Investigating Variable Linguistic Systems in Flux in 17th Century Massachusetts English

Abstract

This thesis reconstructs the variable systems of two linguistic phenomena undergoing change in 17th century English - restrictive relativisation (Ball, 1996; Romaine, 1982), and do-support (Ellegård, 1953; Söderlund, 2017). Innovative here is the use of a digitized compendium of the Salem Witchcraft Papers (Ray, 2018), a repository of surviving textual data from the pre-trial hearings of the Salem Witch Trials, to leverage new insights into the speech of American English in the late seventeenth century, contributing to an area of scholarship that remains relatively understudied (Kytö & Siebers, 2022). Trial data, including depositions, various letters and petitions, and recorded examinations, are used as diachronic surrogates of orality to reconstruct early American English in colonial Massachusetts. Using a historical data source like the Salem Witchcraft Papers necessarily cannot provide an accurate representation of 17th century speech (due to, for example, scribal interference), but may allow the researcher to reconstruct orality by choosing texts that most closely represent the spoken language (Kytö & Walker, 2003).

The application of the variationist framework (Labov, 1972; Weinreich, Labov, & Herzog, 1968) to the variable systems in question provides the opportunity to assess the contribution of social and linguistic factors to the selection of variants, elucidating the nature of the underlying grammar through the use of distributional and multivariate analyses. The thesis critically assesses previous scholarship on relativisation and do-support, identifying a number of methodological infelicities, including non-adherence to the Principle of Accountability (Labov, 1972), inadequate data (too few tokens, drawn from fiction or prose), and consideration of a limited number of possible contexts that do-support could be used in (e.g. affirmative declaratives, or negatives).

Among the key findings to emerge from the investigation of relativisation are: the statistically significant effect of complexity constraints on the selection of variants, disfavouring the zero relative marker in more cognitively complex environments; the use of which as the only wh-relative to be employed to any appreciable extent; and the apparent favourable association of wh-relativisers with younger speakers. The use of relative marker who in the data is negligible, in contrast with findings based on contemporary mainstream varieties of North American English. For do-support, results suggest that the decline of affirmative declarative do at the end of the 17th century has been overstated in previous research (Ellegård, 1953; Kroch, 1989). Sentence type is the single most influential factor in the use of do-support, and interacts with every other factor group tested. Do-support is also subject to social conditioning in the 17th century, particularly age and gender.

This study has implications for the wider field of historical sociolinguistics by demonstrating that a comprehensive and empirically accountable analysis of relativisation and do-support can offer new insights into the social and linguistic conditioning of these variables in Earlier American English. Considered in the aggregate, the major findings of the thesis indicate that judicious exploitation of the Salem Witchcraft Papers confirms them as a unique resource for undertaking large-scale quantitative analysis of 17th century English and reconstructing antecedent varieties of English in their socio-historical context.