How CELPIP Writing Works: 2 Tasks, 53 Minutes
Two tasks, 53 minutes, and one timer that moves on without you. Before any technique is worth learning, you need to know what is coming and what the clock allows.
The section in one page
CELPIP Writing is the shortest section on the test and the one with the least slack in it. There are two tasks. Task 1 is Writing an Email. Task 2 is Responding to Survey Questions. Each asks for a response of roughly 150 to 200 words, and the two together share 53 minutes: about 27 minutes for Task 1 and about 26 for Task 2.
The two tasks are separately timed, and that is the detail that catches people. When the Task 1 timer runs out the test advances on its own. Whatever is unwritten stays unwritten, and there is no borrowing from Task 2 to finish it. You cannot bank leftover time either. Finishing Task 1 in nineteen minutes does not give Task 2 thirty-four.
Both tasks are typed into a plain box on screen. There is a word counter, and there is no spellchecker and no autocorrect. Anything your phone normally fixes for you will sit on the page exactly as you typed it, which is why the last three minutes of each task matter more than most candidates believe.
Task 1: Writing an Email
Task 1 gives you a short scenario and three bullet points. The scenario tells you who you are writing to and why. A neighbour, a landlord, a manager, a friend, a company's customer service desk. The three bullets tell you exactly what the email has to contain, and they are not suggestions. Coverage of all three is scored directly.
The scenario also fixes the register. An email to a moving company's customer service manager about a cracked cabinet is formal. An email to a friend about visiting your new city is informal. Writing the wrong one is not a style preference, it is a task error, and it is penalised under Task Fulfilment even when the English is otherwise strong.
The scenario deliberately leaves gaps. It might say you hired a company last month without giving the date, the booking number or the amount you paid. You are expected to invent those specifics and then stay consistent with them. Inventing detail is correct. Contradicting a detail the prompt has already given you is not, and it costs more than most people expect.
Task 2: Responding to Survey Questions
Task 2 presents a situation and two options, labelled A and B. A town with a budget to spend, a school deciding between two schedules, an office choosing between two policies. You select one option and argue for it in 150 to 200 words.
It is not an email. It takes no greeting, no Dear Parent Council, no sign-off and no name at the bottom. Adding email furniture out of habit from Task 1 is one of the clearest signals that a candidate has not read the task, and Task Fulfilment absorbs the cost.
Neither option is correct. Your opinion is not scored. What is scored is how completely you develop it, so the right choice is whichever option you can supply a concrete example for in the next ninety seconds, not the one you would pick in real life. Selecting an option and then arguing for the other one, or refusing to select at all, is penalised heavily.
What the 53 minutes actually allows
Twenty-seven minutes sounds generous for 180 words until you break it down. A workable split for Task 1 is four minutes reading and planning, twenty minutes writing, three minutes checking. Task 2 is tighter: three minutes deciding and planning, twenty minutes writing, three minutes checking.
That leaves roughly twenty minutes of actual composition for each task, which is about nine words a minute. That is a slow, deliberate pace, and it is the correct one. The section is not testing typing speed. Candidates who finish Task 1 in twelve minutes almost always finish it thin, because they wrote whatever came first rather than what the bullets asked for.
The other thing the arithmetic tells you is that a rewrite is not affordable. If you reach minute eighteen and realise your email is answering the wrong bullet, there is no time to start again. That is what the four planning minutes are buying: not elegance, but the certainty that you will not have to reverse.
Write your three bullets as three words in the answer box before you start composing, then delete them at the end. The most expensive Task 1 error is an unanswered bullet, and a visible checklist makes it almost impossible.
How the two tasks are scored
Both tasks are assessed against the same four criteria, and each criterion is scored from 1 to 12. Content and Coherence carries 30 per cent, Readability 25 per cent, Task Fulfilment 25 per cent and Vocabulary 20 per cent. Those weights are worth memorising, because they say plainly that ideas and organisation are worth more than word choice.
The weighted total maps to a level: 10 to 12 is Highly Effective, 7 to 9 is Effective, 5 to 6 is Developing, 3 to 4 is Limited, and 1 to 2 is Level M. Most candidates land between 6 and 9, and almost every immigration threshold that matters sits on the boundary between 8 and 9.
Because Vocabulary is the lightest of the four, the common instinct is exactly backwards. Hunting for an impressive synonym while a bullet goes unanswered trades a 20 per cent criterion against a 25 per cent one, and loses. Worse, advanced vocabulary used incorrectly does not score high at all, so the trade can cost marks on both sides.
What the section is not testing
It is not testing knowledge. You are never expected to know anything about municipal budgets, school timetables or moving insurance. Every fact you need is either in the prompt or invented by you, and an invented fact is fine as long as it is specific and consistent.
It is not testing length. Going to 230 words to show range is not rewarded, and the extra fifty words are usually the weakest fifty on the page. Length only affects your score when it starts to affect completeness, relevance or clarity, which it does in both directions.
It is not testing formality for its own sake either. An informal email to a friend written in stiff business English scores worse than the same content written naturally, because the register was wrong. What the section is testing is narrow and consistent: did you do what the task asked, in language a reader can follow without effort.
Decide your Task 2 option within ninety seconds, and pick the one you can give a concrete example for rather than the one you agree with. Your opinion carries no marks; the example carries several.
Practise it
Five questions on the shape of the section and the choices it forces on you.