How to start reading before you are ready

Waiting until you are good enough to read is why most learners never start. Here is the threshold that actually matters, why it is lower than you think, and what to read first.

A reader at a table with a book, drawn in flat editorial style

There is a piece of advice every language learner receives, and it is almost always wrong. It goes: finish the course first, build a foundation, and once you have enough vocabulary, start reading real material.

The trouble is that the day never arrives. There is no point at which a course hands you enough words to open a newspaper comfortably, because courses are finite and newspapers are not. People wait, and go on waiting, and eventually conclude that reading in another language is something other people do.

So it is worth being precise about what “ready” actually means, because researchers have measured it, and the number is more useful than the advice.

The threshold

The thing that determines whether you can read a text is not your level or how long you have studied. It is coverage: the proportion of words on the page that you already know.

Hu and Nation put numbers on this in 2000 by giving learners the same text with different proportions of the vocabulary removed from their grasp. The finding has held up in replication since. At around 98% coverage, readers understood what they read and could answer questions about it. At 95%, many struggled. Below that, comprehension fell apart.

Ninety-eight percent sounds close to perfect, but it is not. It means one unknown word in every fifty, which is roughly one per short paragraph. Here is what that actually looks like:

98% coverage, one unknown word in fifty

The ferry leaves the harbour a little after six, when the light is still grey and the cafés along the front have only just started to open. Most of the passengers are regulars who make this crossing twice a week, and they take the same seats without discussing it. The crew know them by name. By the time the boat rounds the headland the sun has come up properly, and for twenty minutes the whole deck is quiet, everyone watching the same stretch of water go past.

You read that. Two words were missing and it barely registered, because the sentences around them carried you straight over the gaps. You could keep going like that for an hour without it costing you anything.

Now the same passage at 90% coverage, one unknown word in ten:

90% coverage, one unknown word in ten

That is the same paragraph. It is now work. You can reconstruct roughly what is happening, but you are solving a puzzle rather than reading, and nobody does that for an hour by choice.

This is the single most useful thing to understand about reading in another language: the difference between pleasant and impossible is a few percent. Not a level, not a year of study. A few percent. (How much vocabulary those percentage points cost is its own guide, and the answer is sobering.)

Why the threshold is lower than the research says

Here is the part that changes what you should do about it.

That 98% figure is for unassisted reading. Paper book, no dictionary, nobody to ask. It is the coverage you need for the text alone to carry you.

The moment you can resolve an unknown word instantly, in the sense it carries in that sentence, without losing your place, the arithmetic changes. An unknown word stops being a wall and becomes a half-second pause. You are no longer reading at 90% coverage and guessing; you are reading at 90% coverage with the other 10% answered as you go.

That is the entire argument for reading with a tool rather than reading with a dictionary. A paper dictionary costs you thirty seconds and your place on the page, so at one unknown word in ten you stop reading and start doing lookups. Remove that cost and the same text becomes readable years earlier than it otherwise would.

It is also why “look it up later” and highlighting words to review afterwards do not work as well as they sound. The meaning you need is the one the word has here, and by the evening you have lost the sentence it came from.

So what should you actually read

Coverage is a property of a pairing, not of you. You do not have one reading level; you have a different coverage figure for every text in the world. The skill is picking texts where yours is high.

Start with graded material. Anything written to a level exists precisely to keep your coverage in the comfortable band. If you know your level from A1 to C2, this is the one case where the label on the tin is accurate. Read at the level you are, not the level you would like to be.

Then move to series rather than standalone pieces. This is the most underrated step. A series reuses its own vocabulary, so your coverage climbs as you go: part four is easier than part one, even though the language has not got simpler. Standalone articles reset you to zero every time.

Then go to the real thing, in the narrowest topic you care about. Do not open the front page. Open the football coverage, or the film reviews, or the recipes. A single domain has a small recurring vocabulary, so your coverage inside it is far higher than your coverage of the language overall. People who “can read the news” almost always started by being able to read one section of it.

And bring things you already wanted to read. An article a friend sent you, the lyrics of a song you already have on repeat, a video you would have watched anyway. If it has audio, read and listen at once. Motivation is not a bonus here, it is the mechanism: it is what makes you finish something slightly too hard, which is the only way coverage ever goes up.

What not to do

Do not look up every word. At the coverage where you should be reading, most unknown words are inferable from the sentence and will turn up again. Look up the ones that block the sentence and the ones that keep recurring. Ignore the rest, and let the ones that matter earn your attention by appearing a second time.

Do not reread until you understand every word. Comprehension of the passage is the goal, not accounting for its parts.

Do not measure a session in pages. Measure it in whether you want to come back tomorrow. Twenty minutes daily beats two hours on Sunday, because coverage is built by volume over time and volume requires that the activity be tolerable.

Do not start with a novel you love. It is the most common mistake and it is a shame every time, because a book that matters to you deserves better than being the thing you gave up on in chapter two. Save it. It will still be there when you can enjoy it.

The honest summary

You are probably ready to read something right now, and that something is not the book you have in mind. It is shorter, easier, and on a subject you would have read about anyway.

Read that. Then read the next one. Coverage is the only thing standing between you and the book you actually want, and the only way it goes up is by reading things slightly below where your pride would like to be.

Sources: Hu, M. and Nation, P. (2000), Unknown vocabulary density and reading comprehension; Nation, I.S.P. (2006), How large a vocabulary is needed for reading and listening?; Laufer, B. (2020), on the 95% minimum and 98% optimum thresholds.