Context-sensitive syntactic source-reordering by statistical transduction

How well can a phrase translation model perform
if we permute the source words to fit target
word order as perfectly as word alignment
might allow? And how well would it perform
if we limit the allowed permutations to ITGlike
tree-transduction operations on the source
parse tree? First we contribute oracle results
showing great potential for performance improvement
by source-reordering, ranging from
1.5 to 4 BLEU points depending on language
pair. Although less outspoken, the potential
of tree-based source-reordering is also significant.
Our second contribution is a source reordering
model that works with two kinds of
tree transductions: the one permutes the order
of sibling subtrees under a node, and the other
first deletes layers in the parse tree in order
to exploit sibling permutation at the remaining
levels.The statistical parameters of the model
we introduce concern individual tree transductions
conditioned on contextual features of
the tree resulting from all preceding transductions.
Experiments in translating from English
to Spanish/Dutch/Chinese show significant
improvements of respectively 0.6/1.2/2.0
BLEU points.