Elamur Research⌕
☰
home / papers / 2013 / Implicitation of Discourse Connectives in (Machine) Translation

Implicitation of Discourse Connectives in (Machine) Translation

Summary

The paper examines how often an explicit English discourse connective has no matching connective in a translation. The authors call this "implicitation" or zero-translation. The meaning of the discourse relation is kept, but it is expressed through other structures or left to context. They take 48 frequent English connectives from the Penn Discourse TreeBank 2.0, each labelled with its most typical level-2 PDTB relation. Dictionaries of French and German equivalents are built from Linguee. These are combined with Giza++ alignments and a small word-window check to flag cases where no target connective appears.

The method is applied to WMT newstest2010+2012 news texts, covering both human reference translations and the output of baseline Moses systems. Human translators drop or rephrase connectives much more often than SMT does. Which relations get implicitated depends on the target language.

Two patterns are studied in detail: French gerund constructions replacing temporal or contrastive connectives, and German verb-initial conditional clauses replacing if. From these, the authors sketch rules and features that could help SMT produce more natural, reference-like output.

Key points

  • Data: newstest2010+2012 contains 5,492 sentences and 117,799 source words, with 2,906 English connective tokens. The baselines were Moses phrase-based systems trained on Europarl v7, tuned on newstest2011, with a 3-gram IRSTLM language model.
  • Zero-translation rates:
    • EN/FR: 17.5% in references vs 7.5% in MT.
    • EN/DE: 13.5% in references vs 4.4% in MT.
  • Most often implicitated relations:
    • In French references: SYNCHRONY, ALTERNATIVE and CONCESSION.
    • In German references: CONDITION, ALTERNATIVE and CONCESSION.
    • German ALTERNATIVE is implicit in 23.3% of references but never in MT.
  • French cases:
    • MT renders if as si in 99.4% of cases.
    • 8 of 13 implicitated while tokens became an "en + gerund" construction.
  • German if and proposed fixes: Of 40 implicitated if cases, 21 used a sentence-initial conditional verb. Suggested remedies include POS-based rules, a Moses sparse word-deletion feature, and syntactic reordering rules.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Discourse analysis · Machine translation

Cited by

29 papers of our library cite it (of the 140673 whose bibliographies we read); 8 of them have a page here, 21 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{meyer2013implicitation,
  title = {{Implicitation of Discourse Connectives in (Machine) Translation}},
  author = {Meyer, Thomas and Webber, Bonnie},
  booktitle = {Proceedings of the Workshop on Discourse in Machine Translation},
  year = {2013},
  url = {https://aclanthology.org/W13-3303/}
}

References

15 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Shoshana Blum-Kulka. 1986. Shifts of Cohesion and Coherence in Translation. In Juliane House and Shoshana Blum-Kulka, editors, Interlingual and Intercultural Communication. Discourse and cognition in translation and second language acquisition, pages 17–35. Narr Verlag, Tübingen, Germany. find
  2. Bruno Cartoni, Sandrine Zufferey, Thomas Meyer, and Andrei Popescu-Belis. 2011. How Comparable are Parallel Corpora? Measuring the Distribution of General Vocabulary and Connectives. In Proceedings of 4th Workshop on Building and Using Comparable Corpora (BUCC), pages 78–86, Portland, OR. find
  3. Marcello Federico, Nicola Bertoldi, and Mauro Cettolo. 2008. IRSTLM: an Open Source Toolkit for Handling Large Scale Language Models. In Proceedings of Interspeech, Brisbane, Australia. find
  4. Najeh Hajlaoui and Andrei Popescu-Belis. 2013. Assessing the Accuracy of Discourse Connective Translations: Validation of an Automatic Metric. In 14th International Conference on Intelligent Text Processing and Computational Linguistics (CI-CLING), Samos, Greece. find
  5. Sandra Halverson. 2004. Connectives as a Translation Problem. In H. et al. (Eds.) Kittel, editor, Encyclopedia of Translation Studies, pages 562–572. Walter de Gruyter, Berlin/New York. find
  6. Silvia Hansen-Schirra, Stella Neumann, and Erich Steiner. 2007. Cohesive Explicitness and Explicitation in an English-German Translation Corpus. Languages in Contrast, 7:241–265. find
  7. Iustina Ilisei, Diana Inkpen, Gloria Pastor Corpas, and Ruslan Mitkov. 2010. Identifcation of Translationese: A Machine Learning Approach. In A. Gelbukh, editor, Computational Linguistics and Intelligent Text Processing Lecture Notes in Computer Science. Springer-Verlag, Berlin, Heidelberg, Germany. find
  8. Philipp Koehn, Hieu Hoang, Alexandra Birch, Chris Callison-Burch, Marcello Federico, Nicola Bertoldi, Brooke Cowan, Wade Shen, Christine Moran, Richard Zens, Chris Dyer, Ondrej Bojar, Alexandra Constantin, and Evan Herbs. 2007. Moses: Open Source Toolkit for Statistical Machine Translation. In Proce… find
  9. Philipp Koehn. 2005. Europarl: A Parallel Corpus for Statistical Machine Translation. In Proceedings of MT Summit X, pages 79–86, Phuket, Thailand. find
  10. Moshe Koppel and Noam Ordan. 2011. Translationese and its Dialects. In Proceedings of ACL-HLT 2011 (49th Annual Meeting of the ACL: Human Language Technologies, pages 1318–1326, Portland, OR. find
  11. Franz Josef Och and Hermann Ney. 2003. A Systematic Comparison of Various Statistical Alignment Models. Computational Linguistics, 29(1):19–51. find
  12. Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Miltsakaki, Livio Robaldo, Aravind Joshi, and Bonnie Webber. 2008. The Penn Discourse Treebank 2.0. In Proceedings of 6th International Conference on Language Resources and Evaluation (LREC), pages 2961–2968, Marrakech, Morocco. find our card
  13. Rashmi Prasad, Aravind Joshi, and Bonnie Webber. 2010. Realization of Discourse Relations by Other Means: Alternative Lexicalizations. In Proceedings of the 23rd International Conference on Computational Linguistics (COLING), pages 1023–1031, Beijing, China. find our card
  14. Sandrine Zufferey, Liesbeth Degand, Andrei Popescu-Belis, and Ted Sanders. 2012. Empirical Validations of Multilingual Annotation Schemes for Discourse Relations. In Proceedings of ISA-8 (8th Workshop on Interoperable Semantic Annotation), pages 77–84, Pisa, Italy. find
  15. EN conn. after also although as as a result as if as long as as soon as because before but by contrast even if even though finally for example for instance however if in addition indeed in fact instead in turn Majority rel. Asynchrony Conjunction Contrast Synchrony Cause Concession Condition Asynchr… find

↑ back to top