CELPIP reading samples: Part 4, Reading for Viewpoints
Two full CELPIP reading samples in the real Part 4 shape: a website article where named people disagree, then a reader's comment with blanks in it. Every option is explained, including which speaker each wrong one had its opinion stolen from.
What Reading for Viewpoints actually asks of you
Part 4 closes the reading section with the longest text you will see: an article from a website, four hundred and fifty to six hundred words, about a decision some Canadian town is arguing over. A bylaw. A pilot programme. A budget line. Three or four people are quoted by name, they disagree, and one of them is usually an outside expert brought in to say that everybody is a little bit wrong.
Ten questions follow, in two blocks. Questions 1 to 5 are sentence stems about the article. Questions 6 to 10 are five blanks inside a comment posted underneath the article by a reader, which introduces a fifth voice with its own partial agreements. You get about 12 minutes, the most generous ratio in the section, and you will use all of it.
Here is what separates Part 4 from every other part: understanding the article is not sufficient. In Part 3 a fact is either in paragraph C or it is not. In Part 4 every fact belongs to somebody, and the question is never just what was said. It is who said it, whether they meant it, and whether the person who conceded it thereby changed their mind. You can follow the whole argument, find it interesting, agree with the sensible one, and still lose four marks because you remembered a position without its owner.
Which is exactly how the distractors are built. The test writer does not need to invent a wrong opinion. There are already three or four real opinions on the page, and moving one of them onto the wrong speaker produces an option that is true, familiar, and wrong. The critic's cost objection appears under the champion's name. The expert's caution is handed to the sceptic. Every one reads as something you definitely saw, because you definitely did.
CELPIP Reading Part 4 format at a glance
What the 2026 format fixes about the viewpoints part, and what every good sample has to reproduce.
- Official name
- Reading Part 4: Reading for Viewpoints
- Questions
- 10, out of 38 in the whole reading section. The largest single block.
- Time
- About 12 minutes, roughly 72 seconds a question. The longest part and the longest text.
- What you read
- An article from a website, roughly 450 to 600 words, quoting several named people who disagree
- The cast
- Usually a champion, a critic, a local voice with a practical worry, and an outside expert who takes neither side
- Questions 1 to 5
- Sentence stems, each with a drop-down of four options, about what the article says and implies
- Questions 6 to 10
- Five blanks inside a comment posted by a visitor to the page, who brings a fifth point of view
- Answering
- Drop-down menus only. Nothing is typed, and there is no penalty for a wrong guess.
How to attack the viewpoints part
Everything in Part 4 comes back to one discipline: never hold a position without its owner attached. These five moves enforce it.
- Keep a ledger
- Name, position in three words, one detail that is theirs alone. Four lines, about twenty seconds, done as you read rather than afterwards. It is the highest-return habit in the reading section, because most Part 4 distractors are simply a line from the ledger filed under the wrong name.
- Never assume the critic is against it
- The person who objects almost never objects to the goal. They object to the costing, the schedule, the honesty of the pitch. Part 4 articles open the critic's quotation with a concession precisely so that the option saying "he does not want the sky darker" can catch you.
- Do not let the expert vote
- The outside expert exists to complicate, not to settle. They endorse one side's facts, reject that side's conclusion, and decline to recommend anything. Options that make them announce a verdict are among the most reliable wrong answers in the part.
- Watch every concession
- "She does not dispute the surrenders." A concession moves a fact across the aisle without moving the argument, and it is the single richest source of distractors, because after it you can find the same fact in two mouths facing opposite ways.
- Give the commenter their own line
- Questions 6 to 10 add a fifth voice with partial agreements of their own. Add them to the ledger before you fill any blank, and keep asking which article voice each blank is really about. The commenter is usually defending someone against a misreading.
The ledger deserves a demonstration rather than a description, so here is one built from the sample below, in the twenty seconds it should take you.
Now read those four lines and notice how much is already answered. Who worries about safety? Achenbach, and only Achenbach, so any option putting that worry in Devereux's mouth dies on sight. Does Trottier think the bylaw should pass? The ledger says she takes no side, so any option in which she recommends something is gone. Is Devereux against dark skies? Line two says the opposite in its first four words.
That is roughly half the wrong options in a typical Part 4 set, eliminated by twenty seconds of bookkeeping done while you were reading anyway. The remaining half turn on the finer distinction between what somebody said and what they meant, which is the part that genuinely requires thought, and which you now have the time to give it.
A Part 4 article, taken apart
Read the article, build the ledger, decide your answers, then read the reasoning. Every option is explained, and for each wrong one you will see whose opinion it borrowed.
Wexford Bay argues about the dark
When the Wexford Bay council voted in April to draft a dark sky lighting bylaw, the local astronomy society had been asking for one since 2011. Ingrid Sorensen, who has chaired the society for six of those years, says she has stopped leading with the stars.
"For a decade I brought photographs of the Milky Way to council and got nowhere," Sorensen says. "Now I bring the hydro bill. Roughly a third of what this town spends on outdoor lighting goes straight up into the sky, where nobody lives. Persuade a councillor that we are paying to light the underside of the clouds, and the stars come free at the end of it."
The draft would require every new outdoor fixture to be fully shielded and no warmer than 3000 kelvin, and would replace the town's existing streetlights on a ten-year schedule.
Councillor Ray Devereux voted to draft the bylaw and says he will vote against adopting it.
"I want the sky back as much as Ingrid does," Devereux says. "What I object to is the arithmetic. The replacement schedule is costed at a fixture price I could not get a single supplier to put in writing, and the money comes out of roads. So the choice is not stars or no stars. It is stars or the Fifth Street culvert, and nobody has said that out loud in a meeting yet."
Bev Achenbach, who has run the marina for eleven years, is worried about something else entirely. "My parking lot is the only lit thing between the highway and the water. I am told a shielded fixture puts the same light on the ground. I am prepared to believe it. I would like to see it on my own asphalt before the old ones come down, not after."
Sorensen accepts the point, and says the society has offered to pay for a demonstration fixture at the marina out of its own budget.
Hélène Trottier, a lighting engineer at a university in Sherbrooke who has surveyed thirty-one Canadian municipalities with dark sky bylaws, says the argument in Wexford Bay is unusually honest and still slightly wrong on both sides.
"The energy savings are real, and they are always oversold in the first year, because the model assumes every fixture is replaced on schedule and they never are," Trottier says. "The safety fear is the reverse: it is raised in every town and it has not shown up in the data anywhere I have looked. What actually decides these bylaws is procurement. A ten-year schedule is written by a council and executed by whoever is still making the part in year six. About a third of the towns I surveyed are behind, and two have quietly stopped."
The council votes in November. Devereux expects the bylaw to pass.
"Good," he says. "I hope it does, and I hope I am wrong about the price. I have been wrong about a price before. It cost us a bridge."
An original sample written for this course, in the shape and length of a real Part 4 article. Roughly 490 words.
Sorensen has changed her approach to council by
"I have stopped leading with the stars ... Now I bring the hydro bill." The swap is stated outright, and the closing clause tells you the goal never moved: "the stars come free at the end of it". Only the pitch changed.
The photographs are real and they are in the same paragraph, which is enough for most readers. They are the tactic she abandoned: a decade of the Milky Way and "got nowhere". This option takes a discarded strategy and reports it as the new one, which is a favourite way to catch someone who registered the topic without the direction.
No members, no attendance, no meetings filled. It is what you would expect an astronomy society to do, which is precisely why it is here. Imported from your knowledge of how lobbying works rather than read from the page.
Reverses the whole article. She is still asking, and the draft bylaw exists because she kept asking for fourteen years.
Devereux's objection to the bylaw is that
"Costed at a fixture price I could not get a single supplier to put in writing, and the money comes out of roads." His complaint is about the credibility of a number and what it is quietly buying, which is why his last line is about a bridge.
The single most common error in Part 4. He is the critic, so readers file him under against-the-thing. He opens with "I want the sky back as much as Ingrid does" and closes with "I hope it does" pass. The person who objects almost never objects to the goal, and the article puts the refutation in his own first sentence.
The safety worry is genuinely in this article, held by a genuinely named person. It is Achenbach's, and Trottier addresses it separately. A straight attribution swap, and exactly what the ledger exists to catch: line three of it says safety, Achenbach, marina.
The society does offer to pay for something, which is what gives this option its foothold. It is one demonstration fixture at the marina, offered by Sorensen, and Devereux has nothing to do with it. A real fact, correct in itself, wired to the wrong speaker and inflated from one fixture to the whole retrofit.
Trottier's view of the argument in Wexford Bay is that
"Unusually honest and still slightly wrong on both sides", and then she itemises it: the savings are real but oversold, the safety fear is raised everywhere and unsupported. One correction aimed at each camp, and no verdict for either.
She is the sceptical outsider with a survey full of towns falling behind and two that have quietly stopped, and expert scepticism reads as expert opposition. She never says what Wexford Bay should do. The Part 4 expert complicates and declines to vote, and options that hand them a verdict are how the part punishes a reader who wants the article to resolve.
She does attack an optimistic number, saying the savings are "always oversold", so criticism of the arithmetic is genuinely hers. But the optimistic number is Sorensen's side, and Devereux is the one already attacking it. The option agrees with Devereux and then blames him for the thing he is right about. Swapped and inverted at once.
The reverse of her most decisive sentence: the safety fear "is raised in every town and it has not shown up in the data anywhere I have looked". If you remember only that Trottier discussed safety, this option looks like a memory.
On the real test this is the second block, questions 6 to 10. Here is one blank from the comment posted under the article.
Posted by K_Ferreira
I moored at that marina for nine years and I want to defend Bev, because this piece makes her sound nervous and she is not. She is not saying 4. She is asking to see the thing working before the old ones come down, which costs the town nothing and which Ingrid has already agreed to pay for.
The commenter is defending Achenbach against being read as an opponent, and the article supports the defence: "I am prepared to believe it," she says, and then asks for a demonstration rather than a defeat. The blank needs the position she is being wrongly assigned, and outright rejection is the only position in the article stronger than the one she holds.
Cost is the loudest objection in the article, so a reader reaching for "the thing Bev is being accused of" grabs the article's main complaint. Cost is Devereux's. Achenbach never mentions money once. The comment block is where the ledger earns its keep, because you are now tracking five people across two texts.
She is saying it: "I would like to see it on my own asphalt before the old ones come down." The sentence is "She is not saying ___", and the very next sentence in the comment confirms she is asking for exactly this. An option the following sentence contradicts is the standard punishment for filling a blank before reading on.
Also something she says, in her first line, word for word. Same failure as C. Two of your four options are real Achenbach positions parked behind a negation, and both only work if you stop reading at the blank.
Common mistakes in Reading Part 4
Six ways to understand a Part 4 article completely and still lose half the marks on it.
Remembering the position, forgetting the owner
The defining Part 4 error. Safety was discussed. Cost was discussed. Both are true, and neither is an answer until it has a name on it. This is what the ledger is for, and it is why twenty seconds of bookkeeping outperforms a second read of the article.
Assuming the critic opposes the goal
He wants the sky back. She is not against chickens. The article usually says so in the critic's own first sentence, and the option that ignores it is still the most-picked wrong answer in the part, because a critic is supposed to be against the thing.
Letting the expert take a side
You want the article to resolve, and the expert is the one who sounds like they could. They will not. They endorse one side's evidence, reject that side's conclusion, and go home. Any option in which they recommend an outcome should be your first suspect.
Misreading a concession as a conversion
"He does not dispute the surrenders." Granting a fact is not joining a side, and after a concession the same fact exists in two mouths pointing opposite ways. That gap is where the hardest options in the part are made.
Answering with your own opinion
Part 4 articles are built on real disagreements, and you will have a view by the third paragraph. The stem never asks for it. When one speaker is obviously more sensible, that is the article working on you, not evidence.
Treating the commenter as a summary
The comment is a fifth opinion, not a recap. They usually agree with somebody in part, disagree in part, and are often correcting a misreading of one speaker. Give them their own ledger line before you touch a blank.
5 more Part 4 questions, no timer
A second original article with four voices. Build the ledger first, commit to an answer, and only then reveal the reasoning. Four names, four positions, and no clock at all.
Halbrook counts its chickens
Three years ago the City of Halbrook began allowing households to keep up to four hens in a registered backyard coop. Two hundred and six households signed up. The pilot expires this fall, and council decides in October whether to make the bylaw permanent.
Priyanka Raman, the bylaw officer who has run the pilot since it began, has the file in front of her. "We received forty-one complaints in three years," she says. "Nine were about noise, and every one of those nine turned out to be a rooster, which the bylaw already forbids. The rest were about smell, and almost all of them came from four addresses. I want to be careful here. Forty-one is a small number, and it is not zero, and both of those things are true at once."
Otis Kwan registered in the first month and has kept four hens since. "My daughter is seven and she can tell you which one is laying and which one is moulting," he says. "I did not expect that to be the part that mattered. I expected eggs. The eggs are honestly a lot of work for what they are."
Marguerite Dubé, who runs a small animal rescue north of the city, has a number of her own. "Sixty-one hens in three years, surrendered to me by people who registered, got bored, and then discovered that a hen lives eight years and lays well for two. I am not against chickens. I have forty. I am against a bylaw that hands a living animal to somebody on the strength of a form and a fifteen-dollar fee and then never asks again."
Kwan does not dispute the surrenders. He points out that the registry is the only reason anybody knows about them.
Dubé agrees, and says that is precisely her complaint: the city built a system that can count its own failures and then declined to do anything with the count.
Ana-Lucia Ferreira, a veterinary epidemiologist who has advised three provinces on backyard flocks, says the debate in Halbrook is happening in the wrong place. "Every council I sit in front of wants to talk about avian influenza, because it is frightening and it is on the news," she says. "At four birds in a fenced yard with no movement between flocks, the risk is real and it is very small, and it is not what will decide whether this works. What decides it is what Ms. Dubé is describing. Abandonment is the actual harm, it is measurable, and Halbrook is one of the very few cities that can measure it. That is an argument for keeping the registry, not for ending the pilot."
Council votes in October. Raman expects a full gallery.
An original sample written for this course. Roughly 440 words. There is no timer here: take as long as you like.
Raman's account of the complaints suggests that she
"Forty-one is a small number, and it is not zero, and both of those things are true at once." She refuses to resolve her own data, which is a position in itself. She is the officer with the file, not an advocate, and the article gives her no opinion anywhere.
The strongest distractor here, because her numbers really are favourable: forty-one complaints across 206 households in three years, every noise complaint invalid, the smell complaints concentrated in four addresses. A reader adds that up and hears success. She very deliberately does not say it, and the stem asks about her, not about what her data implies.
Reversed. All nine noise complaints "turned out to be a rooster, which the bylaw already forbids", meaning the rules were working and the complaints were about something the bylaw had already banned.
She names the four addresses and draws no conclusion about them whatsoever. The option supplies the enforcement action her sentence stops one step short of, which is a reliable way to catch a reader who finishes other people's arguments for them.
Kwan's comment about the eggs suggests that
"I did not expect that to be the part that mattered. I expected eggs." His daughter's fluency with the birds is the payoff he did not plan for, and the eggs are a disappointment he shrugs at. The eggs are the setup; the seven-year-old is the point.
"A lot of work for what they are" sounds like a yield complaint, and a low count is the obvious way to be disappointed by eggs. He never mentions how many he gets. He is complaining about the exchange rate between effort and eggs, not about the quantity, and nobody promised him anything.
He is the enthusiast of the piece and has the warmest lines in it. Nothing regretful anywhere near him.
The cleverest option in this set. Dubé really does make a work-and-commitment argument, Kwan really does say the eggs are "a lot of work", and the phrase genuinely sits in both their mouths. But Dubé is describing people who quit and abandon animals, and Kwan is still keeping his four and delighted about it. Same words, opposite conclusions, different speakers. This is the attribution trap in its purest form.
Dubé's objection is aimed at
"I am against a bylaw that hands a living animal to somebody on the strength of a form and a fifteen-dollar fee and then never asks again," followed by the charge that the city "declined to do anything with the count". Her target is the administration and the absence of follow-up, not the hens and not the keepers.
The standard misread of any Part 4 critic, and sixty-one surrendered hens makes her sound thoroughly anti-hen. She pre-empts it herself in six words: "I am not against chickens. I have forty." The article hands you the refutation in her own voice, one sentence after the thing that provokes the misreading.
The fifteen-dollar fee is real, it is in her own quotation, and a specific dollar figure is exactly what a reader latches onto. She cites it as evidence that the barrier to entry is trivial, not as something she wants raised. Doubling the fee would not touch her complaint, which is about what happens in year two.
Dubé never mentions disease. It is Ferreira's subject, and Ferreira raises it only to dismiss it. The swap is onto the person whose position is closest in temperature, which is what makes it feel plausible: if somebody in this article is worried, surely it is her.
Ferreira's position is that
"Abandonment is the actual harm, it is measurable, and Halbrook is one of the very few cities that can measure it. That is an argument for keeping the registry, not for ending the pilot." She takes Dubé's evidence whole and reverses the conclusion drawn from it: the registry Dubé calls a failure is, to Ferreira, the reason the pilot is worth keeping.
She endorses Dubé's facts completely, and an expert agreeing with the critic reads as an expert backing the critic's demand. Her final sentence refuses it in so many words. A shared fact is not a shared conclusion, and Part 4 builds this option in almost every set, because agreeing with somebody's evidence is the most natural-looking route to agreeing with them.
Reversed twice. She says councils want to talk about avian influenza because it is frightening and in the news, and that at four birds the risk "is very small". It is overweighted in attention, not underestimated. If you remember only that Ferreira and avian influenza appeared together, this looks like a memory of the article.
She does criticise where the debate is happening, and she talks about measurement, so scepticism and numbers are both in her paragraph. She never questions Raman's forty-one. She praises Halbrook for being one of the few places able to count anything at all.
Posted by J_Whitehorse
I kept four hens for the whole three years and gave them up last spring, so I am one of Ms. Dubé's sixty-one, and I would like to say that she is being fair to me. She is not arguing that 5. She is arguing that the city let me sign a form in April and never spoke to me again, which is exactly what happened.
The commenter is one of the surrenderers, defending Dubé against the reading that she blames the keepers. Dubé's own words back it up ("I am not against chickens"), and her fire is aimed at the city, which is precisely what the commenter's next sentence spells out: signed a form in April, never spoken to again.
The fifteen-dollar fee sits in Dubé's quotation, and "the fee is too low" is the most natural repair of a complaint about a trivial barrier. She never asks for it, and the commenter's following sentence is about being signed up and abandoned rather than about price. The blank has to fit the sentence after it as well as the one before.
She is arguing that. "A hen lives eight years and lays well for two" is her line, and it is the whole mechanism of the surrenders. The sentence reads "She is not arguing that ___", so a position the article puts squarely in her mouth cannot fill it.
Also hers, by way of the concession: Kwan raises it, "Dubé agrees", and she builds her complaint on top of it. Same failure as C. Both options take real Dubé positions and hide them behind a negation, which only catches a reader who filled the blank without reading to the end of the sentence.
Reading tips for the viewpoints part
Budget 3, 4 and 5
Three minutes on the article and the ledger, four on the five stems, five on the comment. Twelve minutes is the most generous allowance in the section and the text is the longest, so the surplus is smaller than it looks.
Read the critic's first sentence twice
It is nearly always a concession, and it is nearly always the thing that kills the most tempting wrong option in the set. "I want the sky back as much as Ingrid does." "I am not against chickens. I have forty."
Find the expert's last sentence
The outside expert spends a paragraph agreeing with everybody and then, in one final clause, says what their agreement does not amount to. That clause is where the question lives, and it is the sentence readers skim because the paragraph felt settled already.
Ask who benefits from each option
If an option would delight one of the speakers, check whether it is actually theirs. Options are built by moving a real position one seat to the left. Naming the speaker the option would please is a fast way to spot the seat it came from.
Read past the blank, every time
In the comment block the sentence after the blank routinely settles it, and negated blanks are common: "She is not saying ___." Two of your four options will be things she definitely said, and they only work on somebody who stopped reading.
Answer every question
Nothing is deducted for a wrong drop-down. Part 4 is the last part of the section and the one most likely to catch you tired, so if a blank has beaten you, guess it and take the certainty of a chance over the certainty of a zero.
CELPIP reading samples FAQs
What is CELPIP Reading Part 4?
Part 4 is Reading for Viewpoints. You read an article from a website, usually 450 to 600 words, in which several named people disagree about a local decision. You then answer 10 questions in about 12 minutes: five sentence stems about the article, then five blanks inside a comment posted underneath it by a reader.
What makes a good CELPIP reading sample for Part 4?
One with at least three named people who genuinely disagree, including an outside expert who complicates the argument rather than settling it, and options that were built by moving real positions between real people. A sample that only tests whether you understood the topic is not a Part 4 sample, because Part 4 tests attribution.
Why is Part 4 the hardest reading part?
Because understanding the article is not enough. Every fact in it belongs to somebody, and the wrong options are mostly true statements attached to the wrong speaker. You can follow the entire argument, agree with all of it, and still lose marks by remembering what was said without remembering who said it.
How do I track who holds which view in Part 4?
Build a ledger as you read: name, position in three words, and one detail that is theirs alone. The champion, the critic and the expert each get a line. It costs about twenty seconds of your 12 minutes and it disarms most of the distractors in the part, because most of them are an attribution swap.
What is the comment section in CELPIP Reading Part 4?
Questions 6 to 10 sit inside a comment posted by a visitor to the website page. The commenter is a new voice who usually agrees with one person in part and disagrees in part, so you now have four or five positions to keep straight instead of three. Every blank is settled by what somebody in the article actually said.
Does the expert in a Part 4 article take a side?
Almost never, and that is the point of them. The outside expert typically says both local sides are partly right, endorses one side's facts while rejecting their conclusion, and declines to say what should be done. Options that make the expert recommend an outcome are among the most reliable traps in the part.
How long do I get for CELPIP Reading Part 4?
About 12 minutes for 10 questions, which is the most generous ratio in the reading section at roughly 72 seconds a question. It is also the longest text. A workable budget is around 3 minutes reading the article and building the ledger, 4 minutes on the five stems, and 5 minutes on the comment.
Does this practice course score my Part 4 answers?
No. There is no timer and no evaluation anywhere in the practice course. You settle on an answer, then reveal an explanation that covers why the right option is right and which speaker each wrong option stole its position from. The timed reading tests mark your answers and report a CLB band instead.
Are these CELPIP reading samples free?
Yes. Both articles, both comments, the worked example, the five untimed questions and every explanation on this page are free, with no account and no credit card required.
Are these Part 4 samples updated for the 2026 CELPIP format?
Yes. Both samples follow the 2026 CELPIP reading format: four parts and 38 questions, with Part 4 running 10 questions in about 12 minutes on a website article plus a reader's comment.
Where to go next
That is all four parts. The only thing left is to put the clock back on and do 38 of them in a row.
Part 1: Correspondence
11 questions in 10 minutes, a personal message and a reply with five blanks.
Learn Part 1 →Part 2: Apply a Diagram
8 questions in 8 minutes, and the answers hide in the diagram rather than the email.
Learn Part 2 →Part 3: For Information
Paragraph matching, A to E, where option E is there to punish a forced answer.
Learn Part 3 →Timed reading tests
All 38 questions across the four parts, under the real per-part clock.
Start a reading test →You have all four parts now
Keep the ledger, do not let the expert vote, and read past the blank. The timed reading test puts all 38 questions back on the real per-part clock.