How Many English Words Do You Need to Read a Novel? Vocabulary Size and Coverage

In a 2006 paper, “How Large a Vocabulary Is Needed for Reading and Listening?”, the vocabulary researcher Paul Nation gave a figure that is still widely quoted. If you need to know 98% of the words in a text to understand it without help, then reading written English takes a vocabulary of 8,000 to 9,000 word families. For spoken English the figure was 6,000 to 7,000.

Many learners see 8,000 and lose heart. I don’t think you need to take the number as an entry requirement for English books. This post looks at what the figure means and how far you can rely on it.

The idea of coverage

The 98% is the share of the words in a text that you know, often called coverage. At 98%, one word in fifty is unknown. If a paperback page holds about 300 words, that is around six unknown words a page.

The 98% line comes from a study Hu and Nation published in 2000. They took a short story, replaced some of its words with unknown ones, and gave learners versions in which 80%, 90%, 95% or 100% of the words were known, then compared how well they understood. At 95%, few readers understood it adequately, and the authors concluded that most learners would need around 98% to understand fiction without help. It is worth remembering that 98% itself was not one of the versions tested.

Two lines: 95% and 98%

Another often-cited paper is “Lexical threshold revisited” by Batia Laufer and Geke Ravenhorst-Kalovski (2010). It splits the threshold in two.

• Minimal comprehension needs 95% coverage, which takes 4,000 to 5,000 word families.
• Good comprehension needs 98% coverage, which takes 8,000 word families.

At 95%, one word in twenty is unknown, around fifteen a page. That is a lot to read through without a dictionary, but as I understand it, the line means you can still follow the story. The figures in this paper also count proper nouns, names of people and places, as known words.

Word families are not words

The unit behind these numbers matters. A word family is a word together with its inflected and derived forms. Read, reads, reading and reader count as one family.

The word counts in school curricula and textbooks usually count individual words. So a course that “covers 3,000 words” and a vocabulary of 3,000 word families are not the same thing, even though the numbers match. And a list of words a course covers is not a list of words you have learned.

Where the numbers and real reading differ

Even so, plenty of learners enjoy English books with far fewer than 8,000 word families. A study Schmitt and colleagues published in 2011 found that comprehension rose gradually as the share of known words went up, with no point at which reading suddenly became possible. I think 95% and 98% are best taken as reference points, not walls.

I see three more reasons the numbers and the experience of reading drift apart.

The first is that coverage varies a lot from book to book. The research figures are averages across many kinds of text. Children’s books and dialogue-heavy novels reach high coverage with a small vocabulary. Nineteenth-century novels for adults go the other way.

The second is that a book repeats its own words. A word you didn’t know in the opening pages becomes a word you know after a few meetings. That is why the first chapters are usually the hardest and the later ones easier.

The third is that not every unknown word matters. You can follow the story without knowing the colour of a coat or the name of a piece of furniture.

The NGSL as a reference point

To check the base of your vocabulary, I find the NGSL (New General Service List) a good reference. It is a list of the most useful words, chosen from a large corpus of English. According to the NGSL site, the latest version, 1.2, has 2,809 words and gives an average of 92% coverage of general English texts.

92% is still short of 95%. About one word in twelve is unknown. But knowing these 2,800 or so words is, I think, where reading real English books starts. Beyond them, the most natural way to grow your vocabulary is to meet words in the books you read.

What to read while your vocabulary is small

If your vocabulary feels too small, the quickest route is to match what you read to it. Graded readers are written within a set number of headwords, so you can choose a level that fits. I explain how in Choosing English Books by Level.

If you want real books, children’s books and short stories are the place to start. Read one page. If you can count the unknown words, you can read the book. If counting them becomes tiresome, it is too early, and you can pick another book. The reading method itself is in How to Do Extensive Reading in English.

Seeing the word lists in ABC George

I make ABC George, a free reading app for iPhone and iPad. Its Word Book includes the NGSL (2,801 words), the academic NAWL (963 words) and NGSL-Spoken (721 words). You can look through the base 2,800 words mentioned above as a list.

While reading, long-press a word and tap Look up. You first see a plain-English definition based on the NGSL and NAWL, then the WordNet senses below it. Words you look up are saved automatically to My words in the Word Book, together with the sentence you met them in, so when you finish a book you can see how many words you looked up in it.

There are no ads and no account, and books read offline once downloaded. The steps to your first book are in Getting Started with ABC George.