We de ne a new model of automata for the description of bidirectional parsing strategies for tree adjoining grammars and a tabulation mechanism that allow them to be executed in polynomial time. This new model of automata provides a modular way of describing bidirectional parsing strategies for TAG, separating the description of a strategy from its execution.

Introduction Several parsing algorithms have been proposed for Tree Adjoining Grammars (TAGs), most of them derived from context-free tabular parsers [1]. However, only a few bidirectional parsing algorithms have been proposed [3, 5] although this kind of strategy seems to be naturally adapted for TAG. In this paper, we present a new bidirectional bottom-up parser for TAG derived from the context-free parser dened by de Vreught and Honig [4], which has several interesting characteristics: (i) the bidirectional strategy allow us to implement the recognition of the adjoining operation in a simple way, (ii) the bottom-up behavior allow us to take advantage of lexicalization, reducing the number of trees under consideration during the parsing process, and (iii) in the case of ungrammatical input sentences, the parser is able to recover most of partial parsings according to lexical entries. 2 The parser A Tree Adjoining Grammar is a 5-tuple G =<F11

We define a new formalism, based on Sikkel’s parsing schemata for constituency parsers, that can be used to describe, analyze and compare dependency parsing algorithms. This abstraction allows us to establish clear relations between several existing projective dependency parsers and prove their correctness.