NCLC 7 course

Research notes

TCF Canada: exact current format and scoring

Contents
  1. 0. The ten facts that matter most
  2. 1. Format of each part
    1. 1.1 Summary table
    2. 1.2 Compréhension orale (CO), 39 questions, 35 min [H]
    3. 1.3 Compréhension écrite (CE), 39 questions, 60 min [H]
    4. 1.4 Expression écrite (EE), 3 tasks, 60 min [H]
    5. 1.5 Expression orale (EO), 3 tasks, 12 min including prep [H]
    6. 1.6 Delivery and administration
    7. 1.7 Format conflicts found
  3. 2. How CO and CE are scored
    1. 2.1 What FEI officially says [H] (manual pp.21–23, 25)
    2. 2.2 The community weighting grid (Variant A)
    3. 2.3 Is the grid plausible? (my check) [L]
  4. 3. How many correct answers reach CO 458 and CE 453
    1. 3.1 The two key numbers
    2. 3.2 Example answer patterns
    3. 3.3 Probabilities under a simple model
    4. 3.4 What this means for effort versus reward [L, inference]
  5. 4. How EE and EO are scored
    1. 4.1 How the mark is produced [H] (manual pp.22–23)
    2. 4.2 Official marking criteria [H] (manual pp.22–23)
    3. 4.3 Marks out of 20, CEFR levels and NCLC [H]
    4. 4.4 What separates 9/20 from 10/20 from 12/20
    5. 4.5 Practical consequences for a December 2026 sitting [H on the facts, L on the advice]
  6. 5. Official sample tests and practice materials
  7. 6. Patterns in the official samples (counted)
  8. 7. Open questions (not resolved)
  9. 8. Sources

Prepared 2026-11-04 for a sitting in late December 2026. Target: NCLC 7 in all four skills, which on TCF Canada means CO at least 458, CE at least 453, EE at least 10/20 and EO at least 10/20.

How to read this file. Every claim has a source link and a confidence tag:

Limits of this research.

0. The ten facts that matter most

  1. Format. There are four compulsory parts, taken in one session lasting 2 h 47 in total [H] (manual p.15; FEI TCF Canada page; Concordia):

    • CO: 39 multiple-choice questions in 35 min.
    • CE: 39 multiple-choice questions in 60 min.
    • EE: 3 tasks in 60 min.
    • EO: 3 tasks in 12 min, including 2 min of preparation.

  2. No penalty for wrong answers. A wrong answer and a blank both score 0, so never leave a blank [H] (manual p.21).

  3. CO and CE are weighted by difficulty, but FEI does not publish the weights. FEI applies a scoring scale ("barème") that weights each correct answer by the difficulty of the question and converts the result to 100–699. It does not publish the weights and does not tell candidates how many answers they got right. It also says your level does not depend only on how many answers you got right [H] (manual pp.22–23).

  4. The community grid. Candidates and prep sites widely use this grid:

    Questions Points each
    Q1–4 3
    Q5–10 9
    Q11–19 15
    Q20–29 21
    Q30–35 26
    Q36–39 33

    The bands add up to exactly 699. I found this exact grid on 8 sites and a variant on 1 more. None of them cites an official source. Wikipedia attributes it to an IRCC page that does not contain it [M that this grid is the community standard; L that it is exact] (section 2.2).

  5. 31 correct is enough under the grid. Under the grid, any 31 correct answers out of 39 score at least 463 (459 under the variant grid). That is above both 458 (CO) and 453 (CE), whichever 31 questions you get right. With 30 correct you can score as little as 437 [L, my calculation] (section 3).

  6. The B2 band is the biggest lever. Q20–29 are worth 210 points, 30% of the total. Each extra 10% of accuracy on that band adds about 21 points. Blind guessing on Q30–39 is worth about 72 points on average [L, my calculation].

  7. How EE and EO are marked. Two graders each give each of the three tasks a CEFR level, so each candidate gets six levels. An unpublished rule turns these six levels into a mark out of 20. B1 covers 6–9 and B2 covers 10–13. NCLC 7 is 10–11, so you need the graders' overall picture of you to be B2 [H] (manual pp.22–23 and 26; IRCC).

  8. EE automatic-fail risks. A writing paper can be rated "A1 non atteint" (A1 not reached) if:

    • any task misses its word range,
    • a task is off-topic,
    • a task is missing, or
    • the handwriting is illegible.

    FEI defines a word as anything between two spaces, and tasks must be done in order. Reproducing existing texts, whether memorised or copied, counts as plagiarism and fraud [H] (manual pp.23 and 38; EE demo).

  9. No re-marking for a December 2026 sitting. Requests to re-mark EE and EO are suspended for sessions from 1 September 2026 for about one year. Concordia expects the service back in autumn 2027. A December 2026 result will therefore be final [H] (FEI evaluation page; AF Delhi; Concordia).

  10. Retake rules.

    • You must wait 20 days between two TCF Canada sittings, according to the current FEI page, Concordia and Alliance Française Canada. The 2024 manual said 30 days.
    • TCF Canada is not modular: you cannot retake only one skill.
    • [M–H] (FEI page via search extract; Concordia; AF Canada; manual p.27).

1. Format of each part

1.1 Summary table

Part Items Time Prep Words What it asks Conf.
CO (listening) 39 multiple-choice questions, 4 options, 1 correct 35 min none — Everyday exchanges and announcements, then radio and TV, then talks on concrete and abstract topics, then any speech at normal speed. Difficulty increases through the test. Each recording is played once, and on paper the question comes after the recording. [H]
CE (reading) 39 multiple-choice questions, 4 options, 1 correct 60 min none — Signs and short messages, then everyday documents, then personal letters, then work and daily-life texts, then opinion articles, then long or specialised texts. Difficulty increases through the test. [H]
EE (writing) 3 tasks 60 min for all three no separate prep; draft paper on paper tests T1 60–120; T2 120–150; T3 120–180 (40–60 + 80–120) T1: a message. T2: an account with opinion. T3: compare two viewpoints and give your own view. [H]
EO (speaking) 3 tasks, face to face, recorded 12 min including prep T2 only: 2 min — T1: guided interview, 2 min. T2: interaction, 2 min prep + 3 min 30. T3: point of view, 4 min 30. [H]

Sources:

1.2 Compréhension orale (CO), 39 questions, 35 min [H]

1.3 Compréhension écrite (CE), 39 questions, 60 min [H]

1.4 Expression écrite (EE), 3 tasks, 60 min [H]

The manual's official task definitions (p.20, paraphrased), with the official demo paper EE00demo2022 for TCF tout public, Québec and Canada (mirror PDF; FEI original):

Task Official definition Words Demo example (paraphrased)
T1 A message that describes, narrates and/or explains, to one or more recipients whose status is given in the instructions. 60–120 You land in France and your suitcase is missing. Fill in the lost-luggage form, describing the case and how you lost it.
T2 An article, letter or note to several recipients giving an account of an experience or a story, with comments, opinions or arguments that serve a goal (for example to complain or to make peace). 120–150 Reply to an email from your friend Charlotte, giving advice and justifying it.
T3 A text (for a newspaper, website, colleague or manager) that compares two viewpoints on a social issue taken from two documents, then gives your own opinion. 120–180 Two short web and book extracts about food: weekly baskets from local farmers, and teaching children to eat well. Write an article for your association's newspaper in two parts. Part 1 presents both opinions in your own words (40–60 words). Part 2 gives your position on the common theme (80–120 words).

Rules printed on the demo paper and in the manual:

What EE assesses (manual p.21):

1.5 Expression orale (EO), 3 tasks, 12 min including prep [H]

The manual's task definitions (pp.19–20):

Task Duration Prep Goal (paraphrased)
T1, guided interview (entretien dirigé) 2 min none Show you can talk with a person you do not know, the examiner.
T2, interaction (exercice en interaction) 3 min 30 of dialogue 2 min Get information in an everyday situation. The instructions set your role and the examiner's role.
T3, expressing a point of view 4 min 30 none Speak spontaneously, continuously and convincingly in answer to a question the examiner chooses.

1.6 Delivery and administration

1.7 Format conflicts found

Claim Source Verdict
TCF Canada has 29 questions per section immiglob blog (seen only in a search summary) Wrong for TCF Canada, which has 39 (FEI, IRCC, centres). 29 questions is the TCF tout public format (CO 29/25 min, CE 29/45 min) (France Compétences RS6460). TCF Québec now also has 29 on paper and 34 on computer (FEI TCF Québec page, search extract), although the 2024 manual still gave 39. Do not use 29-question TCF tout public or Québec mocks as CO/CE timing practice.
"2026 reforms": adaptive listening, new NCLC thresholds, a new writing rubric tcfcanadahub Unsourced and contradicted. IRCC's TCF Canada table, modified 2026-10-16, still shows the old thresholds. The page has internal date errors (dated Feb 2026 but reporting April 2026 results) and describes the "old" test as 29 questions. [L], treat as false.
Going more than 10% over the maximum word count brings "deductions" open-exam-prep Unsourced. The official rule is that any task outside its word range risks an "A1 non atteint" rating.
EE is scored out of 699 GlobalExam blog (seen only in a search summary) Wrong. EE and EO are marked out of 20 (manual p.25).

2. How CO and CE are scored

2.1 What FEI officially says [H] (manual pp.21–23, 25)

2.2 The community weighting grid (Variant A)

Questions Level label Questions Points each Band total Cumulative
Q1–4 A1 4 3 12 12
Q5–10 A2 6 9 54 66
Q11–19 B1 9 15 135 201
Q20–29 B2 10 21 210 411
Q30–35 C1 6 26 156 567
Q36–39 C2 4 33 132 699

How often this grid appears. I found Variant A on 8 separate sites:

  1. luckytcf, which says the grid is not official;
  2. go-tcfcanada;
  3. tcfmentor;
  4. Wikipedia;
  5. tcfcanada.site;
  6. paliertcf;
  7. HiTCF reading guide;
  8. tcfcan.com.

None of the eight cites an FEI document for it.

Variants and conflicts.

Verdict.

2.3 Is the grid plausible? (my check) [L]

Assume you get every question up to a given level right and guess at random (25% chance) on everything harder. Under Variant A the expected scores land inside the official CEFR bands:

Mastered up to Expected score Official band Fits?
A1 ≈184 A1 (100–199) yes
A2 ≈224 A2 (200–299) yes
B1 ≈326 B1 (300–399) yes
B2 ≈483 B2 (400–499) yes
C1 ≈600 C1 (500–599) sits on the C1/C2 boundary

This makes the grid plausible as an approximation, but it is not proof.

3. How many correct answers reach CO 458 and CE 453

All figures in this section are my calculations under the community grid [L]. Check them with full mock tests that use the same grid.

3.1 The two key numbers

3.2 Example answer patterns

Each pattern is scored under both grids.

Pattern Correct Variant A Variant B CO 458? CE 453?
S1. Floor first: all of Q1–29, plus 2 from Q30–35 31 463 459 yes yes
S1b: all of Q1–29, plus 1 from Q30–35 and 1 from Q36–39 31 470 465 yes yes
S2. Solid to B1, strong B2: all of Q1–19, 9 of 10 in Q20–29, 3 from Q30–35 31 468 465 yes yes
S3. Some early slips: 9/10 of Q1–10, 8/9 of Q11–19, 8/10 of Q20–29, 2/6 of Q30–35, 2/4 of Q36–39 29 463 459 yes yes
Fail: all of Q1–29, plus 1 from Q30–35 30 437 432 no no
Near miss: all of Q1–19, 8/10 of Q20–29, 2 from Q30–35, 1 from Q36–39 30 454 450 no passes CE only under Variant A

What to aim for, in plain terms.

  1. S1: get every A1–B2 question right (Q1–29) and pick up 2 more from Q30–39.
  2. S2: if you lose one B2 question, win back 3 C1-level questions.
  3. S3: misses on cheap early questions are affordable, but only if you then get 4 of the last 10 right.

3.3 Probabilities under a simple model

The model treats each answer as an independent event with a fixed chance of being right in each band, using Variant A.

How the chance of passing changes with B2 and C1 accuracy. Each cell shows the chance of reaching CO 458 / CE 453, with A1–A2 at 97%, B1 at 95% and C2 at 30%:

B2 accuracy \ C1 accuracy 25% 40% 55% 70%
60% 13/15% 25/28% 40/43% 57/60%
70% 22/25% 38/42% 55/59% 72/75%
80% 35/39% 53/57% 71/74% 84/86%
90% 51/55% 70/73% 84/86% 94/95%
100% 70/74% 85/88% 95/96% 99/99%

Expected points gained from a 10-percentage-point rise in accuracy, by band:

Band Gain
A1 +1.2
A2 +5.4
B1 +13.5
B2 +21.0
C1 +15.6
C2 +13.2

3.4 What this means for effort versus reward [L, inference]

4. How EE and EO are scored

4.1 How the mark is produced [H] (manual pp.22–23)

Writing (EE).

Speaking (EO).

Other official points.

4.2 Official marking criteria [H] (manual pp.22–23)

Criterion type Writing (EE) Speaking (EO)
Linguistic range of vocabulary, grammatical accuracy, spelling, how developed the sentences are range and control of vocabulary, grammatical accuracy, ease, pronunciation, overall fluency
Pragmatic coherence and cohesion, development of the topic interaction, structure of what you say, coherence and cohesion, development of the topic
Sociolinguistic fit with the communication situation (register, recipient) fit with the communication situation

FEI publishes no weighting for these criteria and no per-task weighting. Claims that "Task 3 counts more", or that each task is marked out of 20 and averaged (tcfca.com, seen only in a search summary), have no official source [L].

4.3 Marks out of 20, CEFR levels and NCLC [H]

Mark /20 CEFR level NCLC (TCF Canada)
0 A1 non atteint —
1 A1 —
2–5 A2 4–5 gives NCLC 4
6–9 B1 6 gives NCLC 5; 7–9 gives NCLC 6
10–13 B2 10–11 gives NCLC 7; 12–13 gives NCLC 8
14–17 C1 14–15 gives NCLC 9; 16 and above gives NCLC 10
18–20 C2 NCLC 10

Sources: manual pp.15–16 and 26; France Compétences RS6460; IRCC.

4.4 What separates 9/20 from 10/20 from 12/20

The scores themselves [H]:

Because the rule from 6 levels to a mark is not published, nobody can say exactly which combination of task levels gives 10 rather than 9. The only rule we can safely draw [L, inference]: one B2 task surrounded by B1 tasks is unlikely to be enough. You should aim for B2 on Tasks 2 and 3, the tasks built to show B2-type skills, and solid B1 or better on Task 1. A 12 needs B2 ratings that are consistent across all three tasks and both graders.

Floor rules [H]. For EE, these come before any level judgement: a missed word range, an off-topic answer, a missing task or illegible writing can sink the whole paper to "A1 non atteint" (manual p.23; EE demo). Memorised model texts are risky, because FEI lists reproducing existing documents, memorised or copied, as plagiarism and fraud (manual p.38). Adapt any templates rather than reciting them [H on the rule; how strictly graders apply it to stock phrases is unknown].

B1 versus B2 by task. The table below combines FEI's task definitions with the official CEFR descriptors. The CEFR sources are the self-assessment grid, read in full [H], and the Council of Europe's Table 3 on the quality of spoken language, which I could not fetch in full [M]. FEI's internal marking grid is not public, so treat this as the best available proxy [M].

Task What a 6–9 (B1) performance looks like What a 10–13 (B2) performance adds Moving from 10 towards 12 (more consistent B2)
EE T1, message, 60–120 words Simple connected text on a familiar topic. Covers the points asked. Basic linking (et, mais, parce que). Routine grammar mostly correct. Every requested point developed with precise detail. Past tenses used correctly (passé composé vs imparfait). Relative clauses. Right register for the named recipient. Few errors; varied vocabulary instead of repeated basic words.
EE T2, account plus opinion, 120–150 words Tells an experience and states an opinion simply. Clear, detailed text that gives reasons for or against a view and shows why the experience matters. The argument serves the stated goal (advise, complain, persuade). Paragraphs with clear structure; precise connectors; register suited to several readers.
EE T3, two viewpoints plus opinion, 120–180 words Lifts phrases from the documents; asserts an opinion with basic reasons. Restates both views in your own words (40–60 words) and contrasts them (alors que, tandis que, en revanche). Argues a position with at least 2 developed reasons and an example (80–120 words). This is CEFR B2's "essay giving reasons for or against". Some C1-style features: highlighting what matters most, a concession (même si, bien que + subjunctive), a conclusion that ties back.
EO T1, guided interview, 2 min Connected simple sentences about yourself, family and work; answers the questions. Longer spontaneous answers with reasons and plans, a fairly even pace, and comfortable handling of follow-up questions. Fewer pauses and wider vocabulary.
EO T2, interaction, 2 min prep + 3 min 30 Asks basic questions to get the information. Starts, keeps up and closes the exchange. Asks varied, well-formed questions in the right register (vous). Follows up on answers and compares options. Natural reactions; reformulates when something is unclear.
EO T3, point of view, 4 min 30 Gives an opinion with brief reasons. Explains a view on a current issue with pros and cons of the options. Keeps a structured argument going (position, reasons, examples, conclusion). Defends it when the examiner challenges. A relatively high degree of grammatical control; errors do not cause misunderstanding; a range of linking devices.

4.5 Practical consequences for a December 2026 sitting [H on the facts, L on the advice]

5. Official sample tests and practice materials

Material Where What it contains Conf.
FEI Manuel du candidat, Version J, Sept 2024 Mirror (FEI copy at france-education-international.fr) Format and scoring rules; an NCLC table; 14 example items with answers: 7 CO (2 picture, 2 question–reply, 2 dialogue, 1 monologue), 3 grammar-structure items (that part is not in TCF Canada) and 4 CE (notice, administrative message, personal message, informative article); links to the EE and EO demos; the re-marking form. [H]
FEI "Exemples d'épreuves du TCF" page france-education-international.fr/test/exemples-epreuves-tcf Example questions numbered by section (CO "Exemple 1" onwards, CE "Exemple 11" onwards), matching the manual (search extract). [M]
Official EE demo EE00demo2022 (TCF tout public, Québec and Canada) FEI; mirror A complete 3-task paper with instructions, word ranges, the word-counting rule, "A1 non atteint" rules, and the two-part structure for T3 (section 1.4). [H]
Official EO demo (TCF tout public, Québec and Canada) FEI Exists and is linked from the manual (p.60), but could not be opened (bot block). [M]
Official EO demo (TCF IRN, different timings) FEI 5 subjects per task, examiner chooses one. Task 2 example: a role-play with a real-estate agent (search extract). [M]
TV5MONDE TCF practice, built with FEI apprendre.tv5monde.com/fr/tcf; FEI press release; TV5MONDE support page Over 600 questions written by FEI teams from TV5MONDE video clips. 40-question practice series with no time limit. Simulator: 80 random questions in 90 min, listening played once. Covers CO, grammar structures and CE only; no writing or speaking; follows the TCF tout public format, not TCF Canada's 39 + 39. [H]
RFI "Le français facile" francaisfacile.rfi.fr (recommended in manual p.39) Listening practice for the TCF (site could not be fetched). [M]
FEI resource sheet listing prep books Mentioned in manual p.39 Not retrieved. [L]

6. Patterns in the official samples (counted)

7. Open questions (not resolved)

  1. The exact CO/CE weights. FEI does not publish them, and the grid is a community estimate. It cannot be checked, because FEI never releases raw counts.
  2. Extra unscored items on computer? TCF tout public on computer has 15 extra unscored items (91 instead of 76, manual p.10), and TCF Québec on computer has 34 instead of 29. I found no source saying whether computer-based TCF Canada CO/CE contain extra unscored items.
  3. Going back in computer-based CO. Can you return to earlier CO items, or does each item move on by itself? Is there a visible timer for each question? No official confirmation.
  4. The EE computer screen. Does it show a word counter, and is the keyboard AZERTY or QWERTY? This varies by centre and has not been confirmed.
  5. Choice of EO subjects. Does the TCF Canada examiner impose one subject per task, as in the IRN demo, or can the candidate choose?
  6. The rule that turns 6 levels into a mark out of 20, and whether tasks are weighted. Not published.
  7. Whether FIDELIA automated marking applies to TCF Canada computer-based EE.
  8. The current version of the manual. The FEI site could not be reached directly, so there may be a version newer than J (September 2024). The waiting period (20 vs 30 days) and the TCF Québec item counts already differ from Version J.
  9. The current official EO demo for TCF tout public, Québec and Canada. It could not be opened, so its example subjects are not described here.
  10. Combining sittings for IRCC. Whether IRCC accepts results from two different TCF Canada sittings combined is not covered here. TCF Canada itself offers no single-skill retake.

8. Sources

Official: FEI, IRCC and Council of Europe

Test centres

Prep and third-party sites (used to count claims, not as authority)