Underneath every reading question sits the same transaction: the question says something one way, the passage says it another, and the mark goes to whoever sees that the two are the same. Location, TFNG judgement, option elimination — all of it runs on paraphrase recognition. This lesson makes that hidden skill visible, names its four transformation types, and installs the two-minute daily drill that trains it deliberately.
The skill under every question
You have already met paraphrase in every lesson of this module: confirm-by-meaning in location discipline, the heavily-rewritten TRUE statements in True/False/Not Given, the correct MCQ option that shares no words with the passage. What those lessons treated as a warning — "the test rewrites everything" — this lesson treats as a trainable object.
The insight that makes it trainable: the test does not rewrite freely. It uses four transformation types, over and over. Once you can name them, you stop experiencing paraphrase as fog and start experiencing it as a small set of moves you have seen a hundred times.
The four transformations
| Type | The move | Question phrase → Passage phrase |
|---|---|---|
| Synonym | Word swapped for a same-class equivalent | cheap → affordable; risk → hazard; helped → facilitated |
| Word-form | Same root, different word class; sentence re-arranged to fit | grew rapidly → rapid growth; decided → the decision to |
| Structure | Grammar rebuilt: active↔passive, clause↔phrase, cause reversed | the flood destroyed the bridge → the bridge was destroyed by flooding; because X, Y → Y, owing to X |
| Summary | Many words compressed to few, or a list to its category | walking, cycling and running → physical activity; could not read or write → illiterate |
Real test items usually stack two: "local people opposed the plan" → "the proposal met with resistance from residents" is synonym (plan→proposal, local people→residents) plus word-form (opposed→resistance) plus a structure flip. Stacking is why word-level scanning fails on low-anchor questions, and why the trained response is to match at the level of who did what to what — the skeleton survives every transformation.
The summary type deserves special respect. It is the hardest to see coming, because no word in the question appears in any form in the passage — the link is a category jump. It is also the engine of matching headings and box-type summary completion. When a question phrase is short and abstract (a health benefit, an environmental cost), expect the passage to hold the concrete, listed version.
The two-minute drill
Training is embarrassingly simple, which is why almost nobody does it. After any practice question you have answered and checked:
- Draw two columns. Left: the phrase from the question or correct option. Right: the phrase from the passage that decided the answer. Copy both exactly.
- Name the transformation(s). Write S (synonym), F (word-form), St (structure), Su (summary) — one letter or several — next to the pair.
- Say the pair once as an equation. "met with resistance from residents = local people opposed".
That is the whole drill. Two minutes, one question, one pair. The naming step is not decoration — forcing a classification is what converts a vague "ah, they reworded it" into a reusable pattern. Skipped, the drill decays into passive re-reading.
Worked pair, from a checked question: Question: The materials used were sourced locally. Passage: builders relied on stone quarried within a few kilometres of the site. Name it: Su + S — stone quarried within a few kilometres summarised as materials sourced locally; relied on ↔ used.
Do this for one question per day minimum; after a full practice test, do it for every question you got wrong plus two you got right. Wrong answers show you the transformation that beat you; right answers confirm the patterns you already own. The pairs feed straight into the family lists from vocabulary that scores — a synonym pair is a two-member family waiting for relatives.
Why this compounds
The drill looks too small to matter. It compounds through three channels:
- Location speed. Scanning for funding, your eye learns to trip on investment and grants — near-misses become hits, and hits mean fewer full-passage re-scans.
- Trap immunity. Word-match lures (the passage's exact words, twisted) rely on candidates who equate word overlap with meaning overlap. A hundred named pairs teach the opposite reflex: overlap is suspicious, transformation is normal.
- Prediction. After enough F-pairs, you see the decision to… in a gap and pre-hear the passage's decided. You begin locating by anticipated transformation, which is what fast readers are actually doing when they look telepathic.
Expect the effect within two weeks of daily pairs — first as a strange feeling that questions have become "more obvious", then as measurable time saved per passage.
Your drill
Fifteen minutes today, two minutes a day after that.
- Do one passage of Reading 2026-01 Test 3, any question types, and check it.
- Build the two-column table for six questions: every wrong answer, topped up with right answers to reach six. Copy exact phrases, name every pair — S / F / St / Su, stacked where needed.
- Count your letters. Most candidates discover a skewed profile — e.g. synonyms never fool them but summary pairs caused every miss. That skew is your training target: when a type dominates your misses, hunt that type deliberately in your next passage.
- From tomorrow: one pair per day from whatever practice you do. After your next full test — Reading 2026-04 Test 5 works — run the full wrong-answers-plus-two protocol.
One answered question, two columns, one named transformation. It is the cheapest drill in this course, and inside a month it is quietly paying rent in every single reading mark you earn.