Parameterization of the Interlingua in Machine Translation

What it is about

The UNITRAN system is an interlingual translator for English, Spanish and German. Its pivot is a lexical conceptual structure shared by all three languages. Differences between languages are not coded as separate processors. They are settings of a few parameters attached to universal principles: for example whether the subject must be overt (the null-subject setting), and where adjuncts and heads sit. The syntactic level and the lexical-semantic level each have their own parameters. The paper catalogues the syntactic divergences that these settings produce among the three languages, and states that adding a language means setting its parameters rather than writing a new processor. There are no measured results; the evidence is a catalogue of examples and a description of the modules.

Ideas

• I1. Make the encoder's per-language settings an explicit parameter table in profile.rs: a fixed enum of settings such as adjunct side and head direction, each value counted from the aligned data and not set by hand. Keep the table in a BTreeMap so that its order is deterministic.
• I2. Split the residual by parameter. Ask how much of the encoder's remaining error is explained by word order alone and how much by lexical choice. The order part is what a parameter should absorb, so a parameter that does not reduce it is not earning its place.
• I3. Keep word-order settings in en/src/lin.rs, as side tables that are learned per target language, and leave the Markov chains over word classes as they are. Test each side table with a shuffled-order control that must come out worse.

For Mova

The paper is the origin of the idea that a meta-world plus a small set of settings can serve several languages. For Mova that is the division between tools/translate/src/profile.rs (the standard encoder) and the author delta. The useful part for now is the measure: a residual split by order and by lexis, which measure.rs can already produce. The paper's rule-free parameters should be learned, not copied.

Topics: Machine translation

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/992133.992167

@inproceedings{dorr1992parameterization,
  title = {{Parameterization of the Interlingua in Machine Translation}},
  author = {Dorr, Bonnie},
  booktitle = {COLING 1992 Volume 2: The 14th International Conference on Computational Linguistics},
  year = {1992},
  doi = {10.3115/992133.992167},
  url = {https://doi.org/10.3115/992133.992167}
}

References

16 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Chomeky, Noam A. (1981) Lectures an Government and Binding, Forts Publieations~ Dordrecht. find
2. Choranky, Noam A. (1982) ~Some Concepts and Consequences of the Theory of Government and Binding," MIT PresL Dorr, Bonnie J. (1987) "UNITRAN: A Principle-Ba~d App~ach to Machine Translation," AI Technical Report 1000, MMter of Science thesis, Department of Electrical Engineering and Computer Science, Mauachusetts Institute of Technology. find
3. Dorr, Bonnie J. (1990a) "Solving Thematic Divergences in Machine Tranalationp" Proceedings of the £8th Annual Conference of the Association for Computational Linguistics, Univerlity of Pittsburgh, Pittsburgh, PA, 127-134. find
4. Dart, Bonnie J. (1990b) "A Croda-Linguistic Approach to Machine Translation," Proceedings o] the Third International Conference on Theoretical and Methodological Issues in Ma. chine Translation of Natural Languages, Linguistics Research Center, The Univerlity of Texai, Austin, TX, 13-32. find
5. Dorr, Bonnie J. (1992) "A Parameterized Approach to ]ntegrat-in K Aspect with Lexical-Semantics for Mwehine Translation," Proceedings of the 3Oth Annual Conference of the Association for Computational Linguistics, University of Delaware, Newark, DE. find
6. Hale, Kenneth and M. Laughren (1983) "Warlpiri Lexicon Project: Warlpiri Diction~,ry Entries," Mm~achusetts Institute of Technology, Cambridge , MA, Warlpiri Lexicon Project. find
7. Hale, Kennetb and Jay Keyaer (1986a) "Some qYansitivity Alternations in English," Center for Cognitive Science, MEte. sachuaetts Institute of Technology, Cambridge, MA, Lexicon Project Working Papers #7. find
8. Hale, Kenneth and J. Keyler (1986b) "A View from the Middle," Center for Cognitive Science, Ma~achuaetts Institute of Technology, Cambridge, MA, Lexicon Project Working Papern #10. find
9. Jackendoff, Ray S. (1983) Semantics and Cognition, MIT Pre~a~ Cambridge, M A . find
10. Jackendoff, Ray S. (1990) Semantic Structures, M I T preaa~ Cambridge, M A . find
11. Jaeggli , Osv~ldo Adolfo (1981) Topics in tlomanee Syntax, Forts Publications, Dordrecht, Holland/Cinnaminaon, USA. find
12. Levis, Beth and Malka Rappaport (1986) "The Forn]ation of Adjectival Passives," Linguistic Inquiry 17, 623-662. find
13. S~flr, Ken (1985) "Missing Subjects in German," in Studies in German Grammar, Toman, Jindlfich (ed.), Foria Publications, Dordrecht, Holland/Cinnaminnon, USA, 19::1-229. find
14. Wexler, Kenneth and M. Rite Manzini (1986) "Parameters and l.earnability in Binding Theory," pr~ented at tbe Go#nitiue Science Seminar, M I T , September, Cambridge, MA. find
15. Zubiaarreta, Maria Luis~ (1987) Leveb of Representation in the Lexicon and in the Syntax~ Foris Publications, Dordrecht, Holland/Cinnaminmon, USA. find
16. PROC. OF COLING-92, NANTES, AUG. 23-28, 1992 find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/dorr-1992-parameterization-interlingua-machine-translation/