UC San Diego

This thesis investigates the nature of grammatical patterns through an in-depth study of resultative constructions in Mandarin and Thai. At the heart of the thesis lies the proposal that event structure templates - complex, meaning-based grammatical patterns - must be recognised as primary objects of linguistic analysis. As content-theoretic objects that are independent of form, event structure templates abstract away from language- specific surface encoding of cross-linguistic resultative constructions while providing the toolkit for a language- independent characterisation of the generalisations that exist. As a result, the pattern-theoretic approach developed here provides a happy compromise between mainstream generative approaches, which are typically criticised for focusing on broad cross-linguistic generalisations at the expense of complete descriptions of particular languages, and Construction Grammar approaches, which have been criticised for the complementary problem, i.e. focusing on the complete descriptions of particular languages at the expense of cross-linguistic generalisations. The thesis develops a balance between these proposals by exploring three independent, but interdependent, dimensions of resultative constructions that are crucial for a pattern-theoretic analysis of these constructions: predication relations, argument realisation, and surface encoding. It provides a survey of the myriad patterns of predication relations instantiated by resultative constructions, and carefully and critically evaluates competing theoretical proposals in this well- examined domain. Through a close and fine-grained analysis of Mandarin and Thai resultatives, it argues for a pattern -theoretic approach in which patterns of predication relations are represented in terms of families of event structure templates. It also investigates the regularities governing the alignment between semantic arguments of event structures and grammatical relations in the syntax, and proposes to capture these generalisations by embedding event structures within a correspondence-based theory of argument realisation. Finally, it considers the cross- linguistic variation in the surface encoding of resultative constructions and provides a formal model of how the same event structure template can be realised by different surface encoding strategies in different languages. Treating event structures as complex grammatical patterns provides an insightful way of characterising cross-linguistically recurrent, yet variant, construction types, such as resultative, and identifies a typological research programme with the potential to uncover implicational relations that hold within languages