Welsh Language

Next Selection Previous Selection


In addition to the census enquiries common to the whole of England and Wales a statement was required in respect of every person, aged three years and over enumerated in Wales and Monmouthshire as to whether he or she was "able to speak English only, Welsh only, or both English and Welsh." A similar enquiry with slight modification in the precise terms of the question has been made at each census from and including that of 1891. It is to be observed that the persons in respect of whom the information has been collected and tabulated are those who were enumerated in Wales and Monmouthshire on census night, and include, therefore, visitors and persons of other than Welsh domicile, while residents of Wales and Monmouthshire enumerated in England, or elsewhere, are excluded. No return was required with respect to children under three years of age. Census schedules printed in Welsh were available for such as required them in preference to the more commonly used English form.

As was the case in 1911, no definite rules for the guidance—either of the persons responsible for making the returns or of the local officers whose duty it was to collect and revise the schedules— were laid down as regards the degree of acquaintance with either language which would justify a person to claim or disclaim ability to speak it. It would have been difficult, if not impossible, to define within the limits of a question on the census schedule; a standard of proficiency capable of definite and general interpretation, and there will no doubt have been some lack of uniformity in this respect in the returns of different individuals. The replies may have varied, for example, according to the attitude taken with regard to the object of the enquiry, e.g., as to whether it was intended to elucidate the extent of the habitual use of either or both languages, or whether it was designed from a literary or an educational standpoint.

Language statistics for each Borough, Urban and Rural District by groups of ages, and for each civil parish without distinction of age, are given in the Welsh County series of volumes, and in summary form in tables 54-57 of the General Tables Volume, The civil parish statistics have been introduced for the first time on this occasion to meet the view which has been expressed with regard to the desirability of providing an extended classification showing the detailed geographical incidence of the prevalence of the Welsh language.

The tabulated returns show that of a total population of 2,486,740 persons in Wales and Monmouthshire, aged three years and over, 1,466,211 claimed to be able to speak English only, 155,989 Welsh only, and 766,103 both English and Welsh; in 98,437 cases the language statement was omitted. So that in a general population which has increased by 9.7 per cent. since 1911, English monoglots have increased by 21.3 per cent., while Welsh monoglots and bilinguists have decreased by 18.0 per cent. and 2.7 per cent. respectively. The increase of over 36,000 in the numbers of persons who failed to make the return may be held to disturb the intercensal comparisons slightly, but if, as is not unlikely, some of these were summer visitors who would not have been present if the census had* been taken at the originally appointed date in April (see reference to census inflation in Appendix A), their exclusion will not be a disadvantage, and the disturbance will be reduced to correspondingly smaller proportions. It may be noted that children for whom no statement as to language spoken was made, but who were enumerated on schedules in which the head of the family was returned as able to speak English only, Welsh only, or both English and Welsh, have been distinguished separately in the tables. The first group (head of family speaking English only) amounted to 11,530, the second to 1,006, and the third to 6,085; each of them being very much lower than the corresponding figures of 1911.

The proportion per 1,000 of the population tabulated under the several headings at the present and at the three earlier censuses in which this enquiry has been included, are shown in the following table.


Subject to the defects in the returns, the greatest of which appears from past reports to have been a tendency to an overstatement of the number of monoglot Welsh in 1891, the characteristic feature of this comparison, covering the past thirty years, is the steadily increasing predominance of the use of the English language and the equally continuous disuse of Welsh. Persons able to. speak English now number 90 per cent. of the whole population aged 3 years and over, as compared with 69 per cent. in 1891, 85 per cent. in 1901, and 89 per cent. in 1911. Those able to speak Welsh, on the other hand, were 54 per cent. in 1891, 50 per cent. in 1901, 43 per cent. in 1911, and are reduced to 37 per cent. at the present time. The decline in the latter class, it will be observed, is most marked in respect of those who speak Welsh only, numbering in 1921 between 6 and 7 per cent. of the total population. This is perhaps, inevitable in view of the greater fluidity of population which transport development has brought about and the general necessity, therefore, of a knowledge of English, in even the remotest areas an! the more significant features of the decay of the national tongue are probably those relating to the numbers claming to speak both languages. Even allowing for some understatement in their returns of 1891 the proportion of bilinguists appears to have increased between 1891 and 1901 During the next ten years it remained approximately constant at about 35 per cent of the total, but this appears to have marked a maximum which has not been maintained the latest figures showing that the proportion has now declined to 31 per cent.

Age distribution of the several classes. The returns, which have been tabulated in seven groups of ages, show that in proportion to the total number living in any of the specified age groups, the frequency of monoglot English speakers is highest at the youngest ages, that of bilinguists at 45 to 65, and of monoglot Welsh speakers at the two extremes of life. Generally speaking, the proportion of persons speaking English only decreases, while that of persons speaking both languages rises with advancing age. The age incidence of persons speaking Welsh only is entirely different from those of the other classes, commencing at a relatively high proportion at infancy, decreasing rapidly to a minimum at ages 15-25 and thereafter ascending until at the last group shown in the table it is far higher than at any earlier period of life. It will be observed that though the proportions of defective returns in 1921 are fairly evenly spread over the whole age field after age five, the increase since 1911 is greater at the later ages and will accordingly have affected the intercensal comparison more seriously here than at the earlier periods.


An alternative and perhaps more instructive way of examining the language tendency during the past decennium would be, if it were possible, to compare the numbers of each class at the several age periods in 1921 with the survivors of the corresponding numbers enumerated at ages ten years younger in 1911, and so to ascertain how the changes, indicated by the aggregate figures, appeared to be operating in each class and how they were disposed over the several periods of life. Any precise computation of this nature is, however, impossible owing to the lack of data necessary for making the survivorship calculation, inasmuch as the death records contain no reference to language spoken and the deaths cannot, therefore, be apportioned amongst the three classes with any certainty, while of migration there is no direct record at all. But where the language changes are large in relation to any likely defect in the survivorship estimates, some degree of approximation in the latter can be admitted and in the following table the expected survivors from the numbers recorded at ages ten years younger in 1911 have been ascertained on the basis of survivorship—that is as regards mortality and migration—experienced by the whole population of England and Wales in respect of similar age periods. Mortality is rather higher in Wales than in England, but on the other hand emigration is not so high, and this should afford some compensation for the difference in mortality at the earlier adult ages, so that if the Welsh survivors tend to be overstated in the comparison it will by mainly at the later ages where the effect of migration may be expected to be at a a minimum.


So far as concerns the youngest ages—the first two age periods of the table—the survivorship comparison is not possible since, in 1911, many of the children now under fifteen years of age were non-existent, or were too young to have been included in the language table. This period is, however, of particular importance in that it is chiefly at these ages that language is acquired. The trend of recruitment to the several classes at the school ages will probably have a greater influence on the language distribution of future years than any other factor, and it is interesting to note, therefore, that the numbers under fifteen years of age scheduled in the. monoglot Welsh category, 45,000 approximately in 1921, are more than 10,000 less than they were in 1911, and that in the bilingual class the decrease is from 172,000 in 1911, to 149,000, a drop in the total Welsh speaking population at these ages of 33,000 or about 15 per cent. The number knowing, or acquiring, English only at this period of life, on the other hand, shows an increase from 363,000 to 415,000 in the ten years.

After age 15, when home and school influence tend to give place to individual choice and economic requirements, the survivorship comparison, even in the approximate form given in the above table, is instructive in showing to what extent the Welsh language is gradually being superseded by the increasing prevalence of English, and also how far the language acquired in the earlier years, presumably as a result of the educational policy adopted in various areas in Wales, satisfactorily meets the requirements of later life. The period 15-25, immediately succeeding school years, is most interesting in this connection, for it will be seen that those speaking Welsh only at this period in 1921 are less than half of the survivors of those who were stated to be able to speak Welsh and Welsh only at ages 5-15 in 1911, the majority, about 21,000 having found it desirable, or necessary to acquire the more commonly used English tongue in addition to their native Welsh. The excess in the bilinguists at ages 15-25 over the corresponding estimate of survivors, is, however, only 6,000, though thev must have received the full transfer of the 21,000 odd deficiency in the monoglot Welsh, since emigration, the only other possible explanation of this loss, is inconceivable as a factor of any significance amongst people who are only able to speak Welsh Concurrently with the increase of the bilinguists due to the acquirement of English by Welsh monoglots, there thus appears to be a compensatory movement from the group which might be attributable to migration, but which appears far more likely to be due to the dropping of the Welsh language in the case of many of those who were alleged to be able to speak both languages at the ages 5-15 in 1911. The 1921 return of bilinguists at ages 5-15, though not so numerous in relation to the returns at later ages, as is the case in either of the monoglot classes, is, nevertheless, so high as to suggest that the standard of linguistic proficiency thereby indicated is an academic one applicable to a school subject rather than to the conditions of everyday life, and in spite of such efforts as may have been directed to the preservation of the national characteristic, it is not surprising to find — by inference from the above figures—that a number in the region of 15,000 or about 10 per cent. of those returned as bilinguists in 1911, had lost that acquaintance with Welsh which would enable them to claim ability to speak it ten years later. The transference of this number to the English monoglot class is supported by the excess of the actual over the expected survivors in that class at the same ages. The excess is shown in the table as 20,000, a number somewhat in excess of the transferred bilinguists, and it possibly includes in addition, therefore, an increment from some other source, the most likely being that of immigration from England, since this form of population movement is most marked at the younger adult ages and would begin to be of significance before the age of 25 was reached.

In the next series of groups shown in the above table; that is, amongst the population now aged 25-45 or the survivors of their counterparts aged 15-35 in 1911, the movements are similar in direction but with slightly different resultant effects in the several language groups. The Welsh monoglots are 5,000 less than the expected number, a loss of 14 per cent. only as compared with 55 per cent. in the earlier age period, while the bilinguists, notwithstanding the addition due to the transfer of the majority of this number, also show a net deficiency of 24,000, so that the full loss amongst the bilinguists in this group must have been nearer 30,000 (11 per cent), the dropping of the Welsh language in these cases and their consequent transfer to the English monoglot section again being supported by the excess of actual over survivors in the latter section.

After age 45, the effect of an unsuitable mortality rate in the survivorship calculation has an increasingly disturbing effect upon the comparisons, but the general inferences to be drawn from the figures in no way oppose the tendencies in the earlier years. Persons who have survived early adult ages knowing only Welsh tend to remain in that condition during the remainder of their lives, rarely acquiring a knowledge of English, while amongst those who know both languages the gradual shedding of the Welsh element appears to be steadily maintained up to the latest years of life.

Local Distribution of the several language classes. In the following table the distribution of the several language classes, hitherto discussed for Wales and Monmouthshire as a whole, is extended to the individual counties and to the large towns. The comparable figures of 1911 are inserted with the 1921 proportions in each case, and to avoid the disturbance introduced by the increase in the number of persons who failed to return a language statement in 1921 and by the varying incidence of this factor in the several areas, the "not stated" cases have been excluded, and the proportions calculated throughout per 1,000 of the total population aged three years and over from whom definite language returns were received. The areas are arranged in the order of the frequency of the total Welsh speaking element in the population.


The proportional decline in each of the Welsh speaking sections of the population, that is, amongst Welsh monoglots and bilinguists, is seen to have been shared in varying degree by each of the thirteen counties, with the exception of Radnor. This county, the least populous of those shown in the table, and containing less than 1 per cent. of the total population in Wales and Monmouthshire, ranks last but one in the order of frequency of the Welsh speaking element. It is in no sense a Welsh speaking county, and the small increase in Welsh speakers now shown is a negligible exception in the general decline. Welsh monoglots, who comprise little more than one-sixth of the total Welsh speaking population in the country, are fewer than they were ten years ago in every county, with the negligible exception of Radnor referred to, but the actual reductions in this section are not so great as they were in the decennium 1901-1911. During this period persons able to speak Welsh only fell by more than 20,000 in Glamorgan, and in both Carmarthen and Carnarvon the loss exceeded 10,000. Between 1911 and 1921, the greatest numerical reduction has been one of 9,396 in Carnarvon, followed by 6,179 in Glamorgan, and in no other county does it exceed 5,000, while in Radnor there is an actual increase.

On the other hand, the numbers of persons able to speak both English and Welsh, which were increasing at a slightly higher rate than that of the general population between 1901 and 1911, now show a tendency to decline. In Glamorganshire, in which approximately 45 per cent. of the total bilingual class were enumerated, the number increased by nearly 70,000 between 1901 and 1911, and now shows an apparent decline of 18,616, though it is only fair to point out that this figure is probably higher than it would have been had there not been so large an increase in the number who failed to return the language statement, an increase in this county amounting to nearly 22,000 altogether. Of the remaining counties six out of the twelve register a numerical decrease in the bilingual section as against an increase in the previous decade, and, in the remainder, the current increases are all of a small order except in Carmarthen, where an accession of 11,893 in 1911-1921 compares with the much greater increase of over 27,000 in 1901-1911.

From the point of view of geographical distribution it will be observed that Welsh speakers are located predominantly in the five counties Anglesey, Cardigan, Carmarthen, Merioneth and Carnarvon, a continuous strip of the country on the Western seaboard stretching from the extreme north to the extreme south. The proportions of persons speaking Welsh only in these counties are such as to differentiate them completely from the remainder of the country and the proportions of bilinguists are also higher, though the cleavage is not so sharply marked as in the monoglot section. Of the remaining counties, Denbigh and Montgomery fall into a class intermediate between the extreme types, while Radnor and Monmouth occupy an outstanding position at the end of the series, the Welsh element in each affecting less than 7 per cent. of their populations. The case of Pembroke is interesting, for this county is situated to the west of the Welsh speaking zone, and conflicts, therefore, with a strict geographical criterion of incidence. Notwithstanding its position, it must be regarded as a member, though a detached one, of the group of predominantly English-speaking counties.

In the towns and rural districts within the several counties, of which comparative statistics are given in Table XX of the text portion of the County census reports, the variations in the proportion of the respective language groups are naturally much wider than those shown for the county units in the table above. Generally speaking, the predominance of Welsh monoglots is greater in the rural areas, and English monoglots in the towns, but this rule, as with the one relating language incidence to geographical situation, is subject to many exceptions. The highest proportions of persons speaking only the ancient language are found in—

Penllyn R.D. (Merioneth) 573 per 1,000.
Lleyn R.D. (Carnarvon) 569
Uwchaled R.D. (Denbigh) 560
Glaslyn R.D. (Carnarvon 542
Gwyrfai R.D. (Carnarvon) 514
Llanfyrnach R.D. (Pembroke) 513
Dwyran R.D, (Anglesey) 512

and of total Welsh speakers, monoglots and bilinguists combined, in

Llanfyrnach R.D. (Pembroke) 975 per 1,000.
Dwyran R.D. (Anglesey) 970
Bethesda U.D. (Carnarvon) 966
Gwyrfai R.D. (Carnarvon) 961
Aberayron R.D. (Cardigan) 945
Machynlleth R.D. (Montgomery) 942
Newcastle-in-Emlyn R.D. (Carmarthen) 940
Aberayron U.D. (Cardigan) 937
Ffestiniog U.D. (Merioneth) 936
Lleyn R.D. (Carnarvon) 934
Uwchaled R.D. (Denbigh) 933
Penllyn R.D. (Merioneth) 930

The much greater prevalence of English, whether spoken alone, or in conjunction with the ancient tongue, is shown by the county proportions given in the last columns of Table LXXXVIII. In seven of the counties, containing between them more than three-quarters of the total population under review, the proportion of persons speaking English and English only never falls below 50 per cent. In the remainder, though the proportion of English monoglots falls to as low a minimum as 122 per 1,000 of the population (in Anglesey), there is a compensatory increase in the bilinguists, so that the total English-speaking element is in no county less than two-thirds of the whole, while in eight it exceeds 90 per cent., and in the extreme case of Monmouthshire reaches the very high proportion of 99.8 per cent. of the total.

Next Selection Previous Selection