Friday Seminar: A Distinction between Parse Trees and Constituent Trees in “Deep” Grammar Implementations

September 06, 2013 at 14:15 – 16:00 in HF:217

In this talk I will present a grammar design which makes a distinction between parse trees and constituent trees. While a parse tree simply is the incremental combination of words, one by one, a constituent tree is a regular representation of the constituent structure of a given utterance, inspired by Diderichsen’s Sentence Schema. In the proposed design, the constituent tree is derived from the parse tree.

The parse tree – constituent tree distinction opens for a procedure for building syntactic trees that is neither top-down nor bottom-up, but rather left-right. The aim of this talk is to show how the separation of parse tree from constituent tree (i) gives the grammar more flexibility when it comes to word order, and (ii) has the potential of increasing the psycholinguistic plausibility of “deep” grammar implementations.

Petter Haugereid (Postdoc, LLE,  University of Bergen)