Designing for Swahili: type, line-length and the end of placeholder labels.

Most interfaces here are designed in English first and assumed to work in Swahili. They usually do not break. They just quietly get harder to use. On one sign-in form, three small typographic decisions made a measurable difference for people who do not read English first. None of them were redesigns. All of them were details.

Swahili words run longer, so the layout has to expect it

Swahili builds meaning by stacking prefixes and suffixes onto a root, so a label that is one short word in English is often longer in Swahili. "Settings" becomes "Mipangilio". "Password" becomes "Nywila", or the longer "Neno la siri". If a button is sized to the exact width of its English text, the Swahili version wraps awkwardly or gets clipped. The fix is to build components that flex to their content, and to test with the longest realistic translation rather than the English placeholder.

Line length is a readability lever

Comfortable reading sits at roughly 50 to 75 characters per line. Because Swahili tends to run longer than English for the same meaning, a column tuned to English prose can push the Swahili version past that comfortable range, and reading speed drops. Holding body text to a sensible measure, and letting it sit a little narrower than you would for English, keeps the longer translations easy to read.

Stop using the placeholder as the label

Using a field's placeholder text as its only label is fragile in any language and worse across two. The placeholder vanishes the moment someone types, so anyone who paused to translate the field in their head has just lost the prompt. It also fails for screen readers and lowers contrast. We now keep three rules:

• Always show a visible label above the field, instead of leaning on the placeholder.
• Use the placeholder for an example, not the instruction.
• Write labels in plain, concrete language that translates cleanly, and avoid idioms with no direct equivalent.

Why the small things add up

A sign-in form is the narrowest part of the funnel. Everyone passes through it, so any friction there is paid in full. The improvement we measured did not come from a grand redesign. It came from layout that expects longer words, a readable line length, and labels that stay put. The point generalises. Designing for Swahili is mostly designing honestly for the language people actually read in, instead of treating the local language as a translation you add at the end.