Content associated with: Language spoken in Wales, 1911    Page v

Language questions in Wales

Matthew Woollard

The question of language in the census is related to problems of nationality, nationalism and national identity. However in the United Kingdom, at least, census questions relating to language are well-removed from the traditional philosophies underlying British census-taking. Questions on language were not part of the 'medical' program which epitomised the British censuses. The Irish census was taken on a rather different administrative basis than in the rest of Britain, as a result the Irish census, post 1841 was more advanced than its British counterparts and the range of questions asked was much greater, including education, religion, literacy, house accommodation – more a social survey than a simple population count. The questions on language have to be seen as something rather different to the usual demographic or medical questions.

During the nineteenth-century questions on language were introduced slowly into the British censuses and these questions can be seen as the first which were successfully included through the lobbying of special-interest groups. Language questions had been asked in many European censuses (e.g., in Belgium, Austria, Denmark, Prussia, Sweden and Switzerland all asked questions on the subject during the 1850s or 1860s. Arel has pointed out that most of the discussions in the international statistical community at this time were less concerned with language spoken per se, but with 'cultural nationality'.

The 1881 Scottish census returns on language (see Language questions in Scotland) were possibly the impetus for a similar question being asked in Wales. A Welsh deputation went to London in 1888 to plead for similar provisions for Welsh in the English Revised (education) Code, which had allowed for Gaelic to be taught during ordinary school hours for Scotland.

Between 1851 and 1881 Welsh-language schedules had been used for the householders to complete suggesting that the General Register Office understood that some people were unable to understand English. However, no practical initiative was taken to calculate the extent of Welsh-language speaking. The 1851 report notes "for the use of the poorer native population of Wales a certain number of the Forms was printed in the language of that country". (Population Tables. I. Numbers of the Inhabitants. Report, and summary tables, BPP 1852–3 LXXXV, xiii. )

Pryce suggests that the Welsh language question was first asked as a result of Ravenstein's paper and also a lecture delivered by A. J. Ellis in 1882 to the Honourable Society of Cymmrodorion in London in which he pleaded for a fuller Welsh language survey. While the latter may have been an impetus for this, Ravenstein's paper makes no mention of using the census as an instrument for collecting this form of information.

Discussion in the House of Commons regarding the census Bill for 1891 was more contentious than that for the previous decade. The main reason for this was a Treasury Committee which reported on a number of alterations to the administration of the English and Welsh census which also considered revising some of the questions. There was no discussion at this committee on the language questions, indeed the fact that Scotland had asked a question about Gaelic was almost completely forgotten.

During the second reading of the Bill, Welsh MPs were united over the provision of the question, but not in their reasons for wanting the question. The patriotic MP for Mid-Glamorgan, Mr S. T. Evans, made a spirited defence of the Welsh-language question, noting that the Scottish had their question, that there had recently been a Report of the Royal Commission published in Welsh and that the Welsh were more interested in their language than their Celtic counterparts. (Parliamentary debates, 3rd series, 347 (1890), col. 413–414.) Evans must not be seen as an authority on the census, as he also proposed that the schedules should be issued in the Welsh language, when this procedure had been carried out in Wales for the previous forty years. One of the MPs for Cardiff, Sir E. J. Reed, wanted to know the extent to which Welsh was spoken: "but I wish to bring out clearly the limitations of its use". Reed argued that if Welsh Home Rule were to be discussed, the Welsh language would be of little use. The debate continued with a statement from a Mr Wallace, MP for Edinburgh East who, like many of his counterparts had but a dim recollection of the previous census. He hoped that both the forthcoming Scottish and Welsh (if the question were to be asked) censuses would not phrase the question in terms of "blind, or deaf and dumb, or imbecile, or lunatic or speaks Gaelic, as if all these things were ejusdem generis." Wallace had muddled his columns. The Minister of the Local Government Board said he would consider the language question. (Parliamentary debates, 3rd series, 347 (1890), col. 419–421) In the subsequent Committee stage, the amendment was proposed, magnanimously accepted by Ritchie, voted on and accepted.(Parliamentary debates, 3rd series, 347 (1890), col. 534–535.) Later the House of Lords proposed a subsequent amendment — pedantic, but with hindsight most perspicacious — that the question only be asked in Wales and Monmouthshire, rather than the whole of the England and Wales, as the Bill would have ensured otherwise.

The question, like those asked in Ireland and Scotland was 'closed'. However, it did not allow for a blank space. Thus unanswered questions should have been ignored or demoted to a residual category. The wording used in 1891 was: "If only English, write 'English'; if only Welsh, write 'Welsh'; if English and Welsh, write Both'" (Vol. IV General Report, with summary tables and appendices, BPP 1893–4 CVI (C.7222), 81–3).

The 1891 General report for England and Wales is rather more forthcoming on the subject of this question than its Scottish counterpart. This is perhaps curious as this report was written by Englishmen in London. Their analysis: "that the word 'Welsh' was very often returned, when the proper entry would have been 'Both' on the ground, it may be presumed, that Welsh was the language spoken habitually or preferentially" was perceptive, but if true, would only understate the total numbers who spoke English but not overstate the numbers who spoke Welsh. A further piece of faulty reasoning also affected these English statisticians. They suggested that: desirous do many householders appear to have been to add to the number of monoglot Welshmen, that they not only returned themselves as speaking Welsh, that is Welsh only, but made similar returns as to infants who were only a few months or even a few days old. (Volume IV. General report, p. 81.)

This was probably unfair to these householders and uniquely an apology was forthcoming from the registrar-general (See Letter of registrar-general relative to complain against certain remarks in census report of 1891 as regards inhabitants of Wales, speaking Welsh only). In a letter addressed to the secretary of the Local Government Board Brydges Hennicker wrote: "I understand that some Welsh Members of Parliament complain that, in a paragraph of my Report on the Census Returns of 1891, I imputed to the Welsh people untruthfulness, and charged them with having attempted to increase the numbers of those entered as speaking Welsh only, by fraudulently filling up their schedules." He continued, "I had no intention whatever to accuse the Welsh people of untruthfulness, or of a deliberate intention to make false returns", and explained that "What I did intend to convey...was this: that the people in certain districts in Wales had wrongly understood the requirements of the schedule in the census as meaning that when they spoke Welsh preferentially or habitually, but could also speak English more or less, they were justified in returning themselves as speaking Welsh only. It was intended that such cases should be regarded as bi-lingual, and entered as speaking both Welsh and English. I am quite aware that it is difficult to lay down any definite standard as to the degree of proficiency in speaking English which would warrant the entry under this head, and hence, no doubt, there has been misunderstanding."

Another answer was given in the 1911 census report, which put the blame more squarely on the enumerators, who assumed that if a head of household had returned himself as Welsh speaking, the enumerator, when copying the schedule into his enumerators book, applied the information to the remaining members of the household.

The solution to this problem was, in 1891, to only tabulate those people aged two years and above; but even so, the authors complained that in two particular parishes some 94 per cent of 5 to 15 year olds were reported as being Welsh monoglots despite school attendance, where English would have been a compulsory language.

In the discussions preceeding the 1901 census two Welsh MPs made reference in their speeches to problems arising in the previous census. (See Parliamentary debates, 4th series, 79 (1900), col. 394; Parliamentary debates, 4th series 80 (1900), col. 502–3.) These discussions led to two accusations: First, that in some parts of Wales, English schedules were distributed, that is schedules which did not have the language column; second, that the provision of Welsh-language schedules was not as comprehensive as possible. There was no suggestion that this problem was deliberate, but simply an administrative error on the part of the GRO.

In 1901 the question was repeated. In the committee stage of the Bill for the Census, David Lloyd-George proposed an amendment to provide that the language question should be extended to England. This irritating proposal was forced to a division but was defeated 54 to 164. (Parliamentary debates, 4th series, 80 (1900), col. 1002–1004.) An innovation in the census Act was to prescribe the ages to be tabulated. In the published reports for Wales, figures were tabulated for those aged 3 and over, thus preventing direct comparison with the previous census. Comparative reporting discontent would probably have been exacerbated by the fact that in 1901 reporting was carried out by administrative county rather than registration county. One innovation, however, was the inclusion within the introduction to each county report of an age-specific tabulation, where percentages were given for five age groups: 3–14; 15–24; 25–44; 45–64 and 65+.

The 1911 census saw a departure in the mode of tabulation of the results of the census of England and Wales, which had a considerable impact on the tables produced for language. At all earlier censuses, householders' schedules had been copied into enumerators' books and tabulation had been made from these books. In 1911 tabulation was done directly from the householders' schedules onto punched cards for processing. The effect of this process, as noted above, was that many more people were returned with this question unanswered. A total of 58,517 unanswered questions were returned as opposed to only 2,757 in 1901 (which was less than the number of people who erroneously returned another language.)

In essence, while the report makes great claims about the significance of this problem, it makes very little difference to the final figures when they are calculated out. Of all those people aged over three who returned an answer, 91.1 per cent spoke English and 44.6 per cent spoke Welsh. Adding in those who had no answer returned, but whose language ability was the same as their head of household gives the respective percentages of 90.2 per cent and 44.1 per cent. (These latter figures still include just over 24,000 people for whom no language ability could be inferred, accounting for 1.2 per cent of the population.) It was probably this relatively severe underenumeration which led to the GRO publishing the figures for the language spoken by the head of household of those people who did not return a language.

As in 1901 the published figures were always broken down into their constituent parts, and Welsh-language interest groups would, for all geographical areas, have to calculate the numbers or percentages who spoke Welsh at all. This inconvenience may not have been intentional, but it seems absurd to believe that it was not.

It would be absurd to argue, as Parry and Williams do, that the inclusion of the Welsh language question was wholly fortuitous; rather it was the result of lobbying by Welsh-language interest groups. Once included, however, some of the Welsh attitude towards the census became tinged with nationalism, but more likely cultural patriotism. The backlash on the ground to the 1891 census, as described by Parry and Williams, demonstrates this. The English census-takers reacted to their insensitivity in 1891 but would seem to have been generally uncommitted to this question.


G. Parry and M. A. Williams, The Welsh language and the 1891 census (Cardiff, 1999).

W. T. R. Pryce, 'The British census and the Welsh language', Cambria: A Welsh geographical review, 13 (1986), 79–100.

E. G. Ravenstein, 'On the Celtic languages in the British Isles: a statistical survey', Journal of the Statistical Society of London, 42 (1879), 579–643.

Parliamentary debates, 3rd series, 347 (1890).

Parliamentary debates, 4th series, 79 (1900).

Census of Great Britain, 1851, Population tables, I. Number of the inhabitants in 1801, 1811, 1821, 1831, 1841 and 1851. Vol. I BPP 1852–53 LXXXV. [View this document: Population tables I, Vol. I. England and Wales. Divisions I-VII, 1851]

Census of England and Wales, 1891, Vol. IV General Report, with summary tables and appendices BPP 1893–4 CVI. [View this document: General report, England and Wales, Vol. IV, 1891]