How the TEF Canada is graded: the marking criteria for each section

The TEF Canada is graded against criteria the candidate never sees. You get four numbers back and no explanation of what earned them, which leaves most people preparing for the exam as if effort and word count were the currency. They are not. Each section rewards specific things, and knowing what the marker is looking for is worth more than another hundred flashcards.

This is a tour of what actually gets rewarded in each of the four sections, where anglophone candidates quietly lose marks, and why the famous writing “minimum” is a floor and not a target. The exact criteria and weightings belong to the board, so confirm anything you plan around against lefrancaisdesaffaires.fr; the shape below is stable, the fine print is theirs.

How are listening and reading graded?

Compréhension orale and compréhension écrite are multiple choice and machine-marked, reported on their own scales - listening out of 360 and reading out of 300. Here the “criteria” are simply whether you selected the right option, but two things are worth knowing. Items vary in difficulty, so the sections are not a flat count, and you should confirm whether wrong answers carry any penalty for the current version, because that changes whether blind guessing helps or hurts. When in doubt, treat a wrong answer as a real risk rather than a free lottery ticket.

The practical lesson: comprehension is graded on accuracy under time, not on how much you understood in general. Finishing the section matters, because unreached questions score nothing and the later items are often where the higher bands are decided. Read your converted level, not your tally, off the TEF Canada score chart.

What does the examiner reward in the writing?

Expression écrite is marked by an examiner against a grid, reported out of 450, and this is where knowing the criteria changes your result. Across the two writing tasks, markers are weighing a consistent set of things:

The single most common self-inflicted wound is treating the word count as a target. There is a minimum, not a maximum, and writing far past it to pad length usually adds errors and repetition, which the grid punishes. Hit the floor, then spend the rest of your effort on range and accuracy, not volume.

What does the examiner reward in the speaking?

Expression orale is also examiner-marked and reported out of 450, and it is graded on how you perform two different communicative jobs, not on having no accent. What lifts a speaking band:

• Fulfilling the task’s purpose - getting the information, or genuinely persuading - rather than just talking.
• Fluency that survives a slow start, with self-correction that sounds like control, not panic.
• Range: varied structures and precise words, not a rehearsed monologue bolted onto every prompt.
• Interaction and register appropriate to who you are speaking to.

Anglophones lose marks here less on pronunciation than on structure and register: answering a persuasion task with a flat description, or staying stiffly formal where a natural exchange is expected. The examiner is scoring communication, and a fluent answer to the wrong task still misses.

Why is the writing minimum a floor and not a target?

The TEF Canada writing sets a minimum length, and candidates read it as the length they should hit. That misreads what the grid rewards. A minimum is the point below which a text is considered too thin to assess fairly; it is not the point at which you have earned full marks. Everything above the minimum is judged on quality, and quality does not scale with quantity.

In practice, writing well past the floor tends to lower a score rather than raise it, because a longer text gives you more chances to repeat structures, drift off task, and make errors, all of which the grid penalises. A tightly controlled text a little over the minimum, with varied language and no wasted sentences, reads better to an examiner than a sprawling one that buries a good idea under padding. Hit the floor with room to spare, then spend the remaining time editing for accuracy and range, not adding paragraphs. The word count is a gate you walk through, not a scoreboard you run up.

How do the four scores become an immigration level?

They do not average, and there is no overall pass. Each section converts to an NCLC level on its own, and the level you can claim for a program is your weakest of the four. This is why a candidate with three strong sections and one weak one is graded by the weak one. Put your results through the score converter to see each section’s level, and check the target against NCLC 7, the usual federal benchmark.

So the marking model has a clear message for how you prepare: your worst section is your grade, and each section is scored on task-fit and range, not on effort or length. Spend your time on the section that is furthest from the level you need, and get your writing and speaking marked against real criteria, because those are the two sections where you cannot grade yourself and where the gap between what you produced and what the grid wanted is invisible without an examiner’s eye. The board sets the exact criteria and can revise them, so verify the current grids against lefrancaisdesaffaires.fr before you build a strategy on any one detail.