Tuesday 16th July
Wednesday 17th July
Thursday 18th July
Friday 19th July
Download the conference book
Language-related aspects of surveys |
|
| Convenor | Dr Isabelle Renschler (FORS) |
| Coordinator 1 | Dr Brian Kleiner (FORS) |
| Coordinator 2 | Mr Stefan Buerli (FORS) |
There is increasing recognition that language-related factors can interfere with survey work in variety of ways, and can affect data quality. While survey translation already has an established research tradition, for this session we invite papers on other language-related aspects of surveys, especially in cross-cultural or cross-linguistic contexts. The papers may address issues related to conducting surveys in multilingual settings within and across countries, such as: barriers to survey participation for linguistic minorities (reaching them, establishing contact, getting cooperation); the use of proxies for those who do not speak the survey language; the costs and benefits of using non-majority languages for first contacts with linguistic minorities; the payoff of providing questionnaires for linguistic minorities in their mother tongue (given the costs of translation); comprehension of survey questions among foreign populations. Papers may focus on methodological challenges regarding language, but also on practical solutions in the field. Also invited are papers that address sociolinguistic aspects of survey interviews across cultures or subgroups, such as: levels of formality; dialect variation; on-the-fly adaptation and departures from standardisation; discourse norms (e.g., forms of personal address across cultures, encoding forms of politeness in scripted questionnaires).
Attention has a long time focused on the role of linguistic skills of immigrants. These abilities are extremely important for the social and structural integration of those immigrants. Research takes language ability as an outcome variable but also as an explanatory variable (e.g. income). Our presentation will compare linguistic test scores to parents-assessment and third-party-assessment by an interviewer. We will analyze which these different measurements are most suitable for social research. With data of the project 'Preschool Education and Educational Careers among Migrant Children' we can compare these different types of measurement for children at the age of 3-4 years. First we will descriptively show the correlations between them. We assume that the test score are the most objective measurement, whereas parental assessment should be biased and the assessment of the interviewer should be rather good. Second, we will use the variables as dependent variable. By explaining language ability with efficiency, exposure and incentive (cf. the language model by Chiswick and Miller) we can compare the results between the different measurements of language. Last we will treat the language variables as independent explanatory variables and will see if the objective measurement leads to different (better) results. In sum the results will show if objective language measurement is really necessary or if self- or third-party-assessment is also adaptable for specific research questions and a much cheaper option for huge surveys.
Language of interview is not always obvious choice in many cultures, nations and political contexts. It should be done in regions with multi-lingual population but not only. The decision about the language of interview is taken on two levels - on project planning stage and on the field stage.
On the planning stage several types of considerations can influence the choice of languages for interviews: a) linguistic considerations (suitability of the language for oral and written interaction, linguistic skills of surveyed population and interviewers, functional development of particular language - abstract social and political terms, etc.), b) organizational considerations (costs of the translation, visibility of high quality translations, interviewers with particular language skills, etc.), c) social and political considerations. On the second stage - during the field work, the choice of language depends on the set of languages offered by researcher, cultural norms and traditions in interaction, functional image of the language among target group, linguistic skills of actors, type of interaction between interviewer and respondent, direct influence of interviewer on respondent, personal preferences.
These issues are considered on the example of regions of the former Soviet Union where new independent countries appear on the map in the recent decades and the experience of survey research is still rather limited. The political role of the language in these regions is rather high as well as organizational limitations. The paper describes the current practices and methodological experiments on the choice of language and its consequences.
Most general population and migrant surveys in Germany only conduct interviews in German. Translations of the master questionnaires into foreign languages are not common due to additional costs of (back-) translation or of pretests that have to be conducted in other study languages.
Consequently, if translations are used, the benefits should outweigh the additional costs if migrant groups belong to the population under study. But what are the benefits concerning data quality in this case? The answer to this question requires more information about the survey process than just the rate of completed interviews in foreign languages that is com-monly used as a quality indicator.
In our large-scale study of Muslims and Christians with and without migration background in Germany, we thus additionally asked the respondents about their reasons for choosing the translated version. Here, we differentiate between "insufficient knowledge of German", "pleasant to be interviewed in the mother tongue" and "no special reason".
The translated version is obviously of essential importance for the first group with insufficient knowledge. However, does this group systematically differ from the other two groups who - in theory - could have been surveyed in German? We compare these respondents with the other two groups on demographic characteristics and substantive survey variables of the subgroups. We especially consider differences on attitude questions but also on religiosity and religious affiliation of respondents. Consequently, we can better justify the need of using translated questionnaires for respondents with migration background and demonstrate its importance for data quality.