How many words do you need to know? Coverage explained
Oct 3, 2026 · 12 min read
Ask how many words a learner needs and the answers run from a few hundred to twenty thousand. The gap is less a disagreement about facts than a difference in what gets counted. The number that decides how much you understand is not your total vocabulary but your coverage: the share of the words on a page or in a conversation that you already know.
This page sets out the coverage figures that are usually quoted, the reading and listening thresholds behind them, and how to turn the numbers into a weekly plan.
Coverage, not a total count
Coverage is measured over running text, so every appearance of a word counts, not just the first. A small group of very common words turns up again and again, and each repeat lifts your coverage without changing your vocabulary size. That is why the first thousand word families do far more work than the second thousand, and why a short list can carry a long text.
It also explains a common confusion. A learner with 5,000 words is not five times as well served as a learner with 1,000 words. The gain is real, but it arrives unevenly, with a steep climb early and a gentler slope later.
What the commonest words give you
| Word families known | Rough coverage of everyday text |
|---|---|
| 1,000 | about 75–80 per cent |
| 2,000 | about 85 per cent |
| 5,000 | about 95 per cent |
These are common estimates. Different corpora and different counting methods move the figures by a few points, so treat them as a guide rather than a precise measure.
A paragraph of 100 words in which you know 80 still leaves 20 gaps, and that is a lot of stopping. At 95 per cent the same paragraph has five gaps, which most readers can bridge from the sentence around them. The size of the gap, not the size of the vocabulary, is what decides whether reading feels smooth.
The thresholds for reading and listening
Two figures get quoted more than any others. Spoken language becomes broadly followable at around 95 per cent coverage, where the situation, the tone and the picture in front of you fill the rest. Reading without frequent stops is usually placed nearer 98 per cent, because a written page offers no picture and no tone of voice to lean on.
Both are estimates rather than switches. A supportive listener at 94 per cent may follow everything while a dense technical page at 97 per cent still defeats a reader. The useful part is the direction: each extra per cent of coverage removes a large share of the interruptions, because gaps are what break reading, not unknown words on their own.
These thresholds also explain why graded reading feels so different from a newspaper. A simplified text keeps you above the line; an unsimplified one can drop you well below it, even when the subject is familiar.
Word families and word forms
A word form is a single spelling. Walk, walks, walked and walking are four forms. A word family groups them under one base, and coverage research usually counts families because a learner who knows walk can generally handle the rest. This matters when you read a figure: 2,000 word families is a much larger amount of English than 2,000 forms.
English words also carry several meanings. The most common words tend to have the most senses, so knowing a family at one meaning is not the same as knowing it at all of them. Coverage figures quietly assume the common meanings, which is why a word can be counted as known and still surprise you in a sentence.
CEFR levels as a rough reference
| CEFR level | Rough vocabulary size (word families) |
|---|---|
| A1 | about 1,000 |
| A2 | about 2,000 |
| B1 | about 3,000 |
| B2 | about 4,000 |
| C1 | about 5,000 or more |
Common reference amounts, not exact cut-offs. Different frameworks and tests place the boundaries differently, and the figures describe what you can understand more than what you can produce.
Numbered sets are the easiest way to work at a level. The site keeps a first fifty words for A1, B1 and C1, and each one is a manageable sample of the band rather than a full inventory.
High frequency first, then the long tail
The first few thousand families carry most of what you meet, so they repay study in almost any order. After about 4,000 families the returns per word fall. The next thousand turns up rarely outside a particular subject, so it pays best when you have a reason to meet it: a job, a course, or a book you actually want to read.
That is not a reason to stop at 4,000. It is a reason to choose the next words by purpose rather than by list position. The bigger picture, including how frequency itself is counted, is set out in word frequency and learning.
General words and academic words
Most everyday English sits inside the general high-frequency words. Academic writing adds a layer just outside them, often taught through a list such as the Academic Word List. If your goal is study, the general words come first and the academic layer second. If your goal is conversation, the general words do nearly all the work, and the academic layer is optional.
Word families sit underneath both. Grouping the forms of a base word is explained in word families for vocabulary, and it is the reason a modest number of bases can support a large reading vocabulary.
Turning the numbers into a plan
- Set the target in families, not forms. Around 1,000 families for simple material, 2,000 for everyday texts, and 4,000 for comfortable unsimplified reading.
- Choose a weekly set you can actually keep. Twenty new words a week is about a thousand a year, which puts 2,000 families roughly two years away. The learn twenty words a week plan sets out one version of that routine.
- Measure coverage, not list length. Take a page you want to read and count the words you do not know. That gap, not your total, tells you whether the material is at the right level.
- Review every week, and keep the reviews short. A plan of five short sessions holds up better than one long one, and the reviews are where the earlier sets are kept alive.
Frequently asked questions
How many words do I need to read comfortably?
Common estimates put comfortable reading at about 98 per cent coverage, which for general texts is often placed near 8,000 to 9,000 word families. Much simpler material, such as graded readers, needs far fewer, so the useful answer depends on what you want to read.
Is 95 per cent coverage enough for watching films?
About 95 per cent is often given as the level at which spoken language becomes broadly followable, with context and picture filling the gaps. It will not carry you through every scene, and fast dialogue or an unfamiliar accent raises the bar.
What is the difference between a word family and a word form?
A word form is one spelling, such as walk. A word family groups the forms built from the same base, so walk, walks, walked and walking count as one family. Coverage figures are usually quoted in families, which is why 2,000 families cover far more text than 2,000 forms.
Do I need 20,000 words to read a novel?
No. A very large number usually counts every distinct form across a wide corpus. For reading, what matters is the coverage you reach, and the most frequent few thousand families do most of that work.
How long does it take to reach 2,000 word families?
At about twenty new words a week, roughly two years of steady study, allowing time for review and for words that need several meetings. Shorter, more frequent sessions hold up better than one long weekly one.
This page is a way of planning vocabulary work, not a promise about results. The coverage figures are estimates, and progress depends on how regularly you meet and use the words you choose.