This chapter throws into relief the importance of the link between corpus hits and their sources (i.e. texts or speakers) by comparing study designs uninformed and informed by such metadata. The authors’ argument draws attention to the consequences of uneven distributions of observations across the basic text units of a corpus. In the case study (a distributional analysis of the use of actually in the BNCs of 1994 and 2014), it is demonstrated that disproportionate word counts contributed by individual sources (in this case, speakers) will distort estimates for relevant subsections of corpus data (in this case, demographic groups defined by age and gender) if the analysis assigns the same weight to every observation. The proposed solution is to factor in the text (or speaker) level, but this hinges on the availability of the relevant metadata. Moreover, insights into the hierarchical structure of corpus data are shown to benefit the design stage of a study. Thus, if manual post-processing steps preclude an exhaustive analysis of corpus hits, insights into the organization of data points can direct down-sampling strategies to generate a statistically efficient subset of tokens.

Introduction: design of the study, research questions and hypotheses

This chapter will give an overview of different approaches used in current linguistic research, with a focus on the design of questionnaires and (sociolinguistic) interviews. Theoretical considerations will be illustrated by examples from questionnaire-based studies on regional contact varieties of English (cf. Krug and Rosen 2012; Hilbert and Krug 2012; Krug, Hilbert and Fabri in press for methodological and descriptive detail).

The design of an empirical study obviously varies depending on the research questions one attempts to answer and also on the type of data one intends to analyse. Thus, formulating research questions or precise hypotheses serves as an important guideline both in designing the study and in analysing the data.

If one wants to apply a sociolinguistic approach, i.e. investigate the influence of social factors such as age, gender, ethnicity, etc. on people’s speech, the following issues have to be considered prior to the collection of data. Which variables might be of interest? Which speakers are needed for the analysis, e.g. children, speakers of different age-groups (i.e. cohorts), older people, only males or females, a gender balance, only working-class or highly educated speakers, migrants, native locals? In which circumstances should the data be collected? For example, for an analysis of spoken language, one can conduct interviews. For an acoustic analysis, one should ensure that interfering background noise is kept to a minimum. For sociolinguistic interviews, a relaxed and familiar setting can be an advantage (cf. Tagliamonte 2006: 45). Another important early step is to identify the linguistic variables and their variants in the database. The figures have to be carefully processed and the results cautiously interpreted. Without an interpretation of the data, an important part of the linguist’s work would be missing. These individual steps are crucial for linguistic studies and they tend to form cycles, although their order may vary (cf. Hudson 1996: 150–151). It seems evident that both stereotypes of die-hard ‘number-crunchers’ and ‘armchair linguists’ portrayed in the introductory chapter of this book fail to complete the empirical cycle detailed in Figures 1 and 2 in Krug, Schlüter and Rosenbach (Introduction, this volume).

The importance of variation and change

Language variation and change highlight the fact that language universally involves alternative forms and structures that compete with each other in usage. For instance, speakers of Scottish varieties of English may in certain circumstances front the initial consonant in thing and pronounce it as fing. A speaker from Cumnock in Lowland Scotland or from Portavogie in Northern Ireland may occasionally drop the subject relative pronoun in the man (who) called me was our neighbour. An eighteenth-century speaker and his twenty-first-century descendant may both use kneeled down, although the latter is more likely to use knelt down. As is evident from this arbitrary choice of examples from the present volume, language is inherently variable, both across time (diachronically) and at any specific point in time (synchronically). In the investigation of both synchronic and diachronic linguistic variation, the classic variables relating to the language producer are geographical, stylistic and social in nature. The fact that especially social information (like age, sex, socio-economic class) figures more prominently in the study of more recently produced data follows naturally from the fact that such information is less readily accessible for older data (cf., however, Nevalainen and Raumolin-Brunberg 2003).