Multiple Adjunction in Feature-based Tree-Adjoining Grammar

Abstract

While independent derivations have been shown by Schabes and Shieber (1994) to be essential for correctly supporting syntactic analysis, semantic interpretation and statistical language modelling, the parsing algorithm they propose is restricted to TAG and is therefore not directly applicable to large scale implemented Feature-Based Lexicalised TAGs (FB-LTAG). We provide a recognition algorithm for FB-LTAG which supports both dependent and independent derivations under certain restrictions enforced jointly by feature constraints and by side conditions on the inference rules of the parsing algorithm.  The resulting algorithm combines the benefits of independent derivations with those of Feature-Based grammars. In particular, we show that it accounts for a range of interactions between dependent vs. independent derivation on the one hand, and syntactic constraints, linear ordering, and scopal vs. non scopal semantic dependencies on the other hand.