Everyone asks this question eventually. Usually after a few months in, and likely while staring at a novel they bought optimistically but haven't opened. How many words do I need before I can read this thing?
The number people quote is typically around 8,000 to 9,000 word families for comfortable independent reading. That comes from real research, and it is roughly right. It is also the least useful true fact in language learning, because if you know 800 words this basically says, "not even close. Come back in a few years."
So break this question up a bit, because the interesting part is not in the number itself, but rather what the number is actually measuring.
Coverage is the thing that actually matters
Vocabulary research doesn't really talk about how many words you know. It talks about coverage, which is the percentage of the words on the page that you already recognise.
This distinction matters because words are not evenly distributed. A small number of them do almost all the work. The most common 1,000 word families account for around 80 percent of everything written. The next 1,000 add roughly another 6 or 7 percent. After that the returns fall off a cliff, and you are grinding through thousands of words that each appear once every few hundred pages.
Here is roughly what coverage looks like as you climb. These figures come from Paul Nation's analysis of novels and newspapers, and they are approximate, because the exact numbers shift depending on the text and on whether you count proper nouns:
- 1,000 words gets you to around 80 percent coverage. One word in five is unknown.
- 2,000 words gets you to around 87 percent. Roughly one word in eight.
- 3,000 words gets you to around 90 percent. One word in ten.
- 5,000 words gets you to around 95 percent. One in twenty.
- 8,000 to 9,000 words gets you to around 98 percent. One word in fifty.
That last figure is where that famous number we talked about earlier comes from. Hu and Nation tested readers on fiction at different coverage levels and found that almost nobody reached adequate unassisted comprehension below 95 percent, and that around 98 percent was where most learners got there. Not reconstructing it from context and hoping, but actually reading it.
One honest caveat. A later study by Schmitt, Jiang and Grabe found that comprehension rises fairly steadily with coverage rather than jumping at some magic line, which suggests 98 percent is less a threshold than a convention. That makes the number softer than it is usually presented, and it cuts in the direction of reading sooner rather than later.
Why 90 percent feels so much worse than it sounds
Ninety percent sounds like an A. It is not. It means one word in every ten is unknown.
A line of a paperback holds roughly ten to twelve words. So at 90 percent coverage you are hitting an unknown word on essentially every single line. Not every paragraph, every line.
That is why people bounce off their first novel and conclude they are not ready. They are not wrong about the experience. Reading like that is genuinely awful. You lose the thread of the sentence while you deal with the gap, and by the time you have dealt with it you have forgotten what the sentence was about. Twenty pages of that can be exhausting, and exhausting is what makes people give up.
At 98 percent, an unknown word shows up every four or five lines. You can guess it from context, or ignore it, and the sentence survives either way. That is the difference between reading and decoding.
The trap in the number
Notice what the 98 percent figure assumes: that you are reading unassisted. No dictionary, no notes, no help. The research measured how much vocabulary you need for a text to be comprehensible on its own.
That was the right question to ask about a paper book, because a paper book gives you nothing other than the words on the page. Before, your options were to either guess, skip it, or to get up and find a dictionary, which let's face it, nobody does twice.
So the number is not really "how many words you need to read a novel." It's "how many words you need before a novel stops fighting you when you have no help available."
If you can change the second half of that sentence, then the first half changes with it. And that's where Lexxio comes in.
What changes when the page can answer you
If a word takes half a second to look up, and the sentence is still in front of you when you get back, the math is different. An unknown word stops being a wall and becomes a small pause.
At that point you are not waiting to hit 98 percent coverage anymore. You are reading at 85 or 90 percent and closing that gap while you read, which is the only way anyone has ever closed it. Nobody reaches 8,000 words from flashcards and then starts reading naturally. Like most things, they start reading badly and get better at it. The vocabulary arrives as a side effect of wanting to know what happens next.
This is most of what Lexxio is for. You can quickly double click or tap a word and get the meaning without leaving the page. You can read a passage in your target language and pull up an aligned translation when a sentence defeats you rather than a single word. You can even drop the whole book down to a simpler level for a chapter, then raise it back up when you're ready.
None of that automatically teaches you those 8,000 words. Instead, it removes the reason people quit before they get there.
So what number should you actually aim at?
If you want a target rather than a philosophy:
Around 1,000 to 2,000 words is enough to start reading something real, provided you have help on the page and you pick the book carefully. This is much earlier than most people are told.
Around 3,000 words is where a well-chosen novel starts being pleasant rather than a project, again with lookup available.
Around 8,000 words is where you can pick up an unfamiliar book in a bookshop and read it cold, which is a genuinely different milestone and worth wanting.
The gap between the first and last of those is a few years for most people. The gap between zero and the first is a few months. Almost everyone who fails at this fails in that first gap, and they fail because they were waiting for a number instead of reading.
Picking the right first book matters more than the number
Coverage is not only a property of you. It is a property of you and a particular book together.
The same reader might sit at 88 percent coverage on a literary novel with a large vocabulary and 96 percent on a contemporary thriller with short sentences and repeated words. Those are different reading experiences, and only one of them is worth attempting first.
That's why the advice to "read a book at your level" is less useful than it sounds. Books do not come with levels. What you can do is take a page, count the words you do not know, and see whether you are hitting one every line or one every five. That takes two minutes and tells you more than any estimate of your vocabulary size.
If you want a longer version of that argument, reading levels explained covers what Simple, Guided, Natural, Advanced and Faithful actually feel like on the page. And if you already have a book picked out and want to make it to the end of it, how to finish your first book is about the part that comes after choosing.
The short version
You need about 8,000 words to read a novel with no help. You need about 2,000 to read one with help, which is what you should be doing anyway, because reading is how you get to 8,000.
The number is real. It is just not a gate. It is a description of where the road ends, but people keep reading it as a description of where it starts.
Where these numbers come from
Two things worth saying plainly before the list.
The coverage figures above are approximate. Published tables vary depending on the corpus, on whether proper nouns and compound words are counted, and on how a word family is defined. Treat them as the shape of the curve rather than as precise measurements of you.
And this research is almost entirely English. Spanish, French and the other languages we support show the same steep frequency curve, so the shape holds, but the specific percentages were measured on English texts and have not been transferred here from a Spanish study.
- Hu, M. and Nation, P. (2000). Unknown vocabulary density and reading comprehension. Reading in a Foreign Language, 13(1), 403 to 430. The source of the 98 percent figure.
- Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? Canadian Modern Language Review, 63(1), 59 to 82. The source of the 8,000 to 9,000 word family estimate and the coverage bands.
- 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 to 30. Frames 95 percent as a minimal threshold and 98 percent as an optimal one.
- 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 to 43. Finds a steady relationship rather than a sharp threshold.