Towards a theory of complex predicates in Australian and Oceanic languages: an analysis of coverb constructions in Wagiman and serial verb constructions in Vurës

There is general agreement in the theoretical literature that the concept of ‘complex predicate’ refers to a multi-headed monoclausal structure, which means that at least two predicational constituents are merged to contribute to the overall predication of the clause. Complex predicates therefore challenge syntactic frameworks which define a clause as being headed by a single predicate. The term ‘complex predicate’ is also used in the typological literature to capture various syntactic constructions involving serial verbs, coverbs, preverbs, secondary predication, and noun incorporation. This thesis focuses on coverb constructions (CVC) and serial verb constructions (SVC) in two unrelated languages. The selected languages are Wagiman from northern Australia and Vurës from northern Vanuatu. Wagiman makes extensive use of coverb-verb combinations, while Vurës has productive verb-verb combinations. The morphosyntax of Wagiman is synthetic and typical of a non-configurational language. Vurës is a predominantly analytic language with a strict SVO word order. This thesis evaluates the ability and constraints in forming complex predicates in both languages and calls into question whether there is a motivation to establish distinct categories like complex predicate, coverb construction, and serial verb construction, or whether the underlying conceptual structure is fundamentally identical. I demonstrate that ‘complex predicate’ is not simply an umbrella term to cover various syntactic structures, but that it is a theoretical concept on the syntax-semantics interface. Drawing on previous merger analyses for complex predicates in combination with a VP-shell analysis from the Minimalist Program, it will be shown how the predicational constituents of a complex predicate merge and how this can be represented syntactically. One of these constituents is located in ‘little v’, the other sits in ‘big V’. The v position is filled by the (light) verb in CVCs or by the main verb (V1) in SVCs, whereas the V position in CVCs is filled by the coverb, but in SVCs by a verb in the V2+ position. Cross-linguistically, other potentially predicational parts-of-speech can also occupy the V position, licensed through the Syntactic Position Hypothesis proposed in this thesis. The constituents in v and V undergo a merger, so that their combined argument structure and event structure correspond to those of a single predicate. This confirms previous analyses which state that a complex predicate is conceptualised as a single event. This thesis contributes to a better understanding of the theory of complex predicates by asking the question “What is and what is not a complex predicate?” I show that some structures which have been described as complex predicates are better analysed as adverbial constructions because they do not consist of two predicational constituents. A constituent in the adverbial position acts as a modifier of the predicate. This analysis is reminiscent of Svenonius’ (2016) categorisation of complex predicates into complementation and adjunction, although I arrive at a different conclusion. I argue that only v + V combinations are genuine complex predicates, whereas an adverbial construction is an adjunction, and therefore not a complex predicate. To test various hypotheses of complex predicates, I collected material from previously published works and carried out fieldwork in Australia and Vanuatu. The relevant data have been annotated in ELAN and FLEx and are available in PARADISEC and AIATSIS.