For three decades, John Aitchison and Harold Carter (A&C) have shouldered principal responsibility for interpreting the results of the decennial UK census as it relates to the use of the Welsh language in Wales. In this book A&C give their account of the 2001 census data on Welsh. First-level analysis of the 2001 data is available through an excellent government web site – http://www.statistics.gov.uk/census2001/ – but A&C's further, mainly geographically based analyses have become the standard resource for reading patterns of Welsh language maintenance and shift.

The 2001 census data are widely held to represent a historic upturn in the vitality of Welsh speaking in Wales, reversing a long and ineluctable decline throughout the 20th century. This is “the word on the street,” although careful assessment is needed. Does an increase in the headline number of Welsh speakers constitute “revitalization?” If so, does it index a wider ethnolinguistic and indeed national revival? The devolved Welsh Assembly Government, now in its second elected term but of disputed effectiveness, has national integration, the protection of small “heartland” communities, and language planning initiatives as some of its key policies. So does a reviving Welsh language mark the success of political devolution and the cohering of a new Welsh polity around the Welsh language? In the wider sociolinguistic world, how should we theorize a remarkable reversal of language shift in Wales, with its long history of anglicization and its long and porous territorial boundary with England?

In fact, A&C are very circumspect – to my mind appropriately – about the extent of linguistic revitalization entailed in the new census data. Some of their reticence relates to familiar methodological limitations in census taking. New administrative boundaries and conventions in counting student populations complicate comparison with 1991 data. Then, A&C point out that the 2001 census elaborated the formerly used question, Can you speak, read or write Welsh?, by adding an understand dimension, and allowing discrete responses for each dimension of self-assessed competence. So we have more data on Welsh this time, but it cannot be unambiguously set alongside previous results. The overall new statistic of 20.5% “speaking Welsh” in Wales – that is, of the 2,805,701 people in Wales aged three and over on 29 April 2001 – in fact derives from adding data for three combined-skills statistics: speaks but does not read or write Welsh, speaks and reads but does not write Welsh, and speaks, reads and writes Welsh. It is not at all unreasonable to interpret the 2001 data as showing an overall increase in reported competence in Welsh over the 18.7% reported in 1991, but the comparison is indirect. Owing to a growing overall Welsh population, A&C say that there are 13.3% more speakers of Welsh since 1991. They summarize the numerical increase in Welsh speaking in 2001 as follows:

More important caveats about revitalization, however, relate to how these increases were achieved. They relate to a range of demographic shifts which have unsettled the social arrangements that formerly protected Welsh, as well as those that threatened it. A&C identify “the four core Welsh-speaking areas” of Wales (64), often referred to as Y Fro Gymraeg, which in fact show significant decline and fracture in their numbers of Welsh speakers since 1991: Carmarthenshire (−6.1%), Ynys Môn (−6.1%), Gwynedd (−1.6%), and Ceredigion (+ 4.8% in absolute numbers, but a decline in its proportion of Welsh speakers). Eastern Snowdonia and the Conwy Valley in the north produced among the highest falls, also Ystradgynlais, Pontardawe, and Pontardulais in the south. These losses were more than compensated for by huge percentage increases in reported Welsh language ability, albeit from very low bases in 1991, in historically much more anglicized areas of the southeast of Wales (e.g., +355.4% in Monmouthshire, +342.9% in Torfaen, +338.7% in Newport, +303.2% in Blaenau Gwent, +83.5% in Caerphilly, +76.8% in Cardiff, +64.2% in the Vale of Glamorgan, and +37.2% in Rhondda Cynon Taf ; p. 50ff.). Cardiff, the Welsh capital city, had 18,071 (6.6%) enumerated Welsh speakers in 1991, but 31,944 (10.9%) in 2001, a gain of nearly 77% (p. 75). A&C provide sophisticated geographical analyses of these and similar data, in the form of detailed tables and summative maps.

The data show that “the linguistic centre of gravity in Wales is shifting” (65), with use of Welsh becoming as much an urban as a rural practice. Even more strikingly, it is associated far more with speakers under 16 years of age than with any other age sector, clearly through the impact of compulsory Welsh teaching and learning through the primary and secondary school curricula (to age 16). Under-sixteens account for more than half of Welsh speakers in many parts of southeast Wales. Welsh speakers also have proportionally high representation in the category “professional occupations,” and there is a relatively large preponderance of non-speakers in the lower socio-economic class categories (p. 95). In summary, the overall demographic revitalization of Welsh in 2001 disguises continuing fracture of the old “heartland” zones in the north and west of Wales, where Welsh has been transmitted intergenerationally. It rests on proportionally huge increases in reported Welsh language competence, particularly relating to young speakers in the populous and otherwise anglicized southeast, where competences of a largely unknown extent are made available through compulsory education. A&C comment that “the real depth of the language within society can [therefore] be questioned” (132), and we certainly need to know more about post-age-sixteen tendencies in language choice in Wales.

The authors' demographic analysis – the main rationale for their book – is meticulous and impressive. But their critical response to these sociolinguistic circumstances in their concluding chapter is, to my mind, highly questionable and detracts from the whole. Their assumptions are organicist and modernist; their policy inclinations are protectionist and nationalist. They endorse the organic model of “a truly living Welsh language” (quoting Saunders Lewis), which they see as threatened by and eroded by a more powerful organism, English (133). They argue that resistance to erosion requires a shoring up of independent, dense enclaves of Welsh speakers, while they recognize but regret that cultural isolation of this sort is no longer feasible in a mobile world. They rue the sociolinguistic impact of in-migration, “invasion,” and “penetration” (135–36). In the book's last five pages (where their tone has become markedly pessimistic), A&C invest heavily in “the need to create an association between being Welsh and speaking Welsh” and resisting the “problem” of multiculturalism in Wales (140). In short, they argue that “the language” can best succeed if it is promoted as the unique and necessary hallmark of “a Welsh identity.”

There is far more to be said in response than this brief review allows. First, whatever stance we take on nationalism – its seductive essentialism, its dangerous myopia, or its irrelevance in globalizing late modernity – it is evident that the sociolinguistic robustness of Welsh has not to date centered on its being spoken as a credential of national uniqueness, if that could be defined. “Heartland” Welsh usage in most of Wales has not generally been politicized, and it has readily accommodated English in code-mixed varieties, particularly in the south. “The language” does not have a simple integrity, either in its dialect varieties or in its formal and functional/genre relationships to English. Powerful strains of Welsh identity (of which A&C are quite dismissive) arose, notably in the South Wales Valleys, through industrialization, where the Welsh language was of marginal importance. The new, young Welsh speakers are by no means uniformly exposed to Welsh in education contexts where nationalist ideology predominates; for the most part, parents judge Welsh-medium schools to be educationally superior. In fact, it is clearly the widening of the geographical, age-related, and class-related catchment area for Welsh that has generated the demographic revitalization that A&C document, and this open-access quality of Welsh will surely need to be extended in the interests of wider usage.

A&C's appeal to “identity” misrepresents the language-ideological climate in contemporary Wales. Coincidentally, colleagues and I have substantial recent data (e.g., Coupland, Bishop, Evans & Garrett, in press) showing that Wales already benefits from strongly positive and widely distributed pro-Wales and pro-Welsh ethnolinguistic subjectivities. It is true that higher levels of competence in Welsh, and to some extent older vs. younger age, are positively associated with higher levels of expressed affiliation to Wales. But there is no strong pattern of ideological resistance to the advancing project of Wales, or disaffiliation with Welsh as part of that, on the part of non-Welsh-speakers. So the “identity” infrastructure in Wales, which A&C want to see enhanced, is already strongly consonant with Welsh language maintenance and revitalization. This makes the authors' idea of hallmarking Welsh as the unique criterion for real Welshness far too pessimistic, as well as dangerously restrictive. Engineering more nationalist ideologies around the Welsh language is likely to be counterproductive.

One pattern we detect in our data is an increasing tendency for Welsh-affiliated people to endorse a broad ceremonial function for Welsh, such as using Welsh in names, songs, anthems, and cultural ceremonies. They give ceremonial use rather higher priority than interactional use in homes or workplaces, but without dismissing these either. An interest in ceremonial usage, which might include facility in performing the mini-rituals of social interaction in Welsh, may well motivate some people's current enthusiasm for learning the language. This picks up on A&C's concerns (quoted above) about “the real depth of the language within society,” if we equate ceremonial usage with “shallow” social presence. But we need to ask whether it is feasible to hold on to an ideal of “deeply” authentic language use, in “true bilingualism” or indeed elsewhere. Languages inevitably have to fill niches in whatever new markets open up to them, and ceremonial function may be an increasingly important part of minority languages' profiles, if they are to retain and even grow their vitality in the way that Welsh is doing so successfully.