The most common CELPIP Speaking mistakes

1. Stopping before the answer is finished

64% of answers scored 0 or 1 out of 3 on “concludes with an appropriate closing remark” — the most-missed item on the whole checklist. In the marker’s own notes, 11.9% of every problem it wrote down (about one in eight) is an answer that was cut off, ended mid-sentence, ran out of time or never reached a conclusion, and 42% of answers got at least one such note. This is a timing habit, not an English problem: the language up to that point is often fine.

Fix · Plan the last sentence first. When about ten seconds are left, say it — “So that’s why I’d go with the second option” — even if you have more to add.

2. Saying the same idea twice

52% of answers scored 0–1 on “avoids repeating ideas”. Repetition usually fills time after the speaker runs out of planned points: the first reason comes back in different words.

Fix · Prepare two different points and one example for each during the preparation time; when they are used, close instead of circling back.

3. A narrow range of words

42% of answers scored 0–1 on “accurately uses a wide variety of words”, and Vocabulary is the criterion that most often comes out lowest of the four in seven of the eight tasks (table below). The pattern is the same everyday words carrying the whole answer: good, nice, very, a lot of.

Fix · For each task type, keep five precise words you can use naturally (for a scene: crowded, cluttered, reaching for, about to, in the foreground).

4. Points without detail or sequence

41% of answers scored 0–1 on “uses time indicators and sequencing to add depth”. The point is stated but not developed: no first, after that, by the time, no concrete example.

Fix · Give every point one specific: a time, a place, a number or a short example.

5. Hesitation, fillers and long pauses

34% of answers scored 0–1 on “speaks clearly with minimal pausing and hesitation”. Filler words alone are 14.1% of all marker notes and appear in the notes of 49% of answers.

Fix · Replace “um” with a real phrase that buys time: “Well, the way I see it…”, “What I’d suggest is…”.

6. Ideas in an order that is hard to follow

27% of answers scored 0–1 on “ideas well organized and easy to follow”. It rises sharply in the picture tasks that need a plan — Task 4 (predictions) and Task 5 (comparing) — and is lowest in Task 3 and Task 6.

Fix · Use one fixed frame per task: opening, point one, point two, closing. Say the frame words out loud.

7. Uneven pace

26% of answers scored 0–1 on “appropriate rhythm and pace”. Our marker judges this from timing signals (speaking rate, fillers, long pauses) — it cannot hear accent or intonation, and never claims to.

Fix · Record yourself against the real timer and listen once: a steady pace beats a fast one.

8. Too short for the time allowed

Only 10% of answers scored 0–1 on “long enough for the time available”. Most learners do fill the time — the problem is how the time ends (mistake 1), not how much of it is used.

Fix · If you finish early, add one example rather than repeating a point.

9. Not answering the question asked

Rare: 4% of answers scored 0–1 on “addresses the task”. It is highest in Task 5 (14%), where the prompt has two parts: your choice and why it beats the other option.

Fix · Before you speak, name every part of the prompt to yourself; tick each one off as you say it.

Task by task

The same checklist, split by the eight Speaking tasks. “Most-missed item” is the checklist item the most answers scored 0–1 on; “Unfinished” is the share of that task’s marker notes describing an answer that was cut off or had no conclusion.

• Task 1 (Giving Advice) is the one task where repetition, not the ending, is the most-missed item (56%).
• Tasks 3, 4, 5 and 8 — the scene, prediction, comparison and unusual-picture tasks — have the weakest endings: 79%–85% of answers miss a closing line.
• Task 4 (Making Predictions) also has the thinnest detail (57% scored 0–1) and, level with Task 5, the most disorganised answers (44%).
• Task 7 (Expressing Opinions) is the only task where Listenability, not Vocabulary, is most often the lowest criterion, with the most hesitation (45% scored 0–1 on fluency).

Frequently asked questions