Abstract This study examines the theoretical status as well as the quantitative distribution of subjects with Definite Null Instantiation (DNI) readings in Standard British English, Hong Kong English, and Singapore English from the perspective of Usage-based Construction Grammar. Following an extensive review of previous theoretical treatments, I propose an alternative formalization of DNI as a both schematic and semantically rich construction. Based on spoken data from the International Corpus of English, random forests and logistic regression models were then fitted to complement the theoretical model of DNI with distributional data, taking into account both intra- and extra-linguistic predictors. The models reveal significant differences between L1- and L2-usage; the extent of these constructional reconfigurations is best explained by a combination of substrate influence and the socio-cognitive stability of the L2-varieties.
Vladimir Buskin (Thu,) studied this question.