Clause Initial Null Subjects in Web-based Written Language

Null subjects in English(es) are a phenomenon that has recently received much attention in the specialized literature. However, most studies are based on small datasets and samples of varieties due to the difficulty of extracting null subjects from corpora. The present paper is a first step towards the automatization of the data retrieval process of null subjects and analyzes a much larger sample of cases and varieties than previous research, namely, Australian, Canadian, Jamaican, Singaporean, Nigerian, Indian, Bangladeshi and Pakistani Englishes. By focusing on referential and non-referential third person singular clause initial null and overt subjects, a variationist examination of the data is conducted by means of mixed-effects logistic regression analyses which shows that non-referential null subjects are a much more pervasive and stable phenomenon in World Englishes than their referential counterparts. In addition, a cline of varieties emerges with respect to referential null subjects: these null subjects are more frequent the more advanced varieties are in Schneider’s Dynamic Model.