📄 Simplified Technical English in the 21st Century

Summary

This short workshop abstract surveys Simplified Technical English (STE). STE is a controlled form of English created to make aerospace maintenance instructions easier to read and harder to misread. The author, a writing specialist at Pratt & Whitney, does not report experiments or data. He gives a structured overview in four parts: the history of the standard, its current institutional position, planned extensions, and the practical problems seen in industry. The main point is that STE is moving from an aerospace-only guide to a broader standard for technical documentation. This happens because the S1000D publications specification adopts it as the default language control. The paper also notes where STE struggles in practice: training is hard to get, the rules can be read in different ways, and offshore writers who are not native speakers have trouble with it. The author proposes that automated STE tools should check whole phrases and not only single words, so that output stays idiomatic and tied to its context. This matters for anyone who builds checkers or generators for restricted languages.

Key points

• The standard began in the late 1970s as Simplified English under AECMA. It entered the ATA-100 publications spec in 1987, and its guide reached final form in 1995.
• In 2004 AECMA became ASD (AeroSpace Defence Industries of Europe). SE was renamed STE, and the guide became the ASD STE-100 Specification.
• ASD's S1000D spec uses STE-100 as its default language control. S1000D is expected to replace ATA-i2200 (formerly ATA-100) for both commercial and military writing, which could spread STE to land, sea and air programs.
• Planned work includes:
  • making STE-100 more accessible to defense ministries,
  • a "generic" STE beyond aerospace,
  • adjunct resources such as user helps, Web access and a thesaurus that is still in development.
• Checker software for STE is becoming more common. The reported problems are training access, interpretation, and non-idiomatic word choices by non-native offshore writers.

Interesting ideas

• I1. A controlled language limits vocabulary and grammar so that instructions have one clear reading, and such a restricted register is a clean, low-ambiguity test bed.
• I2. The author argues that compliance checking should work on phrases, not isolated words, to keep idiom and context intact.
• I3. The paper points out that the spec quietly assumes writers know both English nuance and the technical subject, and that this assumption often fails.
• I4. A thesaurus that maps disallowed words to approved ones is planned as a companion to the core rule set.

For Mova

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

• I1 (tree): An STE-style subset of English could serve as an "easy tier" benchmark, where parse errors point to real parser bugs rather than genuine ambiguity.
• I2 (tokens): The lemmatizer and tagger could treat approved multiword expressions as units, so later stages see stable phrase meanings instead of split-up words.
• I3 (learn): Evaluation sets could deliberately include non-native and out-of-domain text as negative controls, to show which rules silently depend on native-speaker intuition.
• I4 (back): The generator could render meaning graphs in a controlled English register, using such a mapping to swap rare or ambiguous words for safer, approved choices.