Small Text Generator

There is no font size in copied text, so small text is made by swapping each letter for a different character that is drawn small. Unicode has three families of them. Small capitals sit on the line and are the height of a lowercase x. Superscript letters float at the top of the line, and subscript letters hang below it. Each family has its own gaps, which is why the rows above do not all look complete.

Small caps keeps the most of your text readable, because the letters are full capitals in shape. Superscript is the one to pick for “tiny text” in a bio or a joke whisper, since it covers every letter but q. Subscript is the patchiest and works best for short words that happen to use its letters, such as ₛₒ ₜᵢₙy, or for the numbers in formulas.

Unicode only encodes a letter when someone shows it is used in real writing, usually phonetics or a minority language. No one needed a superscript q or a subscript b, so for decades they simply did not exist. A few gaps have been filled recently after proposals from phoneticians, but fonts take years to catch up.

The tick box in Fine-tune uses these newer letters. Leave it off if your text must look right for everyone: a font released before the letter existed has nothing to draw, so readers see an empty box or a question mark instead. The subscript w, y and z are so new that almost no device will show them yet.

Worked example. “Good vibes only” has 13 letters. In Small caps and Superscript all 13 shrink: ɢᴏᴏᴅ ᴠɪʙᴇꜱ ᴏɴʟʏ and ᵍᵒᵒᵈ ᵛⁱᵇᵉˢ ᵒⁿˡʸ. In Subscript four stay normal, the g, d, b and y, giving gₒₒd ᵥᵢbₑₛ ₒₙₗy. With the newer letters switched on, the y becomes ₞ and three remain. Superscript caps gives ᴳᴼᴼᴰ ⱽᴵᴮᴱˢ ᴼᴺᴸʸ, where the s and y are small raised letters standing in for missing capitals.

Where a letter has no small form, the generator leaves it as a lowercase letter rather than swapping in a character that only looks similar, such as a Greek ᵦ for b. Look-alikes from other alphabets read as different letters to search, spell-checkers and screen readers. A lowercase x is in fact the same height as a small capital, so the missing small capital X is hard to spot.

Almost every character on this page started out as a sound symbol. The International Phonetic Alphabet uses small capitals for sounds that differ from the plain letter: ʀ is the uvular trill of French and German r, ɢ a g made further back in the throat, ɪ the vowel in “bit” and ʏ the vowel in German “hübsch”. Those were in Unicode from version 1.1 in 1993.

Raised letters mark how a sound is said. English p is pronounced with a puff of air in “pin” but not in “spin”, so phoneticians write [pʰɪn] and [spɪn]; ʷ marks rounded lips and ʲ a y-like glide. Most of the small capitals, such as ᴀ ᴄ ᴅ, and the first subscript letters ᵢ ᵣ ᵤ ᵥ arrived with the Phonetic Extensions block in Unicode 4.0 in 2003, largely for the Uralic Phonetic Alphabet used to transcribe Finnish, Hungarian and related languages. More subscript letters followed in 2005 and 2010.

This is why the sets are uneven. Each letter exists because a transcription system needed it, not because anyone was building a decorative alphabet. Unicode classes the raised letters as modifier letters and gives them a compatibility link to the plain letter; the small capitals have no such link, which matters for search below.

In typesetting, small caps are a feature of the font: capitals redrawn at about lowercase height, with strokes as thick as the rest of the text. Books set acronyms such as ɴᴀꜱᴀ in them so they do not shout from the page, many English Bibles print Lᴏʀᴅ that way where the Hebrew has the divine name, and some style guides use them for ᴀᴍ and ᴘᴍ. The Caps + small caps row copies that look, with capitals full height at the start of each word.

If you can choose a font, use the real thing instead. It keeps your actual letters, so search, spell-check and copy and paste all still work.

When a font has no small caps, browsers and Word fake them by shrinking ordinary capitals. The result has thin, spindly strokes beside the regular letters. In CSS, font-synthesis-small-caps: none turns off that faking so you can see whether your font really has them.

Turning a whole formula into superscript raises the letters too, so the Formula row works differently. It only changes what you mark, using the same signs as calculators and LaTeX, and treats a number straight after a letter or bracket as a chemical count.

The count rule fires on any digit after a letter, so model numbers such as mp3 also become mp₃. Use the Formula row for formulas only, or mark exponents by hand and copy just the part you need.

Footnote marks¹ and powers use the raised digits, but those come from two places. ¹, ² and ³ were in the Latin-1 character set from 1987, while ⁴ to ⁹ and ⁰ were added later in a separate block. Fonts often design them separately, so in a number like 10¹²³⁴ the 1, 2 and 3 can sit at a slightly different height or weight from the 4. If that matters, check it in the place you will paste it.

Word raises ordinals such as 1st on its own (AutoFormat As You Type, “Ordinals with superscript”), using formatting rather than special characters. Its shortcuts for any selected text are Ctrl+Shift+= for superscript and Ctrl+= for subscript. On a web page, the HTML <sup> and <sub> tags do the same.

Software that cleans up text with NFKC normalisation, which many search boxes and username checks do, folds superscript and subscript letters back into plain ones. Small capitals have no such mapping, so a small caps word stays a string of phonetic symbols that a search for the plain word will not find.

Two smaller traps. The raised minus ⁻ folds to the mathematical minus sign (U+2212), not a hyphen, so superscript “x-ray” normalises to x−ray and still misses a search for x-ray. And when this page turns small caps back into plain text, it cannot know which letters were capitals, so ꜱᴍᴀʟʟ ᴛᴇxᴛ always comes back as “small text”.

Screen readers have the same problem as search, often reading each character by its name. The cursive text generator page covers that, and how many characters fancy text costs towards a bio or post limit.