Towards an interlingual treatment of modality

What it is about

The paper argues that modality, the speaker's comment on how valid a statement is, needs its own interlingual module, because many errors in machine translation come from its absence. The author counts modal elements in the EDR corpus of 195,000 Japanese sentences: for example "tai" (2,777 cases), "noda" or "nodesu" (3,119) and "kudasai" (245). Modality is shown to be carried by many categories, among them discourse particles, which verb-mode accounts miss. The proposed Module of Modality has three dimensions: a world (a modal subtype such as belief, duty or plan), a remoteness flag, and the proposition. A world is fired by a combination of trigger elements, not by one element, and a proposition can have only one world. Some elements exclude a world, for example past tense excludes a directive duty. Belief has five strengths. The module was first added to the ALT-J/E Japanese-English system for the belief world, where it stopped stacks of sentence-initial adverbs like "probably surely perhaps" from reaching the output.

Ideas

• I1. Count modal triggers in the verse-aligned pairs, in the same way as the paper's frequency table (shall, will, must, may, and the archaic forms), and report where the meta-world loses modality. Measure first, change second.
• I2. Store trigger combinations as a table in a BTreeMap: a set of meta features fires a modal world, and negative triggers exclude one. The encoder then picks, for example, "thou shalt" or "you will" from the author delta. This makes the modal choice an author feature.
• I3. Singularity for modal adverbs: merge stacked modal adverbs on one meta token, as the paper does for sentence-initial opinion words. Keep the merge only if it lowers the residual bits.
• I4. Give a modal concept a strength index taken from counted strength markers, as the belief world does.

For Mova

The places are tools/translate/meta.rs (the reader's triggers) and profile.rs (the encoder's choice of form). I1 to I4 touch open problem 4 (function words as features) and open problem 5 (number and person from FEATS). The archaic modal forms in the Bible pairs are a real candidate for an author delta, which is open problem 6.

Topics: Machine translation

Cite

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

@inproceedings{neumann1999interlingual,
  title = {{Towards an interlingual treatment of modality}},
  author = {Neumann, Christoph},
  booktitle = {Proceedings of Machine Translation Summit VII},
  year = {1999},
  url = {https://aclanthology.org/1999.mtsummit-1.93/}
}

References

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

1. Aijmer K. (in press), “Epistemic Possibility in An English-Swedish Contrastive Perspective” in Proceedings of the Colloque International, 10.-12.12.1998 Anvers, Les verbes modaux dans les langues germaniques et romanes Bond F., Ogura K. and Uchino H., (1997), “Temporal expressions in Japanese-to-English Machine Translation” in: Seventh International Conference on Theoretical and Methodological Issues in Machine Translation: TMI-97, 55-62 find
2. Calbert Joseph P. (1975), “Towards the Semantics of Modality” in: Calbert J.P./ Vater H. (1975), Aspekte der Modalität, Tübingen, Gunter Narr Verlag, 1 -70 find
3. EDR (1995), Electronic Dictionary Technical Guide (In Japanese), Tokyo, Japan Electronic Dictionary Research Institute, Ltd. find
4. Ikehara S., Shirai S., Akio Y. and Nakaiwa H. (1991), “Toward an MT system without pre-editing – effects of new methods in ALT-J/E” in Proceedings of MT Summit III, 101-106 find
5. Lyons J. (1977), Semantics, Cambridge, Cambridge University Press [Chapter 17: “Modality”] find
6. Masuoka T. (1991): Modariti-no bunpo [Grammar of modality], Tokyo, Kuroshio Neumann C. (1998), “Modality Expressions in Japanese” in: IEICE Technical Report Vol. 98 No. 209, 9-16 (Natural Language Understanding and Models of Communication - NLC 98-8) find
7. Neumann C. and Bond F. (in press), “Problems in Translating Modal Expressions” in: Proceedings of the Fifth Annual Meeting of The Association for Natural Language Processing (Tokyo) Murata Masaki, Ma Qing, Uchimoto Kiyotaka and Isahara Hitoshi (1999), “Yourei-beesu-ni-yoru modariti-no-nichiei-hon'yaku” (The Example-Based Approach to Japanese-to-English Modality Translation) in IPSG SIG Notes Vol. 99 No. 22, 121-128 Nirenburg S., Carbonell J., Tomita M. and Goodman K. (1992), Machine Translation. find
8. A Knowledge-Based Approach, San Mateo, Morgan Kaufman Palmer F.R. (1986), Mood and modality, Cambridge. Cambridge University Press find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/neumann-1999-interlingual-treatment-modality/