More than five million people in England and Wales speak a language other than English at home, and hundreds of thousands cannot communicate confidently in English at all.

Multilingual Care Needs More Than Translation Apps

Language has quietly become one of the more overlooked risk factors in UK care delivery. Census 2021 figures put the number of people in England and Wales who cannot speak English well or at all at over a million, and that is before accounting for the far larger group who manage everyday conversation but struggle with clinical or care-related terminology under pressure, fatigue or distress. 

For years the answer has been telephone interpreting lines and, in hospitals, wheeled tablets pressed into service for video calls. Both have real limitations once care moves into the community. 

A carer visiting a service user’s flat at seven in the morning cannot wait for a phone queue, and there is no wheeled tablet to find in someone’s living room. Software that puts real-time translation directly on the mobile device a carer already carries has started to close that gap, extending language support into exactly the settings where UK social care actually happens: domiciliary visits, supported living, and community nursing rounds that never touch a hospital corridor.

The Scale of the Language Gap in UK Care

The scale is easy to underestimate. Census 2021 recorded more than ninety different main languages spoken across England and Wales, with Polish, Romanian, Panjabi and Urdu the most common after English. Ninety-one per cent of usual residents gave English, or Welsh in Wales, as their main language, which leaves close to five million people who did not. 

Most of that group are proficient English speakers who simply prefer another language at home, but around 880,000 people say they cannot speak English well, and a further 161,000 cannot speak it at all. 

Those figures are averages across the whole population; in some London boroughs and other diverse local authority areas, the proportion is considerably higher, and it is precisely those areas where domiciliary providers and ICBs are trying to deliver consistent, CQC-compliant care.

Why Point-of-Care Translation Still Falls Short

The Care Act 2014 places a duty on local authorities to ensure people can understand information about their care and support, and CQC’s inspection framework expects providers to show that communication needs are identified and met, not assumed away. 

In practice, many providers still lean on family members to interpret, a workaround that raises safeguarding concerns when the family member is also a carer, and one that fails entirely for a service user living alone. On-device AI translation, capable of handling a wide range of languages directly through a carer’s existing phone rather than a separate piece of hardware, addresses the access problem. But access on its own does not satisfy a regulator, and it does not satisfy a family who has learned, sometimes painfully, not to take a clinical conversation at face value.

Data Security Cannot Be an Afterthought

Any technology that processes what a service user says in their own home sits squarely inside the same information governance regime as anything happening on an NHS ward. Domiciliary care providers and residential homes are classed as Category 3 organisations under the NHS Data Security and Protection Toolkit, and completion is effectively mandatory for any provider handling NHS-linked data, using NHSmail, or registered with CQC. 

That obligation does not disappear because a translation tool happens to run on a personal or organisation-issued mobile phone rather than a desktop terminal. Providers deploying this kind of technology need clear answers on where conversation data is processed, who controls software updates across a distributed workforce, and how clinical and personal data stay separated on a device a carer also uses for their own life. Tools that keep language processing on the device itself, rather than routing it through third-party servers, offer a more defensible position under UK GDPR and the Data Protection Act 2018, and a simpler story to tell an auditor.

Trust Is the Variable Technology Cannot Automate

None of this solves the deeper problem on its own. A service user who has had a poor experience with the health and care system, whether through a botched interpretation, a rushed appointment, or a long history of feeling unheard, will not suddenly disengage from that pattern because a carer’s phone can now translate in real time. What changes behaviour is visibility: a person seeing their own words rendered accurately, following the conversation as it happens, and coming away confident that what they said was actually understood. That is a different design goal from simply offering more languages, and it is the one that most determines whether a household actually engages with a care plan rather than nodding along to one.

What the Sector Should Be Watching

Translation capability, information governance and lived trust are not three separate procurement decisions, they are one connected system, and treating any of them in isolation creates exposure somewhere else. A provider with fluent translation but shaky device security risks a DSPT finding and a CQC concern in the same visit. 

A provider with airtight security but clunky, impersonal translation tools risks the service user disengaging anyway. As more ICBs and local authorities push digital tools into domiciliary and community settings under the drive for integrated, home-first care, procurement teams would do well to ask suppliers to demonstrate all three strands together, rather than taking translation coverage as a proxy for the whole problem being solved.

Language access in UK care is not a niche accessibility feature, it is core patient safety and core commissioning risk. The providers and ICBs that treat translation, security and trust as one integrated requirement, rather than three items on separate tender documents, are the ones likely to avoid both the safeguarding incident and the compliance finding.