Retour vers liste

Détail de la contribution

Auteur: Pius TEN HACKEN

Compounding in the Parallel Architecture

Abstract/Résumé: The Parallel Architecture (Jackendoff, Foundations of Language) is based on two fundamental tenets: (1) Phonology, syntax, and semantics are independent generative systems, and their structures are linked by means of interface rules. Syntax is not the prime mover in linguistic structure; in some circumstances phonological and semantic structures can be linked without syntactic intervention. (2) A basic question in describing a language is what is stored and what can be built online by free combination. This leads to a view of the lexicon in which words, idioms, meaningful constructions, and general rules of phrase structure form a continuum of stored structures, with no sharp distinction between words and rules. Compounds are an interesting test case for this architecture: on one hand there are large numbers of conventionalized compounds, many with idiosyncratic semantics, and on the other hand it is possible to coin new compounds on the spot. The Parallel Architecture offers a natural account of these phenomena. The lexicon contains schemas for compounds that function in two ways. First, the conventionalized cases fall into an inheritance hierarchy, inheriting their general properties from the schemas but “paying” for their idiosyncratic properties. Second, the schemas can also function as general rules which license novel compounds. They offer a range of possible interpretations for novel compounds, including interpretations that bring into play the internal semantic structure of the constituents. Such schemas are found elsewhere in the lexicon, so compounds require no new machinery. In addition, English compounds are largely unselective about the syntactic categories of their constituents. This suggests that the rules for compounding may be treated as interface rules that link phonological words with semantics, largely bypassing syntax. In fact, the interpretation of multi-word compounds can be driven largely by prosody and semantics and need not appeal to syntactic constituency. This places compounding (at least in English) in a domain of grammar that is arguably less complex than classical syntactic structure. The conclusion is that the grammar of English involves a mixture of principles of differing degrees of syntactic complexity.