Chief Statistician’s update: a discussion about the Welsh language data from the Annual Population Survey

Today we published the latest results on the Welsh language from the Annual Population Survey (APS) on our StatsWales website. We’ve been publishing these data for some time but in the past year there has been increasing user interest in the data owing to the gradual increase in Welsh speakers suggested by the data, and in light of the Welsh Government’s Welsh language strategy, Cymraeg 2050, which includes the aim of achieving a million Welsh speakers by 2050.

Darllenwch y dudalen hon yn Gymraeg

In this blog we explain the background to these data, what they are showing, and discuss how they could be interpreted.

What do the data show?

The survey reported that for the year ending in December 2018, 898,700 people or 29.9% of people aged three or over were able to speak Welsh.

This is the largest number of Welsh speakers that has ever been reported by the APS, although similar percentages were seen back in 2001, and continues its gradual trend over the past decade of more people reporting to the survey that they are able to speak Welsh.

Can these data be used to assess progress towards a million Welsh speakers?

No. The Census of Population is the main source of information about the number of Welsh speakers in Wales. The last Census in 2011 reported that 562,016 people were able to speak Welsh, which is considerably lower than the APS.

The Welsh language strategy Cymraeg 2050 clearly states that the trajectory towards a million Welsh speakers was based on 2011 Census data and progress towards this target will be monitored using future census data.

Nevertheless, the APS provides us with more regular estimates of the number of Welsh speakers, and we therefore publish the APS data to be able to monitor trends between Censuses.

It’s important to note that the primary purpose of the APS is to provide employment-related statistics, as well as contextual information on social and socio-economic variables at a local level. The question on the Welsh language is included for cross analysis purposes, and not to provide a count of the number of people who can speak Welsh.

Can we believe these APS results?

The sample size for the APS is very large (over 30,000 respondents every year), but we do need to understand more about why the survey results differ significantly from the Census and whether the recent increases seen in the APS are credible.

The APS results have consistently been much higher than the Census results. The chart below shows how the APS results have been gradually increasing each year since March 2010 (25.2%, 731,000), after they had been gradually declining from 2001 to 2007.  The latest results bring the percentage of Welsh speakers back in line with the levels reported by the APS in 2001 (when 30.0%, or 834,500 people were reported to be able to speak Welsh).

The chart shows the results of the APS from 2001 to 2018. In 2001 there were 834,500 Welsh speakers. The trend declines to 2007 and then increases again to 898,700 by 2018

The Census results for 2001 and 2011 have also been plotted on this chart to show the differences between the two sources at the same time. It can be seen that, between the 2001 and 2011 Censuses, there was a decrease of 20,000 Welsh speakers.  The APS in 2001 and 2011 also reported a decline for the same time period, but of 65,500.

While survey data can fluctuate from quarter to quarter, the APS results do seem to show a steady increase over recent years, which indicates that the number of Welsh speakers could be increasing.

When looking at these results by age and over time, we can see that those aged 3 to 15 are the most likely to be able to speak Welsh, and that it is this age group that has seen the greatest increase in the percentage of Welsh speakers over the past decade. A third of the increase in the number of Welsh speakers since 2008 can be attributed to this age group. It is worth noting here that for this age group the responses are given by their parents or other adults within their household.

This chart shows the percentage of people who speak Welsh in 5 age groups. The 3 to 15 year olds have the highest percentages, which fluctuate over the years at around 50% followed by those aged 16 to 24 (at around 35%). Then the three age groups 25 to 44, 45 to 64 and 65 or older are fairly similar to each other at around 22%. By 2018 59% of the 3 to 15 year olds report being able to speak Welsh, 37% of 16 to 24 year olds with 24%, 22% and 21% of the other three age groups respectively.

It is not clear to us whether the increase shown for those aged 3 to 15 over recent years is real or due to a change in parents’ perception of their children’s ability in Welsh for some reason. The increase seen here for this age group is not as evident when analysing administrative data reported by schools.

There have also been smaller increases for other age groups in recent periods, but overall these have recovered these age groups to levels that are similar to what they were in 2001.

Why are the Census results and the APS so different?

The wording of the question about Welsh language ability included in the APS is identical to the question in the Census. However, there are a number of differences between the sources, which could explain why the results differ.

For example, the Census is a statutory self-completion questionnaire, whilst the APS is a voluntary survey, which uses face-to-face interviews. When a respondent is completing the Census, they may not read all of the instructions and may not answer the question as intended. A respondent for the APS, on the other hand, will have an interviewer to assist them.  However, the presence of an interviewer may lead a respondent to provide a more socially desirable response.

It should also be borne in mind that speakers’ assessment of whether they can speak a language is quite subjective, and for some it’s not easy to select either ‘yes’ or ‘no’. The ability to say a few words in the language is enough for some to say that they speak it. Others, despite speaking it regularly, may say that they can’t speak it if they feel that they are more comfortable speaking another language.

For the APS, the respondent can explain and discuss their ability with the interviewer, thus allowing the interviewer to make a judgement on their ability, and with interviewers not wishing to offend the respondent, this might increase their likelihood of being categorised as a Welsh speaker than not. For the Census, the respondent must decide for themselves which box to tick.

More information about the differences between the two sources can be found on in a report presenting National Survey results on the Welsh language (page 4).

We are undertaking further research into the responses to surveys relating to the Welsh language and the differences between people’s responses in surveys and the Census and hope to report on this in the next few months.

What do other surveys tell us?

The National Survey for Wales has also tended to have a higher percentage of Welsh speakers than the Census. The most recent National Survey results (2017-18), published on 20 June 2018, showed that 19% of the population aged 16 and over were able to speak Welsh, with an additional 12% stating that they had ‘some Welsh speaking ability’.

These results reflect a similar trend to that seen in the Welsh Language Use Surveys of 2013-15 – that is, an increasing percentage of people reporting that they have ‘some Welsh speaking ability’ (possibly amounting to just a little Welsh).

What can we conclude?

These results highlight the issues around the subjective nature of self-assessment of language ability and what this means when counting the numbers of Welsh speakers. We have been aware for some time that there is a greater tendency for survey respondents to say they speak Welsh than in the Census, but we can only speculate about the possible reasons for this.

We are clear that the APS results should not be used to measure progress towards the target of a million Welsh speakers. The very fact that the figures now are back to levels previously seen in 2001 when the census figure was 582,368 demonstrates that caution needs to be exercised in interpreting the figures.

There does seem to be a positive trend in the numbers reporting that they speak Welsh in surveys: both in the APS and for those reporting to have ‘some Welsh language ability’ in the National Survey. And as stated above, we need to understand whether the trend observed for those aged 3 to 15 is due to a real change in Welsh speaking ability, or a greater tendency for parents to report that their children are able to speak Welsh.

In conclusion we would advise that trends in the APS data should not be seen as conclusive evidence of trends in the Welsh language, but as a useful indicator of potential trends which should be used in conjunction with other data.

We’ll have to wait for the results of the 2021 Census to understand the progress towards the target of a million Welsh speakers.

What next?

In May we will be publishing a statistical article on the latest results of the APS which will explore how the data on the Welsh language has changed over the past 18 years as well exploring further some of the issues discussed in this blog.

Glyn Jones
Chief Statistician

27 March 2019