Is this book right for my level?

Read one page from the middle of the book and count the words you don’t know. On a page of about 250 words, five or fewer means you know about 98% of them, and you can probably read the book without a dictionary. Around a dozen (95%) is workable with help; more than twenty-five (below 90%) is frustrating for most readers.

By the Readlang Mobile team (an independent app, not affiliated with readlang.com). Updated .

The one-page test

  1. Open the book in the middle. The first pages are not typical: they introduce names, places and the setting, and they can be harder than the rest of the book.
  2. Read one full page at your normal speed. Mark every word you could not explain or translate. Words you can guess with confidence from context count as known. Names of people and places don’t count.
  3. Estimate the words on the page. Count the words in three full lines, take the average, and multiply by the number of lines.
  4. Divide. Unknown words divided by all words gives the share you don’t know. Five unknown words on a 250-word page is 2%, so you know 98%.

An example. On this line from the middle of a detective story, a B1 reader might mark two words:

The inspector scrutinised the ledger for a long minute, then closed it and looked up at the clerk.

Two unknown words in eighteen is 11%. One line is too small a sample, which is why the test uses a whole page: the next five lines may have no new words at all.

What your count means

The last column summarises the studies described below. Hu and Nation replaced real words in one short story with invented ones, gave different versions to four groups of learners, and counted how many understood the story well enough (at least 12 of 14 questions right).

Share of known words, what it means on a page, and what studies found.
Words you knowUnknown per 100 On a 250-word pageWhat studies found
100%00 15 of 17 learners understood the story well (Hu and Nation)
98%25 Estimated level at which most learners understand a story without help (Hu and Nation); the “sweet spot” for extensive reading (Extensive Reading Foundation)
95%512–13 6 of 17 learners understood the story well (Hu and Nation); enough for reading academic texts with some guidance (Laufer and Ravenhorst-Kalovski)
90%1025 4 of 16 learners understood the story well (Hu and Nation); below this, reading becomes frustrating (Extensive Reading Foundation)
80%2050 None of 16 learners understood the story well (Hu and Nation)

For books you read for pleasure and in quantity, the advice is stricter than 98%. Paul Nation and Rob Waring suggest a book is at the right level for extensive reading if it has no more than three or four unknown words per page.

What the research says, and what it doesn’t

The 98% figure is an estimate from one small study. Marcella Hu and Paul Nation (2000) gave 66 adult learners of English in New Zealand a 673-word story in four versions, with 100%, 95%, 90% or 80% of the words known; the rest were replaced by invented words. Nobody read a 98% version. The authors drew a line through the four results and estimated that most learners would need about 98% to understand fiction without help. A 2023 replication with 104 adult learners in Sri Lanka could not fully reproduce the result.

95% and 98% are two different goals. Laufer and Ravenhorst-Kalovski (2010) tested 745 students in Israel on academic reading. They called 95% of known words the minimal level, at which students read with some guidance, and 98% the optimal level for reading on their own. The texts were academic, not novels.

There is no cliff. A study of 661 readers in eight countries by Schmitt, Jiang and Grabe (2011) found that comprehension rises roughly in a straight line as the share of known words goes up. Every extra percent helps a little; there is no point at which understanding suddenly switches on. Treat 98% as a useful target, not a pass mark.

How many words you need to know

Paul Nation (2006) counted how many of the most frequent English word families you need to know to reach 98% of the words in real texts. A word family is a word with its forms and close relatives: “read”, “reads”, “reader”, “readable”.

Word families needed for a share of known words in English texts, including names (Nation 2006).
TextFor 95%For 98%
Novels and newspapersabout 4,0008,000–9,000
Spoken Englishabout 3,0006,000–7,000
The Picture of Dorian Gray, Oxford Bookworms graded reader, 1,000 headwords 2,000 (96.75%)3,000 (98.86%)

These are counts from word lists, not tests with readers, and they are for English. But they explain why the jump to original novels is hard: a graded reader reaches 98% with a few thousand word families, while an ordinary novel needs two or three times as many.

How does that compare with CEFR levels? The most cited estimates come from the X_Lex test, which checks how many of the 5,000 most frequent English words a learner knows. Learners at A2 typically know 1,500 to 2,500 of them, at B1 2,750 to 3,250, at B2 3,250 to 3,750, at C1 3,750 to 4,500 and at C2 4,500 to 5,000. The test cannot go above 5,000, the groups were small, and the numbers differ between languages: learners of French scored lower than learners of English at the same level.

The Council of Europe’s own description of reading for pleasure says much the same thing in words. At B1 a reader can follow the plot of simple novels with a clear storyline, using a dictionary regularly. At B2 they can read novels with a strong plot and straightforward language, taking their time and using a dictionary. Only at C1 do most kinds of literary text become readable with occasional help.

Why the level on the cover is only a guide

If the book is slightly too hard

Sources

  1. Hu, M. and Nation, P. (2000). Unknown vocabulary density and reading comprehension. Reading in a Foreign Language, 13(1), 403–430. doi.org/10.64152/10125/66973
  2. Kremmel, B., Indrarathne, B., Kormos, J. and Suzuki, S. (2023). Unknown vocabulary density and reading comprehension: Replicating Hu and Nation (2000). Language Learning, 73(4), 1127–1163. doi.org/10.1111/lang.12622
  3. Laufer, B. and Ravenhorst-Kalovski, G. C. (2010). Lexical threshold revisited: Lexical text coverage, learners’ vocabulary size and reading comprehension. Reading in a Foreign Language, 22(1), 15–30. Full text (ERIC)
  4. Schmitt, N., Jiang, X. and 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
  5. Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? The Canadian Modern Language Review, 63(1), 59–82. doi.org/10.3138/cmlr.63.1.59
  6. Milton, J. (2010). The development of vocabulary breadth across the CEFR levels. In Communicative Proficiency and Linguistic Development, EUROSLA Monographs Series 1, 211–232. Full text (EUROSLA)
  7. Nation, P. and Waring, R. (2013). Extensive Reading and Graded Readers. Compass Publishing. PDF (Extensive Reading Foundation)
  8. Extensive Reading Foundation (2011). The Extensive Reading Foundation’s Guide to Extensive Reading. erfoundation.org/guide/ERF_Guide.pdf
  9. Council of Europe (2020). Common European Framework of Reference for Languages: Learning, teaching, assessment. Companion volume, p. 59. rm.coe.int

Next: How long will the book take?