Introduction

The present chapter asserts the continuing relevance of a methodology which has gone out of fashion with some segments of the corpus-linguistic community: the careful and – dare we use the dreaded word: philological – analysis of small corpora with a balanced mix of quantitative and qualitative methods. This mix of methods reflects the twin advantages of the use of corpus data in linguistic description. On the one hand, corpus data are good data because they often come in large amounts, therefore encouraging quantification and statistical modelling. Expanding corpus sizes and increasing the sophistication of statistical analyses have been priorities in the past few decades of corpus-linguistic research; and the progress made has been remarkable (for illustration, it is sufficient to refer to the chapters in Parts 1.3, 2.1 and 3.2 of this volume). On the other hand, corpora are also good data because they present authentic records of individual written or spoken acts of communication in their original discourse contexts and thus encourage the qualitative study of complex verbal interaction. The current corpus-linguistic mainstream tends to be more aware of the first advantage than the second. However, if properly analysed, even a small number of examples obtained from small corpora can add considerably to our knowledge of change and variation in the English language, as will be shown below using the example of a currently ongoing shift from the to-infinitive to the bare infinitive in constructions of the type what I did was to find him a new job → what I did was find him a new job. If you can only obtain dozens rather than thousands of examples – a situation not uncommon in the study of spoken language or of historical data – this need not be a cause for despair.