1. Main points
In Wales, a question about Welsh language has been included in the census since 1891; people self-assessed their own Welsh language skills, or the skills of others if they were responding on someone else's behalf, such as parents or guardians reporting on behalf of children.
In Census 2021, people self-assessed whether they could understand spoken Welsh, and speak, read or write Welsh; people could select a single skill, multiple skills, or no Welsh language skills.
In 2021, an estimated 538,000 usual residents in Wales aged three years and over (17.8%) reported being able to speak Welsh, which is a decrease since 2011 (562,000, 19.0%).
One of the main factors contributing to the overall decrease in the percentage of people who reported being able to speak Welsh between 2011 and 2021 was the decrease in children and young people aged 3 to 15 years who reported this skill.
The percentage of usual residents aged three years and over able to speak Welsh decreased between 2011 and 2021 in all local authorities except Cardiff, Vale of Glamorgan, Rhondda Cynon Taf, and Merthyr Tydfil.
Among children and young people aged 3 to 15 years, the percentage who could speak Welsh decreased in all local authorities between 2011 and 2021.
Tell us what you think about this publication by answering a few questions.
Back to table of contents2. Welsh language data
Welsh language skills (speaking)
Dataset | Released 6 December 2022
Census 2021 estimates that classify usual residents aged three years and over in Wales by their ability to speak Welsh. The estimates are as at Census Day, 21 March 2021.
Welsh language skills (speaking) by single year of age
Dataset | Released 6 December 2022
Census 2021 estimates that classify usual residents aged three years and over in Wales by their ability to speak Welsh by age. The estimates are as at Census Day, 21 March 2021.
Welsh language skills (understanding)
Dataset | Released 6 December 2022
Census 2021 estimates that classify usual residents aged three years and over in Wales by their ability to understand Welsh. The estimates are as at Census Day, 21 March 2021.
Welsh language skills (reading)
Dataset | Released 6 December 2022
Census 2021 estimates that classify usual residents aged three years and over in Wales by their ability to read Welsh. The estimates are as at Census Day, 21 March 2021.
Welsh language skills (writing)
Dataset | Released 6 December 2022
Census 2021 estimates that classify usual residents aged three years and over in Wales by their ability to write Welsh. The estimates are as at Census Day, 21 March 2021.
Welsh language skills (detailed)
Dataset | Released 6 December 2022
Census 2021 estimates that classify usual residents aged three years and over in Wales by their overall Welsh language skills. The estimates are as at Census Day, 21 March 2021.
3. Measuring the data
Reference date
The census provides estimates of the characteristics of all people and households in England and Wales on Census Day, 21 March 2021. It is carried out every 10 years and gives us the most accurate estimate of all the people and households in England and Wales.
We are responsible for carrying out the census in England and Wales but will also release outputs for the UK in partnership with the Welsh Government, the National Records of Scotland (NRS) and the Northern Ireland Statistics and Research Agency (NISRA). The census in Northern Ireland was also conducted on 21 March 2021, whereas Scotland's census was moved to 20 March 2022. All UK census offices are working closely together to understand how this difference in reference dates will impact UK-wide population and housing statistics, in terms of both timing and scope.
Response rate
The person response rate is the number of usual residents for whom individual details were provided on a returned questionnaire, divided by the estimated usual resident population.
The person response rate for Census 2021 in Wales was 96.4% of the usual resident population, and over 94% in all local authorities. The proportion of returns submitted online was lower in Wales (68%) than in England (90%). This is likely because Wales had a higher percentage than England of households in which initial contact was with a paper questionnaire rather than an online access code (50% in Wales compared with 9% in England), as they were in areas where the take-up of the online option was expected to be low.
Read more about question-specific response rates for England and Wales in Section 6 of our Item editing and imputation process for Census 2021, England and Wales methodology.
Self-identified assessment
Welsh language data from Census 2021 come from a self-assessment of ability.
The Census 2021 question asked people to assess their ability to:
understand spoken Welsh
speak Welsh
read Welsh
write Welsh
Everyone will assess their own language skills differently. This means that two people with the same skills in Welsh language may give different answers about their ability. It could also be an assessment of another person’s ability if someone is answering on behalf of someone else. For example, a parent could be answering on behalf of a child. The parent may not know how well a child can understand, speak, read, or write Welsh.
Read more about quality assurance in our Welsh language quality information for Census 2021 methodology.
Back to table of contents5. Cite this statistical bulletin
Office for National Statistics (ONS), released 6 December 2022, ONS website, statistical bulletin, Welsh language, Wales: Census 2021