Capturing Language-Specific Semantic Distinctions in Interlingua-based MT

What it is about

The paper describes an English-Spanish system (KBNL) that uses a Discourse Representation Structure (DRS) as its interlingua. The source text is parsed into a source DRS, which is the interlingua itself; a target DRS is then built, and a generator produces the target text. The authors argue that translation mismatch and divergence are generation problems rather than transfer problems. The target generator should decide what to say, and the source text can guide it. Three devices do this work. Assertions forced by the source grammar are marked, so that the generator can drop them when the target does not force them. Assertions that came from one source word are bracketed together. A target word is blocked from covering several assertions when the source used a single word that would have covered them. Examples are Spanish "pez" versus "pescado", and the rice case, where a Japanese word implies "cooked". The text gives no counts or scores; it reports a working system whose Spanish grammar handles these distinctions.

Ideas

• I1. Mark the forced features in the meta-world: a number or status distinction that the source grammar forces (plural in English, the Japanese status distinction) is marked on the meta token. The encoder drops it unless the target standard also forces it. Measure how many marked features are dropped per author pair, run twice.
• I2. Use the bracketing as evidence in the encoder: when meta tokens came from one source word, a single target chunk is fine; when they came from several source words and the source had a one-word form, the encoder should not merge them. Count how often the current chunk choice would differ.
• I3. Marked and unmarked pairs in profile.rs: when the standard holds a marked and an unmarked form for one concept, pick by the source form's marking (informal, formal, archaic). Start with counts from the labelled author data.
• I4. In the residual, split the mismatch cases from the reader and encoder losses. The paper says mismatches belong to generation, so Mova's per-class residual should put them under the encoder.

For Mova

The places are tools/translate/profile.rs (the encoder) and measure.rs (the residual classes). The forced-feature marks touch open problem 4 (function words and boundaries) and open problem 5 (number and person from FEATS). The paper has no numbers, so every idea above needs a measure of its own before any claim.

Topics: Machine translation

Cite

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

@inproceedings{barnett1991capturing,
  title = {{Capturing Language-Specific Semantic Distinctions in Interlingua-based MT}},
  author = {Barnett, James and Mani, Inderjeet and Rich, Elaine and Aone, Chinatsu and Knight, Kevin and Martinez, Juan C.},
  booktitle = {Proceedings of Machine Translation Summit III: Papers},
  year = {1991},
  url = {https://aclanthology.org/1991.mtsummit-papers.4/}
}

References

26 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. A. Andrews, “Unification and Morphological Blocking”, Natural Language and Linguistic Theory, 8(4), November 1990, pp. 507-558. ↗
2. C. Aone, “Classifiers and an Interlingua for Multilingual Machine Translation”, Proceedings of the International Conference on Current Issues in Computational Linguistics, Penang, Malaysia, 1991. find our card
3. J. Barnett, K. Knight, I. Mani, and E. Rich, “Knowledge and Natural Language Processing”, Communications of the ACM, August, 1990. ↗
4. J. Barnett, D. D’Souza, K. Knight, I. Mani, P. Martin, E. Rich, C. Aone, J. Blevins, W, Bohrer, S. Luper-Foy, J.C. Martinez, and D. Newman, “Knowledge-Based Natural Language Processing: the KBNL System”, MCC Technical Report ACT-NL-123-91, 1991. find
5. J. Barnett, I. Mani, P. Martin, and E. Rich, “Reversible Machine Translation: What to Do When the Languages Don’t Line Up”, Proceedings of the ACL Workshop on Reversible Grammars in Natural Language Processing, Berkeley, 1991. find our card
6. J. Barnett, E. Rich, and D. Wroblewski, “A Functional Interface to a Knowledge Base for Use by a Natural Language Processing System”, MCC Technical Report ACT-NL-OI9-91, 1991. find
7. J. Barnett and I. Mani, “Using Bidirectional Semantic Rules for Generation”, Proceedings of the Fifth International Workshop on Natural Language Generation, pp.47-53, Dawson, Pa., 3-6 June, 1990. find
8. J. Barnett and I. Mani, “Shared Preferences”, Proceedings of the ACL Workshop on Reversible Grammars in Natural Language Processing, Berkeley, 1991. find
9. J. Bresnan, “The Mental Representation of Grammatical Relations”, MIT Press, 1982. find
10. J. Calder, M. Reape, and H. Zeevat, “An Algorithm for Generation in Unification Categorial Grammar”, Proceedings of the 4th Conference of the European Chapter of the ACL, pp. 233-240, Manchester, 10-12 April, 1989. find
11. J. Crawford, “Access-Limited Logic-A Language for Knowledge Representation”, Ph.D. Thesis, The University of Texas as Austin, 1990. find
12. C. DiMarco and G. Hirst, “Accounting for Style in Machine Translation”, in Proceedings of the Third International Conference on Theoretical and Methodological Issues in Machine Translation of Natural Language, Austin, Texas, 1990. find
13. B. Dorr, “Solving Thematic Divergences in Machine Translation”, Proceedings of the 28th Annual Meeting of the ACL, Pittsburgh, 1990. find our card
14. K. Goodman and S. Nirenburg, “KBMT-89 Project Report”, Center for Machine Translation, Carnegie Mellon University, 1989. find
15. I. Heim, “The Semantics of Definite and Indefinite Noun Phrases”, University of Massachusetts, Ph.D. Dissertation, 1982. ↗
16. E. H. Hovy, “Generating Language with a Phrasal Lexicon”, in D. D. McDonald and L. Bolc, eds., Natural Language Generation Systems, Springer-Verlag, New York, 1988. ↗
17. P. S. Jacobs, “PHRED: a generator for natural language interfaces”, Computational Linguistics, 11(4), 1985. find
18. R. Jakobson, “Zur Struktur des Russischen Verbums”, in Hamp, Householder, and Austerlitz, eds,, Readings in Linguistics, II, Chicago: The University of Chicago Press, 1966. find
19. M. Kameyama, R. Ochitani, and S. Peters, “Resolving Translation Mismatches with Information Flow”, Proceedings of the 29th Annual Meeting of the ACL, Berkeley, 1991. find
20. H. Kamp, “A Theory of Truth and Semantic Representation”, in M. Groenendijk, J. Janssen, and M. Stokhoff, eds., Formal Methods in the Study of Language, Dordrecht: Foris, 1984. find
21. K. Knight, E. Rich, and D. Wroblewski, “Integrating Language Acquisition and Knowledge Acquisition”, Proceedings of Workshop on Automated Tools for Lexical Acquisition, IJCAI-89, 1989. find
22. G. Mounin, Les Problèmes Théoriques de la Traduction, Gallimard, 1963. find
23. E. Rich and S. Luperfoy, “An Architecture for Anaphora Resolution”, Proceedings of the Second Conference on Applied Natural Language Processing, Austin, 9-12 February, 1988, find
24. H. Uchida and M. Zhu, “An Interlingua for Multilingual Machine Translation”, Shizengengoshori, 1989. 5. 19., Japan, 1989. find our card
25. G. Van Noord, “Reversible Unification-Based Machine Translation”, in Proceedings of COLING-90, Helsinki, 1990. (Wittenburg and Barnett, 88] K. Wittenburg and J. Barnett, “Canonical Representation in NLP System Design: A Critical Evaluation”, Proceedings of the Second ACL Conference on Applied Natural Language Processing, 1988. find our card
26. D. Wroblewski and E. Rich, “LUKE: An Experiment in the Early Integration of Natural Language Processing”, Proceedings of the Second ACL Conference on Applied Natural Language Processing, 1988. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/barnett-1991-capturing-language-specific-semantic-distinctions/