Non-standard morphosyntactic variation in L2 English varieties world-wide: a corpus-based study

Previous research has identified and studied a wide range of non-standard syntactic constructions in global varieties of English. However, there is currently a lack of large-scale corpus-based evidence on non-standard syntactic variation in English from a global perspective. The present study seeks to fill this gap through an analysis of a set of syntactic constructions across the fourteen L2 and six L1 varieties of English included in the Corpus of Global Web-based English, totalling 1.9 billion words. We conceptualise non-standard features as lying outside the “core” of the language that represents the object of description in English grammars, including those that might be regarded as “colloquial” or “vernacular”, such as there-existentials with singular agreement. The methodological approach is onomasiological, modelled on the alternation between non-standard features and corresponding standard forms. Across 28 morphosyntactic features drawn from eWAVE 3.0, a total of 289,827 non-standard and more than 37 million standard occurrences were analysed in a logistic mixed effects regression model. Register variation was accounted for by means of a co-variate for involved discourse following Biber's Multidimensional Analysis. Results indicate a relatively low degree of non-standardness across both L1 and L2 varieties. The register dimension of involved discourse is the most important variable governing non-standard variation, followed by differences between world regions. These results were further confirmed by a hierarchical clustering model and a multidimensional scaling analysis.