Elamur Research⌕
☰
home / papers / 2016 / Error Typology and Remediation Strategies for Requirements Written in English by Non-Native Speakers

Error Typology and Remediation Strategies for Requirements Written in English by Non-Native Speakers

Paper · 2016 · Marie Garnier and Patrick Saint-Dizier

Summary

Garnier and Saint-Dizier study why requirements documents in English, written by engineers whose first language is French, still contain many language errors even when authors follow controlled-language norms. They first show the problem with the LELIE alert checker. LELIE runs on the Prolog-based <TextCoop> platform. On about 35,000 words of already proofread technical text from three companies, it flags roughly one issue every 50 words. The authors then apply classic error analysis from Second Language Acquisition. A single trained linguist manually screens a requirements corpus and judges errors by both grammaticality and acceptability. Each error is sorted into comparable categories, mainly by its grammatical domain. The distribution is compared with errors by French speakers in student essays and in scientific papers. Requirements show a distinctive profile: errors cluster in the Noun Phrase, especially stacked modifiers, alongside signs of simplification strategies. This matters because generic L2 tools or controlled languages alone will not fix these errors. The paper outlines two remedies: genre-aware automatic detection inside LELIE and certification-oriented training material.

Key points

  • Selected LELIE alerts on about 35,000 words include:
    • 261 for overly complex sentences
    • 182 for fuzzy lexical items
    • 135 for heavy N+N or noun complements
    • 127 for complex discourse structures
    • 126 for negation
    • 80 for deverbal nouns
  • These alerts occur even in texts proofread by several trained writers. This suggests that controlled languages such as ASD-Simplified Technical English, SBVR Structured English and INCOSE norms are not sufficient.
  • Errors are identified with two criteria. Grammaticality asks whether rules are broken. Acceptability asks whether a native speaker would produce the form in context.
  • Errors in requirements cluster in the Noun Phrase, notably modifier stacking. This differs from comparison corpora of student essays and scientific papers by French L1 writers.
  • LELIE is implemented in Prolog with a freely available kernel for French and English. The work was funded by IREB, and future work targets NP errors and two more industrial domains.

Interesting ideas

  • I1. The authors compare the same L1 group across three genres (requirements, essays, papers) to separate genre-specific errors from general learner errors.
  • I2. They treat L1 transfer as evidence for what the author meant to say, which then guides the correction.
  • I3. They judge errors on two axes: strict grammaticality versus native-like acceptability.
  • I4. Their checker counts every repeated instance of an error, because each occurrence carries the same risk for the reader.
  • I5. They flag stacked noun modifiers (heavy N+N chains) as a key source of ambiguity.

For Mova

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

  • I1 (learn): Evaluate repair rules or parser errors on several genres at once, so a fix tuned to one register is not mistaken for a general improvement.
  • I2 (frames): When a sentence is malformed, a model of likely L1 interference could help recover the intended who-did-what-to-whom frame instead of discarding the parse.
  • I3 (tree): Tag parser input anomalies as ungrammatical or merely unidiomatic, so the parser can stay robust on unidiomatic text and flag only true breakage.
  • I4 (learn): Weight evaluation errors by frequency in real text, not only by distinct type, so common failure patterns get repaired first.
  • I5 (tree): Add diagnostics for long compound chains, where compound/nmod attachment is most fragile, and route them to stricter witness agreement.

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

 Institutions

🔬Centre National de la Recherche ScientifiqueParis · France168 works · 62,369 citations🔬Institut de Recherche en Informatique de ToulouseToulouse · France35 works · 1,872 citations🎓Université Toulouse - Jean JaurèsToulouse · France2 works · 3 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksDevelopmental and Educational Psychology · 499 worksLiterature and Literary Theory · 519 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Marie Garnier and Patrick Saint-Dizier. 2016. Error Typology and Remediation Strategies for Requirements Written in English by Non-Native Speakers. Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16).

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

@inproceedings{garnier2016error,
  title = {{Error Typology and Remediation Strategies for Requirements Written in English by Non-Native Speakers}},
  author = {Marie Garnier and Patrick Saint-Dizier},
  booktitle = {Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)},
  year = {2016},
  doi = {10.63317/43ptauh329wn},
  url = {https://doi.org/10.63317/43ptauh329wn}
}

References

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

  1. Barcellini F., Albert, C., Saint-Dizier, P. (2012). Risk Analysis and Prevention: LELIE, a Tool dedicated to Procedure and Requirement Authoring. In Proceedings of LREC 2012,p. 698-706. find
  2. Biber, D., Johansson, S., Leech, G., Conrad, S., Finegan, E. (1999). Longman Grammar of Spoken and Written English. Harlow: Pearson Education. find
  3. Fuchs, N.E. (2012). First-Order Reasoning for Attempto Controlled English. In Proceedings of the Second International Workshop on Controlled Natural Language. Berlin:Springer-Verlag. find
  4. Grady, J. O. (2006). System Requirements Analysis. Orlando: Academic Press. find
  5. Garnier, M. (2012). Automatic Correction of Adverb Placement Errors for CALL. In L. Bradley, S. Thouësny (Eds.), CALL: Using, Learning, Knowing, EUROCALL Conference, Gothenburg, Sweden, 22-25 August 2012, Proceedings. Research-Publishing.net, p. 99-103. find
  6. Garnier, M. (2014). Utilisation de méthodes linguistiques pour la détection et la correction automatisées d'erreurs produites par des francophones écrivant en anglais. PhD Thesis in English Linguistics. Université de Toulouse. https://tel.archives-ouvertes.fr/tel-01257640. ↗
  7. Granger S., Dagneaux E., Meunier F. and Paquot M. (2009). The International Corpus of Learner English. Version 2. Handbook and CD-Rom. Louvain-la-Neuve: Presses Universitaires de Louvain. find
  8. Hull,E., Jackson,K. and Dick, J. (2011). Requirements Engineering. 3 rd ed. NY: Springer-Verlag. find
  9. Kuhn, T. (2013). A Principled Approach to Grammars for Controlled Natural Languages and Predictive Editors. Journal of Logic, Language and Information, 22(1). p. 33-70. ↗
  10. Kuhn, T. (2014). A Survey and Classification of Controlled Natural Languages. Computational Linguistics, 40(1). p. 121-170. ↗
  11. Lennon, P. (1991). Error: some problems of definition, identification and distinction. Applied Linguistics, vol. 12, n°2, p 180-196. ↗
  12. Jarvis, S., Pavlenko, A. (2007). Crosslinguistic Influence in Language and Cognition. London, New York: Routledge. ↗
  13. Pastor-Gomez, I. (2011). The Status and Development of N+N Sequences in Contemporary English Noun Phrases. Bern: Peter Lang. ↗
  14. Saint-Dizier, P. (2012) Processing natural language arguments with the TextCoop platform. Journal of Argumentation and Computation, vol 3(1). p. 49-82. find
  15. Saint-Dizier, P. (2014). Challenges of Discourse Processing: the case of technical documents. Cambridge: Cambridge Scholars. find
  16. Saint-Dizier, P. (2015). Features of an error correction memory to enhance technical texts authoring in LELIE. International Journal of Knowledge Content Development and Technology,, vol 5(2). p. 75-101. ↗
  17. Tschichold, C. (1999). Grammar checking for CALL: Strategies for improving foreign language grammar checkers. In K. Cameron (Ed.) CALL: Media, Design and Applications. Lisse: Swets et Zeitlinger. p. 203-222. find
  18. Wyner, A., Angelov, K., Barzdins, G., Damljanovic, D., Davis, B., Fuchs, N., Hoefler, S., Jones, K., Kaljurand, K., Kuhn, T., Luts, M., Pool, J., Rosner, M., Schwitter, R., Sowa, J. (2010). On Controlled Natural Languages: Properties and Prospects. In Proceeding of Controlled Natural Language Conference. Berlin/Heidelberg: Springer. p. 281-289. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/garnier-2016-error-typology-remediation-strategies-requirements/