One family emoji, four different counts

Chatcument · September 8, 2026 · Original reproducible fixture study

A character limit is incomplete until it says what is counted. The family emoji below has 11 UTF-16 code units, 7 Unicode code points, 1 grapheme cluster and 25 UTF-8 bytes. These are four measurements of the same input, not competing answers to one defined question.

Download and reproduce

This page supplies 24 deliberately chosen inputs, their exact code-point sequences and four measurements. The data is useful for teaching and small regression tests. It is not a market survey, a benchmark of competing tools, or the full Unicode conformance test suite.

Choose the unit before enforcing a limit

• UTF-16 code units: JavaScript string length reports storage units. Some code points occupy two units.
• Unicode code points: string iteration separates encoded code points; a combining mark is still its own code point.
• Grapheme clusters: segmentation groups sequences such as combining accents and supported emoji sequences. This is useful for many user-facing character limits, but it does not perfectly define every reader’s perception.
• UTF-8 bytes: useful when a receiving system imposes a UTF-8 byte limit. A grapheme count cannot substitute for a byte budget.

Use the receiving form or API’s documented rule. Do not assume that a social platform, application form or database applies this page’s grapheme rule. Spaces and invisible characters remain in these measurements.

The 24-fixture results

The table is rendered in the page HTML and can be read without running JavaScript. Code-point labels make invisible and visually similar inputs distinguishable.

Why Korean text is a useful test

In this fixture set, composed 한글 and its decomposed conjoining-jamo form both contain 2 grapheme clusters. Their UTF-8 sizes are 6 and 18 bytes. Compatibility jamo ㄱㅏㄱ is a separate fixture: it produces 3 clusters here, while conjoining 각 produces 1. Visually related input is not necessarily represented by the same sequence.

Normalization changes representation. Decide whether the application should normalize before comparing or storing text; do not silently normalize identifiers or signed payloads merely to obtain a preferred count. The measurements above preserve each supplied input.

Run a single example

const s = "👨‍👩‍👧‍👦";
console.log({
  utf16Units: s.length,
  codePoints: Array.from(s).length,
  graphemes: Array.from(new Intl.Segmenter("en", {
    granularity: "grapheme"
  }).segment(s)).length,
  utf8Bytes: new TextEncoder().encode(s).length
}); // { utf16Units: 11, codePoints: 7, graphemes: 1, utf8Bytes: 25 }

To run the full set, download the JSON and script into the same folder and run node unicode-character-count-reproduce.mjs. The script checks the four measurements against the stored expected values and reports the runtime.

Check your own text locally

This demonstration computes in your browser and sends no text. It contains no analytics or external script. This statement describes this standalone page, not every feature on the Chatcument domain.

How this relates to Chatcument

The Chatcument character counter uses grapheme segmentation when Intl.Segmenter is available and shows UTF-8 bytes separately. Its fallback counts code points, so older runtimes can differ for combined sequences. This report is published by Chatcument; it is not an independent review of the service.

한국어 요약: 글자 수와 바이트 수는 다른 기준입니다

가족 이모지 한 개는 이 자료의 실행 환경에서 글자 묶음 1개, 유니코드 코드 포인트 7개, UTF-16 단위 11개, UTF-8 바이트 25개입니다. 완성형 한글과 분해된 자모는 보이는 모양이 비슷해도 바이트 수가 달라질 수 있습니다. 입력을 받는 서비스가 공백·줄바꿈·이모지를 어떻게 세는지 먼저 확인하세요. 한국어 글자 수 도구와 이 표를 이용해 작은 입력부터 재현할 수 있습니다.

Method, limits and reuse

Generated on September 8, 2026 with Node v24.15.0, ICU 78.2, Unicode 17.0 and locale en. Each expected tuple was explicitly specified and checked by the generation script. The suite excludes malformed lone surrogates, line-ending conversion, script-specific tailoring and many newer or complex sequences. Font rendering does not change the encoded input; runtime segmentation data can affect results. Use the Unicode conformance data for standards testing.

The original fixture data and reproduction script are released under CC0 1.0. Reuse them without a required backlink. For readers who want a citation: Chatcument (2026), One family emoji, four different counts: 24 Unicode fixtures, this page.

Primary references

• Unicode Standard Annex #29, revision 47 — segmentation definitions and conformance.
• MDN: String length — JavaScript UTF-16 length.
• MDN: String normalize — normalization forms.