Complex predicates can be defined as predicates which are composed of more than one grammatical element (either morphemes or words), each of which contributes a non-trivial part of the information of the complex predicate. The papers collected in this volume, which were presented at a workshop at Stanford in 1993, represent a variety of approaches to the question of the range and nature of complex predicates, and draw on data from a wide spectrum of languages. This collection develops a better understanding of the range of phenomena that a general theory of complex predicates would have to account for, and to see what kinds of linguistic ideas and methodologies would be necessary for such a task.
At a very basic level, the papers collected here attempt to define the term 'complex predicates'. One point of view can be found in the paper by Hale and Keyser, who argue that many surface monomorphemic verbs in the lexicon are internally complex (e.g. 'clean (the house)' 'make clean'), and thus that complex predicates are the norm, rather than defining some special area. A related view can be found in Baker's work, for his assumptions are roughly that each morpheme with thematic structure projects its own syntactic domain, and he does not countenance any kind of argument "fusion" of the kind found in many of the papers here. Similarly, Rosen's paper treats complex predicates in Italian as complementation structures.
Other researchers, such as Alsina and Butt, take it that complex predicates are formed by syntactically independent elements whose argument structures are brought together by some kind of argument fusion mechanism that differs from the usual types of complementation (it is more a kind of co-complementation). The resulting composite argument structure then shows some of the behavior of a simplex predicate, and can be contrasted with a biclausal complementation structure. This leads to the question of what a word is, and how wordhood should be defined. Several papers address the issue of the boundary between lexicon and syntax, and whether such a distinction can usefully be drawn at all, and how strongly it can be drawn (for Ackerman and LeSourd, the syntactic integrity of items internal to a lexical word is only a markedness principle, for example). A rather different perspective, familiar to prosodic phonologists, is adopted by some of those working within the conceptual framework of LFG. The idea is that the concept of word is simultaneously defined at multiple parallel levels or dimensions of representation--morphemic, prosodic, structural, and functional--which may mismatch or overlap (Matsumoto 1992, Mohanan 1995, Bresnan and Mchombo 1995). For example, a word may be a functional unit at one level, but syntactically complex at another.
Mohanan's is the most explicit multi-dimensional analysis, arguing for generalizations at different linguistic levels. Such an approach places a heavy burden on the argument structure and/or semantic structure, and several of the papers here emphasize the semantic nature of (certain types of) complex predicate formation (see the contributions by Durie, Evans, Foley and Kiparsky). This connects back to the question of whether complex predicates require some kind of argument composition in their formation, where independent thematic structure are 'merged' in some way, or whether they can all be understood as complex subordinating structures with surface incorporation, an approach developed in the body of work building on Baker (1988).
Finally, there is the issue of what theoretical mechanisms are necessary in an account of complex predicates. Baker, and Hale and Keyser, adopt the view that complex predicates are just types of complementation structures, and that no mechanisms beyond those needed elsewhere in the syntax are necessary. However, this necessarily entails a quite abstract approach to syntax and morphology. The other papers in this volume do assume that there is some process of fusion or sharing that allows complex predicates to exist. Typically, these involve mechanisms for sharing in terms of argument structure or semantic structure, and/or the existence of 'incomplete' predicates which must necessarily be involved in complex predicates. In her contribution, Goldberg argues that complex predicates must inherit properties from other constructions in the grammar, and that their properties cannot be located solely in properties of (sets of) lexical items. This possibility follows naturally from the Construction Grammar framework she adopts, but would seemingly require extra mechanisms for other approaches.