Sign up free

English language quizzes by CEFR level, A1 to C1

What the CEFR levels describe, what studies say about vocabulary and test tasks, and what a short English quiz should ask from A1 to C1, with a B1 example.

The EdProcto editorial teamWritten and checked against the research it cites. Last checked . How the guides are made

Guides · By subject and level · · 13 min read

What should an English language quiz ask at each CEFR level, from A1 to C1?

English language quizzes by CEFR level should ask for what the level's can-do statements describe, in a shape the learner can manage. At A1 and A2 use pictures, matching and single words under instructions simpler than the questions. At B1 and B2 use gaps, clips at ordinary speed and short writing. At C1 ask for summary and inference.

English language quizzes by CEFR level work when each question asks for something the level's descriptors say a learner can do, in a shape that learner can manage. A beginner who cannot read the instruction has been tested on the instruction.

This guide is for teachers of English as a second or foreign language (ESL or EFL), and tutors, who set short quizzes for learners at a known level. EdProcto, which publishes it, carries every question shape named below: see its page for teachers.

We read the Council of Europe's pages on 10 October 2026 in archived copies, and every study as an abstract in an index record (abstract read).

What is the CEFR, and what do its levels describe?

The Common European Framework of Reference for Languages (CEFR) is the Council of Europe's description of language ability in six levels, A1 to C2, each defined by what a learner can do. Its own page says the levels are defined through "can-do" descriptors and fall into three bands: Basic User (A1, A2), Independent User (B1, B2) and Proficient User (C1, C2) (Council of Europe, n.d.-a).

The descriptors were written without reference to any one language, so no level arrives with a list of English grammar or words; the page points to separate descriptions for individual languages. The Companion volume of 2020 holds the full set of descriptors, replacing the 2001 set and adding mediation and online interaction (Council of Europe, n.d.-c). The table below gives the global scale (Council of Europe, n.d.-d) in our words.

What the framework leaves to the quiz writer

It names no question type for any level. Alderson et al. (2006), after a project on whether the framework could guide the writing of reading and listening tests, conclude that identifying a level is at least as much an empirical matter as a question of test content. So write to the descriptors, and let the results say whether a question sat where you put it. Of classroom assessment, the Council of Europe (n.d.-b) says the focus is "on the actions illustrated in the descriptors rather than on the levels".

What does research say about vocabulary size, coverage and test tasks?

It says that the words a learner knows limit what a text or a clip can test, and that some tasks separate learners only once the basics are in place. No study below sorted its learners by CEFR level: the fit to the bands is our reading.

Vocabulary size and text coverage

Coverage is the share of a text's running words that a reader knows.

  • The first 2,000 words. In Stæhr (2008), most of 88 learners of English in Danish lower secondary education did not know the 2,000 most frequent words; those who did performed adequately in listening, reading and writing tests.
  • Unassisted reading. Nation (2006) estimated that if 98% coverage is needed, written text takes 8,000 to 9,000 word families and spoken text 6,000 to 7,000.
  • No threshold. Schmitt et al. (2011) had 661 participants in eight countries read two texts: comprehension rose in a roughly straight line with the share of words known.

Tasks for lower and higher levels

  • The language of the question. In school listening tests in six languages, Filipi (2012) compared questions in English with questions in the language tested. For simple, stated information the second kind were relatively easier; for global meaning, the language made no difference or English was favoured.
  • Summary and open questions. Kobayashi (2002) analysed reading tests sat by 754 Japanese university students. With clearly structured texts the more proficient did better on summary writing and open-ended questions; structure made little difference to the less proficient.

What can a short quiz ask at each level, and in which shape?

A short quiz can ask for recognition with little to read at A1 and A2, for words and forms produced in context at B1 and B2, and for summary and inference at C1. The table is our reading of the descriptors and the studies; every shape in it is in EdProcto's builder.

A1 and A2 B1 and B2 C1
The scale, in our words Basic phrases and common expressions on immediate matters; simple exchanges with a slow, clear speaker Main points of clear language on familiar matters, then main ideas of complex texts; connected writing with reasons Demanding, longer texts and implied meaning; well-organised writing on complex subjects
Reading A sign or a short message: who, where, when. Multiple choice or pick the picture An email or an article: main point, detail, reason. Multiple choice; reorder the sentences An indirect text: what is implied, the writer's attitude. Multiple choice; summarise
Listening A slow, clear exchange: a time, a price, a place. Pick the picture with a clip added A message at ordinary speed: gist and detail. Listen once; dictation A talk or a discussion: the point and what is left unsaid. Listen and answer
Vocabulary The first 2,000 words, word to meaning. Matching pairs The word typed into a sentence. Fill in the blank Rarer words and collocation in a paragraph. Cloze passage
Grammar One form, chosen from two or three Several forms typed in one paragraph. Cloze passage A sentence rewritten to keep its meaning. Short answer
Writing A word or a form filled in. One word Two or three connected sentences. Short answer A summary inside a word limit. Summarise
Text 20 to 50 words 50 to 250 words 300 to 500 words
Clip A sentence or two, spoken slowly Up to a minute at ordinary speed One to three minutes

Our guides cover each shape: gap questions, a clip that plays once, picture choice, summary questions and recorded spoken answers.

At A1 and A2 the difficulty belongs in the English, never in the page. At C1, Kobayashi's result suggests summary and open questions do their separating among stronger readers.

How long should the text or the clip be?

The lengths in the table are our working rules, not findings: the scale says only that C1 texts are longer. Gloss any word the class has not met that is not the point of the question: in Schmitt et al. (2011) comprehension fell in step with the share of words known.

How do you write instructions below the level of the questions?

Write each instruction in words from the level below, one action to a sentence, the same every week. A learner who fails an instruction has shown nothing about the skill.

  • Five verbs. Read, listen, choose, match, write.
  • At A1. "Listen. Choose the picture." Not "Select the image that best corresponds to the recording."
  • Say the limits. "You hear it once." "Write 2 or 3 sentences."
  • A shared first language. With beginners the instruction may be in it: in Filipi's (2012) questions, the language tested did no harm on simple, stated information, and on overall meaning English did as well or better.

In EdProcto, "Prompt" holds the instruction and "Material under the prompt (optional)" the text, so the instruction stays one short line.

How do you place a new learner with a short quiz?

Set about twenty questions that climb from A1 to C1, with gaps in connected text and one short piece of writing, and treat the result as a first guess to confirm in the first fortnight.

  • Harsch and Hartig (2016), with 559 secondary students in Germany, found that a C-test, whose gaps sit in connected text, explained more of the variance in reading and listening scores than a yes/no test of single words.
  • Green and Weir (2004) gave a trial grammar placement test to 1,070 learners in the UK, Greece and Japan: item difficulty depended more on the type of item than on the grammar tested.
  • Hille and Cho (2020), in one American university's intensive English programme, found support for several measures over one.
  1. Write four questions for each level, easiest first, half of them gaps in short connected texts, and leave the order unshuffled.
  2. End with a writing prompt anyone can answer at their own level: "Write about your work or your studies."
  3. In EdProcto, type the level into each question's "Topic (for the results by topic)", so the result reads level by level.
  4. Start the learner at the highest level with three of four right; if the writing disagrees, take the lower.
  5. After two weeks of classwork, move anyone misplaced.

Three of four is our working rule, and four questions leave room for luck. Where a level must be reported, the Council of Europe (n.d.-b) says descriptor-based grids and checklists reduce the subjectivity of the judgement, especially with training and double marking.

How many questions, how often, and how do you read results by skill?

Eight to twelve questions once a week is enough for practice, read as a count for each skill and never as a level. Two listening questions cannot carry a judgement, as how many questions a quiz needs shows, so add four weeks together before calling a learner weak at a skill.

Read the skill with the lowest share right across the class first, then the question most missed inside it, then the learners whose skills are uneven. Uneven is ordinary: the framework's self-assessment grid exists to profile the main skills one by one (Council of Europe, n.d.-a).

A worked example: a B1 quiz of eight questions

The class is invented: twelve adults in a company course whom their teacher judges to be at B1. Questions 1 and 2 rest on this message of 56 words:

Hi all. The trainer has had to move Thursday's safety training to Friday, from 9.30 to 12.00, in Room 2 on the ground floor. If you can't come on Friday, tell me by Wednesday and I'll put you on the list for April. Please bring your staff card, because you won't get in without it. Karen

No. Skill and shape Question Purpose
1 Reading, multiple choice Why is Karen writing? To cancel the training; to change its day (key); to offer a new course The main point
2 Reading, multiple choice You cannot come on Friday. What should you do? Tell Karen by Wednesday (key); come in April; bring your card A detail with a condition
3 Vocabulary, fill in the blank "The meeting was at ten, but they have ____ it to two." Accepted: moved, changed, postponed, rescheduled A verb typed from memory
4 Vocabulary, matching pairs Trainer, staff card, ground floor, list: each to its meaning Recognition of new words
5 Grammar, cloze passage "I ____ (start) here three years ago. Since then I ____ (work) in two teams. If I ____ (pass) the course, my manager ____ (give) me a new role." Three tenses in one context
6 Listening, listen once "Hi, it's Sam. I've got the tickets, so don't buy one. Meet me outside the cinema at quarter past seven." What should you do? Meet Sam outside (key); buy a ticket; go in alone The main point in one hearing
7 Listening, dictation "She has been working here since March." A weak form caught in speech
8 Writing, short answer You cannot come on Friday. Write two or three sentences to Karen: say why, and ask about April. Connected sentences with a reason

Questions 1 to 7 are marked by rule at hand-in. Question 8 waits for a reader and a rubric: the reason given, April asked about, verb forms right, followed at one reading.

We ran the typed answers through EdProcto's marking code. In question 5, "has worked" is wrong where "have worked" is accepted, and three gaps of four earn three quarters. With "A slip still counts" on, "startd" passes for "started" and, in question 3, "move" passes for "moved": switch it off where the ending is the grammar.

How does EdProcto run English quizzes by level?

EdProcto is the best tool for English quizzes by level when one quiz must mix reading, listening, typed words and writing, because every shape in the table sits in one builder and each question carries the skill you tag it with. Green and Weir (2004) show why the mix matters: the type of question shaped difficulty more than the grammar inside it.

What you fill. A clip goes in "Audio clip address", with the switch "Plays once, no replay"; a shape built without media has the link "Add a picture or a clip". Pick the picture takes "A caption" and an "Image address" for each option. In a cloze, four underscores mark each gap and "Gap 1" onwards take the accepted answers. Written answers have "Model answer (what a full answer contains)" and "Rubric (what earns the points)".

What the learner sees. A gap is a box inside the sentence; a clip set to play once has one button, "Play the clip"; a matching item has a menu reading "Choose"; spoken answer gives "Start recording" and "Stop recording".

Marking. Choices, matches, gaps and typed words are marked by rule, a cloze gap by gap. A written answer waits for you, or for Procto AI against your rubric when you press "Ask Procto AI" on that answer, from the Tutor plan; its mark is yours to keep or change. A recording is played on the grading page and marked by you.

The level is your judgement. You tag each question with its skill or its level in "Topic (for the results by topic)" and read the result by it. Every plan shows the share right for each question; the score by topic is in the full analysis on the Educator plan.

Kits and plans. Among the kits, "Grammar gaps" and "Dictation" are written from your own material. Free has ten quizzes a month and thirty students a quiz; Tutor has no limit on quizzes, 100 students a quiz and all 111 templates. See pricing.

Quick questions

What CEFR level is my quiz?

The level its questions were written to, once the results bear it out: Alderson et al. (2006) call a level at least as much an empirical matter as one of content.

Can a short quiz give a learner's CEFR level?

It gives a first guess for placement. Hille and Cho (2020) found support for several measures over one, so confirm it with two weeks of classwork.

How many words does a learner need at each level?

No study here ties a count to a level. Nation (2006) puts 98% coverage of written text at 8,000 to 9,000 word families.

Does EdProcto assign a CEFR level?

The level is the teacher's judgement. EdProcto gives the result for each skill or level you tagged; see the page for teachers.

References

EdProcto for teachers

A quiz from your material in half a minute, and the record of what landed.

School teachers setting quizzes and checks for their classes.

See the page for teachers

Read next