The Rosetta project

What it is about

The paper tells how the Rosetta project grew out of a Philips question-answering system and a Montague-style grammar that could analyse sentences without a transformational stage. Rosetta translates by isomorphic grammars: the source and target grammars are written side by side, so that two sentences are translations when they share one semantic derivation tree. The intermediate language is not universal; it is fixed for a chosen set of languages (Dutch, English, Spanish). The author then asks whether Rosetta is interlingual or transfer-like, and concludes that its architecture is interlingual, while the grammars are tuned to each other, which makes adding a third language cost extra rules. The paper gives no benchmark numbers.

Ideas

• I1. Measure the isomorphism directly: for each verse pair, count how often the reader trees of the two languages have the same shape after the meta-world step. A low rate names the places where the meta-world is not really shared.
• I2. Keep the general part of the meta-world apart from the parts that exist only for one author pair, as the paper separates general parts from pair-specific ones. In Mova this means that adding a new author changes only its delta, and the test is that the shared residual does not move.
• I3. Test the cost of the third-language problem: translate a fixed set of verses with the meta-world built from two authors, then add a third and check that the residual on the first two does not change.

For Mova

The code is tools/translate (meta.rs, profile.rs), with the frames of prag. The isomorphic idea is not a grammar tuning method for Mova, since the meta-world is learned from data; the useful part is the test of what is shared against what belongs to one pair. This touches open problem 1 (reader attachment) and the choice between a general and an author-specific layer.

Topics: Machine translation

Cite

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

@inproceedings{landsbergen1989rosetta,
  title = {{The Rosetta project}},
  author = {Landsbergen, Jan},
  booktitle = {Proceedings of Machine Translation Summit II},
  year = {1989},
  url = {https://aclanthology.org/1989.mtsummit-1.15/}
}

References

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

1. Bronnenberg, W.J.H.C., H.C. Bunt, S.P.J. Landsbergen, R.J.H. Scha, W.J. Schoenmakers and E.P.C. van Utteren, The question-answering system PHLIQA 1, in: L. Bolc (ed.), Natural Language Question-answering systems, Carl Hanser Verlag, München, 1980. find
2. Thomason, R.H. (ed.), Formal Philosophy, Selected Papers of Richard Montague, Yale University Press, New Haven, 1974. find
3. Friedman, J. and D.S. Warren, A parsing method for Montague grammars, Linguistics and Philosophy, 2, 1978. find
4. Landsbergen, J., Adaptation of Montague grammar to the requirements of parsing, in: Groenendijk, J.A.G., T.M.V. Janssen, and M.B.J. Stokhof, Formal methods in the Study of Language Part 2, MC Tract 136, Mathematical Centre, Amsterdam, 1981, pp 399-420. find
5. Partee, B.H., Some transformational extensions of Montague grammar, in: Partee, B.H. (ed.), Montague Grammar, Academic Press, New York, 1976, pp 51 - 76. find
6. Landsbergen, J., Montague Grammar and Machine Translation, in: Whitelock, P. et al. (eds.), Linguistic Theory and Computer Applications, Academic Press, London, 1987. find
7. Landsbergen, J., Isomorphic grammars and their use in the Rosetta translation system, presented at the Tutorial on Machine Translation, Lugano, 1984. In: M. King (ed), Machine Translation the state of the art, Edinburgh University Press, 1987. find
8. Odijk, J., The organization of the Rosetta grammars, Procs of European ACL Conference, Manchester, 1989. find
9. Appelo, L., C. Fellinger and J. Landsbergen, Subgrammars, Rule Classes and Control in the Rosetta Translation System, Procs of European ACL Conference, Copenhagen, 1987. find
10. Appelo, L., A Compositional approach to the Translation of Temporal Expressions in the Rosetta System, Procs of the 11th Conference on Computational Linguistics, 1986, Bonn. find our card
11. Schenk, A., Idioms in the Rosetta Machine Translation System, Procs of the 11th Conference on Computational Linguistics, 1986, Bonn. find our card
12. Van Munster, E., The Treatment of Scope and Negation, Procs of the 12th Conference on Computational Linguistics, 1988, Budapest. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/landsbergen-1989-rosetta-project/