CELPIP Reading Task Types: All 4 Parts, Decoded

You've read the overview. You know CELPIP reading has four parts and about 40 questions in 55–60 minutes. Then you open Part 2, meet a recreation-program schedule with seven columns, and discover that knowing the format and knowing what to do are different skills entirely.

This post is the deep dive — one full section per part. Each gets the task's job, the trap built into it, the move that beats it, and a micro-example you can hold in your head. By the end, you'll know each part's failure mode and which one to practice first. To see which part is costing you the most points right now, take a free CELPIP mock test with instant section scores → free mock test

The short answer

Part 1 (correspondence) tests whether you can track who said what across an email chain. Part 2 (applying a diagram) tests whether you can locate a cell instead of reading the whole visual. Part 3 (reading for information) tests skim-then-scan. Part 4 (reading for viewpoints) tests whether you can separate opinion from fact and keep two authors apart. Each part has one trap — quotes, real-world logic, word-by-word reading, and confident opinions, in order.

Part 1: correspondence

The task's job. You get a chain of three or four emails — the classic case is a dispute, like neighbors arguing about noise. Statements follow, and you match each to the person who wrote or believes it. This part moves fast.

The trap. Replies quote each other. Tenant A complains; the manager replies quoting the complaint. The quoted words sit in the manager's email, but they're still A's position. Match a statement to whoever's email the words appear in, and you've credited the wrong sender — the trap's favorite harvest.

The move. Before touching any question, write each sender's stance in five words: "A: noise until midnight. Manager: rules exist. B: gym is fine." Then answer by matching meaning to that map, not keywords to emails.

Micro-example. A thread about a shared driveway:

Manager: "Ravi asked whether guests can park overnight." Ravi: "My cousin visits monthly. One night, that's all I'm asking." Priya: "Ravi's guests blocked my garage last spring."

Question: Who thinks the parking rule is too strict? Ravi — he's asking for an exception. The manager quoted Ravi's question without owning it. Three senders, three stances, no keywords needed.

Part 2: applying a diagram

The task's job. A schedule, table, or set of images — program guides, drop-in hours, membership options. You match statements to what the visual actually says. This is the lookup part: the answer exists in one or two cells, and your job is to find them.

The trap. Real-world logic. You "know" kids' programs need registration, that pools close by 10, that weekends cost more. The diagram overrules everything you know. Options are written to match the world instead of the visual, because the writers know you'll want them to.

The move. Read the statement first. Pull out its two deciding details — the day, the age group, the price band. Then scan the rows and columns for those cells only. Never read the diagram start to finish. It's a tool, not a story.

Micro-example. A community swim schedule:

Statement: Family swim is available twice a week. True — Wed 4–6 pm and Sat 1–3 pm, two cells, decided. Flip one cell to "—" and it's once. The whole question lives in two cells. That's Part 2.

Part 3: reading for information

The task's job. One longer informational text — how a program works, what a policy covers, how a service is organized. Questions ask where specific information lives, usually matching statements to sections. It's the longest read in the section, and the pacing math gets tighter here (see the time management post for the per-part budget).

The trap. Reading word by word, start to finish, before looking at the questions. It feels responsible — it's how school taught reading. Under a clock, it's how people spend four minutes on a passage and answer from memory anyway.

The move. Structure first. Skim the headings and the first sentence of each paragraph — that's usually the topic. Build a rough map: "para 2 is about eligibility, para 4 about fees." Then scan to the section each question needs. You read each part of the text once, for a reason.

Micro-example. A text about a city's compost program, paragraphs on: what goes in, pickup days, apartment rules, seasonal changes, and complaints.

Question: Where would you find out what to do with pizza boxes? "What goes in" — that's the map doing its job. The trick is that the words "pizza box" never appear in the heading; you matched the question's meaning to the section's job. That's the trained skill, and it's why Part 3 rewards structure over vocabulary.

Part 4: reading for viewpoints

The task's job. Two authors, one topic, different opinions. Questions match claims and opinions to each writer. This is the deepest reading in the section — not hardest, deepest. It comes last, when you're an hour into the test and your focus is thinnest.

The trap. Confidence. A strongly stated opinion reads like fact when you're tired. "This policy will destroy the neighborhood" feels like information about the policy; it's one author's belief. The test counts on you filing opinions as facts.

The move. Tag attribution language as you read — "argues," "claims," "suggests," "worries," "points out." Verbs of claiming mark opinions; verbs of reporting mark facts. Note each author in five words, like Part 1's sender map, then match by meaning.

Micro-example. Two neighbors on a new bike lane:

Author A: "The lane will cut through-traffic, and the data on other streets suggests families walk more afterward." Author B: "I worry the lane removes parking, and I've argued that shop owners will suffer."

Question: Which author presents a personal concern rather than a claim about data? B — "I worry" is first-person concern; "I've argued" is a stance. A leans on data ("suggests" — still hedged, still A's reading of it). Notice the verbs doing all the work. That's the whole part in two lines.

The format at a glance

One caveat that belongs here: formats are periodically revised — confirm current structure on the official CELPIP site before test day.

3 mistakes that cross all four parts

Practicing your favorite part. People grind Part 4 because it feels like "real reading" and skip Part 2 because it looks easy. Then Part 2 quietly costs them four points. Your error log — not your preference — picks your practice parts.

Matching keywords instead of meaning. Every part's wrong options are built from the passage's own words. The right answer is usually a paraphrase. If an option shares three words with the text, trust it less.

Changing answers on instinct. A change with a reason (reread, found the cell) is a correction. A change because "C felt sudden" is a coin flip. Know which one you're doing.

What good looks like, part by part

Part 1: a stance map in the margin before question one — sender names, five words each, quotes credited to their original authors. Part 2: statements read first, two deciding details pulled, then a cell hunt. Part 3: a heading map built in ninety seconds, then every scan aimed at a question. Part 4: attribution verbs tagged, two author stances named, opinions never filed as facts.

None of that is reading talent. It's four checklists, run in order, on a clock. Practice works on checklists in a way it can't work on talent.

This week's plan

• Monday: one Part 1 set. Build the stance map before answering anything.
• Tuesday: one Part 2 set untimed. Practice pulling the two deciding details from each statement.
• Wednesday: one Part 3 set. Heading map first — no question until the map exists.
• Thursday: one Part 4 set. Tag every attribution verb you meet.
• Friday: your weakest part, again.
• Log every miss as misread / vocabulary / trap — the loop that makes this stick lives in the daily practice method.

Want these part-by-part fixes applied to your actual answers? Get a detailed evaluation with band-level feedback → evaluation

The part nobody tells you about

Here's the thing I'd want someone to tell me: Part 2 looks like the easy part and quietly decides scores. The pattern shows up over and over (details combined from real student cases) — a student confident in English walks out sure the hard part was Part 4, and the breakdown says the diagram section cost them more than anywhere else. Why? Because "easy-looking" parts get no strategy. Nobody writes a stance map for a swim schedule; they wing it on general knowledge — and general knowledge is exactly the trap. The test writers aren't cruel; they're just accurate about which parts people prepare honestly.

Tools and resources

• The official CELPIP site's samples. The closest thing to the real question style and interface. Do the Part 2 and Part 4 samples even if you feel ready — they're the two that surprise people.
• Our reading practice materials guide — what a set needs for each part.
• The pillar CELPIP reading guide — pacing and strategy across all four parts.

Frequently asked questions

What are the 4 parts of the CELPIP reading test?

Correspondence (email chain), applying a diagram (schedule or visual), reading for information (longer text), and reading for viewpoints (two opinion authors). About 40 questions in 55–60 minutes.

Which CELPIP reading part is hardest?

The one you haven't practiced with a strategy. Part 4 feels hardest — it comes last and reads opinions. But Part 2's diagram questions quietly cost more points, because people treat them as easy and answer from world knowledge instead of the visual.

What does "applying a diagram" actually mean?

It means using a schedule or table as a lookup tool. The answer sits in one or two cells. Your job is finding the cells — not reading the diagram like a passage.

Are the questions multiple choice?

Yes — the section runs roughly 40 questions, and the matching questions are multiple-choice style, answered on screen.

How do I tell opinion from fact in Part 4?

Follow the attribution verbs. "Argues," "claims," "suggests," "worries" mark opinions; "states," "reports," "points out" mark information. When a verb of claiming shows up, what follows belongs to that author's stance, not to the world.

Do the four parts come in a fixed order?

Yes — correspondence, diagram, information, viewpoints. Formats are periodically revised, so confirm current structure on the official CELPIP site, but the order has been stable.

What to do next

1. Today: pick your weakest part from your last practice set — honest, not comfortable.
2. Tomorrow: run that part with its move and nothing else.
3. This week: cycle all four parts once, one move each.
4. Log every miss with its category.
5. Weekend: one timed rotation of all four.

Related reading

• CELPIP reading: the complete guide — the pillar: pacing, strategy, the 7-vs-9 difference
• CELPIP reading practice: the daily 30-minute method — the error-log loop for these four parts
• CELPIP reading time management — the 2-pass system and per-part budgets
• CELPIP reading score 9 — turning the four moves into a target score

The bottom line

Four parts, four traps, four moves. The test isn't asking whether you can read — it's asking whether you brought the right procedure to the right text. Learn the trap, run the move, log the miss. That's the whole game, one part at a time.

Ready to find out which of the four is actually holding your score? Take a free CELPIP mock test with instant section scores →