Strategies
Core Reading Skills¶
CELPIP Reading rewards three kinds of processing: global (big picture), local (one precise detail), and inferential (a conclusion the text points to but doesn't state). These four skills wire those habits into your approach.
Skill 1: Skimming for Gist (30–40 seconds)¶
Capture the passage's core idea, structure, and stance fast — so later choices have context.
- Lead lines first — Read the first sentence of each paragraph (they announce the topic or shift)
- Flag contrast markers — however, although, whereas, on the other hand, still
- Name the sections — Tag each paragraph with one word (e.g., Problem / Evidence / Counterpoint / Recommendation)
- Check the edges — Title, caption, footnotes; in diagrams: legend + units
- One-line gist — Say it: "This is mostly about , arguing that ___ because "
Skim Pitfalls
- Detail drag: Skip examples and numbers during skimming — you'll verify later
- Scope drift: Your gist should cover the whole passage, not one paragraph you liked
Skill 2: Scanning for Anchors¶
Jump straight to the line that answers a Specific Information item and prove it with the text.
What counts as an anchor (won't paraphrase): - Proper nouns: names, streets, programs, agencies - Numbers & formats: 12:30 pm, $14.50, June 21, 3 lanes, Unit 5B - Exact labels: FoodSafe Level 1, Permit 27-B, Line 980
Micro-routine: 1. Extract the constraint string from the stem: e.g., "≥3 lanes, after 7 pm, weekdays" 2. Hunt the anchor: scan vertically for the number/label/time 3. Read ±1 sentence around it 4. Verify every constraint — one mismatch = reject
High-Frequency Traps
- Digit flips: 12 vs 21; $13 vs $31
- Time windows: before/after/until/from; am vs pm; weekday vs weekend
- Unit swaps: km vs m; per day vs per week
- Scope creep: a rule for "temporary staff" becomes "all staff" in a wrong option
Skill 3: Paraphrase Recognition¶
Recognize when a choice is the same idea in new words vs. a meaning change disguised as a synonym.
Safe Paraphrases (meaning preserved)¶
| Category | Swaps |
|---|---|
| Cause/effect | because ↔ due to ↔ owing to; therefore ↔ so ↔ thus |
| Contrast | however ↔ but ↔ nevertheless ↔ even so |
| People | residents ↔ people who live here; staff ↔ employees |
| Money/time | cost ↔ fee ↔ charge; free ↔ at no charge; schedule ↔ timetable |
| Change | increase ↔ rise ↔ grow; decrease ↔ drop ↔ decline |
False Friends (meaning changed)¶
| Looks Similar | Actually Different |
|---|---|
| must / required | ≠ should / recommended |
| will | ≠ may / might / could |
| free | ≠ discounted / reduced |
| exactly | ≠ at least |
| only (restrictive) | ≠ especially (emphasis) |
| some (at least one) | ≠ most (more than half) |
| by 5 pm (≤ 5:00) | ≠ until 5 pm (period ends at 5:00) |
| not until 5 pm (nothing before) | ≠ no later than 5 pm (any time up to) |
The Three Switches¶
Every option has three switches that must match the passage:
- Strength — must → should → may (don't upgrade or downgrade)
- Scope — all → most → some → few (don't widen or narrow)
- Polarity — watch not, only, except, unless, before, after, by, until
If any switch moves, the option is wrong — even if the topic matches.
Skill 4: Tone and Purpose Detection¶
Tone signals: - Hedges (soften): may, might, tends to, appears to - Boosters (strengthen): clearly, definitely, certainly - Attitude verbs: appreciate, regret, recommend, oppose - Register cues: "Hi Alex… Thanks!" (informal) vs. "Dear Sir/Madam…" (formal)
Purpose quick-labels: - Inform — explain a process, announce a change - Request — ask for action, information, confirmation - Persuade — argue for a policy, recommend a decision - Complain — report a problem + request remedy
Function Test
"If I reply to this writer, what would I do?" If you'd perform an action → request. If you'd form an opinion → persuade.
When to Infer vs. "Not Stated"¶
| Infer | Not Stated |
|---|---|
| Text gives clues pointing to a conclusion | The idea genuinely doesn't appear anywhere |
| Takes one small logical step | Would require outside knowledge or stacked assumptions |
| "Evenings are crowded" → crowding is common after work | "A pilot runs for two weeks" → the city will expand it next year |
Three filters: 1. Text-only rule — If you need outside knowledge, it's not safe 2. One small step — If you're stacking assumptions, stop 3. Global sweep — Check all paragraphs; confirm absence before choosing Not Stated
Answer Engineering: Eliminate Like a Pro¶
You don't win by "feeling" the right option — you win by disqualifying wrong ones fast.
The Four Wrong-Answer Types¶
| Type | What It Does | Tell |
|---|---|---|
| Outside Scope | Goes beyond what the text discusses | all, always, everywhere where passage is narrower |
| Over-Strong | Upgrades hedges (may → will) | Absolute language passage doesn't use |
| Misaligned Speaker | Assigns a claim to the wrong voice | Right idea, wrong person |
| Time/Quantity Mismatch | Breaks number, time, or unit logic | Looks right at a glance, fails digit-by-digit |
The Five Kill-Tests (stop at first fail)¶
- Finger-on-Text — Can you point to the exact line/cell that supports it?
- Strength Match — must/should/may levels identical?
- Scope Match — Same people, time, place, quantity?
- Polarity & Logic — not, only, except, before/after/by/until all correct?
- Speaker/Stance — Right voice, right intensity?
The ACE Quick Filter¶
On first read, tag each option: - A (Aligned) — Matches text. Keep and verify. - C (Contradicts) — Conflicts with the text. Eliminate. - E (Extra) — Adds claims not in the text. Eliminate.
Trash all C's and E's, then run Kill-Tests on surviving A's.
Tie-Break Rules (when two options survive)¶
- Closest wording to the actual text wins
- Fewer assumptions wins
- Specific support beats "whole passage vibe"
- Moderate beats extreme
Every-Item Playbook¶
- Tag the question G/S/I
- ACE filter on all options
- Finger-on-text for survivors
- Check Strength, Scope, Polarity
- Verify speaker/stance
- Choose (tie-break if needed)
- Over 90 seconds? Pick best guess, mark for Pass 2
Speed Moves When Time Is Tight
- Numbers first — one digit mismatch = instant kill
- Nuke absolutes — always/never/only/must unless the text matches
- Prefer hedges — options with may/might/some when the passage hedges
- Nearest neighbor — between two survivors, pick the one using more of the passage's own words