Diachronica - Online First

On the basis of just under 5,050 examples of perfect constructions, this paper traces the development of the be:have perfect alternation in English between the 1620s and 1750s. For a core group of 18 verbs, the study investigates the role that language-internal and language-external predictor variables played in the choice of auxiliary. Multifactorial modelling reveals that language-internal factors such as modality, negation, clause-type and tense are among the most important predictors favouring the choice of have as auxiliary; there is also some indication of diachronic, lexical and idiosyncratic variation within Early Modern English. A close investigation of perfects that combine both auxiliaries strengthens the view that ambiguity-avoidance did not play a major role in the loss of the be-perfect. The results of the multifactorial model suggest greater independence of negation and counterfactuality as factors than previously claimed. The study thus contributes a novel perspective on the demise of the be-perfect, with paradigmatic variability taking centre stage.