CELPIP® Writing: the mistakes to check for in your own draft

How CELPIP Writing is judged, and what a list of mistakes can do

The Writing component of the CELPIP General test takes 53 minutes and has 2 tasks, as the test provider's test format page sets out. Each response is judged as a whole. The dimension names come from the test provider's Performance Standards for Writing and Speaking: Content/Coherence, Vocabulary, Readability and Task Fulfillment.

The names are published, but the level-by-level descriptors behind them are not. That affects how you should use this article. The mistakes below are not the test provider's criteria or sub-factors, and fixing them does not add up to a level. They are things to check in your own draft, grouped by the dimension each one most clearly touches. A draft without them is easier for any reader to follow and stays closer to what the task asks. celpip9 has not yet measured which of these mistakes are most common or which cost the most, so they are not ranked.

Content/Coherence: points left without support

The mistake to look for here is a point made with no example or detail to support it. A draft can state a reason, a preference or a request clearly and still leave the reader wondering why, or what it would mean in practice. If every point is a bare statement, the response reads like a list of claims, not an explanation.

To check, go through the draft point by point. For each point, find the sentence that supports it. That might be a reason, a detail from the situation the task describes, or an example from your own experience. If you cannot find one, add it or cut the point. While you are at it, check that the points come in an order a reader can follow, so each one leads to the next instead of appearing out of nowhere.

Readability: linking words and run-on sentences

Readability is about how easily a reader moves through your text. Linking words can go wrong in more than one way. They can be missing, which leaves the reader to guess how sentences relate. They can be repeated, so the same connector carries every relationship in the text. Or they can be misused, signalling a contrast, a cause or an addition that the sentences do not contain.

Run-on sentences are the other thing to look for: complete thoughts strung together with commas or with no punctuation at all, so the reader loses track of where an idea ends. To check, read each long sentence slowly and ask whether it holds several complete thoughts. If it does, decide whether they belong in separate sentences or need a linking word that says how they connect. Then read only the joins between sentences and between paragraphs, and ask whether each linking word names the relationship that is really there.

Vocabulary: the same word where a more precise one fits

The mistake here is reusing a word, often a general one, where a more precise word would say what you mean. Repetition is not always wrong, and a key term from the task may need to come up again. The problem is a general word standing in for a precise one over and over, so the reader gets less information than you had in mind.

To check, scan the draft for words that keep coming back and for vague words that could describe almost anything. For each one, ask what you mean in that sentence and whether you know a word that says it more exactly. Choose words you are confident using. A less familiar word used with the wrong meaning or in the wrong form can create a new problem while fixing the old one.

Task Fulfillment: tone, length and staying on task

Task Fulfillment is about whether the response does what the task asked, for the reader the task describes. Tone is part of that. A message to a manager written the way you would write to a close friend, or the reverse, does not suit its reader even if every sentence is correct. Before rereading, remind yourself who the reader is and what relationship the task sets up, then read the draft as that person would.

Length is the next check. Each task states the length it asks for, and a response outside that length does not do what was asked. The test shows a live word count while you write, so this check is quick. The test drive lets you try that word count and the test's other mechanics before test day, untimed and unscored.

The last check is drift. A response can start on the task and slowly turn into an answer to a nearby question, or leave out a point the task lists. The guide to Writing Task 1 and the guide to Writing Task 2 explain what each task asks, and that is what your draft has to meet here.

How to check your own draft in the time you have

Checking only works if you leave time for it. Before you start writing, decide that the end of each task is for rereading, and stop drafting when you get there. How to divide the clock between planning, writing and checking is covered in using the clock in CELPIP Writing and Speaking.

Reread in separate passes, each looking at a different dimension, instead of trying to catch everything together. When you are hunting for run-on sentences, it is easy to miss a point the task asked for, and the other way round. A practical order is to start with Task Fulfillment. Open the task again and tick off every point it lists against your draft, then confirm the tone and the word count. After that, read for Content/Coherence, then Readability, then Vocabulary.

The passes get faster with practice. Build them into every Writing response you practise, under the clock, so the routine is automatic on test day.

Fix it in your own words, and get feedback on your own sentences

When a check turns something up, fix it in your own words. Swapping in set phrases, or rebuilding the draft around a structure learned by heart, does not fix anything. The test provider warns that responses that are not original can put a result at risk, for example by having it cancelled. That is why celpip9 publishes no templates, openers or phrases to learn, as the guide and the methodology explain.

Feedback helps most when it points at specific sentences in your own draft, such as a point with no support, an unclear join or a vague word. A teacher or study partner who marks exact places in your text gives you something to fix. A general impression does not.

celpip9 does not score Writing yet. Your responses are kept in your report but not scored, as the status page shows. The scoring now being built will judge a response against the published dimension names in independent passes and report the middle result as a range. Like every level celpip9 shows, it will be an estimate on the CLB scale, not a CELPIP score or result, and it cannot promise a level on test day.