Most people using the products I work on are preparing for a US licensing exam in a language that isn't
their first. Many are studying after a full shift. Some have not sat a formal written test in twenty
years.
The thing I keep relearning: they are not bad at the test. The test is being delivered through a
layer they have to fight. Every bit of that layer you remove goes straight into their score.
Here is what I've learned about removing it.
Negation is the single most expensive construction
"Which of the following is not required when…"
A fluent reader handles this without noticing. A second-language reader has to hold the entire option
list in working memory and invert it, in a language they are translating as they go, while a timer
runs. The question stops testing the subject and starts testing processing load.
Official exam material is full of negation and you often can't change that — but everything around it
is yours. Explanations, hints, onboarding, settings labels. I write those in the positive wherever
there's a choice, and when I can't, the negation gets visual weight so it is impossible to skim past.
Double negation — "which is not permitted unless…" — I treat as a defect.
Short sentences are not dumbing down
There's a reflex, especially when the subject is regulatory, to write in the register of the source
material. Long clauses, stacked conditions, formal connectives.
That register exists for legal precision, not comprehension. You can keep the precision and lose the
sentence length: one idea per sentence, conditions as a list rather than a chain, the subject near the
verb.
What this is not is simplification of content. These are adults preparing for a professional
qualification. Patronizing them is its own failure, and they notice it faster than you'd think. The
goal is to make the language transparent so the difficulty lives entirely in the subject, which is
where it belongs.
Translation is not the same as localization
Translating a question correctly and making it answerable are different jobs.
A term of art may have no equivalent in the target language. The honest handling is often to give the
translation and keep the English term alongside it — because the test and the
examiner will both be in English. A learner who only ever sees the translated term will meet the
English one for the first time under exam conditions.
That means some things deliberately stay untranslated. Working out which is domain judgment, not a
language decision, and it's the part that can't be outsourced to a translator who doesn't know the
exam.
Icons help less than you want them to
Icons feel like the obvious fix for a language barrier. In practice they are their own vocabulary,
and one that is strongly culturally conditioned. A floppy disk means save to people who have never
seen a floppy disk, and means nothing to people who have never seen the convention.
Where icons genuinely help: reinforcing a label you have already written in words. Where they fail:
replacing it. I use them as redundancy, not as compression — the exception being domain symbols the
learner must recognize anyway, where showing them is the content.
Time pressure multiplies everything
Every problem above gets worse under a clock, and the exam has one. A sentence that costs a fluent
reader two seconds and a second-language reader eight is a rounding error in practice and a
catastrophe across a full test.
So I treat reading time as a real budget. If a screen can be understood in one pass without
re-reading, it's done. If my eye goes back, it isn't — and my eye is the easy case.
Don't make them read anything they don't have to
The highest-leverage work isn't rewriting sentences. It's removing screens.
Every piece of interface copy is a tax paid in a second language. Onboarding that explains a feature
the user could discover in two taps, a confirmation dialog for a reversible action, a tooltip
restating the label it's attached to — in English these are mild clutter. Here they are a cost with
no return.
Why this matters beyond this audience
Everything above is just good interface writing with the volume turned up. Short sentences, positive
constructions, words doing real work, fewer screens — none of it is special pleading for non-native
speakers. It's what helps a tired native speaker on a phone at eleven at night too.
Designing for the harder case makes the easy case better. It just makes the failures impossible to
ignore, which is the useful part.
I build and operate multilingual mobile applications, shipped across ten languages.
More at akmalu.com.
Top comments (0)