How is the TCF Canada speaking test scored?
Checked November 1, 2026
The examiner marks you once on the day. The interview is recorded and a second marker trained by France Éducation international (FEI) marks it separately, without seeing the first mark. Each gives a level to each of the 3 tasks, and a calculation rule over these 6 levels gives the mark out of 20. NCLC 7 needs 10 out of 20.
Two independent marks
- The examiner, trained and certified by FEI, runs the interview, records it and gives a first level to each task. At the end they check in front of you that the recording worked.
- A marker at FEI listens to the recording and marks it too, double-blind.
- A calculation rule takes the 6 levels (3 tasks, 2 markers) and gives a mark out of 20 and a final level.
- If the two marks are far apart, a third marking takes place.
FEI does not publish how much each task weighs. Task 1 lasts 2 minutes and task 3 lasts 4 min 30, but nothing says the longer one counts more.
The criteria
- Linguistic, range and control of vocabulary, grammar, ease, pronunciation, fluency.
- Pragmatic, interaction, structure, coherence, development of the topic.
- Sociolinguistic, fit to the situation, for example register in the role play.
Reciting a text learned by heart could get the test marked “A1 non atteint”.
From mark to NCLC
| Mark out of 20 | CEFR level (FEI) | NCLC (IRCC) |
|---|---|---|
| 18–20 | C2 | 10 and above |
| 16–17 | C1 | 10 and above |
| 14–15 | C1 | 9 |
| 12–13 | B2 | 8 |
| 10–11 | B2 | 7 |
| 7–9 | B1 | 6 |
| 6 | B1 | 5 |
| 4–5 | A2 | 4 |
What you won’t get
You get a single mark out of 20 for speaking, not one per task. Recordings are not commented and FEI gives no improvement tips. You only get the mark and the level. Re-marking of speaking is suspended for sessions from 1 September 2026, for about a year (details). A disappointing speaking mark cannot be reviewed, only a new sitting changes it.
Writing gets the same double marking, and it is the one productive skill you can get marked before the exam, here, as an estimate.