Aligning linguistically motivated phrases

In this paper, we describe the architecture of a sub-sentential alignment system that links linguistically motivated phrases in parallel texts. We conceive our sub-sentential aligner as a cascade model consisting of two phases. In the first phase, anchor chunks are linked on the basis of lexical correspondences and syntactic similarity. In the second phase, we will focus on the more complex translational correspondences based on observed translation shift patterns. The anchor chunks of the first phase will be used to limit the search space in the second phase. We present the first results of our sub-sentential alignment system, which links linguistically motivated chunks. In our baseline system, the obtained recall scores range from 44% to 59% and precision scores from 90% to 98% depending on text type. We experimented with two different types of bilingual dictionaries to generate the lexical correspondences: a handcrafted bilingual dictionary and probabilistic bilingual dictionaries. We demonstrate that although the handcrafted dictionary is twice the size of the probabilistic dictionary, the obtained recall scores are lower.