Constituency and Linearization in Functional Discourse Grammar

Functional Discourse Grammar stands out among functional linguistic frameworks in availing itself of both a highly formalized approach to pragmatic and semantic structure and a sophisticated set of morphosyntactic placement rules. What is still largely missing is an equally thorough account of the principles that inform the mapping of pragmatic and semantic units onto morphosyntactic constit-uents: it goes without saying that, before deciding which slot a constituent occupies in morphosyntactic structure, one must determine whether we are dealing with a constituent in the first place. This paper attempts to bridge this gap by establishing a restricted set of principles that govern the interaction between morphosyntactic constituency and linearization. My proposal takes its cue from a refinement of the Domain Integrity principle (the preference for functionally related units to form a single morphosyntactic constituent) and suggests that this be supplemented with two other principles, which I refer to as Maximal Flatness and Positional Economy. These principles are then ranked in different orders for words that do and do not form a single semantic and/or pragmatic unit, after which their interaction is illustrated with an application to a handful of languages with different basic word orders (Dutch, English, Italian, Tahitian and Tamil).