Knowledge baseExam Center
ENEnglish
Choose language13

CKTIS / JavaScript

Intl.Segmenter: segment() and isWordLike

Split text into locale-sensitive segments and use `isWordLike` to distinguish segments treated as words from the other segments in the result.

Version: ECMAScript 2024 and the current web platform; runtime support varies

Syntax

const segmenter = new Intl.Segmenter(locales, { granularity: "word" });
const segments = segmenter.segment(input);
for (const { segment, index, input: source, isWordLike } of segments) {
  // Use the segment data here.
}

Intl.Segmenter.prototype.segment and isWordLike serve different roles: segment(input) is an instance method, while isWordLike is a property on individual segment records in word mode. The method returns a new iterable Segments object, not an array. Boundaries use the resolved locale and the granularity chosen in the constructor: "grapheme" (the default, for grapheme clusters), "word", or "sentence". The method converts its argument to a string. A Symbol argument or an invalid this receiver causes a TypeError.

Each record yielded by iteration contains segment (the substring), index (its zero-based start in UTF-16 code units), and input (the entire converted string). The exclusive end is index + segment.length. The index is not a count of visible characters.

Only in "word" mode does each record have a Boolean isWordLike property: true means the segment is treated as word-like, and false means it isn't. Spaces and punctuation generally remain in the result with false, so filter the records if you want only word-like segments. In the other modes, the property is absent rather than set to false.

The ECMAScript 2024 edition of ECMA-402 specifies this API, but runtime support varies. Both boundaries and classification can vary by locale and implementation. Don't assume identical results across runtimes.

Examples

const segmenter = new Intl.Segmenter("en", { granularity: "word" });
const words = [];

for (const { segment, index, isWordLike } of segmenter.segment("Hi, Sam!")) {
  console.log(JSON.stringify([index, segment, isWordLike]));
  if (isWordLike) words.push(segment);
}

console.log(JSON.stringify(words));

Result

[0,"Hi",true]
[2,",",false]
[3," ",false]
[4,"Sam",true]
[7,"!",false]
["Hi","Sam"]

The first five lines show the index, substring, and classification, including the space and punctuation. The last line contains only the segments marked `true`. This is the expected output for this English text, not a guarantee across runtimes.

Related lessons

  • JavaScript · Lesson 24 · Advanced · Version: 4

Sources