A Grammar of South Efate

9. Verb combinations

9. Verb combinations Verb combining in Oceanic languages is the subject of an extensive literature, and is usually analyzed as verb serialization within a theoretical framework based on nuclear, core, and clause-layer predication. As we will see in this chapter, South Efate has relics of serial verb constructions with traces of what may once have been serial verbs but are now auxiliary verbs, adverbs, or directional particles. Work on serial verbs of languages of Vanuatu includes that on Paamese (Crowley 1987), Lewo (Early 1993; 1994), Namakir (Sperlich 1991; 1993), Tamambo (Jauncey 1997), Lolovoli (Hyslop 2001), Araki (Franc;ois 2002), and Mwotlap (Franc;ois 2004). Crowley's (2002) general survey of serial verbs in Oceanic languages also focuses on several languages of Vanuatu. These studies all reflect the need to deal with multi-predicate structures but there is no single approach that satisfies each of these authors. Essentially, all of these approaches reveal a distinction between contiguous verbs which, for South Efate, are called compound verbs, and less tightly bound structures (illustrated in [1]) which are usually called core-layer serial verbs by the authors listed above. Franc;ois (2002; 2004) is the exception as he treats contiguous verb stems as serialized, but then regards any combination of verbs that each bear subject marking as being a clause chain rather than a core-layer verb series. This is an attractive approach as it dispenses with the need to distinguish clause chains from the sequence of structures as schematized in (1), in which any number of sequential inflected verbs can occur with subjects and, optionally, objects. 1 S=V(-O) (S=V(-O))n In South Efate I regard any structure ofthe type outlined in (1) as representing two or more clauses. I analyze a predicate (made up of a simple or compound verb) as being the head of a verb complex which makes up a clause. In general, the presence of the subject argument in the form of a subject proclitic is diagnostic of a clause boundary. However, its absence does not necessarily preclude the presence of a clause boundary, as, in the absence of a subject argument (by what I call subject omission), there is a further diagnostic for clause boundaries in the pre-verbal complex (PVC) (see §9.1.2.1.) which initiates a new clause. Sequential clauses with subject omission (which I call clause-chains) have a structure like (2) because we can identify a new clause starting when a new preverbal complex begins. 2 S=(PVC) V 0=(PVC) V The reason that other studies have included core-layer serial verbs in their analysis rather than regarding them as sequences of clauses is that these structures are regarded as having commonalities that would be missed by treating them as 221 Chapter 9 separate clauses. Some of these commonalities, summarized from Jauncey (1997:368) and Aikhenvald (1999:470), are reproduced in (3). 3 Features of verb series include that they: a) encode sub-parts of a single overall event. b) share at least one core argument. c) share the same TMA and polarity value. d) have the intonational properties of a mono-verbal clause. e) can be distinguished from complex predicates and other Verb+Verb sequences, neither verb of which can be a predicate on its own (e.g., compound verbs). None of these criteria have provided a diagnostic tool for the structures under discussion in South Efate, as Aikhenvald (1999:470) points out: "no one of these characteristics is defining per se, since exceptions can be found to each of them." It is unclear how we can define an 'event' (a above) that is described differently by mono-clausal or multi-clausal elements. Crowley (2002:263) discusses the problem of equating events with individual predicates and Foley (2003), in a careful analysis of the notion of 'event' in a cross-linguistic survey of SVCs, suggests that SVCs (however defined) cannot be considered to express a single event. His example of the verb 'kill' in four languages shows that it is encoded as a lexicalized root, a serial verb construction, and as coordinated clauses, and he concludes that these formal differences cannot be equated to the number of events encoded. In none of these languages, he argues, is the semantic structure of 'kill' a single event. Distinct clauses can also share core arguments and TMA and polarity values (features band c above), so this criterion does not distinguish sub-clausal verb linkage unless the arguments are obligatorily...