Book Difficulty ·· 4 min read · By The VocaPages team

95% or 98%: what “reading comfortably” really means

Research says you need to know 95–98% of the words in a text to read it well. Here is what those numbers look like on a real page, and how to use them to choose books.

Our Book Difficulty articles say things like “you need 5,000 words for 95% coverage and 10,000 for 98%”. This article explains where those two numbers come from, what they feel like when you are actually reading, and why the gap between them is so large.

What “coverage” means

Lexical coverage is the percentage of the running words in a text that a reader knows. If a 300-word page contains 15 words you don’t know, your coverage of that page is 95%.

It is a simple idea, but it predicts reading comprehension surprisingly well. Two thresholds come up again and again in the research:

  • 95% — Batia Laufer’s studies in the late 1980s found that below about 95% coverage, learners’ comprehension dropped sharply. It is often described as the minimum for reasonable understanding.
  • 98% — Hu and Nation (2000) had learners read a fiction text at different coverage levels and found that most needed around 98% to understand it well without help. Later work by Laufer and Ravenhorst-Kalovski (2010) described 95% as the minimal and 98% as the optimal threshold.

Paul Nation (2006) then estimated how much vocabulary it takes to reach 98% coverage of real texts: around 8,000–9,000 word families for novels and newspapers, and somewhat fewer for spoken English.

What the numbers look like on a page

Three percentage points sound small. On a page, they are not.

Coverage Unknown words on a 300-word page Roughly
90% 30 one in every line
95% 15 one every two lines
98% 6 one every five lines
99% 3 one every ten lines

At 90%, almost every sentence contains a word you don’t know, and guessing from context stops working because the context is full of gaps too. At 95%, you can follow the story but you will stop often. At 98%, unknown words are rare enough that most can be skipped or guessed — this is what reading for pleasure feels like.

Why the last 3% costs so much

Here is how coverage grows with vocabulary size in three novels we measured (figures are the share of running words, names excluded):

Words you know Alice in Wonderland Pride and Prejudice Moby-Dick
1,000 86.4% 84.5% 73.6%
2,000 91.3% 89.8% 80.1%
4,000 95.4% 94.4% 86.1%
6,000 96.8% 96.2% 90.0%
8,000 97.7% 97.2% 92.1%
10,000 98.2% 98.0% 93.8%
15,000 99.0% 98.9% 96.0%

The first thousand words do most of the work: they cover more than 80% of a typical novel. Each extra thousand adds less. For Pride and Prejudice, going from 2,000 to 4,000 words adds almost five percentage points; going from 8,000 to 10,000 adds less than one. That is why the step from 95% to 98% often means doubling your vocabulary — 5,000 to 10,000 words for Austen.

This is just how language works. A few hundred words — the, and, was, said — make up a huge share of any text. Rarer words are individually rare but collectively common: any one of them is unlikely to appear, but some of them appear on every page.

How to use this when choosing a book

  • Aim for 95% as a minimum, 98% for pleasure. Our classics ranking gives both figures for twenty well-known novels.
  • Check a page yourself. Count the unknown words on a page or two. Fifteen or more on a 300-word page means the book will be hard work; six or fewer means you can read freely.
  • Between 95% and 98%, learn the repeated words. Every novel has a set of less common words it uses again and again. Learning those first is the cheapest way to raise your coverage of that particular book — which is why our book analyses list them.
  • Don’t treat the numbers as gates. Motivation, background knowledge and familiarity with the genre all change how a book feels. A story you love at 94% may be easier going than a dull one at 97%.

A note on our figures

Research usually counts word families (happy, happily and happiness count as one). Our measurements group common derived forms with their root but do so conservatively, so our “words needed” figures tend to be a little higher than Nation’s. For example, we put 98% coverage of Pride and Prejudice at about 10,000 words, compared with the 8,000–9,000 families that Nation’s research suggests for novels in general. Because every book is measured the same way, comparisons between books are fair. The method is described at the end of each of our book analyses.

References

  • Hu, M. & Nation, P. (2000). Unknown vocabulary density and reading comprehension. Reading in a Foreign Language, 13(1).
  • Laufer, B. (1989). What percentage of text-lexis is essential for comprehension? In C. Lauren & M. Nordman (Eds.), Special Language: From Humans Thinking to Thinking Machines. Multilingual Matters.
  • Laufer, B. & Ravenhorst-Kalovski, G. C. (2010). Lexical threshold revisited. Reading in a Foreign Language, 22(1).
  • Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? The Canadian Modern Language Review, 63(1).