Towards a corpus of eighteenth-century English phonology

This paper gives an account of plans for constructing a searchable database of eighteenth-century English phonology, an area which has hitherto received little attention from corpus linguistics. The project draws on a sample of eighteenth-century primary sources to construct a searchable database which will eventually provide visualisations of the distribution of phonological variants in time, space and social class. The project incorporates data from pronouncing dictionaries and other texts dealing with pronunciation published in the second half of the 18th century. The data will be recoded in the form of Unicode transcriptions of as many of the approximately 1,700 words used to exemplify John Wells’ (1982) Standard Lexical Sets as appear in the texts chosen, together with supplementary sets chosen to represent consonantal variants such as /hw/~/w/ in WHICH, etc. The use of these sets and their associated keywords is standard practice in studies of variation and change in English, and including the full range of example words allows for differences in lexical distribution between the eighteenth-century texts, and between these and the contemporary accents described by Wells. Although all the eighteenth-century texts purported to describe the ‘best’ English, they were compiled by authors from different parts of the English-speaking world (mainly different regions of England, Scotland and Ireland but including some from North America) and so can provide evidence for geographical diffusion of innovations. (Beal 1999, C. Jones 2006). The entries will be tagged according to the main lexical set to which they belong. Thus, a researcher interested in the distribution of words in Wells’s (1982) PRICE and CHOICE sets will be able to find how each of the example words from these sets was transcribed in each of the 18th-century sources included in the database. There will also be links to descriptive and prescriptive comments included in the primary sources. The database will also include metadata providing background information on the texts, such as place of publication, birthplace, occupation and social class of author, and bibliographical references to published work referring to these sources. This paper provides an account of the design of this database and presents the results of a pilot study demonstrating how such a database can be used to answer questions concerning the chronological, social, geographical and phonological distribution of variation between /hw/ ~/w/ ~ /h/ in WHICH, WHO, NOWHERE, etc. which is of interest to sociolinguists, dialectologists and historical phonologists.