Grammatica e Quantificazione in LFG

This work presents a unified treatment of extraction and scope phenomena based on LFG. It is shown that only this theoretical framework can attain this aim, basically owing to the bipartition of non argumental operator-like functions, into TOPic and FOCus. LFG also allows to reach a powerful generalization: i.e. the position purported by GB according to which in a language like Italian S' and not S must be assumed as bounding node (or as cyclic node) is untenable. S is the only node which can be claimed to be the bounding node, together with NP, for explaining island phenomena at a universal level. The parameterized version introduced by Rizzi(1982) on facts of Italian is shown to be a by-product of theory-internal motivations. Two values of the same feature are used [+BOUND] for syntactic island, and [-BOUND] for semantic islands, and are enough to explain island phenomena, together with the usual LFG apparatus of rules and principles. Of the two non argumental functions TOPic and FOCus, only constituents assigned FOCus function can create [+BOUND] domains in Italian. This automatically carries on a number of welcome consequences, such as, the uselessness of abstract operators in Comp position for topicalized structures, which in LFG version is not available as a syntactic node at c-structure level, hence not required for interpretation. Also the uselessness of Subjacency as a theory based on or founding movement versions of extraction phenomena. The interaction between syntax and semantics is carried out through the use of semantic features which capture in a simple and intuitive way aspects of the interpretation process: these features enter the representation from the lexicon and/or from the Phrase Structure Rules and are propagated according to LFG internal percolation mechanisms. We also discuss the relevance for LFG of Logical Form, a level of representation where syntactic information is made transparent for quantifier scope and for semantic interpretation. The proposal we make is strictly tied up to the syntactic level for computing syntactic control for quantifiers and quantified NPs, and is in this sense alternative to Halvorsen's semantic approach to the problem of quantifier scope.