Elamur ResearchβŒ•
☰
home / papers / 2012 / Machine Translation of Labeled Discourse Connectives

Machine Translation of Labeled Discourse Connectives

Summary

Statistical MT works at the phrase or sentence level, so it often mistranslates discourse connectives. Words like since or while can mark different relations (temporal, causal, contrastive, concessive), and each sense needs a different French equivalent. The authors label a small set of highly ambiguous English connectives with their senses before translation. They use a MaxEnt classifier trained on Europarl annotations. The labels feed English/French SMT, trained and tested on WMT10 data, in two ways:

  1. as an extra factor in phrase-based and hierarchical factored models;
  2. by replicating each training and tuning sentence according to the classifier's discretized label probabilities.

Connective translation is assessed with a new semi-automated metric (ACT) as well as BLEU. Labeling corrects a noticeable share of connective translations, while BLEU stays roughly unchanged. This is expected, since connectives are rare. The work matters because a wrong connective can silently distort a text's argument. The same method might also be applied to further function words that operate across the text rather than within a sentence, for example pronouns or tense markers.

Key points

  • The sense classifier covers seven ambiguous connectives: although, meanwhile, since, though, even though, while, yet. It was trained on 1,449 annotated Europarl instances and tested on 328, reaching a macro-averaged F1 of 0.94 overall.
  • With factored models, 11–18% of connective translations improved and 14–24% got worse compared with the baseline. Gains in BLEU (+0.48 for POS+DL, +0.55 for POS) came mainly from the POS factors, not from the discourse labels.
  • The label-probability (LPD) setup makes ten copies of each sentence, labeled in proportion to the classifier's confidence across 12 possible senses. It gave +0.3 BLEU, with about 11% of connectives improved, 85% unchanged and 4% degraded.
  • In sentence-level BLEU on segments containing connectives, LPD improved 29% of segments and degraded 16%. POS+DL improved 34% and degraded 26%.
  • Connectives are infrequent: they make up 1.8% of tokens in the Penn Discourse Treebank, and about 2.6% of Europarl sentences contain a potentially ambiguous one.

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

Fields: Linguistics & languages

Topics: Discourse analysis Β· Machine translation

Cited by

19 papers of our library cite it (of the 140673 whose bibliographies we read); 2 of them have a page here, 17 not yet.

Cite

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

@inproceedings{meyer2012machine,
  title = {{Machine Translation of Labeled Discourse Connectives}},
  author = {Meyer, Thomas and Popescu-Belis, Andrei and Hajlaoui, Najeh and Gesmundo, Andrea},
  booktitle = {Proceedings of the 10th Conference of the Association for Machine Translation in the Americas: Research Papers},
  year = {2012},
  url = {https://aclanthology.org/2012.amta-papers.20/}
}

References

32 references, 8 of them with a card of ours. β†— leads to the work, Β«findΒ» searches for one we have no id for yet.

  1. Kathryn Baker, Bonnie Dorr, Michael Bloodgood, Chris Callison-Burch, Nathaniel Filardo, Christine Piatko, Lori Levin, and Scott Miller. 2012. Use of Modality and Negation in Semantically-Informed Syntactic MT. Computational Linguistics, 38(2):411–438. find
  2. Alexandra Birch, Miles Osborne, and Philipp Koehn. 2007. CCG Supertags in Factored Statistical Machine Translation. In ACL 2007 Workshop on Statistical Machine Translation, pages 9–16, Prague. find
  3. Marine Carpuat and Dekai Wu. 2007. Improving Statistical Machine Translation using Word Sense Disambiguation. In Joint Conference on Empirical Methods in Natural Language Processing (EMNLP) and Computational Natural Language Learning (CoNLL), pages 61–72, Prague. find
  4. Pi-Chuan Chang, Dan Jurafsky, and Christopher D. Manning. 2009. Disambiguating β€˜DE’ for Chinese-English Machine Translation. In Fourth Workshop on Statistical Machine Translation at the 12th Meeting of the European Chapter of the ACL, Athens. find
  5. Chris Dyer et al. 2010. cdec: A Decoder, Alignment, and Learning Framework for Finite-state and Context-free Translation Models. In 48th Conference of the ACL, Demonstrations, pages 7–12, Uppsala. find
  6. Vladimir Eidelman, Jordan Boyd-Graber, and Philip Resnik. 2012. Topic Models for Dynamic Translation Model Adaptation. In ACL 2012 (50th Annual Meeting of the ACL), pages 115–119, Jeju. find
  7. Liane Guillou. 2012. Improving Pronoun Translation for Statistical Machine Translation. In EACL 2012 Student Research Workshop (13th Conference of the European Chapter of the ACL), pages 1–10, Avignon. find
  8. Christian Hardmeier and Marcello Federico. 2010. Modeling Pronominal Anaphora in Statistical Machine Translation. In International Workshop on Spoken Language Translation (IWSLT), Paris. find
  9. Christian Hardmeier, Joakim Nivre, and Jörg Tiedemann. 2012. Document-wide Decoding for Phrase-based Statistical Machine Translation. In Conf. on Empirical Methods in Natural Language Processing and Natural Language Learning (EMNLP-CoNLL), Jeju. find
  10. Eduard H. Hovy, Margaret King, and Andrei Popescu-Belis. 2003. Principles of Context-based Machine Translation Evaluation. Machine Translation, 17(1):43–75. find
  11. Philipp Koehn and Hieu Hoang. 2007. Factored Translation Models. In Conf. on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL), pages 868– 876, Prague. find our card
  12. Philipp Koehn et al. 2007. Moses: Open Source Toolkit for Statistical Machine Translation. In 45th Annual Meeting of the ACL, Demonstration Session, pages 177–180, Prague. find
  13. Philipp Koehn. 2005. Europarl: A Parallel Corpus for Statistical Machine Translation. In MT Summit X, pages 79–86, Phuket. find
  14. Ronan Le Nagard and Philipp Koehn. 2010. Aiding Pronoun Translation with Co-reference Resolution. In Joint 5th Workshop on Statistical Machine Translation and Metrics (MATR), pages 258–267, Uppsala. find our card
  15. Ziheng Lin, Hwee Tou Ng, and Min-Yen Kan. 2010. A PDTB-styled End-to-End Discourse Parser. Technical Report TRB8/10, School of Computing, National University of Singapore. find
  16. Jianjun Ma, Degen Huang, Haixia Liu, and Wenfeng Sheng. 2011. POS Tagging of English Particles for Machine Translation. In 13th Machine Translation Summit, pages 57–63, Xiamen. find
  17. Christopher Manning and Dan Klein. 2003. Optimization, MaxEnt Models, and Conditional Estimation without Magic. In Tutorial at HLT-NAACL and 41st ACL conferences, Edmonton and Sapporo. find
  18. Daniel Marcu, Lynn Carlson, and Maki Watanbe. 2000. The Automatic Translation of Discourse Structures. In 1st North American chapter of the ACL, pages 9–17, Philadelphia. find
  19. Thomas Meyer and Andrei Popescu-Belis. 2012. Using Sense-labeled Discourse Connectives for Statistical Machine Translation. In EACL 2012 Joint Workshop on Exploiting Synergies between IR and MT, and Hybrid Approaches to MT (ESIRMT-HyTra), pages 129–138, Avignon. find our card
  20. Thomas Meyer, Andrei Popescu-Belis, Sandrine Zufferey, and Bruno Cartoni. 2011. Multilingual Annotation and Disambiguation of Discourse Connectives for Machine Translation. In 12th SIGdial Meeting on Discourse and Dialogue, pages 194–203, Portland. find our card
  21. George A. Miller. 1995. WordNet: A Lexical Database for English. Communications of the ACM, 38(11):39– 41. find our card
  22. Eleni Miltsakaki, Nikhil Dinesh, Rashmi Prasad, Aravind Joshi, and Bonnie Webber. 2005. Experiments on Sense Annotations and Sense Disambiguation of Discourse Connectives. In 4th Workshop on Treebanks and Linguistic Theories (TLT), Barcelona. find
  23. Franz Josef Och and Hermann Ney. 2003. A Systematic Comparison of Various Statistical Alignment Models. Computational Linguistics, 29(1):19–51. find
  24. Franz Josef Och. 2003. Minimum Error Rate Training in Statistical Machine Translation. In 41st Annual Meeting of the ACL, pages 160–167, Sapporo. find
  25. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. BLEU: A Method for Automatic Evaluation of Machine Translation. In 40th Annual Meeting of the ACL, pages 311–318, Philadelphia. find our card
  26. Emily Pitler and Ani Nenkova. 2009. Using Syntax to Disambiguate Explicit Discourse Connectives in Text. In ACL-IJCNLP 2009 (47th Annual Meeting of the ACL and 4th International Joint Conference on NLP of the AFNLP), Short Papers, pages 13–16, Singapore. find our card
  27. Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Miltsakaki, Livio Robaldo, Aravind Joshi, and Bonnie Webber. 2008. The Penn Discourse Treebank 2.0. In 6th Int. Conf. on Language Resources and Evaluation (LREC), pages 2961–2968, Marrakech. find our card
  28. Kristina Toutanova, Dan Klein, Christopher Manning, and Yoram Singer. 2003. Feature-rich Part-of-Speech Tagging with a Cyclic Dependency Network. In Human Language Technology Conference and the North American Chapter of the ACL (HLT-NAACL), pages 252–259, Edmonton. find
  29. Marc Verhagen and James Pustejovsky. 2008. Temporal Processing with the TARSQI Toolkit. In 22nd International Conference on Computational Linguistics (COLING), Companion volume: Demonstrations, pages 189–192, Manchester. find
  30. Rui Wang, Petya Osenova, and Kiril Simov. 2012. find
  31. Linguistically-augmented Bulgarian-to-English Statistical Machine Translation Model. In EACL 2012 Joint Workshop on Exploiting Synergies between IR and MT, and Hybrid Approaches to MT (ESIRMT-HyTra), pages 119–128, Avignon. find
  32. Billy T.-M. Wong and Chunyu Kit. 2012. Extending Machine Translation Evaluation Metrics with Lexical Cohesion to the Document Level. In Conf. on Empirical Methods in Natural Language Processing and Natural Language Learning (EMNLP-CoNLL), Jeju. find

↑ back to top

This page in Markdown β†’ https://md.research.elamur.ai/free/papers/meyer-2012-machine-translation-labeled-discourse-connectives/