The words a text holds, as a person counts them: the segments Intl.Segmenter cuts at the word granularity and
marks word-like (letters, numbers, ideographs), so don't and 1.2M are one word each and punctuation is none.
Chinese and Japanese are cut by a dictionary, so two runtimes can differ by a word or two on such a text. Where the
runtime has no Intl.Segmenter, or no Intl at all, the runs between white space.
The words a text holds, as a person counts them: the segments
Intl.Segmentercuts at thewordgranularity and marks word-like (letters, numbers, ideographs), sodon'tand1.2Mare one word each and punctuation is none. Chinese and Japanese are cut by a dictionary, so two runtimes can differ by a word or two on such a text. Where the runtime has noIntl.Segmenter, or noIntlat all, the runs between white space.