The contributions to this volume, all based on careful examination of data from a variety of mostly lesser-known languages, take up the continuing challenge of mapping semantic structure to morphosyntactic expression in complex predicates, a challenge that persists despite the considerable attention that complex predicates have received in the last two decades. The particular interest of this volume lies in the attention it devotes to several types of complex predicate (CP) which are less widely recognised, including converb constructions (consisting of a subordinate verb form and a main/light verb), coverb or ideophone constructions (consisting of a distinct, non-verbal predicative part of speech and a main/light verb), and some constructions which are restricted in their lexical inventory and/or semantics (for example, to activity plus associated motion).
The ‘Introduction’, co-authored by the three editors, Mengistu Amberber, Brett Baker & Mark Harvey, sets the scene by providing a useful overview of some important contributions to the field in recent years and some of the issues for debate, followed by a chapter-by-chapter overview.
Chapter 2, ‘Complex predicate formation’, by Brett Baker & Mark Harvey, is in many ways the most ambitious chapter in the volume. It argues for a distinction between ‘merger’ constructions and ‘coindexing’ constructions as distinct types of CP (in monoclausal structures). Merger constructions are semantically equivalent to simple verbs, while coindexation constructions can include additional semantic components (never lexicalised within monomorphemic predicates) as well as introduce additional arguments. The need for a distinction between merger and coindexing structures is convincingly argued, though the details of the syntactic vs. semantic criteria employed for establishing the types are certainly open to investigation. One also cannot help wondering in how far the distinction proposed by Baker & Harvey differs from the influential distinction between nuclear and core serialisation made in Role and Reference Grammar and first introduced by Foley & Van Valin (1984). A footnote (35) suggests that nuclear and core junctures should be regarded as subtypes of coindexation, probably because most of the chapter equates coindexation with serial verb constructions. However, a full discussion of this issue is lacking. Neither do the authors address a similar distinction, made by Butt (Reference Butt, Alsina, Bresnan and Sells1997), between event and argument fusion, which is explicitly equated with ‘merger’ and ‘coindexation’, respectively, in Rachel Nordlinger's chapter (247). Nevertheless, the chapter by Baker & Harvey makes an important contribution by considering coverb constructions, prominent in Northern Australian languages, in terms of the distinction between subtypes of complex predicates.
Most importantly, the authors propose a new, empirically testable hypothesis about constraints on the semantics of simple verbs and complex predicates of the ‘merged’ type – testable, that is, if one regards the predicates in their semantic representations based on Jackendoff's (Reference Jackendoff1990, Reference Jackendoff2002) Lexical Conceptual Structure (or any other decompositional representations) as well defined. In particular, the general conceptual function MOVE, which according to Baker & Harvey's proposal is part of the event structure of motion predicates (‘run’) as well as of agentive (‘do’) and non-agentive (‘tremble’) activity predicates, may well turn out to be too general to successfully constrain ‘merged’ CPs cross-linguistically.
The next two chapters also take a cross-linguistic perspective on complex predicates, while the remaining contributions focus on individual languages. Chapter 3, by Miriam Butt, ‘The light verb jungle: Still hacking away’, discusses the status of light verbs (LVs) in complex predicates. This contribution partly summarises arguments made in earlier publications regarding the monoclausality of LV constructions (as opposed to, for example, control constructions) and the difference between LVs and auxiliaries, for which interesting evidence is provided from the historical stability of LV constructions in Indo-Aryan languages. The sections on LVs as a separate syntactic class (59–64) and on the semantics of LVs (71–74), which are the most interesting sections in my view, are unfortunately very brief and sketchy. I did not find the arguments for LVs as a syntactic class entirely convincing, especially regarding the data from Northern Australian languages; I am more inclined to side with Butt & Geuder's (Reference Butt, Geuder, Riemsdijk and Corver2001: 363) conclusion that ‘the existence of syntactic distinctions between lexical and semi-lexical (i.e., light) verbs can vary across languages’.
The contribution that poses perhaps the biggest challenge for any proposal for a straightforward typology of complex predicates is Chapter 4, ‘Events and serial verb constructions’, by William Foley. Focusing on languages of New Guinea, Foley presents a number of examples of the many-to-many relationship between semantic and morphosyntactic structures. For example, translation equivalents of ‘kill’ in Yimas, Numbami, Watam, and Mangap Mbula can be analysed as simplex verb, (lexicalised) serial verb construction (SVC), SVC with potentially complex (that is, multi-verb) causative ‘subevents’, and clause chaining construction (with an overt conjunction between causing and resulting event), respectively. All of these still constitute ‘unitary events’ in the sense that temporal operators have scope over all subevents (93), but, as Foley argues, singling out any subset of them as corresponding to ‘single events’ either on semantic or syntactic grounds proves difficult. Applying a number of morphosyntactic tests to serial verbs of various semantic types in a single Papuan language, Watam, Foley further shows that what looks like a single structure (viz. a SVC) on the surface can in fact exhibit stronger or weaker cohesion between subsets of its elements. He argues that patterns of lexicalisation (including CP formation) are possibly more variable for ‘verbal’ than ‘nominal’ concepts, reflecting the relative absence of a cognitively dominant bundle of perceived properties in the domain of events (82–84). Still, he concludes (107–108) that those SVCs displaying the strongest syntactic cohesion (in language-specific terms) will be those whose semantic composition obeys constraints that also hold for monomorphemic verbs and the most frequent types of CP cross-linguistically.
Neridah Jarkey, in ‘Cotemporal serial verb constructions in White Hmong’ (Chapter 5), explicitly addresses the question of whether or not coindexation structures (which she equates with core junctures) can be regarded as ‘single events’ on some level. She proposes a distinction between ‘simplex events’ corresponding to merged structures and ‘single conceptual events’ corresponding to coindexed structures, arguing, like Foley, that the latter still display semantic coherence reflected in the scope of temporal and modal operators over the entire unit. Jarkey then demonstrates that most SVC types in White Hmong, a Miao-Yao language, are core junctures which indeed display the properties of ‘single conceptual event’. Of all the contributions, this is the one that most succeeds in bringing to life the language under discussion, for example by the illustration of the wide-spread use of parallelisms (somewhat misleadingly labelled ‘reduplication’ of verbs) in multi-verb serial expressions. This is not accidental, since Jarkey's contribution endorses the argument that constraints on complex predicate formation are partly culture-specific.
Keren Rice's contribution, ‘Activity incorporates in some Athabaskan languages’ (Chapter 6), discusses a strongly constrained type of CP in Ahtna and Koyukan. An event-denoting nominal (often derived from a verb stem) is combined, in a single grammatical word, with a verb of motion or body position (for example, ‘return whistling’, 144; or ‘stay mourning’, 145). Rice accepts the characterisation of these items as coindexation structures (in the sense used by Baker & Harvey) on semantic grounds. Rice's generalisations are based on dictionaries, which, while yielding an impressive and interesting range of data, limits the probing into constraints on combinatorial possibilities. One robust generalisation that nevertheless emerges is that the nominal component is restricted not just to activities (150–152) but specifically to oral activities, thus providing an interesting point of comparison, regarding linguistic vs. cultural constraints on event composition, with similar expressions of associated motion or stance in this cross-linguistically frequent CP type.
With Chapter 7, ‘Warlpiri verbs of change and causation: The thematic core’, by Mary Laughren, the volume returns to the coverb construction in Northern Australian languages that was a prominent feature of Chapter 2 (the relevant category is here referred to as preverb (PV)). This chapter is twice as long as most of the others (64 pages excluding appendices) and written in a dense style, which occasionally makes the line of argumentation difficult to follow. The undeterred reader is rewarded with a rich account of the contribution of the individual lexical items in the formation of the innermost layer of complex predicates in Warlpiri (Laughren's ‘thematic core’), as well as in contrasting constructions, such as aspectual light verb constructions employing motion verbs, an ‘intended result’ construction, and aspectual particles. Laughren's chapter focuses on complex predicates of change and causation. It emphasises the need for semantic characterisations which incorporate both Aktionsart (lexical aspect) and argument structure, and proposes a hierarchical representation of their interaction. In particular, the chapter provides evidence for the validity of a distinction between internal and external causation, as proposed by Levin & Rappaport Hovav (Reference Levin and Hovav1995). It further supplies evidence that preverbs (coverbs) exhibit a syntactic behaviour distinct from nominals.
Rachel Nordlinger's contribution, ‘Complex predicates in Wambaya: Detaching predicate composition from syntactic structure’ (Chapter 8), is a nicely argued case study, demonstrating (as the title suggests) that two morphosyntactically distinct types of CP found in this Australian language – one a serial verb construction and the other an associated motion construction combining a main verb with one of just two deictic directional suffixes – can overlap in their semantics. Both constructions can have a concomitant motion (‘go while V-ing’, mostly restricted to motion verbs as V) and an anterior motion (‘go and then V’) interpretation. One somewhat doubtful aspect of the analysis is the interpretation of the second type as a ‘dual event’ structure, which highlights again the problematic nature of the term ‘single event’. While the two subevents of motion and associated event indeed are sequential under the anterior motion interpretation, there still is (presumably) a requirement of direct temporal contiguity, just like in the cause-result combinations discussed by Laughren and by Baker & Harvey (30).
Chapter 9, by Azeb Amha, is entitled ‘Compound verbs and ideophones in Wolaitta revisited’. It offers an overview of two distinct types of CP in an Omotic language of Ethiopia; both are analysed as ‘merger’ constructions. The first consists of a same-subject converb form of a verb from an open class and a main (light) verb from a closed class. This type can be distinguished by syntactic criteria from a sequence of clauses, where converbs are also employed as predicates of co-subordinate clauses. In the second type of CP, an ideophone (again from an open class) combines with just one of two light verbs, selected on the basis of the semantic transitivity of the ideophone. Amha convincingly argues that both types are true CPs, that is, the light verb is not an auxiliary. In contrast to many other contributions in this volume, one would have wished in this chapter for a more elaborate and explicit account of the semantics of (at least a subset) of the light verbs in the converb construction and the mechanisms for event and argument fusion.
Chapter 10, ‘The structure of the light verb construction in Amharic’, by Mengistu Amberber, deals with a construction reminiscent of the second, ideophonic, type of CP in Amha's chapter, employing one of just two highly generic light verbs and an uninflected lexical item from a distinct class (that is, a ‘coverb’). Interestingly, in Amharic and a few other languages in the same linguistic area in Ethiopia, coverbs can be derived productively from verbs by one of two morphological templates, which in addition contributes either an attenuative or an intensifying semantic component. Like Butt (who in her contribution refers to LVs as ‘verbal passepartouts’), Amberber argues that LVs are not necessarily semantically bleached, but rather semantically generic (‘underspecified’) in both their light verb and simple verb uses (314–315). His analysis of the construction in terms of Distributed Morphology is, however, designed to highlight the similarity of LVs to derivational markers such as the inchoative/reflexive and causative rather than to simple verbs.
A recurring theme throughout the entire volume is the variability of the mapping of semantic components onto elements in syntactic structure. As many of the contributions show, semantic cohesion between predicates is only partially dependent on their syntactic structure. The contributions by Foley and Nordlinger, and the section in Butt's paper (69–71) on the diachronic ‘trade-off’ between the use of preverbs and light verbs to express notions of lexical aspect in Indo-Aryan languages, are particularly nice case studies in this respect. Baker & Harvey's very specific proposal notwithstanding, one is left with the impression that any robust generalisations about the relationship between semantic composition and morphosyntactic type will require much further research. As indicated in several chapters – most explicitly in Foley's and Jarkey's contributions – the term ‘single event’ is not particularly helpful in this endeavour.
An unresolved issue concerns the semantic metalanguage that best captures mechanisms of, and restrictions on, complex predicate formation (and verbal semantics more generally). The contributions to the volume reflect the general disagreement in the field about whether (or to what extent) to employ semantic decomposition; and if employing semantic decomposition, which primitive predicates to assume. One cannot help noticing that the current preoccupation with modelling the mechanisms of syntagmatic combination by means of decompositional structures often leads to a proliferation of polysemous entries for individual lexical items (so, for example, in the case of the associated motion bound morphemes in Wambaya in Nordlinger's contribution; and in Baker & Harvey's representation of a single light verb in Jaminjung which depending on its contexts is assigned either MOVE or BECOME as its main component, 34). This distracts from a need to examine in more detail the role of pragmatics in event construal, and in particular, if there is a closed class of light verbs, the role played by the (language-specific) oppositions between members of this class.
While one of the merits of this volume is undoubtedly the emphasis on the coverb construction as a CP on a par with the better-known types, one of its greatest shortcomings is the considerable degree of inconsistency in the terminology applied in particular to the coverb construction and related constructions. This could have been reduced by a more heavy-handed approach by the editors. For example, in the introduction, a coverb is defined as a distinct predicative part of speech (1) – but given this definition, it is doubtful whether the term can be applied to Iranian languages, as is claimed in Baker & Harvey's chapter (15, fn. 1), since descriptions of Persian and other Iranian languages usually treat the lexical component of CPs as nominal. Confusingly, the introduction also equates converbs in Wolaitta with coverbs (8), although the former are quite obviously dependent verb forms and not a distinct part of speech. The reader will be further confused by the fact that Laughren not only uses the term ‘preverb’ for what is referred to as a coverb in other contributions (albeit for good reason, respecting traditional usage in Warlpiri linguistics), but also suggests the term ‘coverb’ for a more adverbial type of preverb in Warlpiri (186, fn. 38). Terminological differences might in fact obscure the similarity between the ideophone-light verb type of CP in Wolaitta (Chapter 9) and coverb constructions (in particular for Amharic, Chapter 10). The volume also does not offer a conclusive answer to the question of what counts as a complex predicate and what does not. Serial verb constructions are explicitly or implicitly treated as a subtype of complex predicates in most contributions, but are excluded from the definition in Butt's chapter (49). And finally, with the discussion of ideophones as parts of CPs in Wolaitta in mind, one is tempted to analyse the sound-symbolic expressives in White Hmong, designated as manner adverbials by Jarkey (124), as parts of complex predicates instead.
Even if it does not offer the last word on complex predicates, this volume succeeds in advancing both empirical and theoretical research. It is strongly recommended to anybody interested in the topic of event construal cross-linguistically.