Acquiring Syntactic Translation Rules from a Parallel Treebank

Abstract

The presented study addresses the issue of machine translation system development,as it makes use of a parallel syntactic phrase extraction algorithm in order to obtaina rich and robust set of syntactic translation patterns. To make this approach feasible, only phrase-to-phrase alignments from a bilingual treebank with syntactic constituents are considered. The knowledge enclosed in this study is defined by means of syntactic motivated translation sequences, written in terms of the morpho-syntactic specifications developed during the MULTEXTEast Project (Erjavec, 2010).