Speak to everyone you need to hear from.

Why multilingual surveys matter.

The 2021 Census recorded Polish as the most common main language in England and Wales after English, with Romanian, Panjabi and Urdu close behind. Welsh and Scottish Gaelic have statutory standing in their nations. Afghan resettlement brought Dari and Pashto speakers into services that had never needed either. Eritrean communities need Tigrinya.

If a survey only exists in English, you don't just lose responses, you lose them non-randomly. The people you didn't hear from are likely to be the people who are least likely to be well served already. Your results misstate how things are going.

It starts as a data quality problem. It is also a compliance one:

• Welsh Language Standards. Public bodies in Wales operate under standards made under the Welsh Language (Wales) Measure 2011, which require Welsh and English to be treated equally in dealings with the public. A consultation offered only in English is not a defensible position.
• Gaelic Language Plans. Public authorities in Scotland with a plan under the Gaelic Language (Scotland) Act 2005 carry commitments about the use of Gaelic in their services and communications.
• Language rights in Northern Ireland. The Identity and Language (Northern Ireland) Act 2022 gives official recognition to Irish, alongside Ulster Scots.
• The Public Sector Equality Duty. Under section 149 of the Equality Act 2010 you need to show you gave people a genuine opportunity to take part, and native language access is how that is evidenced in practice.
• NHS communication expectations. The Accessible Information Standard (DCB1605) covers disability-related communication needs, and NHS bodies are separately expected to arrange interpreting and translation for people whose first language isn't English.

Handling this properly enables consultations to survive challenge. “We consulted” is a much weaker statement than “we consulted, in the languages our population speaks, and here is the response rate by language.”

Running a consultation for a UK public body? See how SurveyOptic fits the GOV.UK common platforms, including invitations and reminders through GOV.UK Notify in the respondent's own language.

The languages other platforms get wrong.

• Welsh, Scottish Gaelic and Irish are all in. With special features to support their unique features — Cymraeg, Gàidhlig and Gaeilge are three of the 67.
• Both Kurdish languages, kept separate. Sorani (Central Kurdish, ckb, right-to-left in Arabic script) and Kurmanji (Northern Kurdish, kmr, left-to-right in Latin script) are managed as different languages in their own scripts running in their correct directions. A Sorani speaker handed a Kurmanji form is being asked to read a language they may not read at all.
• Punjabi in both scripts. Gurmukhi (pa, left-to-right) and Shahmukhi (pnb, right-to-left) are the same spoken language written two different ways. Many platforms list a single “Punjabi” and leave you to discover which script you have after it's presented to you.
• The resettlement and asylum languages. Somali, Tigrinya, Amharic, Oromo, Afar, Dari and Pashto are all in the list — these are languages statutory services are most often asked to reach, but are most often missing from an enterprise language menu.
• Right-to-left that actually works. Seven languages run right-to-left: Arabic, Sorani Kurdish, Dari, Farsi, Pashto, Shahmukhi Punjabi and Urdu. They are shown in a fully mirrored interface.

All languages.

Every language below ships with translated interface text. Search by language, by the name people use for it themselves, or by locale code.

Showing 67 of 67 languages

Codes follow BCP 47 — the same codes a browser sends, so language detection works without configuration. All of these languages are available on the platform; you choose to which ones you want to enable on your instance and surveys.

Reporting across languages.

One survey in twelve languages is still one survey.

• Compare by language. Ask a language question or issue separate links, then break results down by language to see whether one group's experience differs from another's.
• Or ignore language entirely. Aggregate trends, benchmarks and dashboards work across the whole response set regardless of which version someone completed.
• Report in the manager's language too. Reports follow the language preference of the person receiving them — useful when your survey managers are spread across countries.
• Track representativeness. If your Urdu-language responses came back at a third of the rate of your English ones, you want to know that before you publish, not afterwards. You can even track response rates by language in real time as the survey progresses.

Multilingual results feed the same action planning and reporting as any other survey, and free-text responses in every script are held under the same UK-only data residency and security terms. For multinational staff programmes, see employee engagement.

Localisation, not just translation.

Translating the words is most of the job, but not all of it. Before you go live:

• Repoint your links. A link to an English-language support page is a dead end in a Ukrainian survey, so be sure to provide the local equivalent.
• Review your images. Check that photographs, icons and illustrations read the way you intended in each culture.
• Re-translate anything you change. If you edit the English after translation, the translated versions no longer match — re-export the manifest for the changed items.

The full operational detail — manifest format, upload, language switching and parallel surveys — is in the translation overview in our documentation.