Argument structure and the syntax-morphology interface. A case study in Latin and other languages

In this dissertation I put forth a syntactic theory of argument structure and the syntaxmorphology interface and I apply it mainly to Early and Classical Latin, although comparison with other languages and discussion of the facts in these languages is conceded a considerable weight. Drawing on Mateu 2002f., Borer 2005b and Marantz 1995f., I argue for a view of argument structure where a basic distinction is drawn between the elements carrying encyclopaedic content, the roots, and the syntactic configuration built around functional heads. Argument structure properties exclusively depend on the latter. Furthermore, the syntactic configuration provides the structural semantics of the linguistic expression. I endorse a theory of the syntaxmorphology interface like the one proposed in the Distributed Morphology framework: morphology is, by default, syntax, although some specific PF operations can disrupt the basic syntax-morphology isomorphism —an isomorphism which, I argue, is inherent to the syntax-semantics interface. Crucially, cross-linguistic variation shall be defended to depend exclusively on that lack of isomorphism between syntax and morphology. In particular, it is triggered ultimately by languagespecific morphophonological properties of functional heads. The empirical focus is set on the domain of events expressing a transition. I consider Talmy’s (2000) theory of the cross-linguistic expression of events of change, where a basic divide is drawn between the languages in which the transition can be encoded by a non-verbal element —satellite-framed languages— and the languages in which the transition must be encoded by the verb —verb-framed languages. I couch Talmy’s theory of transition events within a syntactic theory of argument structure, and I explore a wide range of constructions in Latin —either presenting new data or giving a new perspective on data from the Latin linguistics tradition— to show that Latin pertains to the class of satellite-framed languages. Following an idea in Mateu 2002:160 and Mateu & Rigau 2002, I propose that the s-/v-framed distinction is explainable in purely morphophonological terms. In particular, I make use of the theory of PF operations developed by Embick & Noyer (1999, 2001) within the Distributed Morphology framework, together with Hale & Keyser’s (2002:60f.) and Harley’s (2004) theory of conflation, to account for the distinction. Thus, in v-framed languages the eventive v head lowers, at PF, to the head encoding transition —Path— and fuses with it, yielding a unique locus for phonological realisation. On the other hand, in s-framed languages this Fusion operation does not take place, so v and Path are free to be phonologically realised independently from each other. Finally, I propose a refinement of Talmy’s typology within the class of s-framed languages. First, there are strong s-framed languages, like the Germanic languages, where v and Path are not required to form one word, and, thus, allow constructions like complex adjectival resultative constructions. Second, there are weak s-framed languages, like Latin, where v and Path must form one word (if both are realised independently from each other) and disallow, hence, constructions like adjectival resultative constructions. This distinction is accounted for in terms of a v-to-Path (PF) Lowering operation for weak s-framed languages, which creates a complex head. A three-way, gradual typology emerges encompassing strong s-framed languages (no v-to-Path Lowering and no Path-v Fusion), weak s-framed languages (v-to-Path Lowering, no Path-v Fusion) and v-framed languages (v-to-Path Lowering and Path-v Fusion).