Reversing Language Shift: Celtic Languages Today – Any Evidence ?
(paper to: 12 th International Congress of Celtic Studies University of Wales: Aberystwyth 24 th. – 30 th. August 2003)
Kenneth MacKinnon
Introduction
The concept of reversing language-shift was introduced by Fishman (1990, 1991), and is now widely used in the field of the sociology of minority languages. The idea of language-shift itself goes back to earlier work of Fishman on bilingualism and diglossia (1971), and has been used a conceptual basis for application to the cases of Celtic languages (MacKinnon, 1977, 154-167, 184-188). Such minority language situations can be seen as occupying legitimate sociolinguistic space within a given society, and commanding a range of social uses (‘demesne-extension': MacKinnon 1977, 148 - 154, 170), or in a more dynamic way engaging in ‘struggle' for speakers. (Williams, 1992, 113f) The displacement of one language by another in a particular domain or range of uses goes beyond ‘demesne-extension' (or its converse: ‘demesne-retraction'), and may result in a change in numbers of actual speakers of the language in question. Language-shift is here taken to comprise more specifically the resultant contraction in actual numbers of speakers of a given language which results when the language is no longer used as a language of child-raising or intergenerational transmission. The concept of reversing language shift is the process whereby the contracting language-group reverses this process and gains in numbers.
Population censuses in the United Kingdom and the Isle of Man enable this phenomenon to be assessed empirically. Questions on ability to speak Celtic languages have been asked on population censuses since 1851 in the case of Irish, 1881 for Gaelic, and 1891 for Welsh. From 1921 age-analysis of speakers has been presented, providing some further impression of language maintenance and viability. Following partition, questions on Irish were not asked in Northern Ireland in 1921 – nor again until 1991. In the Isle of Man , there are census data on Manx in 1991 and 2001. These data all provide an empirical basis on which the assessment of reversing language shift may be attempted.
Age-analysis and impressions of language maintenance and viability
Age-analysis of language-groups in recent censuses enables language-maintenance to be assessed. Figures 1 – 4 indicate comparisons of numbers of speakers by age-group in the 1991 and 2001 censuses. The census data for Gaelic are most detailed of the four cases considered here, These are illustrated in Figure 1
Figure 1: Gaelic speakers by age: comparison of numbers 1991 – 2001.
Source: GROS Census 1991 Table LBS67S; Census 2001 Tables S206, T27.
Although there have been increases in numbers of Gaelic speakers in the 0-4, 5-11 and 12-15 age-groups, numbers have fallen in almost every other age-category. Numbers have fallen overall ( from 65,978 to 58,969 speakers), and the increases amongst young people do not seem to have the potential for reversing language shift.
In the case of Irish in Northern Ireland, assessment is made doubly difficult as the age-groups in the 2001 census results (Table S1372 ) are far less detailed than in 2001 (Irish Language Report, Table 1). However by combining age-categories in 1991 data, corresponding age-groups to those in 2001 can be aggregated. This results in a far less detailed picture for Irish in Northern Ireland than for Gaelic in Scotland . A further difficulty results from the inclusion of some of the actual speakers of Irish in an ‘Other Combinations' category (totalling 24,167 persons of whom 15,280 were Irish speakers.) The results as they are presented are illustrated in Figure 2. The age-categorisation for 2001 as detailed in Table S372 provides a poor impression of the actual age-structure of the Irish language-group. Apart from ‘burying' actual speakers in an ‘Other Combinations' category, the age-ranges are irregular and do not match those for 1991. The situation is illustrated in Figure 2A
Figure 2A: Irish speakers in Northern Ireland by age: comparison of numbers 1991 – 2001.
Source : N.I Census 1991: Irish Language Report, Table 1; 2001 Census, Table S372.
Figure 2A presents totals of all columns in Table S372 which state that they include actual speakers. The ‘Other Combinations' column also contains actual speakers, ands a total including this category is also added to the diagram. The actual total of speakers lies somewhere between these two values – but is unknown until such time as the Northern Ireland census presents data on actual Irish speakers by age-group.
Aggregation of speaker categories from province and district totals based on Table UV14 (to date unpublished) indicates that total Irish speakers have reduced from 131,974 to 115. The impression from Figure 2A is that the age-profile builds up fairly evenly from the 3-11 age-group to the 25-39 age-group and at this level of analysis does not seem to be ‘spearheaded' by conspicuous increases in the school-aged category – despite the development of Irish-medium primary and secondary schooling during this period. This impression is misleading and results from the uneven age-groupings. A different impression results from re-calculation based upon mean single years of age in each category, and is illustrated in Tfigure 2B.
It now becomes clear that there has been considerable effect resulting from the development of Irish-medium education. The results manifestin the themselves conspicuously at the secondary stage.12-15 age-group, and carrying over into the successive age-group of 16-24.
This is in contrast with the contrasts with the Gaelic case where the gains at the primary stage are reduced at the secondary stage and beyond as the result of poor follow-through. The Manx and Welsh cases , although on different scales, both show considerable relative gains in speaker numbers in the age-range of school-attendance. In the Welsh case there are also gains in the preschool age-group. Between 1991 and 2001 there have been notable advances for Manx in the Isle of Man school system: both in terms of second-language learning, and as of Manx as a teaching medium. The resultant situation is illustrated in Figure 3.
Figure 3: Manx speakers by age: comparison of numbers in 1991 and 2001.
Source: Isle of Man Census Report 1991 Vol 1, Table 3; 1991, special tabulation 15.08.03;
Isle of Man Census 2001, special tabulation 11.08.03, 15.08.03.
The age-profile shows a dramatic increase amongst school-aged children. Here, as elsewhere, it must be emphasised that this is a census effect resulting from the return of these children as able to speak Manx by their parents. The extent of this ability is not questioned by the census – neither is actual use of the language in family and community. To date there has been only one survey of Manx usage (Gawne, 1999). It did have some questions on Manx and the family, but these were more in terms of what the respondent would wish rather than on actual patterns of language-usage. There is now a need for a study of the language-abilities and usage patterns of the young people who have acquired Manx through their school experience.
In terms of numbers, the Welsh language-group achieved some stability between 1981 and 1991, with total numbers of Welsh speakers increasing from 503,549 to 508,098. Since 1971 there had been a steady increase in both numbers and percentages in all young person age-groups under 15, and from 1981 the 15-24 age-group also. These trends have continued to 2001, with increases in the numbers of young people returned as Welsh speakers overcoming losses of numbers through mortality and migration. This trend continued even more strongly between 1991 and 2001, with numbers of Welsh speakers increasing overall from 508,098 to 582,368 (Census 2001, Table UV84). The considerable gains amongst 3-15 year-olds show up particularly conspicuously in Figure 4A.
Figure 4: Welsh speakers by age: comparison of numbers 1991 - 2001
Source; OPCS 1991 Census LBS 67W; ONS 2001 Census, Table S133
This analysis however actually under-enumerates total Welsh speakers. By aggregating the various speaker categories, a national total of 575,640 can be derived from Table S133. Some 6,728 Welsh speakers have inexplicably been ‘buried' in an ‘Other combinations' category. This is apparent from Table T 39 which categorises the entire 582,368 total of speakers, although it has fewer age-categories. The un-evenness of the age-categories in both tables can be overcome by calculating a mean single year-of –age equivalent for each category. The results are presented as Figure 4B
Figure 4: Welsh speakers by age: comparison of numbers 1991 - 2001
Source; OPCS 1991 Census LBS 67W; ONS 2001 Census, Table T39
In terms of numbers of children returned as Manx or Welsh speakers, compared with numbers in the parental age-groups, reversing language shift seems clearly to have occurred. More precise measures of this phenomenon can now be considered.
Reversing Language Shift (RLS) from Census Results. .
A measure of RLS can be devised by comparing age-data on speakers for the parental generation and for the generation of childhood. To some extent the age-limits of these groups will be arbitrary, but in this analysis the age-group of childhood is taken as the age-group of statutory school attendance, from 5 th to 16 th birthdays, i.e. the 5-15 age-group, which is accounted as such in 1991 and 2001 census age-tables. This age-range is probably more reliable than 0 – 15s (which would be possible to calculate for Gaelic and Manx, but not for Welsh or Irish), or 3 – 15s (which would be possible to calculate for Welsh, Gaelic, and Irish, but not for Manx ). Tracing cohorts of young speakers through successive censuses has indicated that the under-5 group does not comprise all potential future speakers. In recent censuses numbers have increased within this cohort as it ages into the next two or three quinaryl age-groups. (MacKinnon 1984, 2) The choice of parental age-group presents greater difficulties. The conventional age of parenting can be regarded as occurring between ages 20 – 44. But of course there are many cases of both younger and older parents. Until such time as tables for speakers in family and household groups might become available in the cases of all the Celtic languages, more precise assessment and comparison between them cannot be attempted.
This analysis is an interim measure put forward as an approximation and indication of potential. This is rendered even more of an approximation in that age data for Irish speakers in Northern Ireland are not presented in the same age-categories as for Welsh and Gaelic The closest categorisation for the childhood group of Irish speakers in Northern Ireland in 2001 is 3 – 15, which is broader than the other language cases, and the parenthood group 25 – 39, which is somewhat narrower. In the case of Manx the age-groupings for young people are 0 - 4, 5 - 9 and 10 - 14, These categories do not match those for Welsh, Gaelic or Irish in Northern Ireland . These are however consistent between the 1991 and 2001 censuses and enable intra-group comparison to be made. Comparison on this basis with the other language groups is thus impressionistic rather than strictly like with like. (For further discussion of these problems, see MacKinnon 2003).
Operationalising the concept of RFS from census data
The 20 – 44 age-group could be regarded as a working approximation to the parental age-group, and can be taken from census data for Welsh, Gaelic and Manx. Unfortunately the 2001 age-categorisation in Northern Ireland does not enable this to be done for Irish. The best age-group is 25-39, which is not really very satisfactory. The age-group of childhood can be taken as 5-15 in the cases of Welsh, Gaelic and Irish in Northern Ireland . In the case of Manx the age-categorisation is 5-14.
Reversing Language Shift can be said to occur when the language is stronger amongst young people than their parents. Various measures can demonstrate this. A measure of salience of the language in a particular age-group can be demonstrated by the proportion of speakers amongst the total age-group. Where this is greater amongst young people than their parents RLS may be said to occur. The difference in incidence may stand as a measure of Intergenerational Gain – or Loss. External factors may of course affect this and conceal actual gains or losses, such as differential migration of age-groups.
A more reliable and valid measure may be found in comparing the mean values of both age-groups of speakers expressed in terms of single years of age. If these are compared as a ratio, for example by dividing the mean childhood value by that for parenthood, an Intergenerational Ratio measure may be derived. Where greater than unity RLS may be said to have occurred, and where less than unity language shift is still taking place. Either may of course be the case in circumstances where the salience in a local population is quite small, and the values are unaffected by external factors.
The two measures may be plotted together on a graph enabling different areas and language groups to be compared. Figure 5 compares these measures for the four language-groups in the two censuses of 1991 and 2001, and thus gives some impression of trends. The cases of Manx and Welsh stand out dramatically in this comparison. Manx demonstrates the most considerable gains, even though its salience in its local society is small. Welsh has much greater salience, but it too has made considerable gains in terms of both measures. Gaelic and Irish in Northern Ireland have both made gains on these two measures, but they are slight. This analysis may however be helpful in monitoring progress.
Figure 5 : Reversing Language Shift 1991, 2001.
Sources : OPCS 1991 LBS 67W, GROS 1991 , LBS 67S, NI 1991 Census Irish Report, Table 1, IOM 1991 Census Report, Vol 1 Table 3, special tabulation 19.05.98, 11.08.03;
ONS 2001 Census S133, T27, GROS 2001 S206, NISRA S372, IOM 2001 special tabulations 11.08.03, 15.08.03.
The concepts may also be used in comparing local areas graphically and in mapping. Figures 6 and 7 show the different strengths of RLS in local authority areas in Wales and Scotland . Figure 6 clearly indicates the strength of RLS throughout Wales . Both values are positive in every area. The gains amongst young people show up very clearly in southeast Wales on both measures. In the Welsh ‘heartland' areas (Y Fro Gymraeg) the gains are less conspicuous, as in these areas there is not so much ‘leeway' to be made up anyway.
Ideally it would have been preferable for T39 data, which include all Welsh speakers, to have been used as the source for this analysis. However neither this table, nor S133 was available at local authority level at the time of preparation of this paper. Thus the age-categorisation is not strictly comparable with that in Figure 7 for Gaelic speakers. Nevertheless it is quite apparent that considerable strides for Welsh RLS are in process, In comparison with Welsh only six local authority areas in Scotland register positive results on these two measures. The overall picture is shown in Figure 7.
Figure 7: Comparison of Gaelic speakers in the 5-15 and 20-44 age-groups: Intergenerational Gain / Intergenerational Ratio in the 1991 Census
Source: GROS 2001 Census: S206.
Figure 6: Comparison of Welsh speakers in the 5-15 and 20-44 age-groups: Intergenerational Gain / Intergenerational Ratio in the 2001 Census.
Source: ONS 2001 Census: S137
In terms of scale, the Gaelic plot could be contained within the three bottom left-hand cells of the Welsh diagram, and a further three negative cells below them. In comparison with Welsh, any gains for Gaelic RLS are miniscule. Nevertheless six local authority areas are positive on both measures: Argyll & Bute, Highland and Perth & Kinross in the traditional Gaelic area, together with three Central Belt authorities: Stirling, Clackmannanshire and North Lanarkshire . These authorities have developed Gaelic-medium primary education, and have Gaelic development staff. The strongest ‘heartland' area for Gaelic is the weakest in terms of Intergenerational Loss – and just fails to register positively in terms of Intergenerational Ratio. Although it has developed Gaelic-medium schooling, only 25% of its primary children are enrolled. It is however in course of developing a public outreach through a local area language plan. If these measures can be calculated for 2011, shall we see this area move forward positively on both counts – and other Scottish areas make similar strides to those in south-eastern Wales ?
Prospects for RLS – some conclusions
Co-ordinated policies in education, the media and public life have been the basis of the successful achievement of RLS in Wales. Earlier policies developed were already showing results between 1971-81. Subsequently the trends have strengthened and are gaining critical mass and momentum. This has important lessons for language planning in other cases
In recent years The Isle of Man has developed Manx within its school system, and a potential for RLS is demonstrable amongst young people as a census effect. For this to mature into genuine RLS, further policies in Manx administration and life would need to be developed and gain a place in local society.
In Scotland Gaelic-medium policies have been successfully developed at the primary level. They hardly exist in the secondary sector. If the gains achieved in the primary sector are to realised, a strong Gaelic-medium presence needs to be developed at secondary level. In October 2003 a Gaelic Language ( Scotland ) Bill was announced. If it is to be the means of achieving genuine RLS for Gaelic it will need to include a clear declaration of the legal status for Gaelic, such as the principle of equal validity which is enjoyed by Welsh, initiatives for secondary education, and a basis for the new Gaelic Board to engage in language planning not just in the public sector but in wider commercial and public life also. Present policies are failing to deliver RLS for Gaelic. The new institutions now being created for the language will need to develop more appropriate and effective policies beyond the ‘school and other higher order props' (Fishman 1991, 380) which have proved insufficient in the past.
In Northern Ireland there is an immediate need for census data which enable the progress of Irish in education and public life to be demonstrated and monitored. Cross-border language authorities have been created under the ‘Good Friday Agreement' and there has been considerable development of Irish-medium education and media. Overall numbers seem to have slipped, but without a single years of age table for under-25s as in 1991, the progress of the language amongst young people cannot be monitored.
The population census has enabled the prospects for RLS to be revealed. It can be demonstrated to have occurred for Welsh – and maybe for Manx also. The further analysis of these and other aspects of the census results can provide a firmer basis on which to plan for Gaelic in Scotland and Irish in Northern Ireland .
References
Fishman, J. A. (1971) ‘The Sociology of Language: an Interdisciplinary Social Science Approach to Language in Society', in Fishman, J.A. (ed.) (1971) Advances in the Sociology of Language, Vol 1.The Hague: Mouton (no ISBN) pp.255 – 404.
Fishman, J. A (1990) What is reversing Language Shift – and How Can it Succeed? Pp. 5- 36 in: Gorter, D., Hoekstra, J. F., Lammert, G., and Ytsma, J. (eds.) Fourth International Conference on Minority Languages, Vol 1. General Papers , ( Journal of Multilingual and Multicultural Development Vol 11, Nos. 1 & 2 1909) Clevedon: Multilingual Matters ISSN 0143-4632
Fishman, J. A. (1991) Reversing Language Shift (RLS) : Theoretical and Empirical Foundations of Assistance to Threatened Languages. Clevedon: Multilingual Matters ISBN 1-85359-121-1 pbk
Fishman, J. A. (2001) Can Threatened Languages be Saved? Reversing Language Shift, Revisited: a 21 st century perspective, Clevedon: Multilingual Matters ISBN 1-85359-492-X
Gawne, P. (1999) Survey of Manx Speakers and Learners – Preliminary findings, Douglas IOM: Manx National Heritage and Heritage Foundation.
General Register Office for Scotland , 1991 Census, Local Base Statistics, Table 67S Edinburgh
General Register Office for Scotland , 2001 Census, Tables S206, T27.. Edinburgh .
Isle of Man Government (1996) Isle of Man Census Report 1991 (Table 3) Douglas : Economic Affairs Division, Treasury.
Isle of Man Government (2003) Special tabulations 11.08.03, 15.08.03, Douglas : Economic Affairs Division, Treasury.
MacKinnon, K. (1977) Language, Education and Social processes in a Gaelic Community, London : Routledge ISBN 0 7100 8466 8 pp.148- 154, 170.
MacKinnon, K. (1984) Gaelic Language Regeneration amongst Young People in Scotland 197-1981 from Census Data (paper to Future of Celtic nations in EEC, Heriot-Watt University 23.11.84) Hatfield Polytechnic Social Sciences Reports Series SSR 15. ISBN 0 900458 29 1
MacKinnon, K. (2003) Celtic Languages in the 2001 Census: how population censuses bury Celtic speakers, paper to 4 th Language and Politics Symposium, Queen's University of Belfast (forthcoming in proceedings)
Northern Ireland Research and Statistics Agency (2002, 2003) 2001 Census Tables S372, UV14 Belfast .
Office for National Statistics for England and Wales , 2001 Census, Tables S133, T39. Titchfield, Fareham , Hants.
Office for Population Censuses and Surveys for England and Wales , 1991 Census. Local Base Statistics, Table 67W. Titchfield, Fareham , Hants.
Registrar General for Northern Irealnd ( 1993) The Northern Ireland Vensus 1991 Irish Language Report (Table 1) Belfast : Her Majesty's Stationery Office (No ISBN)
Williams, G. (1992) Sociolinguistics –a Sociological Critique , London : Routledge ISBN 0-415-06514-3
Kenneth MacKinnon 20 th. October 2003
____________________________________________________________________________
Affiliation: Professor Kenneth MacKinnon is Visiting Professor and Emeritus Reader in the Sociology of Language at the University of Hertfordshire , Honorary Fellow in Celtic at the University of Edinburgh , and Associate Lecturer in Social Sciences, Education and Language Studies of the Open University.
Contact details: Ivy Cottage, Ferintosh, The Black Isle, by Dingwall, Ross-shire IV7 8HX
Tel: 01349 – 863460 E-mail: kenmackinnon@enterprise.net Website: www.sgrud.org.uk
____________________________________________________________________________