Cross-disciplinary issues in compounding (review)

Cross-disciplinary issues in compounding (CIC) brings together nineteen papers by thirty-three different authors on 'compounds and issues related to compounds' (vii) from a 2008 conference at Bologna, Italy, the home of both the Componet compound project and the European Network for Linguistic Morphology project. The editors aim 'to yield a well-rounded book, not just a series of conference papers' (vii) in a research field that is 'rich and complex' (12) and 'currently at the center of attention in all areas of linguistics—both theoretical and applied' (1). As a core issue they identify 'the question of where and how compounds fit into the grammar, and ... the ways in which compounds challenge our view of the organization and architecture of the language capacity' (vii).

The editors begin with 'Why compounding?'(1-18), which is not a question addressing the issue of why languages have the compounding process; rather, this introduction offers an overview of central issues concerning compounds that have been discussed in the literature, especially in recent years. Compounds are words exhibiting 'internal syntax' that is intriguingly 'invisible'. In the English example truck driver, the two members exhibit a relation similar to that occurring in the phrase driver of a taxi, but such a relation remains unexpressed in the compound. Compounds have rich interpretive freedom, as is exemplified by the possible compound tree man. The authors in CIC appear to agree on a ('prototype') definition of compound: [X r Y]_Z, in which X, Y, and Z are lexical categories, and r is the (hidden) grammatical relation between X and Y. In endocentric compounding Z equals eitherXorY(X: Italian cassaforte '(box + strong =) safe',Y: English deep sea); in exocentric compounding Z equals neither X nor Y (Italian sali scendi 'elevator', English to wrongfoot). Endocentricity implies the presence of a head, while exocentricity the absence of one, where headhood implies supplying relevant features to the full word. If compounds are words (as the contributors of this volume largely appear to agree on) and if there is evidence for a separate word-formation component in the grammar, then compound formation will take place in that component, that is, it is MORPHOLOGICAL (as opposed to syntactic).

The volume is divided into four sections: 'Delimiting the field', 'At the core of compounding', 'Typology and types of compounds', and 'Quantitative and psycholinguistic aspects of compounding'.

Arguably the most sweeping and challenging contribution is PETER ACKEMA and AD NEELEMAN's 'The role of syntax and morphology in compounding' (21-36) in the first section. They propose that syntax and morphology, when both viewed as offering operations that combine categories, are in direct competition: all other things being equal, syntax wins, acting as a blocking device. Using the notorious case of 'synthetic compounding', they argue that morphological [truck][drive(-r)] is blocked by syntactic [drive (a) truck] and [drive-r (of a) truck]: they have different word orders but an identical head, which in English is right in morphology but left in syntax. However, [[[truckdrive] -er]] is not blocked because [[driver] (of a) truck] is structurally different. Based on their investigation of Saramaccan (Suriname), they show that the proposed analysis correctly predicts that if a language lacks N + V compounding, it will also lack synthetic compounding.

In 'Constraints on compounds and incorporation' (37-56), MARIANNE MITHUN investigates noun incorporation in Kapampangan (the Philippines), Mohawk (North America), and Yup'ik (Alaska). This paper is appealing for its non-Western material in a crosslinguistic approach, showing that incorporation is not crosslinguistically homogeneous but exhibits a continuum. In Kapampangan it is 'more syntactic', whereas in Mohawk and Yup'ik it comes from word formation: that is, incorporation behaves as compounding in the former but as derivation in the latter. The status of Kapampangan remains least clear because 'more syntactic' incorporation is not explicitly defined. Their claim that the continuum represents historical stages that single languages can go through (rather than...