In short
- The quick method takes ten minutes: one text, an unknown-word count, five questions, then a text one level up or down.
- The careful method adds a vocabulary size test and a levelled reading test, then checks both against two weeks of classroom reading.
- No single test is clean. Comprehension rises steadily with words known, so a pupil near a boundary may land on different sides on different days.
- Record what you saw, not a label alone: 'reads A2 texts with ease; B1 with support', with the date and the method.
You can find a pupil's reading level in ten minutes or in two lessons. The ten-minute version is good enough to form groups. The two-lesson version is for decisions that are hard to undo, such as streaming or an exam entry. This article gives both, explains what each measures and what it misses, and includes a short A2 text with five questions you can use tomorrow. It ends with young beginners, who cannot yet be tested on paper.
The short answer: two methods
The quick method: guess a level, give the pupil a 100-word text at that level, and ask them to underline every word they do not know. Then ask five questions about the text. Two or fewer unknown words and four right answers means try the level above. More than five unknown words or two or fewer right answers means try the level below. The careful method: a vocabulary size test, a reading test with one short text per CEFR level, and a check against what the pupil reads in class over two weeks.
The cut-offs in the quick method are based on coverage research. Hu and Nation (2000) gave 66 adult learners a story with 80%, 90%, 95% or 100% of its words familiar. At 95%, most still did not reach adequate comprehension. The authors put unassisted reading at about 98% of words known, or two unknown words in a hundred. Treat that as a working rule for pupils, since the study used adults and one story.
The quick method, step by step
- Guess a level from age, years of English and the class the pupil is joining. When in doubt, guess low.
- Give a text of about 100 words at that level. Ask the pupil to underline any word they do not know. Count the underlines.
- Ask five questions: two about facts stated in the text, one about order or cause, one about a word's meaning in context, one about why a character did something.
- Decide: move up, stay, or move down. Give a second text at the new level and repeat once.
- Write down both texts' levels, the unknown-word counts and the scores. That record is the level.
An original A2 text (90 words)
Sami's grandmother lives in a small village by the sea. Every summer, Sami goes to stay with her for two weeks. In the morning, they walk to the market and buy fish and bread. In the afternoon, Sami swims while his grandmother reads under a tree. One day, the sea was very rough, so they stayed at home. Sami was bored. His grandmother showed him an old box of photos. 'This is your father,' she said. 'He was eight, like you.' Sami looked at the photo for a long time.
Five quick questions
1. Where does Sami's grandmother live?
2. What do they buy at the market?
3. Why did they stay at home one day?
4. What does 'rough' mean in 'the sea was very rough'?
5. Why do you think Sami looked at the photo for a long time?
The text uses the present and past simple, short sentences and everyday words. That fits the A2 descriptor in the Companion Volume (Council of Europe, 2020): short, simple texts on familiar, concrete matters. Question 5 separates readers. An A2 reader may say 'because it was his father'; a stronger reader will add more.
The careful method: three measures, then a check
A vocabulary size test
Nation and Beglar (2007) published a Vocabulary Size Test with 140 multiple-choice items, ten from each of fourteen 1,000-word-family levels. The score multiplied by 100 estimates the pupil's written receptive vocabulary. In Read's (2000) terms it is a discrete, selective, context-independent test: it measures vocabulary on its own, in isolation from reading. That is its strength and its limit. It tells you how many words a pupil recognises, not how well they read.
A reading test built from CEFR descriptors
Take the overall reading descriptors from the Companion Volume (Council of Europe, 2020). At A1 a reader understands very short, simple texts a single phrase at a time. At A2, short, simple texts on familiar, concrete matters. At B1, straightforward factual texts on subjects of interest, with satisfactory comprehension. At B2, the reader reads with a large degree of independence. Write or choose one short text that fits each description, with five questions each. The pupil's level is the highest text they handle with ease.
Alderson (2000) surveys the techniques for testing reading, including multiple choice, short answers, cloze, ordering, information transfer and recall, and offers no single best method. So mix them. A short-answer item catches the pupil who guesses well on multiple choice. CEFR Reading Can-Do Statements, A1 to B2 lists the descriptors in full.
The classroom check
For two weeks, give the pupil ordinary class reading at the level the tests suggest. Watch three things: whether they finish, whether they can retell, and whether they look words up more than once or twice a page. If the tests and the classroom disagree, the classroom wins. It has more texts, more days and more evidence.
What each method measures, and what it misses
| Method | Time | What it tells you | Its weakness |
|---|---|---|---|
| Quick text check with five questions | 10 minutes per pupil, or 20 minutes as a class | Whether this pupil can read this level now | One text; a familiar topic makes the level look easier than it is |
| Vocabulary size test | One lesson for a class | How many words the pupil recognises in isolation | Measures words, not reading |
| Reading test built from CEFR descriptors | One lesson for a class | The highest level of text handled with ease | Depends on your texts being well levelled |
| Two weeks of classroom reading | No extra time | How the pupil reads when nothing is at stake | Slow, and needs a written record |
| Published test or reading inventory | Varies | A score on someone else's scale | Evidence varies: Nilsson (2008) found that only one of eight published inventories gave enough reliability data to support alternate forms |
Nilsson (2008) reviewed informal reading inventories published for US schools, not tests of English as a foreign language. Her finding does not transfer directly. The lesson does: a printed kit is not reliable because it is printed. Ask what evidence it carries. English Reading Placement Tests for Schools explains how to read those claims.
Why one test is a noisy measure
Schmitt, Jiang and Grabe (2011) tested 661 learners in eight countries on two texts. Comprehension rose in a roughly straight line as the share of known words rose, with no point where it jumped. A pupil who knows 96% of one text and 98% of the next is the same pupil, but a single test may label them A2 one day and B1 the next.
The texts add noise too. A story about football is likely to be easier for a pupil who plays it, whatever its level. So two results from texts on different topics can differ without either being wrong. The fix is not a longer test. It is a second text, a second day, and a record of both.
- Test twice, on different days, with texts on different topics.
- When the two results disagree, place at the lower level and set a date to look again.
- Treat any result older than a term as out of date. A pupil who reads daily moves.
How to record a level honestly
A label such as 'B1' hides what was measured. Write the evidence next to it, in one line that a colleague can act on next term.
An example record that a receiving teacher can use
Reads A2 texts with ease; B1 with support. 14 September.
Quick check: A2 text, 1 unknown word in 100, 5 of 5. B1 text, 6 unknown in 100, 3 of 5.
Vocabulary size test: 2,800 (same test as the rest of Year 7).
Class reading, two weeks: finishes A2 stories alone; needs a partner for B1. Look again in December.
The record names the texts' levels and the test, so the next teacher can repeat the check and compare. It also separates 'with ease' from 'with support': a story for home, or a story for class. How to Measure CEFR Progress by Cohort shows how such records add up across a year.
Young beginners who cannot be tested on paper
A seven-year-old who is still learning to read cannot sit a written test. The Companion Volume (Council of Europe, 2020) adds a Pre-A1 reading descriptor for this stage: the child can recognise familiar words accompanied by pictures, such as a picture book using familiar vocabulary. It also notes a separate collation of descriptors for learners aged 7 to 10 and 11 to 15.
- Listen and point. Say a word from the story; the child points to the picture. Ten words, two minutes.
- Read one word. Show a word card from a picture book; the child says it or finds the picture.
- Read one line, with the picture. The A1 descriptor is a single phrase at a time, with rereading allowed. Let the child reread.
- Retell in the first language. Read a very short story aloud, then ask what happened. This measures understanding without penalising weak spoken English.
- Record it the same way. 'Recognises 8 of 10 picture words; reads single lines with the picture; retells the plot in their first language.'
Reassess without a testing event
A level set in September is a guess about June. The light way to keep it current is to use what the class already produces: quiz scores on each story, how often a pupil finishes, and how many words they underline on a new text. Once or twice a term, look at those three for each pupil and move anyone who has outgrown the group.
Keep the formal method for the few decisions that are hard to reverse: a change of stream or an exam entry. Everywhere else, a ten-minute check and an honest record do the job. Formative Assessment for Reading in EFL describes the daily version of this habit.
How iRead handles this
In iRead, every story, quiz and game is levelled to the CEFR, from A1 to C2. A quiz runs after every session and practice questions are available between sessions, with questions generated from the story itself. Teachers see retention per word, per game and per class. School leaders see, by class, grade and term, the levels reached, words retained and stories completed.
The core principle: A reading level is a record of what a pupil read and how, on a named day with a named text. Write that down, and the label takes care of itself.
See it in iRead: Every story, quiz and game in iRead is levelled to the CEFR from A1 to C2, and teachers see retention per word, per game and per class.
See CEFR inside iReadKey takeaway
Start with a text and a count, add a vocabulary test and a levelled reading test when the decision is hard to undo, and record the evidence rather than the label.
Frequently asked questions
How do I quickly find a student's English reading level?
Give a 100-word text at a guessed level. Ask the pupil to underline unknown words, then ask five questions. Two or fewer unknown words and four right answers: try the level above. More than five unknown or two or fewer right: try the level below. Write both results down.
How many unknown words per page is too many?
Hu and Nation (2000) found that adult learners needed about 98% of a story's words known to read it unaided, which is two unknown words in a hundred. At five in a hundred, most readers in that study fell short. Use two as the mark for independent reading and five as the limit for guided reading.
Is a vocabulary size test enough to place a student?
No. The test by Nation and Beglar (2007) measures written receptive vocabulary, which is only part of reading. A pupil can recognise 3,000 words and still fail to follow a B1 story. Use the size for planning what to teach, and a reading check for the level you record.
How often should a reading level be reassessed?
Lightly once or twice a term, from quiz scores, finished stories and unknown-word counts on a new text. Formally only when a decision is hard to reverse, such as a change of stream or an exam entry. Treat a level older than a term as out of date.
How do I assess a young child who cannot read yet?
Use the picture. Say a word and have the child point; show a word card and have them find the picture; let them read one line with rereading allowed. The Council of Europe (2020) Pre-A1 descriptor is recognising familiar words with pictures. Record what the child did, not a label.
Sources
- Hu, M. H.-C., & Nation, P. (2000). Unknown vocabulary density and reading comprehension. Reading in a Foreign Language, 13(1), 403-430. doi.org/10.64152/10125/66973
- Nation, I. S. P., & Beglar, D. (2007). A vocabulary size test. The Language Teacher, 31(7), 9-13. jalt-publications.org/tlt/issues/2007-07_31.7
- Read, J. (2000). Assessing Vocabulary. Cambridge University Press, Cambridge. doi.org/10.1017/CBO9780511732942
- Alderson, J. C. (2000). Assessing Reading. Cambridge University Press, Cambridge. doi.org/10.1017/CBO9780511732935
- Council of Europe (2020). Common European Framework of Reference for Languages: Learning, teaching, assessment. Companion volume. Council of Europe Publishing, Strasbourg. rm.coe.int/common-european-framework-of-reference-for-languages-learning-teaching/16809ea0d4
- Schmitt, N., Jiang, X., & Grabe, W. (2011). The percentage of words known in a text and reading comprehension. The Modern Language Journal, 95(1), 26-43. doi.org/10.1111/j.1540-4781.2011.01146.x
- Nilsson, N. L. (2008). A critical analysis of eight informal reading inventories. The Reading Teacher, 61(7), 526-536. doi.org/10.1598/RT.61.7.2