Dave Sayers | 21st January 2025 | Policy Papers
In the Welsh Government’s language policy, a predominant focus on increasing the number of Welsh speakers has crowded out concerns with social inclusion and wellbeing.
This means a lack of attention to key unintended consequences that are revealed by a raft of recent studies – including unequal educational attainment and heightened social divisions.
Meanwhile, a headline focus on creating a ‘bilingual Wales’ overlooks Wales’ unique potential for not just bilingualism but multilingualism – there are some promising policy signs here, which could be built upon to Wales’ advantage.
To remedy all these issues, this paper proposes a novel ‘doughnut language policy’ model, borrowed from recent advances in economics, designed to balance a wider range of factors.
The doughnut model would mean a very different structure of policy text, with equally weighted sections covering a range of factors other than language itself – understanding that these all contribute to linguistic vitality in different ways.
Correspondingly, there is much potential for research funding to be diversified into these broader avenues for supporting bilingualism and multilingualism; and for encouraging teachers to embrace more fluidity and mixing of languages.
Introduction
Some schools in Wales use only Welsh, some English and Welsh, some mostly English; the proportion in Welsh is steadily growing (data here and here).
My analysis of flagship Welsh Government language policies in 2003, 2012, and 2017 shows a consistent focus on increasing the number of Welsh speakers to protect the Welsh language. The language is seen as an entity to be protected, and schooling is the primary means to this end. Much less attention is paid to whether all this actually increases people’s wellbeing. Such benefits might feel obvious, but they are neither planned for nor evidenced at policy level.
Meanwhile a growing body of research in education, economics, sociology, and linguistics, shows unintended negative consequences. These are not inevitable in bilingual education, but they are missed by policies centred tightly on the language. All this grates against the Welsh Government’s “absolute duty” to equality.
Complex social issues cannot be effectively addressed in isolation. I propose a ‘doughnut’ model borrowed from economics, holistically encompassing a range of social factors alongside language, aiming to benefit all at once.
The primary focus on the Welsh language
The Welsh Government publishes flagship multi-year policies across all policy briefs. For language policy, the three most recent are:
To weigh up the priorities in these texts, I adapted a framework from Belgian political scientist Helder De Schutter (e.g. 2007). He notes three broad goals in any language policy:
to promote people’s identity (he calls this ‘constitutive’)
to raise wellbeing (‘instrumental’)
to protect the language as a good in and of itself (‘intrinsic’)
See here and here for more detail of my method, and for why ‘constitutive’ and ‘intrinsic’ are split in two in Fig. 1; see also this study and this study for similar analyses on Welsh social policy. A bigger framework of ideologies in language policy is available in Sayers (2024).
Figure 1. The frequency of each goal in the three flagship Welsh language policy texts (as a percentage of the total).
Here is an example of the ‘intrinsic’ goal, where the language is a priority in and of itself (more examples of all three goals are available here and here):
(1) ‘The Assembly Government is clear about the crucial importance of maintaining Welsh as a living community language if the language is to thrive and flourish’ (Iaith Pawb p.21 – emphasis added).
The language is clearly the beneficiary. People’s wellbeing is absent. One could argue wellbeing is implicit; indeed, I have actually heard that from Welsh Government policymakers. But that is a remarkably veiled and tacit way to write government policy. There is not even some statement early in the policy text to the effect of ‘whenever we say promote Welsh, we mean X,Y,Z to raise wellbeing’. Explicitly the language is foregrounded alone.
Next, an example of the ‘constitutive’ goal (foregrounding identity):
(2) ‘Its [Welsh’s] cultural influence and traditions remain relevant today and are embraced by new generations learning and using the language.’ (Iaith Fyw p.7 – emphasis added).
And now the ‘instrumental’ goal (wellbeing). Although the census records no monolingual Welsh speakers since 1981, for first-language speakers, Welsh provision can be instrumental in certain contexts, for example:
(3) ‘Strengthening Welsh-language services in health and social care is … a priority since, for many, language in this context is … a matter of need.’ (Iaith Fyw p.42 – emphasis added)
Weighing these three together across all the texts, Fig. 1 shows a consistent and growing emphasis on the intrinsic goal, promoting the language as a good in and of itself, separate from wellbeing or identity.
There are other policies that link language and wellbeing, for example the periodically renewed ‘More than just words’ (latest in 2022) on Welsh speakers in health and care. The Well-being of Future Generations (Wales) Act 2015 also includes assertions that increased Welsh use would improve wellbeing (though the evidence on that point is mixed). But still, the three broader overarching Welsh language policies listed earlier remain. All this is significant, because, as we will see, there are negative outcomes arising, missed by this central policy focus on the language.
What are the unintended consequences?
Below I review some growing social ills in Wales. I do not claim that Welsh language policy causes these, only that it misses them, amid a tight focus on the language itself. These issues could, however, be addressed if language policy were more holistic and intersectional.
Number of Welsh speakers (vitality)
Linguists refer to speaker numbers as the ‘vitality’ of a language, and this is the headline goal for Welsh. Before we consider issues that are being missed, what of this, the actual stated aim? To measure vitality, the Welsh Government prefers the UK census to the Annual Population Survey. (This is due to reliability of their methodologies – for example the APS allows only adults to answer for all under 16s, see e.g. here and here.) The census demonstrates a sustained decline: most recently from 562,000 in 2011 to 538,300 in 2021, the sharpest drop among ages 5–15 (-6.0%), with only small gains in older groups, the biggest a mere +0.9% among ages 20–44. So Welsh language policy is not so far delivering even its main stated goal.
This is often explained by a shortage of qualified teachers (p.15 of that report). But there are other factors. To start with, motivation and personal agency. Children in Welsh-medium schools often feel their use of language is overbearingly patrolled. They have “the opportunity to choose the language they speak outside the classroom [but] on school grounds … this choice is suspended; pupils must speak Welsh at all times or face consequences.” This and a similar study find that such strictness actually reduces motivation to use Welsh outside school.
Like language use, language structure is also policed, reinforcing standard forms over Welsh local dialects or mixtures with English. Such overly scholastic methods speak to the drive to fortify Welsh, but they further erode not only motivation but also perceptions of everyday relevance. According to a Wales-wide survey of 849 Year 8 & 10 pupils (age 12–15), Welsh was the most disliked compulsory subject.
Diversity within Welsh
Following on from the last point, schooling favours standardised materials and correct answers. M.C. Jones (1998) shows how these normative pressures towards standard Welsh, as well as new social pressures on regional dialects, thin out diversity within Welsh. A more recent study finds the “formal register” of school “is wholly inappropriate for casual application” socially (p.201), meaning “linguistic idiosyncrasies indicative of a living language progressively disappear” (ibid.). The resulting Welsh suits “the classroom rather than the chatroom” (ibid.). Those “idiosyncrasies” include mixing with English – as noted above, strictly prohibited.
The belief that protecting Welsh requires fencing it off is desperately counterproductive for motivation and diversity. Coupland pointedly asserts “the whole apparatus” of language policy in Wales “is built on a questionable view of languages as both bounded linguistic systems and bounded cultural commodities” (2014:144). Perhaps diversity within Welsh is a price worth paying to increase its speakers; but, as above, it is being paid not even for that.
Social cohesion within Wales, and with the rest of the UK
Singaporean linguist Lionel Wee argues that promoting the languages of particular groups “encourages social fragmentation along ethnic lines” (2010: 96). Indeed within Wales, two ethnographic studies (one in a rural area, one in Cardiff) both demonstrate social rifts between English- and Welsh-medium schools, with language explicitly used to self-segregate.
There is also tension between first-language Welsh speakers and the growing number of ‘new speakers’ often seen as less authentic – a lively and sometimes divisive civic debate, fuelled by normative pressures noted above.
Meanwhile, Evans points out the galvanisation of “Welsh language ability … as ‘proof of Welshness’” (2019: 175). The Joseph Rowntree Foundation presents survey evidence showing Welsh speakers as more likely to identify as Welsh rather than British. That in turn implies weakening of pan-British affiliations if Welsh usage grows. From a UK perspective this counsels caution. For those favouring Welsh self-determination, this might motivate fresh thinking since Welsh speakers are decreasing. Either way, change is needed.
Educational attainment
Increasing Welsh-medium education inevitably draws in pupils for whom Welsh is a weaker second language. Sometimes parents choose this – indeed, it’s often prized – while in other cases it’s the only option. But those second-language speakers of Welsh attain significantly lower grades. Content is harder. Motivation suffers. Their grades lag.
Meanwhile, children with special educational needs “may encounter systemic disadvantages … regarding Welsh language opportunities” (Davies 2023: 218) – again these institutional barriers are missed amid a focus on the language itself.
Zooming out Wales-wide, these disparities contribute to lower average grades in Welsh-medium than English-medium schools (see p.120 of this report), despite broadly equal funding.
Employment and mobility
Welsh speakers are more employable in Wales (see p.84 here, and a wider overview here). But does promoting Welsh boost employment overall? It is straightforward to count numbers of Welsh speakers employed, and infer economic benefit; but a more important test is whether unique jobs are created – which would not exist in English, and whose existence does not come at the cost of other jobs – thereby adding to overall gross value added. Requiring Welsh for employment often changes the criteria for jobs (particularly for teachers), but few entirely unique jobs are created. Urdd Gobaith Cymru (see next section) shrewdly publishes an annual report on its own economic value; but this simply itemises their incoming budget (from finite state coffers) against money raised through selling services and products. That fails the above test. Unique costs are introduced though, principally for translation; and given finite budgets, this has to come from somewhere. These expenditures can absolutely be justified – not everything must net a strict financial return – but if Welsh language policy is failing its own main aim and in other areas, this is harder to argue.
Meanwhile Welsh holds no such employment cachet outside Wales. Welsh speakers appear statistically less likely to migrate out of Wales (for their movements within Wales, see this report). We simply do not know whether this signifies a free choice (they prefer to stay) or a limitation (they received on average lower grades – as above – and are less employable outside Wales). Ultimately it is likely a mix of both. S.L. Jones finds “students who envisaged that they would move outside of Wales in the future … felt that they would rarely use Welsh” (2019: 269). Further research is urgently needed on this.
One might (perhaps cynically) argue Wales has something to gain by preventing skilled workers from leaving, but this neglects the economic value of a skilled diaspora.
Cultural enjoyment and wellbeing
Quite apart from whether people speak Welsh, do they gain fulfilment and joy in the presence and use of this rich and evocative language? Is enjoyment measured and prioritised?
The Welsh Government’s overall Welsh language budget was £53.6m in 2024–25. Within this is £2.2m for Urdd Gobaith Cymru, the main body promoting Welsh through youth activities, sport, music etc.; and an additional £1m for its annual Eisteddfod festival. Other funded organisations include Mentrau Iaith (encouraging Welsh in communities), and Papurau Bro (Welsh-medium community newspapers). But within that £53.6m is also a notably wider range of efforts still focused on education – teacher training, higher education, pre-schools, and so on. And all these various organisations share the same rather blunt focus on sheer numbers. The Urdd’s annual report is replete with numbers of participants; but, despite some mention of disadvantaged groups like disabled people, ethnic minorities, etc., there is nothing specific about wellbeing. Of course, my point is not that nobody enjoys these things (they are voluntary after all), but that enjoyment is not a reporting priority. Numbers are. Crucially, it’s also not clear that the same activities could not achieve the same outcomes in English. The core focus is on the numbers of participants speaking Welsh. All these organisations ultimately serve the same master: a policy framework heavily dominated by the language.
But there is a much bigger picture here. The £53.6m Welsh language budget is dwarfed by direct expenditure on Welsh-medium schooling (which falls within the education budget). Education expenditure on Welsh specifically is not separately accounted for, but we can calculate a broad estimate of around £1.1 billion based on proportions of Welsh-medium instruction. (As above, Welsh- and English-medium schooling is funded about equally.) That calculation is not terribly precise – it ignores pupil numbers per school for example – but even as a broad benchmark, it makes the few million for culture, youth activities, sport, etc. look like little more than a trifling token gesture.
But again, this simply mirrors policy. Cymraeg 2050 (82 pages) has only a 1.5-page section on ‘Culture and media’ containing no targets at all. Moreover, Welsh language policies have generally not kept pace with cultural change, for example the shift to socialising online.
Overall, it appears anathema to policymakers that any activity should not, at its heart, steer towards increased speaker numbers. But simply celebrating Welsh, using it symbolically even if not conversationally, is a huge area of promise. One rare study showed “systematically higher” commitment to Welsh’s iconic, symbolic value than use in everyday interaction. Another study outside Wales playfully highlights the “principle of Cool” in minority language use. Even education can find ways to embrace playfulness and creativity. This is possible, and of real human value; it just falls largely outside the current paradigm.
‘Bilingual Wales’, or ‘multilingual Wales’?
Welsh language policy revolves ceaselessly around the totemic watchword “bilingual Wales”. But, why only bilingual? Learning two languages clearly facilitates further languages, so Wales has a head start. But despite the Global Futures plan for foreign languages, the British Council finds over three in five Welsh secondary schools reporting under 10% of students picking an International Language GCSE, and over half of schools without any such GCSE. Levels are low across the UK nations, but Wales is lowest of all. Welsh primary schools have just begun increasing provision; and in secondary schools there is a promising new mentoring scheme. Wales’ future curriculum plans could support ‘plurilingual’ language teaching; time will tell if this helps Wales exploit its head start.
There is also the question of growing immigration, and Wales’ mixed record of recognising these diverse incoming languages. There is scope to support these for everyone’s benefit. Scotland has forged ahead with its 1+2 policy, investing equal status in traditional and newer immigrant foreign languages. Overall, ‘multilingual Wales’ could be a much more dynamic and fruitful aim, building on already strong potential.
Recommendations and policy implications
The above unintended consequences are not simply collateral damage from increasing the number of Welsh speakers. The number of Welsh speakers is not increasing. Indeed, all these inequalities and hostilities could very well be reasons for the continued decline. Whether or not these issues are created by Welsh language policy, they are neither foreseen nor remedied by it, given its constrained focus on the language in and of itself. This has come fully to light only in recent years with the raft of authoritative studies cited above.
If Welsh language policy held sacrosanct equal educational outcomes, equal career mobility, social cohesion between language groups, and a positive and celebratory view towards any and all use of Welsh – if language policy was not allowed to proceed without all these things – then perhaps things could be different. One is reminded of the Finnish education model, which famously strove only for equality and ended up accidentally with the highest grades in the world. A multi-faceted, intersectional approach could make for better ambient social conditions as well as increased Welsh use.
So, what would such a renewed Welsh language policy look like?
Figure 2. The doughnut model of economics.
As flagged here, useful inspiration comes from the ‘doughnut model’ of economics. This was originally proposed by Oxford economist Kate Raworth to snap economics out of its blinkered focus on GDP, and to lever in other factors like sustainability and social justice (Fig. 2). If GDP only rises with spiralling inequality, climate catastrophe, and ecological collapse, then economics has failed. This is part of a broader policy movement towards intersectionality – understanding that no issue is solvable in isolation.
Intersectionality informed the Welsh Government’s 2022 report on racism, Running Against the Wind. Like Raworth’s critique of economics and GDP, my textual analysis above shows Welsh language policy focused on one metric – speaker numbers – and missing other problems arising. What if it were also intersectional?
Figure 3. Doughnut language policy.
(Creative Commons BY-SA-4.0 license)
In Fig. 3, I take the template of Fig. 2 and map on the issues in each section above, to create ‘doughnut language policy’. This does include speaker numbers (‘vitality’) but balances it against all other factors. A ‘safe and just’ zone exists around the doughnut, in which each the above factors are attended to. Straying outside this zone leads to each of the unintended consequences outlined above. This simple visualisation, like doughnut economics, helps spur a balanced focus on these diverse issues.
The following is a list of recommendations for a more intersectional future Welsh language policy:
Use the doughnut model to create equally sized, cross-referenced sections, each containing equally specific goals, rather than e.g. the blurry priorities of the short ‘culture and media’ section noted above. In each section, display the doughnut model as a visual aid to intersectionality.
When that intersectional policy is in place, fund research into the above social factors in equal measure. The current tendering model, based on current policy, offers funding to work out how to drive up speaker numbers, and so researchers explore how to do that. Their alternative is no funding. This unduly skews their priorities, when they could instead expertly explore a much wider range of issues without such a predetermined end point.
Once researchers have evidence on those wider factors, include those researchers substantively in refining policy to equally balance all those factors – of course, many of them are already included in policy consultations and development, so they would be ready to apply new intersectional insights.
In teacher training, critically unpack the purposes and effects of policing Welsh and banning English in schools. Base this on the research evidence, which clearly shows such approaches to be both counterproductive and demoralising. Even if one still sticks to the singular aim of increasing Welsh speakers, this is not the way.
Relatedly, train teachers to embrace the fluidity of each student’s linguistic repertoire, not enforce singular standard language norms. The Welsh homegrown concept of translanguaging is already included in teacher training; it should be further mainstreamed.
Altogether this could soothe lingering tensions and facilitate not just increased use but also enjoyment of Welsh, so far diffused and stymied amid a constrained focus on countable, assessable proficiency. There is great promise in a more diverse intersectional approach, both for Welsh and for people in Wales.
References
Davies, Eleri Nia. 2023. ‘Are Children and Young People with Additional Learning Needs at a Systemic Disadvantage Regarding Welsh Language Opportunities?’, Educational Psychology in Practice 39.2: 217–234, https://doi.org/10.1080/02667363.2023.2186835.
De Schutter, Helhder. 2007. ‘Language Policy and Political Philosophy: On the Emerging Linguistic Justice Debate’, Language Problems and Language Planning, 31.1: 1–23, https://doi.org/10.1075/lplp.31.1.02des.
Jones, Mari C. 1998. Language Obsolescence and Revitalization: Linguistic Change in Two Sociolinguistically Contrasting Welsh Communities. Oxford: Clarendon Press.
Lee, Rachelle. 2016. Language Ideologies in The Secondary School: Attitude and Identity in Bilingual Wales. Doctoral dissertation, University of York. https://etheses.whiterose.ac.uk/16439/.
Sayers, Dave. 2024. ‘Using Language to Help People, or Using People to Help Language? A Capabilities Framework of Language Policy’, International Journal of Applied Linguistics 34.1: 3–29, https://doi.org/10.1111/ijal.12463.
Wee, Lionel. 2010. Language Without Rights. Oxford: Oxford University Press.
Cite this article
Sayers, Dave. 2025. ‘Doughnut Language Policy: How to Promote Welsh without the Unintended Consequences’, Languages, Society and Policy, https://www.lspjournal.com/post/doughnut-language-policy-how-to-promote-welsh-without-the-unintended-consequences.
About the author
Dave Sayers (ORCID number 0000-0003-1124-7132) is Senior Lecturer in the Department of Language and Communication Studies and Docent in the Centre for Applied Language Studies, at the University of Jyväskylä, Finland. He has previously worked at Åbo Akademi University and the University of Turku in Finland; and at Sheffield Hallam University in the UK. He has also held honorary positions at Swansea University and Cardiff University in the UK. He was chair of the COST Action ‘Language In The Human-Machine Era’ (2020-24), (https://cost.eu/actions/CA19102/). His research interests cover the moral and ideological bases of governmental language policy and contemporary conceptualisations of language rights.
Comments