Treebanks in Machine Translation

We present an approach using treebanks in machine translati on. Our experiment in Czech-English machine translation is an attempt to develop a full machine translation system based on dependency trees (Dependency Based Mac hine Translation, DBMT). We use the following resources: Prague Dependency Tr eebank, a newly created Czech-English parallel corpus of Penn Treebank, En glish monolingual corpus, and translation lexicons. The fully automatic process includes analysis of the Czech input into tectogrammatical (semantic) representat io , lexical and structural transfer, a simple rule-based system for generation to Engl ish surface realization, and an -gram language model for scoring and choosing from translat ion hypotheses. The results are evaluated quantitatively with BLEU score.