top of page

Censoring multilingualism? Language questions in the 2021 Census

Mark Sebba and Wendy Ayres-Bennett | 23 April 2021 | Opinion Articles

In this article, Mark Sebba, Reader Emeritus of Linguistics and English Language at Lancaster University, and Wendy Ayres-Bennett, Professor of French Philology and Linguistics at the University of Cambridge, reflect on the inclusion of language questions in the 2021 Census in England. They discuss the shortcomings of the current approach and argue that new and better questions are needed to capture the true extent of England’s multilingualism.

In England in March this year, the Twittersphere was alive with comments about the national census questions asking about language use. On the one hand, there was negative reaction to the advice from groups such as British Hindu leaders to enter ‘Gujarati, Tamil, Hindi’ in response, on the grounds, to cite one Tweet,‘ how many British-born Hindus actually use anything but English as their “main language”?’. On the other hand, many bi- and multi-linguals complained bitterly that they found the question about their main language impossible to answer.

As only the second time in history that the census has asked language questions in England (in the other nations of the UK, questions had been asked about Welsh and Gaelic for over a century), inclusion of a language question might seem an affirmation of the importance of the role of language in an increasingly diverse society. Unfortunately, the wording of the questions means that the data collected will be nowhere near as informative as they could and should be. Whatever form the next census takes, new and better questions are needed to capture the true extent of England’s multilingualism. The 2011 and 2021 censuses in England both asked the same two questions: (1) ‘What is your main language’ and (2) – only for those who did not answer ‘English’ to the first question – ‘How well can you speak English?’. For the first question only one answer was permitted, reflecting the monolingual mindset which also underlies the assumption that everyone must have a ‘main language’. For the second question, the respondent was required to choose between ‘very well’, ‘well’, ‘not well’ and ‘not at all’. Both questions were, in our view, problematic. As was clear from the 2011 census, they result in the collection of incomplete and unreliable data about England’s rich multilingualism and a missed opportunity to capture accurate knowledge of the population’s linguistic and cultural competence in languages other than English, a valuable (but undervalued) resource for British business, trade and the economy, diplomacy and soft power, integration and social cohesion.

The decision to ask about ‘main language’ and to limit the answers to a single response ensured that no one could register the ability to speak more than two languages. Furthermore, those who chose English as their ‘main language’ were unable to report speaking other languages at all. The printed form offered very brief guidance as to how to respond, ‘If you're not sure what your main language is, think about the language you use most naturally. It could be the language you use at home’; online, this was reduced to the simple suggestion that this might be your home language. For many bilinguals, the choice of one main language is not just difficult, but nonsensical, leading to partial and misleading answers, no matter which language they chose. Their answer might reflect the language they speak most often, speak most fluently, their first language or the official language of their place of origin (Matras and Robertson 2015).

Respondents who were used to functioning regularly in two (or more) languages – for example, in English and in a community, heritage or indigenous language – were thus left to decide for themselves which to declare as their ‘main language’. How were you to respond if you are a Welsh student using predominantly English at university in England but Welsh at home, or a Polish administrator functioning in English at work but in Polish at home and with the Polish community? What if you were a receptionist who has learnt Urdu as a heritage language and been brought up bilingual, who mostly uses English at work, but can put Urdu speakers at ease by addressing them in their language? The wording of the question precluded capturing these and numerous other multilingual realities and thus drawing any kind of detailed or nuanced picture of the numerous multilingual households of England. Equally neglected were speakers of languages such as French, Spanish, German or Chinese that had been acquired at school and university, sometimes to a very high level.

The question about ability in English – although similar to questions previously used in the US census – was likewise problematic. Whilst self-reporting is at the heart of census returns, in linguistic matters it is known to be particularly problematic (Carr-Hill et al. 1996, Edele et al. 2015), especially when coupled with insisting on viewing ‘proficiency’ as measurable separately from its context, which fails to recognise the interactive nature of communication (Zentella et al. 2010). Instead of using the vague, relative terms ‘well’, ‘very well’, etc. more concrete and objective formulations might have helped, such as ‘do you need an interpreter at the doctor’s?’ or ‘can you hold a conversation in English with your neighbour’, particularly in view of the stated goal of eliciting information about which services are needed for those whose first language is not English.

Following the 2011 census, evidence began to emerge of these shortcomings. In Manchester, a large, ethnically diverse city, Matras and Robertson (2015) compared the data derived from the national census with data from the Schools Census, an annual survey carried out in every English school. For the Schools Census, schools record a single ‘first language’ for each child, so while the multilingual dynamic of households is largely overlooked, it provides a general idea of multilingualism within the school and community. The 2013 Schools Census in Manchester recorded 34% of school students as having a ‘first language’ other than English, and listed 152 different languages altogether. Yet the national census recorded that only 16.6% of the Manchester population over 3 years of age had a ‘main language’ other than English, and listed a total of 67 specified languages in addition to a number of ‘other’ language groupings (such as ‘all other South Asian Languages’). The national census, although it had much more comprehensive coverage than the Schools Census, had clearly failed to pick up a lot of linguistic diversity. Matras and Robertson also pointed out that the ‘main language’ question seemed to have missed many speakers of particular languages. For example, the census found 559 people with the Nigerian language Yoruba as ‘main language’ in Manchester; however, the Schools Census records 427 Yoruba-speaking pupils. This gives a very unrealistic ratio of adult to school-age speakers of Yoruba. It seems likely that most Yoruba-speaking children have at least one Yoruba-speaking parent and in many cases, two; but the ‘main language’ question did not elicit this information.

Many other countries ask language questions in their national censuses or household surveys, and while these may ask for the respondent’s ‘home language,’ ‘first language,’ ‘mother tongue’ or similar, the term ‘main language’ is not generally used. In Scotland, the corresponding question in the 2011 census was ‘Do you use a language other than English at home?’. While this might have been expected to produce broadly similar results to a question on ‘main language’, Sebba and Turner (forthcoming) show that in fact some minority languages like British Sign Language and Italian were much more widely reported in Scotland than in England, showing that asking about ‘home language’ is a more effective way of finding out about other languages in use. There remains the issue of only allowing a maximum of two languages per speaker. Other countries ask questions which invite the respondent to show their linguistic repertoire: for example, New Zealand asks ‘In which language(s) could you have a conversation about a lot of everyday things?’, allowing for multiple answers. Such questions have the potential to reveal a much more detailed picture of multilingualism within households.

The Office for National Statistics in England has prioritised identifying one ‘main language’ as the language in which government can communicate with residents whose English is limited (Sebba 2017: 272). In so doing, they have overlooked not only linguistic realities – the fact that a ‘main language’ is difficult or impossible for many people to identify – but also have missed the opportunity to find out more about the variety of linguistic resources which the population has available. We believe that future censuses should take up this opportunity, by asking more probing questions about language and allowing for multiple answers.


Carr-Hill, Roy, Steve Passingham, Alison Wolf and Naomi Kent. 1996. Lost Opportunities: The Language Skills of Linguistic Minorities in England and Wales (London: Basic Skills Agency)

Edele, Aileen, Julian Seuring, Cornelia Kristen and Petra Stanat. 2015. 'Why bother with testing? The validity of immigrants’ self-assessed language proficiency', Social Science Research, 52: 99–123

Matras, Yaron and Robertson, Alex. 2015. ‘Multilingualism in a post-industrial city: policy and practice in Manchester’, Current Issues in Language Planning, 16(3): 296-314. DOI: 10.1080/14664208.2015.1048925

Sebba, Mark. 2017. '"English a foreign tongue": The 2011 census in England and the misunderstanding of multilingualism’, Journal of Language and Politics, 16(2): 264-84. DOI:

Sebba, Mark and Graham H. Turner (forthcoming). '"Home language", "Main Language" or no language: questions and answers about British Sign Language in the 2011 British censuses'

Zentella, Ana Celia, Bonnie Urciuoli and Laura R. Graham. 2007. 'Problematic language assessment in the US Census', Anthropology News, September 2007: 10-11

Cite this article

Sebba, Mark and Wendy Ayres-Bennett. 2021. ‘Censoring multilingualism? Language questions in the 2021 Census’, Languages, Society and Policy

Download P • 143KB


bottom of page