in with an hour to spare. 
I tend to work that way even now, for better or worse.
Drill E — would versus will
I would prefer to work somewhere quieter, if I had the choice.01
I would love to learn the language properly at some point.02
If I had more time, I would travel far more than I do.03
If that were to happen, the consequences would be considerable.04
68

Drill F — prepositions
Drill G — advanced structures, one at a time
Drill H — Part 1 calibration
Answer each in 27–39 words. Answer, reason, one detail — then stop deliberately.
Do you prefer to work in the morning or the evening?
I would recommend it, provided the circumstances were different.05
I would imagine most people feel the same way about it.06
Every summer we would drive to the coast. (repeated past action)07
I used to work night shifts. (past state — not "would")08
I am used to working under pressure. (accustomed to)09
The results were different from what we expected.01
I arrived at the clinic just before the handover.02
The conclusion is based on a limited sample.03
Access to care depends on where you happen to live.04
The programme consists of three distinct phases.05
We discussed the proposal at length. (no preposition)06
She is responsible for the whole department.07
The figure rose by fifteen percentage points, to just over half.08
Spending peaked at sixty per cent in the final year.09
There has been a sharp increase in demand over recent years.10
Not only does the policy reduce costs, but it also improves access.01
Rarely had I seen a system fail so predictably.02
Only when I started did I understand how much was involved.03
Had I known the constraints, I would have planned differently.04
Feeling completely out of my depth, I asked for help.05
Rejected by the committee, the proposal needed a full rewrite.06
Having worked in that setting for years, I came to see it differently.07
Absent a reliable transport network, congestion becomes inevitable.08
It is the cost of housing that drives the decision, not preference.09
While there is genuine merit in that argument, the evidence points elsewhere.10
• 
69

How often do you spend time outdoors?
Have you ever learned to play a musical instrument?
Would you like to live in a different city?
Did you enjoy your studies when you were younger?
Do you usually cook at home or eat out?
Are you interested in photography?
What kind of weather do you like best?
Drill I — Part 3, the four-move spine
Answer each with position, mechanism, illustration, qualification. Record, transcribe, self-mark.
Why do you think people are working longer hours than in the past?
Should governments be responsible for funding cultural institutions?
How has the relationship between generations changed in recent decades?
Is technology making professional training easier or harder?
What would happen if universities were free for everyone?
Are large cities becoming less liveable, in your view?
Who should bear the cost of environmental regulation?
Is it possible to measure the quality of education objectively?
Drill J — Task 1 sentence frames, produced against real data
HOW TO RUN THESE
Block by drill, not interleaved, when a test date is close — interleaving adds fatigue faster than benefit
under time pressure. Space the drills across days rather than massing them in one session; spacing is
what produces retention, and it is the one variable that costs nothing.
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
The figure began at just under a fifth and climbed steadily to roughly half by the close of the period.01
Consumption fell by almost a third, from approximately 60% to a little over 40%.02
Coal accounted for the largest proportion, while renewables comprised a negligible minority.03
Both categories declined, though the former fell more sharply than the latter.04
The figure for Spain was roughly double that of Portugal throughout.05
Having risen sharply until 2010, the figure levelled off and held steady thereafter.06
The raw material is extracted, is transferred to a processing facility, and undergoes a heating stage.07
The original warehouses were demolished, and the northern section has since been converted into
residential housing.
08
70

The Complete IELTS Academic Master Document. Covers Writing Tasks 1 and 2, Speaking Parts 1
to 3, Listening, Reading, grammar priorities, lexical banks, exam-day protocol and production drills,
organised around the errors most frequently reported in real test scripts.
Sources for the error data in Part II. Error rates, band-by-band figures and first-language
comparisons are drawn from Müller, A. & Han, W. (2022), IELTS Writing band scores 5.5–7.5:
Grammatical error rates, stakeholder perceptions, and risk, IELTS Research Reports Online Series
No. 1/22, published by the British Council, Cambridge Assessment English and IDP: IELTS Australia.
Supporting observations on fossilisation, skill regression and rate of improvement are drawn from the
second-language acquisition literature cited in that report. Band descriptor summaries paraphrase the
publicly available IELTS assessment criteria.
Two things to verify independently. The exact English-language requirement set by your own
receiving institution — including whether it uses an overall figure or a per-module minimum, and
whether a single sitting is required — and current One Skill Retake availability and acceptance at your
test centre. Both are set by external bodies and both change.
Everything else is technique, and technique transfers. The rest is production.
71