QA & testing · · 9 min read
A practical checklist for testing multilingual forms
Most form testing happens with the same handful of English test values: "John Smith", "test@example.com", "123 Main St". That's enough to catch obvious breakage, but it misses an entire category of bugs that only shows up once real international users start typing in their own language.
Below is a field-by-field checklist for catching those bugs before your users do — ten specific fields, each with a real bug pattern and how to catch it before it ships.
- Cyrillic addresses run 20–40% longer than English — fixed
maxlengthvalues truncate them silently. - Turkish's dotless ı/İ breaks naive case-folding used for email/username deduplication.
- Ambiguous dates (01/02) hide day/month swap bugs — always test with day > 12.
- RTL layout bugs usually aren't about translation — they're about a forgotten
dir="rtl".
- Name fields: Cyrillic, CJK, and multi-part names
- Address fields: Cyrillic and diacritic overflow
- The Turkish dotless-i problem
- Phone number fields: format and length variance
- Date fields: format ambiguity
- Currency and number fields: separators
- RTL layout: Arabic and Hebrew
- Text expansion in UI chrome
- Postal codes: format, length, and "no postal code at all"
- Name order and honorifics
- Running this checklist without typing it by hand
1. Name fields: Cyrillic, CJK, and multi-part names
Test with names in the scripts your actual users will use — Cyrillic (Иван Петров), CJK (田中太郎), and names with more parts than "first + last" (Spanish double surnames, Arabic patronymics). Common bugs: character-count limits set for Latin text truncate CJK names mid-character; regex validators like ^[A-Za-z\s]+$ reject anything non-Latin outright; name-splitting logic that assumes exactly two words breaks on names with three or more parts.
2. Address fields: Cyrillic and diacritic overflow
Addresses are one of the most locale-sensitive fields in any form. Russian and Ukrainian addresses in Cyrillic are frequently 20–40% longer in character count than the English equivalent for the same real-world address, because street-type words and abbreviations don't compress the same way. If an address input has a fixed maxlength tuned against English test data, it will silently truncate real Cyrillic addresses. Test German addresses too — umlauts (ä, ö, ü) and ß are simple to type but easy to mishandle in uppercase-transform CSS (text-transform: uppercase turns ß into SS in proper German typography, but most browsers just render it unchanged) and in sorting/search logic that doesn't normalize diacritics.
3. The Turkish dotless-i problem
This one catches almost everyone. Turkish has two distinct pairs of the letter i: dotted İ/i and dotless I/ı. Standard case-folding — the kind built into String.prototype.toUpperCase() in most languages when running under a Turkish locale — turns lowercase "i" into dotless "I" and uppercase "I" into... well, it depends on the runtime, and that's exactly the bug. If your form does client-side or server-side case-normalization on an email or username field (e.g., lowercasing before a uniqueness check) and your server happens to run with a Turkish system locale, "Istanbul" and "istanbul" can normalize to different strings than you'd expect from testing in an English locale, breaking deduplication or login. The fix is to always use locale-invariant case folding (e.g., explicitly pass a locale, or use a Unicode-aware case-fold function) rather than relying on the runtime's default locale.
This bug is invisible in almost every test environment, because most CI runners and dev machines default to an English or "C" locale. It only shows up in production if a server happens to be provisioned with tr_TR — which makes it a classic "works everywhere except one customer's exact setup" report.
4. Phone number fields: format and length variance
Phone number length and formatting conventions vary enormously by country — not just digit count, but grouping (spaces, dashes, parentheses) and whether a leading country code or trunk prefix is expected. A validation regex built against a US 10-digit format will reject valid German (up to 11 digits after country code), Chinese (11 digits, no area code), or Belarusian numbers outright. Test with real-format numbers per target country rather than a single template.
5. Date fields: format ambiguity
01/02/2026 means January 2nd in the US and February 1st almost everywhere else. If a form accepts free-text date entry rather than a proper date picker, test with an unambiguous date like the 13th of a month (13/05/2026) in every supported locale to catch day/month swapping — an ambiguous date like 01/02 will pass silently even when the parsing is backwards.
6. Currency and number fields: decimal and thousands separators
English uses a period for decimals and a comma for thousands (1,234.56); most of continental Europe does the reverse (1.234,56); some locales use a space or narrow no-break space for thousands grouping. A price or quantity field that only accepts a period as the decimal separator will reject perfectly valid input from users who type numbers the way their own OS and keyboard produce them.
7. RTL layout: Arabic and Hebrew
Right-to-left languages don't just need translated labels — the whole form layout should mirror direction (labels, icons, the visual order of "Next" vs "Back" buttons). A very common bug is a form that translates the copy to Arabic but leaves dir="ltr" on the container, so RTL text renders left-aligned and reads awkwardly, or bidi punctuation (mixing an English email address inline with Arabic text) renders in the wrong visual order.
8. Text expansion in UI chrome, not just inputs
Translated button labels, placeholder text, and validation error messages are often 30–200% longer in German, Finnish, or Russian than in English. A "Submit" button sized exactly to that word's English width will clip or wrap awkwardly once localized. Test layout with real translated strings, not just translated input data.
9. Postal codes: format, length, and "no postal code at all"
US ZIP codes are 5 digits (or 9 with a hyphen), which is what most postal-code regexes are quietly built against — but UK postcodes are alphanumeric with a variable-length format (SW1A 1AA), Canadian ones interleave letters and digits (K1A 0B1), Japanese postal codes use a fixed NNN-NNNN format, and several countries (Ireland historically, and a number of small nations) don't use postal codes at all. A required field with a digits-only pattern silently locks out every one of these; the safer default is to treat postal code as free text unless you're validating against a specific known-country format.
10. Name order and honorifics
Not every culture writes given-name-first. Japanese, Hungarian, and Chinese names conventionally put the family name first in their own language (though often reordered to given-first when romanized for an English-language form — which is itself a source of mismatches between a user's legal documents and what they type). Forms that hardcode a "First name / Last name" pair and concatenate them in that fixed order for anything user-facing (a greeting email, a printed label) can produce a name in the wrong order for these users. Similarly, honorific/title fields hardcoded to a fixed English list (Mr./Mrs./Ms./Dr.) don't map cleanly onto every locale's own conventions, and an increasing number of products are dropping the field entirely rather than trying to localize it correctly — worth considering if it's not load-bearing for your form.
Running this checklist without typing it by hand every time
Manually typing Cyrillic addresses, Turkish dotless-i names, and RTL Arabic text into every field of every form you test is tedious enough that most teams just skip it — which is exactly why these bugs make it to production. This is precisely the gap Auto Fill Fake Data is built to close: pick a locale, and every field on the page fills with realistic, correctly-formatted data for that locale — Cyrillic names and addresses for Russian, umlaut-correct German text, properly formatted phone numbers and dates per country — in one click, repeatable across as many locales as you need to support.
FAQ
Further reading
- Why form fillers break on React (and Vue, and Angular) forms — the technique that makes filled values actually stick.
- Auto Fill Fake Data vs Fake Filler: an honest comparison — how the two extensions differ on localization specifically.
Stop hand-typing test data in eight languages. Pick a locale, click once, and every field fills correctly.
See supported locales