Capturing particulars and universals in clause linkage

Cross-linguistic variation in adjoined clause linkage is higher than what is allowed by universal concepts like ‘coordination’ or ‘subordination’ which entail sets of strictly correlated properties. This chapters uses statistical techniques to uncover probabilistic correlations and clusters in a pilot database. For this, a set of variables is developed that ranges in coverage from the scope of illocutionary force operators to extraction constraints and that allows both detailed qualitative analysis of language-specific clause linkage structures and large-scale quantified measurement of the similarities of such structures within and between languages. The study tentatively suggests that there is a prototype of subordination which is closer to ‘and’-like than to ‘chaining’ constructions, and that there is a continuum between more vs. less tightly constrained types of converb and chaining constructions, but no general prototype of ‘cosubordination’.