Machine Translation with Many Manually Labeled Discourse Connectives

Abstract

The paper presents machine translation ex-periments from English to Czech with a large amount of manually annotated dis-course connectives. The gold-standard discourse relation annotation leads to bet-ter translation performance in ranges of 4–60 % for some ambiguous English con-nectives and helps to find correct syntacti-cal constructs in Czech for less ambiguous connectives. Automatic scoring confirms the stability of the newly built discourse-aware translation systems. Error analysis and human translation evaluation point to the cases where the annotation was most and where less helpful. 1