Intended for healthcare professionals


Why the next census needs to ask about language

BMJ 2005; 331 doi: (Published 11 August 2005) Cite this as: BMJ 2005;331:363
  1. Peter J Aspinall, senior research fellow (p.j.aspinall{at}
  1. Centre for Health Services Studies, University of Kent, George Allen Wing, Canterbury CT2 7NF

    Delivery of culturally competent health care and other services depends on such data

    Estimates of the numbers of people in England who have difficulties with the English language vary widely from around 400,000w1 to 1.7 millionw2; evidence from surveys shows that only 16% of Bengalis, 44% of Gujaratis, 29% of Punjabis, 41% of Chinese, and 32% of refugees would reach a "survival level" of competence.1 A recent skills audit of adult refugees from the Far East, Iraq, Somalia, and other parts of Africa found that just a quarter understood spoken English and a fifth could speak English.2 Unmet language need is one of the key drivers of social exclusion, barriers to employment, education and training, and inequity in access to services. Information on this is therefore urgently needed, to plan interpretation and translation services and the delivery of culturally competent health care.

    It is surprising, then, that in spite of demand no comprehensive information is available for the UK population on the use of community languages and proficiency in English. Many users in consultations expressed a requirement in the 1991 census for a question on language other than English usually spoken at home. The 2001 census, too, eschewed question(s) on language, even though the business case was strong: an effective question on main language spoken at home was devised, but this did not meet the stated primary need relating to ability in English.3 The initial view of the Office for National Statistics (ONS) on the content of the 2011 census for England and Wales is that proficiency in the English language is among "those topics where there is insufficient evidence of user demand to justify inclusion"4; it does not even make those where further work is to be undertaken before a decision is made on inclusion. This position is anomalous in the United Nations Economic Commission for Europe (UNECE; the region of Europe, central Asia, and North America), where 76% of countries asked one or more questions on mother tongue, main language, most spoken language, or knowledge of language(s) in the 2000-1 round of censuses.w3 w4 In contrast, the census agency in Scotland is consulting specifically on a question on languages spoken and understood.5

    The ONS evaluates the strength of the case for questions against three key user requirement criteria. Firstly, that of user need can be justified in terms of substantial resource allocation: the NHS weighted capitation formula for resource allocation now uses data on language difficulty that are 10-15 years out of date.6 Further, primary care trusts and other healthcare organisations need such information to plan interpretation and translation services locally, thus fully meeting the second criterion that census data should be required for small population groups and at detailed geographical levels. Thirdly, such information is not available from alternative sources such as administrative records. Questions on language are omitted from most of the government's major social surveys including the harmonised question sets (which could provide a mechanism to help update information between censuses), nearly all NHS core datasets—except diabetes user, national joint registry, and older people single assessment process—and much ethnic monitoring in primary care. The only comprehensive data, from the pupil level annual school census—on whether the first language of pupils is known or believed to be other than English (but omitting proficiency in English)—showed the expected high proportions among South Asians, Africans, and Chinese people, but also 46% in "any other white background" and 77% in "any other ethnic group."7

    Given the intense competition for space on the census form, a strong case for language has to be made. A survey is urgently needed of the major users of language data, to determine the language variables for which there is a demand (for example, main language spoken at home, first language, proficiency in English), so that questions and response categories can be prepared, tested, and evaluated. Censuses in the United States in 2000 and Australia in 2001 asked: "Does the person speak a language other than English at home?" (and what language) and "How well does the person speak English?" (in four response categories); both types of information are needed for planning interpretation and translation services. Other examples are available from the 2001 censuses in Canada (which asked five questions on language), New Zealand, and South Africa and a few UK national surveys, such as the 1999 health survey for England, the 1994 fourth national survey of ethnic minorities, and millennium cohort study 2001-2 sweep. The utility for public services of comprehensive census data is nicely illustrated in analyses undertaken by the US Census Bureau, including the conceptualisation of "linguistically isolated" people and households8 and cross tabulations of ability to speak English by language spoken at home.w5

    Increasing ethnic and cultural diversity in the population and the government's commitment to multiculturalism make the omission of language in the UK censuses conspicuous. In addition to established areas of need, substantial inflows of refugees and asylum seekers (exceeding 630 000 principal applicants during 1991-20039) and a major programme of dispersal are requiring much increased planning and provision of language services. The targeted intervention needed to tackle the current substantial underuse of NHS Direct by non-English speaking callers—who account for just 0.1% of the 0.5 million calls made monthly10—is equally urgent. Examples from specific care areas abound: a recent audit found that less than a third of titles of HIV information materials targeting African people were published in languages other than English,11 and work on African HIV prevention in London has been forced to use data for primary school pupils on language.12 Given the persistent and pronounced ethnic disparities in health and health care and demands of the Human Rights Act 1998 and Race Relations (Amendment) Act 2000, NHS healthcare organisations are urged to make a case to the ONS for including language questions in the 2011 census.


    • Embedded Image Additional references w1-w5 are on

    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    View Abstract