CELPIP Reading: All 4 Parts, and the Half of the Test Nobody Studies

What the Reading section actually is

Three of the four Reading parts are two tasks wearing one number. Part 1 is an email plus a response with blanks. Part 2 is a diagram plus an email with blanks. Part 4 is an article plus a reader's comment with blanks. Only Part 3 is what people picture when they say "reading comprehension" — one text, one set of questions, done.

That second text is not a footnote. It scores the same as everything else, and it tests something else entirely.

The four parts, side by side

Part 3 is also the only part with an option E, for "information that is not stated in the reading". Do not go hunting for a not-stated escape hatch in Parts 1, 2 or 4 — there isn't one.

Do the arithmetic

Blanks: 5 in Part 1, 3 in Part 2, 5 in Part 4. That's 13. Questions: 11 + 8 + 9 + 10. The four parts come to 38, which is what your Reading raw score is reported out of.

So 13 of 38 marks — a third of the section — are not comprehension questions at all. They are "does this sentence still make sense" questions. The comprehension half asks did you find the fact. The blanks half asks does the sentence hold together. Different skill, different failure, different practice. Most study plans treat Reading as one lump and drill only the first kind.

The failure: grammar-first blank filling

You reach the response email, read the sentence with the gap, open the dropdown, and pick the option that sounds right in the sentence. That instinct is correct in almost every English exam you have ever taken. It is wrong here.

Call it grammar-first blank filling. The options are usually all grammatical — the sentence reads well with any of them. Only one is consistent with a fact stated in the other text, sixty words away, which you read once and half-forgot.

Worked example (original AlmiWorld practice item)

Main text — an email from a neighbour, Part 1 style:

Plots in the Riverbend community garden are handed out by lottery every March. The one exception is volunteers: if you put in four hours or more at any garden work day last season, your plot is assigned before the lottery is drawn. You were at the fall clean-up for most of a Saturday, so you're in good shape.

Response text — one sentence, with the blank:

Thanks for explaining all this. Since I did six hours at the fall clean-up, I ________ in March.

The options:

Answer: B.

Now the part that matters — why the others fail, given that all four are grammatical and all four produce a natural-sounding sentence.

A is a straight contradiction. Six hours clears the four-hour threshold, so the exception applies. A is what you choose after skimming "lottery every March" and never reaching the exception.

C is a meaning reversal dressed in a word from the passage. "Eligible" feels like passage vocabulary, and being exempt from the lottery and being ineligible for it are both ways of not being in it. But one is a benefit and one is a bar. C quietly turns the benefit into a penalty. That is the commonest distractor design in the section: right topic, right register, inverted relationship.

D invents a rule. Nothing mentions multiple entries. It is grammatical and it fits the friendly register perfectly — and it is unsupported. In Parts 1, 2 and 4, unsupported means wrong.

The shape that works

1. Read the main text once for structure, not detail. Who is writing, to whom, and what one decision or event sits at the centre of it.
2. Answer the comprehension questions first. They send you back into the text, and that second pass loads the facts the blanks will need.
3. Before opening a dropdown, say the fact in your head. "Volunteers skip the lottery." Now you are matching options against a fact, not against a sound.
4. Treat every option as grammatical. If two both fit the sentence, you have learned nothing yet — go back to the main text.
5. Watch the relationship words: still, no longer, instead, unless, except, before, only. Reversals live there.
6. Answer everything. There are "no points deducted for wrong answers", so a guess is free and a blank is not.

Timing and one-way doors

The section runs 43–56 minutes — the range exists because the test "may contain unscored content used for test development". Older PDFs still in circulation quote 55–60 minutes; that figure is stale.

Reading is timed per part, and the test moves on by itself: "the test will automatically move forward to the next part", and "you can't go back to a previous part." Once Part 1 closes, those 11 marks are settled. One caveat — the free official sample test lets you go back and change answers. The real test does not. Do not build a habit on the sample's behaviour.

What actually costs you the mark

• Picking the blank that sounds best instead of the one the main text supports.
• Reading the response text on its own, as if it were a standalone grammar exercise.
• Missing an exception clause — unless, except — in the main text, then contradicting it.
• Choosing a distractor because it reuses a word you recognise from the passage.
• Treating Part 2's diagram as decoration. The diagram is the main text.
• Looking for a not-stated option in Parts 1, 2 or 4. Only Part 3 has one.
• Overspending on Part 1 and getting auto-advanced out of a later part.
• Leaving anything blank.

Try one

[Try a Reading task →] — one full task, marked, no card required.