# A Chinese Bible Character Carries 8.77 Bits

> Measured over the Chinese Union Version (和合本, simplified), a single Chinese character carries 8.77 bits of information and a segmented word 9.62 bits — the Shannon entropy of how often each one occurs.

Question: information per Chinese character
Answer: 8.77 bits
Page: https://chinese-bible.com/en/reference/information-per-chinese-character

Entropy here is the standard unigram measure: take how often every character occurs, treat those frequencies as probabilities, and compute −Σ p·log₂p. It answers "how unpredictable is the next character", counting nothing but frequency — no grammar, no context, no memory of the previous character.

8.77 bits per character is close to the theoretical maximum for an inventory this size: with 2,996 distinct characters used perfectly evenly the figure would be 11.55, so the Bible's characters are unevenly used but not nearly as unevenly as a small alphabet would be. A word carries 9.62 bits, more than a character — which is the arithmetic behind the claim that Chinese is dense: fewer symbols, each one worth more.

The practical corollary for a learner is the coverage curve: because usage is uneven, 73 characters cover half the text and 971 cover 95%. Entropy is the compact way of saying how much room there is left over in the tail — 2,025 characters share the last twentieth.

## Unigram entropy

| Unit | Entropy (bits) | Distinct units | If used evenly (bits) |
| --- | --- | --- | --- |
| Character | 8.77 | 2,996 | 11.55 |
| Word | 9.62 | 11,168 | 13.45 |
| Characters for 50% of the text | — | 73 | — |
| Characters for 95% of the text | — | 971 | — |

Unigram entropy only: it ignores context, so it is an upper bound on how unpredictable real reading is.

## Questions

**Is this the same as compression ratio?**

No. Unigram entropy ignores context, and real compressors exploit context heavily, so a compressed file is much smaller than this figure implies.

**How does it compare with English?**

Per symbol, far higher — an English letter's unigram entropy is about four bits. Per unit of meaning the two languages are much closer, which is the honest way to compare them.

**Which corpus is this?**

和合本 (Chinese Union Version), simplified script — public domain — the text this site publishes, segmented as described in the method below.

## Method

- Text: 和合本 (Chinese Union Version), simplified script — public domain
- Words: forward maximum matching against CC-CEDICT (this site's own segmentation)
- HSK levels: HSK 3.0 levels from the complete-hsk-vocabulary list; band 7 covers HSK 7-9
- Reading times: 260 characters a minute (an assumption)
- Computed: 2026-09-22

These are reproducible counts over a public-domain text, not a peer-reviewed study. Difficulty ranks are one stated formula, not a validated readability score.


---

- Definitions: CC-CEDICT, licensed CC BY-SA (https://creativecommons.org/licenses/by-sa/4.0/).
- If you share this file onward, keep this notice and share alike.
- Character data: Make Me a Hanzi (Arphic Public License / LGPL).
- Bible text: public domain — 和合本 · World English Bible · Louis Segond 1910.
- Built with https://chinese-bible.com