Choices over time : methodological issues in investigating current change 1

The fact that English is changing is immediately apparent to a modern reader of, say, 18th or 19th century literature, or indeed to a teenager speaking to an elderly relative. However, as Mair (2006) points out, anecdotal evidence for linguistic change is unreliable. The systematic study of language change requires large, evenly balanced, and reliably annotated corpora with texts sampled over a period of time. These considerations are accepted by many linguists working on current changes in English. However, with regard to methodology we observe that within the field of diachronic corpus linguistics there are still a number of issues that generate a certain amount of discussion and debate. One of these concerns the issue of variability. Bauer (1994: 19) highlights the importance of this concept in studies of language change when he states that “change is impossible without some variation”. Variation within a set of linguistic choices, including the idea that there may be ‘competition’ between these variants, is fundamental to studies in current change. In this paper we will argue that an important methodological task for corpus linguists studying language change is to focus on linguistic variation where there is a choice. Many factors are likely to influence the use of particular words, phrases or constructions. If we wish to study and explain variation found in a corpus as being the result of factors affecting variation over time, then we need to eliminate as many potential alternative sources of variation as possible. This, we contend, calls for a restricted definition of the variants involved in a perceived change, and a consideration of any ‘knock-out’ contexts, i.e. contexts where variation may be impossible, or constrained in a different manner to the general case. We use the Diachronic Corpus of Present-day Spoken English (DCPSE) as a database. This corpus is unique in two important respects: it exclusively contains spoken English and is fully parsed, and as such is suitable for studying current change in English from the late 1950s to the early 1990s. It complements other resources, including major historical corpora of writing, notably A Representative Corpus of Historical English Registers (ARCHER) 2 which contains written texts sampled from the late 17th to the late 20th century, as well as corpora of earlier speech derived from written sources such as A Corpus of English Dialogues (CED; Kytö and Culpeper 2006) and the Old Bailey Corpus (Huber 2007). 3 In the next section we briefly present the functionality of DCPSE.