Chains in Syntax and Morphology

Chains in Syntax and Morphology Thomas Groß Faculty of International Communication, Aichi University Machihata-cho 1-1, Toyohashi-shi, 441-8522 Aichi-ken, Japan tmgross@vega.aichi-u.ac.jp Abstract. This paper argues for the existence of a deeper and more primitive structural unit of syntax and morphology than the constituent. Data from ellipsis, idiom formation, predicate complexes, bracketing paradoxes, and multiple auxiliary constructions challenge constituency-based analyses. In chain-based dependency grammar, however, constituents are seen as complete components. Components are units that are continuous both in the linear and in the dominance dimension. A unit continuous in the dominance dimension is called a chain. Evidence suggests that chains constitute the fundamental structural relationship between syntactic and morphological units, and that constituents are just a special subset of chains. If these assumptions are correct, linguistic research may need to change direction. Keywords: bracketing paradox, chain, constituent, ellipsis, idiom 1 Introduction In current linguistic research, the vast majority of work assumes that constituency is the ordering principle of syntax and morphology. Items that belong together form constituents. If items are understood as belonging together, but do not appear together, one assumes displacement of (at least) one item. This is then explained by movement (internal merge) in syntax, or by lowering in morphology. Evidence from three areas in syntax, however, poses significant problems for a constituency- based analysis. These areas are ellipsis, idiom formation, and predicate verb complexes Examples of these phenomena are given in section 2.1. In morphology, bracketing paradoxes in particular have shown considerable resistance to constituency-based analyses. Further, multiple auxiliary constructions challenge the constituency position. These areas are taken up in section 2.2. This paper argues that languages need not necessarily build constituents, but rather chains.1 The concept of the chain is introduced in section 3. In section 3.1, chains are applied to the syntactic phenomena introduced in section 2.1. Section 3.2 shows that chains equally apply in morphology. The bracketing paradox examples need to form proper chains, not proper constituents. Multiple auxiliary constructions also form chains, not constituents. The paper concludes that, based on the evidence presented, the constituent may not be the central structural unit of syntax and morphology, but rather the chain is. If the aforesaid holds, research in syntax and morphology, but also in language acquisition and computer-based language modeling, should start to at least ponder the possible existence of a more central unit, and what this could mean for theory formation. 1 To date, the system employed here has been referred to as “chain-based dependency grammar”. Due to a number of criticisms, the contributors to this system have decided to change the terminology. “Chains” are now referred to as catenae (the Latin word for „chains‟) in Osborne & Putnam (forthcoming). In this paper however, I will continue to use the older terminology. Any instance of “chain” in this paper may be substituted against “catena” in order to yield the new terminology. PACLIC 24 Proceedings 143 2 Difficulties for Constituency-based Approaches This section details some data that present evidence challenging the assumption that languages build …

Content truncated to 500 words