[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/papers/wilks-1976-semantics-world-knowledge-mt/)

[home](https://md.mova.elamur.ai/free/en/) / [docs](https://md.mova.elamur.ai/free/en/docs/) / [papers](https://md.mova.elamur.ai/free/en/docs/papers/) / [Machine translation](https://md.mova.elamur.ai/free/en/docs/papers/mtrans/) / Semantics and world knowledge in MT

# Semantics and world knowledge in MT

**Authors:** Yorick Wilks · **Year:** 1976 · **Venue:** Foreign Broadcast Information Service Seminar on Machine Translation **Link:** [aclanthology.org/1976.earlymt-1.20](https://aclanthology.org/1976.earlymt-1.20/) · **Licence:** [CC BY-NC-SA 3.0](https://creativecommons.org/licenses/by-nc-sa/3.0/) — summary only

[Foreign Broadcast Information Service Seminar on Machine Translation](https://aclanthology.org/volumes/1976.earlymt-1/)[licence](https://creativecommons.org/licenses/by-nc-sa/3.0/)

- [Publisher's page](https://aclanthology.org/1976.earlymt-1.20/)
- [PDF](https://aclanthology.org/1976.earlymt-1.20.pdf)

**Authors (1)**

1. Wilks, Yorick

Citations and author metrics from OpenAlex (); authors with the same name are not always told apart.

## What it is about

A position statement for a seminar. The author shows that even simple newspaper paragraphs need more than syntax to translate: knowledge about the world, inference, and what he calls preference rules, which choose between readings that are all grammatical. The problems he names are word sense ambiguity, case structure and pronoun reference. He describes a small English-to-French system that reads paragraphs and translates them through an interlingua of deep meaning structures with inference rules, and he argues for a middle path between brute-force systems and fully knowledge-based ones. The text is short and gives no numbers.

## Ideas

- I1. Preference rules as counts: when a concept has several frame readings in `prag`, choose by the counts of the same predicate with the same argument class in the data, and log which preference decided. This is a count-based form of a preference rule and it can be measured by how often the choice matches the translation in the held-out verses.
- I2. Pronoun reference through the head chain: the reader can link a pronoun (the *thou* or *he* of a verse) to the nearest antecedent in the same frame, and the link can be counted as a feature of the meta token (open problem 4, function words and boundaries).
- I3. Each inference step recorded: where the world makes an inference, store the rule used and the tokens it touched, so an inference can be questioned like any other step in the stack.

## For Mova

The preference rule is the nearest idea to a world-level choice in `prag` and `world`, and it can be built from counts rather than hand rules. The pronoun link is part of the function-word and boundary work in the translator's ranked problems. The recorded inference is in line with the transparency rule of Mova. The paper gives no evaluation, so each idea has to be measured against the Bible pairs first.

Topics: [Machine translation](https://md.mova.elamur.ai/free/en/docs/papers/topics/machine-translation/)

## Cite

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

```
@inproceedings{wilks1976semantics,
  title = {{Semantics and world knowledge in MT}},
  author = {Wilks, Yorick},
  booktitle = {Foreign Broadcast Information Service Seminar on Machine Translation},
  year = {1976},
  url = {https://aclanthology.org/1976.earlymt-1.20/}
}
```

- [← previous Shake-and-Bake Translation](https://md.mova.elamur.ai/free/en/docs/papers/whitelock-1992-shake-bake-translation/)
- [next → Generation by Inverting a Semantic Parser that Uses Statistical Machine Translation](https://md.mova.elamur.ai/free/en/docs/papers/wong-2007-generation-inverting-semantic-parser-that/)
- [↑ back to Machine translation](https://md.mova.elamur.ai/free/en/docs/papers/mtrans/)