Cursive to normal text

Why the usual fixes do nothing

Cursive text from a generator is not formatted text. Each letter is a separate Unicode character that happens to look like a script letter — 𝓗 is U+1D4D7, a character in its own right, not an H in a fancy font. That is why it survives being pasted into an Instagram bio, and also why none of the usual tools can undo it. Clear Formatting in Word or Google Docs, choosing another font, or retyping the surrounding text leaves the characters as they are, and a search for Hello will not find 𝓗𝓮𝓵𝓵𝓸, because the two share no characters at all.

Undoing it means swapping each character for the ordinary letter it stands for. You can do that by hand for a word; the box above does it for a paragraph, and tells you which alphabets it found on the way.

How the conversion works

Unicode records, for most fancy letters, what they are a variant of. The entry for 𝓗 in the Unicode Character Database reads <font> 0048: a font variant of U+0048, the capital H. Circled ⓗ is marked <circle> 0068, fullwidth ｈ is <wide> 0068, tiny ʰ is <super> 0068. Unicode’s compatibility normalisation, NFKC, follows those notes, and that is the core of the converter.

It is not applied to the whole text, though, because NFKC also rewrites things nobody would call fancy. It turns ™ into TM, ½ into 1⁄2 and the fi ligature into fi, and it would reduce the ℹ️ emoji to a plain i. So the converter only follows the notes for characters from the blocks fancy-text generators draw on — the maths alphabets, circled and squared letters, fullwidth forms, superscripts — and within the Letterlike Symbols block only when the answer is a single letter, which catches ℌ and ℬ but not ™.

Three popular families carry no such note, so they are matched from a table instead: small capitals such as ꜱᴍᴀʟʟ, the white-on-black letters 🅐 and 🅰, and the dingbat numbers ❶ and ➊. Underline and strikethrough marks are dropped, as are the stacks of marks in “glitch” text, while a single real accent such as the one in é stays where it is.

What it leaves alone, and why

• Accented letters and other languages. é, ø, Привет, Ωμέγα and 漢字 are already normal text for someone, so they pass through untouched.
• Cyrillic and Greek look-alikes. Some generators swap a Latin a for the Cyrillic а, which looks identical. Converting every Cyrillic а would wreck Russian text, so the converter names any it finds inside a Latin word and leaves the retyping to you.
• Emoji. 🅰️ with its emoji presentation, keycaps such as 1️⃣, and the letter pairs that make flags such as 🇺🇸 stay as they are, unless you tick the box to remove emoji and symbols altogether.
• Upside-down and mirrored text. These are built from phonetic letters such as ɐ and ǝ, in reverse order. They are real characters with meanings of their own, and there is no record in Unicode of which letter each was meant to stand in for.
• Spacing made with ordinary spaces. The invisible hair and zero-width spaces that some styles use are removed, but a generator that put a normal space between every letter produced text the converter cannot tell apart from separate words.

When plain text is the better choice

Some places read the characters rather than looking at them. A screen reader announces 𝓗as “mathematical bold script capital H” or skips it, a search box will not match a cursive name to its plain spelling, and a device without the right font shows empty boxes instead of letters. For a job title, a name people will search for, or anything someone needs to read aloud, convert it back and keep the cursive for the flourish around it. The compatibility guide goes through which profile fields take cursive at all.