Disambiguating Temporal-Contrastive Discourse Connectives for Machine Translation

Thomas Meyer
2011
Conference paper

Abstract

Temporal–contrastive discourse connectives (although, while, since, etc.) signal various types of relations between clauses such as temporal, contrast, concession and cause. They are often ambiguous and therefore difficult to translate from one language to another. We discuss several new and translation-oriented experiments for the disambiguation of a specific subset of discourse connectives in order to correct some of the translation errors made by current statistical machine translation systems.

Official source

https://infoscience.epfl.ch/record/165908?ln=en

About this result

This page is automatically generated and may contain information that is not correct, complete, up-to-date, or relevant to your search query. The same applies to every other page on this website. Please make sure to verify the information with EPFL's official sources.

Related concepts

Related publications

The paper presents machine translation experiments from English to Czech with a large amount of manually annotated discourse connectives. The gold-standard discourse relation annotation leads to better translation performance in ranges of 4–60% for some ambiguous English connectives and helps to find correct syntactical 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.

2013

Implicitation of Discourse Connectives in (Machine) Translation

Thomas Meyer

Explicit discourse connectives in a source language text are not always translated to comparable words or phrases in the target language. The paper provides a corpus analysis and a method for semi-automatic detection of such cases. Results show that discourse connectives are not translated into comparable forms (or even any form at all), in up to 18% of human reference translations from English to French or German. In machine translation, this happens much less frequently (up to 8% only). Work in progress aims to capture this natural implicitation of discourse connectives in current statistical machine translation models.

2013

Translation Error Spotting from a User's Point of View

Thomas Meyer

The evaluation of errors made by Machine Translation (MT) systems still needs human effort despite the fact that there are automated MT evaluation tools, such as the BLEU metric. Moreover, assuming that there would be tools that support humans in this translation quality checking task, for example by automatically marking some errors found in the MT system output, there is no guarantee that this actually helps to achieve a more correct or faster human evaluation. The paper presents a user study which found statistically significant interaction effects for the task of finding MT errors under the conditions of non-annotated and automatically pre-annotated errors, in terms of the time needed to complete the task and the number of correctly found errors.

Idiap2012