Search

The present paper aims at further investigating the relative degree of flexibility of constituent order in Greek, which has been classified as a free word order language (Tzartzanos 1946, 1963; Greenberg 1963; Philippaki-Warburton 1982, 1985, 1987; Lascaratou 1984, 1989, 1994, among others). In this study we present additional evidence in support of the view expressed in Lascaratou (1989, 1994, 1999) that, though very flexible, Greek word order is not completely free, but rather it is the result of tension between competing forces determining linear arrangements. Focusing on the analysis of various VP structures drawn from the Hellenic National Corpus (HNC)", we observe that no single (syntactic and/or lexical/semantic) factor appears to override the others in a salient manner, the relative strength of each individual factor not always being clearly and reliably measurable. In particular, it seems that constituent length does not constitute a more important factor than the syntactic and/or lexical/semantic relations holding between elements in the Greek VP. What is more, with respect to the concept of "adjacency" put forward by Hawkins (2001, 2004) to interpret the preference for certain linearization patterns vs. others, we propose that constituent length essentially operates in a substantially different way and direction than syntactic and/or lexical/semantic dependency relations. More specifically, while-for parsing efficiency-constituent length may often be responsible for the postposition of heavy constituents thus resulting in the disruption of adjacency by separating elements which "belong together", syntactic and/or lexical/semantic relations, on the other hand, intrinsically motivate the adjacency of elements which "belong together", thus resisting any rearrangement that would bring them apart. [ABSTRACT FROM AUTHOR]

Pustejovsky’s well-known analysis of so-called logical metonymy, such as she began the book (= ‘reading’ or ‘writing’ the book), fails in several respects (cf. e.g. Godard & Jayez 1993, Lascarides & Copestake 1998, Verspoor 1997). The main problem is that his account is based on just a few prototypical examples of logical metonymy. When more examples with other verbs and other objects are analysed, his theory runs into problems. Since verb-concrete object combinations diverge widely in their interpretations, one can even ask to which extent metonymy is actually involved. With the help of FrameNet, this article analyses the semantics of these verbs and their combinations with metonymical objects. On this basis, a new analysis of logical metonymy is proposed. [ABSTRACT FROM AUTHOR]