Skip to main content

Data on people’s ability in Welsh and how often they speak the language for April 2020 to March 2021.

The census of population is the key source used to measure the number of Welsh speakers in Wales. However, as the Annual Population Survey (APS) provides quarterly results, it is a useful source to look at trends in Welsh language ability between censuses.

The chart shows the results from the Annual Population Survey from 2001 to the end of March 2021. In 2001 there were 834,500 Welsh speakers. The trend decreases until 2007 and then increases again to 883,300 by the end of March 2021. The results of the 2001 and 2011 Census have also been plotted on the same chart to show that the Census estimates for the number of Welsh speakers are significantly lower; over 200,000 lower.

Main points

The Annual Population Survey was reweighted by the Office for National Statistics (ONS) in August 2021, for the period January to December 2020. The changes that the ONS have made to the weighting should reduce the bias of estimates at high levels of aggregation. However, some smaller breakdowns may be impacted negatively, and more extreme changes could be seen given the reduced size of the underlying sample since the start of the pandemic.

The comparisons with previous data, below, are based on the data that have been reweighted. Further information about the impact of the reweighting on the data is available in the ‘Changes to the survey’ section.

  • For the year ending 31 March 2021, the Annual Population Survey reported that 29.1% of people aged three or over were able to speak Welsh. This figure equates to 883,300 people.
  • This is 0.8 percentage points higher than the previous year (year ending 31 March 2020), equating to around 24,600 more people.
  • The chart shows how these figures have been gradually increasing each year since March 2010 (25.2%, 731,000), after they had been gradually declining from 2001 to 2007. The number of people reporting to be able to speak Welsh decreased from December 2018 to March 2020, before increasing again in the most recent four quarters. This increase should be treated with caution due to the change of survey mode since mid-March 2020 because of the coronavirus (COVID-19) pandemic. See ‘Changes to the survey’ below.
  • The highest numbers of Welsh speakers continue to be found in Carmarthenshire (94,600) and Gwynedd (90,700).
  • The lowest numbers of Welsh speakers are in Blaenau Gwent (10,900) and Merthyr Tydfil (11,600).
  • The highest percentages of Welsh speakers can be found in Gwynedd (76%) and the Isle of Anglesey (68%).
  • The lowest percentages of Welsh speakers are in Blaenau Gwent (16%), Monmouthshire (16%), and Bridgend (18%).
  • 15.7% (477,100) of people aged three or over reported that they spoke Welsh daily, 4.9% (149,000) weekly and 6.9% (209,300) less often. Around 1.6% (47,900) reported that they never spoke Welsh despite being able to speak it, with the remaining 70.9% not able to speak Welsh.
  • 33.5% (1,014,700) reported that they could understand spoken Welsh, 26.0% (788,100) could read and 23.8% (720,000) could write Welsh.


The Annual Population Survey (APS) is a UK-wide survey conducted by the Office for National Statistics (ONS). The APS, which began in 2004, is compiled from interviews for the Labour Force Survey (LFS), along with additional regional samples reaching an approximate sample of 320,000 respondents across the UK. In Wales, around 35,000 people are included in the survey each year. Details about how the survey is developed and carried out can be found on the Office for National Statistics website.

The census results and APS results for 2001 and 2011 have been included on the chart above to indicate the discrepancies between the two sources at the same time periods. There are a number of possible explanations for why census results would be lower than survey results. For example, the census is a statutory self-completion questionnaire while the APS is a voluntary survey, which uses face-to-face and telephone interviews. 

The APS results should not be compared with census results, nor used to measure progress towards the Welsh Government target of a million Welsh speakers by 2050. The Welsh language strategy, Cymraeg 2050, clearly states that this target was based on census data and that progress towards this target will be monitored using future census data.

A blog published by the Chief Statistician in 2019, gave a brief discussion on how to interpret the Welsh language data from the APS. More information about the differences between the APS and the census can be found in a bulletin presenting more detailed results on the Welsh language from the APS from 2001 to 2018.

Changes to the survey

Following government advice regarding the coronavirus (COVID-19) pandemic, the Annual Population Survey as well as all other Office for National Statistics (ONS) face-to-face studies about people, families and households were suspended. Further details of these changes can be found in this statement on the ONS website.

From mid-March 2020, the APS survey has been carried out by telephone only. A change in how a survey is administered can affect survey results. This set of results cover the period from April 2020 to March 2021, therefore all the interviews were carried out by telephone.

By comparing those who completed the survey over the telephone with those who completed the survey face-to-face in the period before March 2020, respondents did appear to be more likely to state that they could speak Welsh when answering the survey over the telephone.

The Office for National Statistics has continued to monitor the impact of this change to the survey method on the estimates, and has reweighted the estimates accordingly. It found that the change had led to an increase in the non-response trend, particularly from those with a country of birth or nationality outside the UK, and a higher proportion of owner-occupiers participating in the survey and a lower proportion of renters participating in the survey than before the pandemic.

Another reason for further revision of the estimates was that survey responses were weighted to national population projections. The current projections are based on demographic trends prior to the coronavirus (COVID-19) pandemic. The Office for National Statistics has now reweighted the Survey data to new populations derived using growth rates from HM Revenue and Customs (HMRC) Real Time Information (RTI) for January to December 2020 (i.e. quarters 1, 2, 3 and 4 of 2020) to adjust the estimates as a result of these changes.

The new estimates about the ability of the population aged three or over to speak Welsh are very similar to the original estimates, varying by between 0 and 0.2 percentage points (or between -600 and 4,800) from the original estimates, at a Wales level. This varies at a local authority level and for the other Welsh language skills (including frequency of speaking Welsh).

At present, it is not possible to say whether any changes in Welsh language ability between March 2020 and March 2021 is as a result of the change in the way the survey is conducted, or real changes in the population’s ability in Welsh. The results should therefore be interpreted with caution.


Datasets and interactive tools


Llio Owen

Telephone: 0300 025 5530

Rydym yn croesawu galwadau a gohebiaeth yn Gymraeg / We welcome calls and correspondence in Welsh.


Telephone: 0300 025 8099

Rydym yn croesawu galwadau yn Gymraeg / We welcome calls in Welsh.