Chinese Word Count: Choose Words or Characters First

For a Chinese essay, first confirm whether the limit means characters or words. Then use a counter that measures that unit. A label such as CHINESE WORDS does not establish the rule for your assignment.

1. Get the required unit from the teacher or organization.
2. Confirm whether punctuation, spaces, digits, and English text contribute to the total.
3. Select the required part of the document.
4. Record the counter name and its result.

If the instructions specify a particular application, use that application for the final result. Do not replace an unclear requirement with a fixed characters-to-words ratio.

The question occurs before the calculation

In a discussion about a study abroad essay, an applicant asks what a 500-word requirement means for Chinese. The author compares three characters with one word. The replies suggest a character total, but they do not establish the institution's rule.

A separate discussion about Google Docs reports confusion after a document language change. A reply also raises punctuation as part of the total. This historical report does not establish how Google Docs operates today.

These cases show why the unit needs a name. A correct calculation can still answer the wrong question.

Characters and word segments measure different things

Chinese text does not need spaces between each word. A counter that only separates text at spaces cannot identify all Chinese word boundaries.

Unicode's text segmentation standard explains why Chinese word boundaries need language-specific treatment, such as a dictionary. Segmentation is the process that divides text into units. Its output depends on the method.

A code point is a numbered unit in Unicode. A grapheme cluster groups code points to approximate a character that a reader perceives. These units can differ. Neither unit is automatically a Chinese word or the unit that an assignment requires.

Six fixed samples, different totals

On November 4, 2026, we ran six fixed strings in Node.js 24.14.1 on Windows. The runtime used ICU 78.2 and Unicode 17.0 data. We selected Chinese word segmentation through Intl.Segmenter('zh', { granularity: 'word' }).

The table shows the actual results. The code-point total includes spaces and punctuation. The Han column counts code points with the Han script property.

Input	Han code points	All code points	Nonempty space-separated chunks	Word-like segments
图书馆	3	3	1	图书, 馆 — 2
我爱中文。	4	5	1	我, 爱, 中文 — 3
中文 test 123	2	11	3	中文, test, 123 — 3
𠮷🙂	1	2	1	𠮷 — 1
!!!	0	3	1	None — 0
Empty input	0	0	0	None — 0

For 图书馆, this runtime returned two word-like segments. That result is a software observation, not a linguistic reference answer. The first experiment incorrectly required one segment. We removed that assumption and kept the observed output.

The ECMAScript internationalization specification states that word-like classification depends on the implementation. In this sample, the number 123 qualified. Thus, the segment total was not a total of Chinese words alone.

The fourth sample also had four UTF-16 code units but two code points. A program's string length can therefore differ from the other totals.

All 20 final assertions passed. They verify selected unit totals, complete reconstruction of each input, and basic output conditions. They do not certify linguistic accuracy. The experiment did not test Enfect, Google Docs, Microsoft Word, or an essay submission system.

Procedure: compare the labels before the totals

1. Open the Enfect Word Counter.
2. Enter a short sample in INPUT.
3. Record CHARS NO SPACES.
4. Record CHARS WITH SPACES.
5. Record CHINESE WORDS.
6. Compare each label with the required unit.
7. If the units differ, use the method that the recipient specifies.

Enfect displays these separate fields on its current page. This article does not assign the local experiment's numbers to those fields. A label alone does not document every rule for punctuation, rare characters, or mixed text.

For example, 中文 test 123 contains Chinese text, English text, digits, and spaces. A total without spaces can still include the English letters and digits. It is not necessarily a Han-only total.

The useful record contains the selected text, the unit, the inclusion rules, and the application. Those details make a disagreement between counters possible to investigate.