The 95% and 98% Coverage Rules for Reading Chinese

Two numbers come up again and again when people talk about how hard a text is: 95 percent and 98 percent known-word coverage. They are useful rules of thumb for choosing what to read, but they are guideline figures from the reading literature, not laws, and they are about word coverage, not full comprehension. This guide explains what they mean and how to use them without overclaiming.

What “coverage” is

Coverage is the share of the running words in a text that you already know. If a passage has 100 words and you know 95 of them, your coverage is 95 percent. It is measured over tokens, meaning every word as it appears, so common words that repeat count every time. That is the number the analyzer estimates for any text you paste, at each HSK level.

The two thresholds, and where they come from

• Roughly 95 percent for comfortable extensive reading. At about this level, you can read for pleasure and follow the thread, guessing many unknown words from context. You still meet unfamiliar words, roughly one in twenty, but not so often that reading stalls.
• Roughly 98 percent for comfortable independent reading. At about this level, you can read without leaning on a dictionary, because unknown words are rare enough, roughly one in fifty, that context carries you.

These thresholds trace back to vocabulary-and-reading research, most often associated with studies by Hu and Nation and by Laufer, and they are widely repeated in the Chinese-learning community, including by Hacking Chinese. We cite them as figures from that literature, never as our own finding. The precise cutoffs vary between studies and between readers; treat 95 and 98 as useful landmarks, not exact switches. A deeper look at the underlying studies is in the 98% research, honestly.

The honest caveat: coverage is not comprehension

This is the part most repetitions of the rule leave out. Knowing 98 percent of the words does not mean understanding 98 percent of the meaning. The words you do not know are often the content words that carry the point of a sentence, and grammar, idiom, and cultural reference sit on top of vocabulary. Coverage tells you how often you will hit a wall, not how well you will understand what is between the walls. It is a planning estimate for choosing texts, and nothing more.

How to use the rules in practice

1. Measure, do not guess. Paste a real sample of what you want to read into the analyzer and read your coverage at your level. A page or two is enough to get a stable estimate.
2. Aim for about 95 percent for reading you enjoy. If a text lands below roughly 90 percent for you, it will feel like decoding rather than reading. That is fine for short, intensive study, but it is the wrong choice for building reading stamina.
3. Use 98 percent as the “no dictionary” bar. If you want to read something start to finish without stopping, look for material where your coverage is up near 98 percent. Graded readers by level are designed to sit in this zone.
4. Close the gap in frequency order. The analyzer lists the words you are missing, most common first. Learning those, ideally with spaced repetition so each one returns right before you forget it, raises your coverage on the next text of the same kind. Wordbrush, from our publisher Urban Algorithm, does that drilling; the analysis here is free and needs no app.

Used this way, the 95 and 98 percent rules stop being trivia and become a practical filter: they tell you whether a given piece of Chinese is reading practice or study homework today, and what to learn to move it from one to the other.