Dutch causative constructions
This chapter is a multivariate corpus-based study of two near-synonymous periphrastic causatives with doen and laten in Dutch. Using multiple logistic regression and classification trees, the study explores the conceptual differences between the constructions. The results support the existing definition of doen as the direct causation auxiliary, and interpretation of laten as the indirect causative (e.g. Verhagen and Kemmer 1997). However, the analyses also reveal more specific patterns: the most distinctive semantic pattern of doen is affective causation, whereas the contexts with the highest probability of laten refer to inducive causation. These differences remain valid when we control for geographic and thematic variation, as well as for the individual Effected Predicates treated as random effects in a mixed model.