Published online by Cambridge University Press: 18 July 2016
This article investigates the occurrence and distribution of referential null subjects in Middle English. Whereas Modern English is the textbook example of a non-null-subject language, the case has recently been made that Old English permits null subjects to a limited extent, which raises the question of what happens in the middle period. In this article we investigate Middle English using data drawn from the Penn–Helsinki Parsed Corpus of Middle English Prose and the new Parsed Corpus of Middle English Poetry, aiming to shed light on the linguistic and extralinguistic factors conditioning the alternation between null and overt subjects. Generalized mixed-effects logistic regression and random forests are used to assess the importance of the variables included. We show that the set of factors at play is similar to that found for Old English, and we document a near-complete disappearance of the null subject option by the end of the Middle English period.