Home Language & Literature Historical Sociolinguistics: Language Change in Tudor and Stuart England
VARBRUL Analyses of Five Historical Changes
Our aim in this section is to explore whether regularities can be observed in the way in which external factors correlate with changes in progress. Region, gender and register constitute the independent variables or factor groups of the analysis. The five dependent variables are ye/you; -th/- s; the object of the gerund; mine, thine/my, thy; and the which/which. The VARBRUL application GoldVarb 2 developed by Rand and Sankoff (1990) was used to compute the relative weightings of region, gender and register in the different phases of the linguistic changes. In no one case were all their five temporal phases from incipient to completed available in sufficient quantities, but in all but one at least three of them were.
The results of the VARBRUL analysis will be tabulated for each change separately, and two aspects of the analysis will receive particular attention. What interests us, first, are the factors that particularly favour the change in progress at a given point in time. As the probabilities vary between 1 and 0, a factor with a probability weighting over 0.5 favours the incoming form or structure, and one with a weighting below 0.5 disfavours it. If a factor has a weighting of 1, it determines the selection of the form categorically, while a factor with a weighting of 0 would totally inhibit it. When a change is nearing completion, factors that favour the incoming form most are the first to become categorical ('knockout' factors).
Secondly, systematic attention will be paid to the range of variation of the three factor groups. Their ranges, 'variation/variety spaces' as they are called by Preston (1991), are calculated by subtracting the minimum from the maximum weighting a factor receives within each factor group. These ranges will be compared in order to establish the relative impact of the external factors on the change in progress. However, as not all these variables need to be monotonically aligned with the change in progress - crossovers are in fact expected on the basis of previous research - we cannot hypothesize that the ranges for the three external variables would always get progressively smaller as the process nears completion.9
|< Prev||CONTENTS||Next >|