The change in words-n-number is as follows:
No more "words" formed by several emojis. This because each emoji in a sense are words. Will also work better with search engines as each can be used as a filter if you have an emoji facet to filter on.
So instead of:
cons stringOfWords = 'A ticket to 大阪 costs ¥2000 👌😄 😢'
extract(stringOfWords, { regex: emojis})
// ['👌😄', '😢']...you will get:
cons stringOfWords = 'A ticket to 大阪 costs ¥2000 👌😄 😢'
extract(stringOfWords, { regex: emojis})
// ['👌', '😄', '😢']