Elamur Research⌕
☰
home / papers / 2003 / Controlling controlled English

Controlling controlled English

Paper · 2003 · Sharon O'Brien

Summary

O'Brien asks whether Controlled English (CL) rule sets share enough rules to form a common core that a newcomer could adopt instead of starting from scratch. She asked sixteen organisations for their rule sets and analysed eight: AECMA Simplified English, Attempto Controlled English (ACE), COGRAM, IBM's Easy English, GM's CASL and the CLs of Océ, Sun and Avaya. Six were under non-disclosure, so they appear only under code names.

Every rule was sorted by its main purpose into one of three classes: Lexical, Syntactic or Textual (split into Text Structure and Pragmatic). Each class has fine-grained sub-categories. The sets were then compared for completeness, shared rules and unique rules.

The overlap turned out to be tiny. All eight agree on just one rule, keeping sentences short, and only seven rules appear in at least half of the sets. The author explains the divergence by:

  • different goals (readability vs. translatability)
  • the target MT system and language pair
  • inherited corporate style guides
  • plain subjectivity

The study matters because it shows there is no de facto standard. To compensate, the author offers her own starter list of rules for machine translatability.

Key points

  • Corpus: eight Controlled English rule sets. AECMA SE is the only human-oriented CL; the other seven aim at machine translatability.
  • Only one universal rule (short sentences), and seven rules are shared by 50% or more of the sets.
  • Widely shared rules:
    • use the active voice (SE plus six others)
    • place an article or demonstrative before nouns (SE plus six)
    • avoid gerunds (six CLs)
    • cap noun clusters, usually at three nouns (SE plus five)
    • do not drop relative pronouns (five CLs)
  • Unique-rule share ranges from 30% (CL 3) and 32% (CL 4) to 83% for ACE. ACE is the outlier because it maps English onto a formal language.
  • Completeness varies across the sets. CL 3's rules had to be reconstructed from its checker's error messages, and CL 5 is known to be partial.

Interesting ideas

  • I1. Sorting each rule by its single primary function (word choice, syntax, text layout, pragmatics) gives a clean way to compare heterogeneous rule collections, despite the admitted fuzziness of linguistic categories.
  • I2. Measuring the "proportion of unique rules" per set is a simple divergence metric between guidelines written by different groups.
  • I3. A hidden rule set can be reconstructed from a checker's error messages, provided some omissions are tolerated.
  • I4. The shared CL rules target known ambiguity sources: dropped relative pronouns, long noun clusters, gerunds, passives and missing articles.
  • I5. A small starter rule set built from cross-set consensus can serve as an output style for machine-friendly text.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (learn): self-induced repair rules could be tagged by primary function, so we can see which layer generates most repairs and audit them by class.
  • I2 (tree): applied to annotation conventions of different treebanks or UD dialects, it would show which conventions are idiosyncratic and need explicit conversion rules.
  • I3 (learn): when a witness parser or LLM teacher has implicit conventions, its rejections and corrections could be mined into an explicit, explainable rule list.
  • I4 (tree): these phenomena make a ready-made stress-test checklist for the perceptron parser and gparse, for example by evaluating with and without each construction.
  • I5 (back): constraining generation to short, active sentences with explicit articles and relative pronouns should make text → graph → text round-trips easier to reparse and verify.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,379 works

Cite

Cite as Sharon O'Brien. 2003. Controlling controlled English. EAMT Workshop: Improving MT through other language technology tools: resources and tools for building MT.

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

@inproceedings{obrien2003controlling,
  title = {{Controlling controlled English}},
  author = {Sharon O'Brien},
  booktitle = {EAMT Workshop: Improving MT through other language technology tools: resources and tools for building MT},
  year = {2003},
  url = {https://aclanthology.org/2003.eamt-1.12/}
}

References

22 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. AECMA. 2001. AECMA Simplified English – A Guide for the Preparation of Aircraft Maintenance Documentation in the International Aerospace Maintenance Language, AECMA Document PSC-85-16598, Issue 1, Revision 2. find
  2. Adriaens, Geert, Jeffrey Allen, Arendse Bernth, Kurt Godden, Teruko Mitamura, Eric Nyberg, Rick Wocjik, Rémi Zajac (eds). 2000. Proceedings of the Third International Workshop on Controlled Language Applications (CLAW 2000), Seattle, Washington, 29-30 April. find
  3. Adriaens, Geert, Roger Havenith, Rick Wocjik, Bruno Tersago. 1996. Proceedings of the First International Workshop on Controlled Language Applications (CLAW 96), Centre for Computational Linguistics, Leuven, Belgium. find
  4. Adriaens, Geert. 1994. “ Simplified English Grammar and Style Correction in an MT Framework: the LRE SECC Project” , in Proceedings of Translating and the Computer 16, London, 78-88.
  5. Allen, Jeffrey. 1999. Different Types of Controlled Language, Electronic publication in http://www.tc-forum.org, February 16 (last checked October 28, 2002) ↗
  6. Bernth, Arendse. 2000. “ EasyEnglish: Grammar Checking for Non-Native Speakers” , in Adriaens et al (eds), 33-42. find
  7. Bernth, Arendse. 1999. “ EasyEnglish: A Confidence Index for MT” , in Proceedings of the 8 th International Conference on Theoretical and Methodological Issues in Machine Translation, Chester College, England, 120-127. find
  8. Bernth, Arendse. 1998a. “ EasyEnglish: Pre-processing for MT” , in Mitamura et al (eds), 30-41. find
  9. Bernth, Arendse. 1998b. “ EasyEnglish: Addressing Structural Ambiguity” , in Proceedings of AMTA-1998, Association for Machine Translation in the Americas, 164-173. find
  10. Bloor, Thomas and Meriel Bloor. 1996. The Functional Analysis of English: A Hallidayan Approach, Edward Arnold. find
  11. Cremers, Lou. 2001. “ Towards an Automated Translation Workflow at Océ Technologies” , International Journal for Language and Documentation, Issue 9, May/June 2001. find
  12. Fuchs, Norbert, Uta Schwertel, Rolf Schwitter. 1999. Attempto Controlled English Language Manual, Version 3.0, Institut der Informatik der Universität Zürich, Nr. 99-03. find
  13. Fuchs, Norbert, Rolf Schwitter. 1996. “ Attempto Controlled English (ACE)” , in Adriaens et al (eds), 124-136.
  14. Godden, Kurt. 1998. “ Controlling the Business Environment for Controlled Language” , in Mitamura et al (eds), 185-189. find
  15. Godden, Kurt. 2000. “ The Evolution of CASL Controlled Authoring at General Motors” , in Adriaens et al (eds), 14-19. find
  16. Huijsen, Willem-Olaf. 1998. “ Controlled Language: An Introduction” , in Mitamura et al. (eds), 1-15. find
  17. Mitamura, Teruko, Eric Nyberg, Geert Adriaens, Linda Schmandt, Rick Wojcik, Remi Zajac (eds). 1998. Proceedings of the Second International Workshop on Controlled Language Applications (CLAW 98), Language Technologies Institute, Carnegie Mellon University, Pittsburgh, Pennsylvania. find
  18. Mitamura, Teruko and Eric Nyberg. 1995. “ Controlled English for Knowledge-Based MT: Experience with the KANT System” , in Proceedings of the 6 th International Conference on Theoretical and Methodological Issues in Machine Translation (TMI 95), Leuven, Belgium, July 5-7.
  19. Reuther, Ursula. 1998. “ Controlling Language in an Industrial Application” , in Mitamura et al (eds), 174-184. find
  20. Reuther, Ursula and Antje Schmidt-Wigger. 2000. “ Designing a Multi-Purpose CL Application” , in Adriaens et al (eds), 72-82. find
  21. Schreurs, Dirk and Geert Adriaens. 1992. “ From COGRAM to ALCOGRAM: Toward a Controlled English Grammar Checker” , in Proceedings of COLING ’92, 595-601.
  22. Schütz, Jörg. 2001. “ Controlled Language Deployment: New Challenges and Opportunities for Translation Professionals” , in Proceedings of the Federcentri Conference, Bologna, 26-28 October. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/brien-2003-controlling-controlled-english/