Synchronous Linear Context-Free Rewriting Systems for Machine Translation
Miriam Kaeshammer · 2013
We propose synchronous linear context-free rewriting systems as an extension to synchronous context-free grammars in which synchronized non-terminals span k ≥ 1 continuous blocks on each side of the bitext. Such discontinuous constituents are required for inducing certain alignment configurations that occur relatively frequently in manually annotated parallel corpora and that cannot be generated with less expressive grammar formalisms. As part of our investigations concerning the minimal k that is required for inducing manual alignments, we present a hierarchical aligner in form of a deduction system. We find that by restricting k to 2 on both sides, 100 % of the data can be covered. 1