Elamur Research⌕
☰
home / papers / 2025 / A Platform for Generating Educational Activities to Teach English as a Second Language

A Platform for Generating Educational Activities to Teach English as a Second Language

Paper · 2025 · Aiala Rosa, Santiago Gongora, Juan Pablo Filevich et al.

Summary

Rural schools in Uruguay often lack teachers who speak English well. To help, the authors built a free web platform that uses NLP to create English practice games and exercises for children. They developed it over five years with Uruguay's National Public Education Administration, working alongside teachers, linguists and undergraduate students. It began as a desktop app bundling three games. In 2021 it became a unified, low-bandwidth web hub modeled on Steam.

The platform works in two ways:

  • Pre-curated mode. Games are assembled on demand from shared resources: categorized word lists, definitions and an image bank. These were built semi-automatically and then checked by hand, so children can play right away without review.
  • Text-based mode. A teacher enters a text, and the system derives Q&A exercises, crosswords, word searches and a story-reordering game. The teacher reviews and edits them before students use them.

The team is also testing LLM-based sentence generation and image generation, but these are not yet integrated, mainly because of compute limits. The paper matters as a real-world case of making generative NLP safe for children through curation and teacher review.

Key points

  • The project began with a desktop app combining a crossword generator, a word-search generator and a Battleship-like game (Percovich et al., 2019). Every update required reinstalling it in schools, which motivated the move to the web in 2021.
  • Activities from curated resources:
    • Word-search, Crosswords and Memory
    • the Story game (reorder a story's sentences)
    • multiplayer Sea Animals (Battleship) and Categories
    • multiple-choice drills, often with pictures and sound to help pronunciation practice
  • Privacy design: no personal data is stored, and children don't need to register. Only teachers log in, to edit activities generated from their texts.
  • Experimental sentence generation: a few-shot prompt gives three ⟨word, short sentence⟩ pairs, and the LLM completes a fourth for a vocabulary word. Five candidates are produced, one is picked at random, and an illustration is generated for it.
  • Server migration: moving to a stronger server (still without a GPU) cut some response times from over a minute to a few seconds.

Interesting ideas

  • I1. Limiting generation to a curated, categorized vocabulary acts as a cheap safety filter against unsuitable or hallucinated output, especially for young users.
  • I2. Trust is assigned by where content comes from: material built from hand-curated resources skips review, while material derived from arbitrary input text always gets human checking.
  • I3. Generating several candidates and then selecting one separates quality control from the generator itself.
  • I4. The story-reordering exercise is in effect a test of whether a reader grasps the order of events in a narrative.
  • I5. A minimal few-shot format (three example pairs, complete the fourth) yields short, controlled outputs from an LLM with little prompt engineering.

For Mova

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

  • I1 (back): Generation from the meaning graph could be held to a level-specific curated lexicon, so produced text stays checkable and matches the reader's level.
  • I2 (learn): Silver data could be tiered the same way, accepting items built from vetted lexicons or agreed parses directly and routing free-text-derived items to stricter review.
  • I3 (back): The text → graph → text path could over-generate realizations and keep one that re-parses back into the same graph.
  • I4 (world): Shuffling the sentences of fables and asking Mova to restore the order from its event timeline would give an explainable test of event ordering.
  • I5 (learn): LLM teacher tiers could use this pattern to produce short targeted example sentences for a given lemma or verb class, which the witness parsers then vet.

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 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksInformation Systems · 1,437 works

🧭 Topics 2Research subjects and languages the work is about.English · 2,263 worksGrammatical error correction & learner language · 156 works

Cite

Cite as Aiala Rosa, Santiago Gongora, Juan Pablo Filevich, Ignacio Sastre, Laura Musto, Brian Carpenter and Luis Chiruzzo. 2025. A Platform for Generating Educational Activities to Teach English as a Second Language. arXiv:2504.20251.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2504.20251 · arXiv 2504.20251

@misc{rosa2025platform,
  title = {{A Platform for Generating Educational Activities to Teach English as a Second Language}},
  author = {Aiala Rosa, Santiago Gongora, Juan Pablo Filevich, Ignacio Sastre, Laura Musto, Brian Carpenter and Luis Chiruzzo},
  year = {2025},
  doi = {10.48550/arxiv.2504.20251},
  eprint = {2504.20251},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2504.20251}
}

References

34 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Manex Agirrezabal, Begoña Altuna, Lara Gil Vallejo, Josu Goikoetxea, and Itziar Gonzalez Dios. 2019. Creating vocabulary exercises through nlp. Digital Humanities in the Nordic Countries. Proceedings, 2019. ↗
  2. David Alfter, Lars Borin, Ildikó Pilán, Therese Lindström Tiedemann, and Elena Volodina. 2019. Lärka: From language learning platform to infrastructure for research on language learning. In Selected papers from the CLARIN Annual Conference 2018, Pisa, 8-10 October 2018, 159, pages 1–14. Linköping University Electronic Press. find
  3. Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. 2021. On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’21, page 610–623, New York, NY, USA. Association for Computing Machinery. ↗
  4. Emily M. Bender and Alexander Koller. 2020. Climbing towards NLU: On meaning, form, and understanding in the age of data. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5185–5198, Online. Association for Computational Linguistics. ↗
  5. Gonzalo Berger, Tatiana Rischewski, Luis Chiruzzo, and Aiala Rosá. 2022. Generation of english question answer exercises from texts using transformers based models. In 2022 IEEE Latin American Conference on Computational Intelligence (LA-CCI). IEEE. find
  6. Samuel R. Bowman. 2023. Eight things to know about large language models. ↗
  7. Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language models are few-shot learners. ↗
  8. Chi-Jen Lin and Husni Mubarok. 2021. Learning analytics for investigating the mind map-guided ai chatbot approach in an efl flipped speaking classroom. Educational Technology & Society, 24(4):16–35. find
  9. Christopher Bryant, Mariano Felice, Øistein E. Andersen, and Ted Briscoe. 2019. The BEA-2019 shared task on grammatical error correction. In Proceedings of the Fourteenth Workshop on Innovative Use of NLP for Building Educational Applications, pages 52–75, Florence, Italy. Association for Computational Linguistics.
  10. Martín Morón, Joaquín Scocozza, Luis Chiruzzo, and Aiala Rosá. 2021. A tool for automatic question generation for teaching english to beginner students. In 2021 40th International Conference of the Chilean Computer Science Society (SCCC), pages 1–5. IEEE. find
  11. J. Burstein, J. Sabatini, J. Shore, B. Moulder, and J. Lentini. 2013. A user study: Technology to increase teachers’ linguistic awareness to improve instructional language support for english language learners. In Proceedings of the 2nd Workshop on Natural Language Processing for Improving Textual Accessibility (NLP4ITA), Atlanta, Georgia. Association for Computational Linguistics.
  12. Luis Chiruzzo, Laura Musto, Santiago Góngora, Brian Carpenter, Juan Filevich, and Aiala Rosá. 2022. Using nlp to support english teaching in rural schools. In Proceedings of the Second Workshop on NLP for Positive Impact (NLP4PI), pages 113–121.
  13. Mirjam De Haas, Paul Vogt, and Emiel Krahmer. 2020. The effects of feedback on children’s engagement and learning outcomes in robot-assisted second language learning. Frontiers in Robotics and AI, 7:101. ↗
  14. Alena Fenogenova and Elizaveta Kuzmenko. 2016. Automatic generation of lexical exercises. In Proceedings of the Workshop on Computational Linguistics and Language Science. find
  15. Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Yejin Bang, Andrea Madotto, and Pascale Fung. 2022. Survey of hallucination in natural language generation. CoRR, abs/2202.03629. ↗
  16. Anna Jordanous. 2022. Should we pursue SOTA in computational creativity? In Proceedings of the 13th International Conference on Computational Creativity, Bozen-Bolzano, Italy, June 27 - July 1, 2022, pages 159–163. Association for Computational Creativity (ACC). find
  17. Anisia Katinskaia, Javad Nouri, and Roman Yangarber. 2018. Revita: a language-learning platform at the intersection of its and call. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japón. European Language Resources Association (ELRA). find
  18. Bing Li and Miaomiao Peng. 2022. Integration of an ai-based platform and flipped classroom instructional model. ↗
  19. Scientific Programming, 2022. find
  20. Diane Litman. 2016. Natural language processing for enhancing teaching and learning. In Thirtieth AAAI conference on artificial intelligence. find
  21. Hwee Tou Ng, Siew Mei Wu, Ted Briscoe, Christian Hadiwinoto, Raymond Hendy Susanto, and Christopher Bryant. 2014. The CoNLL-2014 shared task on grammatical error correction. In Proceedings of the Eighteenth Conference on Computational Natural Language Learning: Shared Task, pages 1–14, Baltimore, Maryland. Association for Computational Linguistics. find
  22. Analía Percovich, Alejandro Tosi, Luis Chiruzzo, and Aiala Rosá. 2019. Ludic applications for language teaching support using natural language processing. In 2019 38th International Conference of the Chilean Computer Science Society (SCCC), pages 1–7. find
  23. Naiara Perez and Montse Cuadros. 2017. Multilingual CALL framework for automatic language exercise generation from free text. In Proceedings of the Software Demonstrations of the 15th Conference of the European Chapter of the Association for Computational Linguistics, pages 49–52, Valencia, Spain. Association for Computational Linguistics. find
  24. Richard Heinz Patrick Prager. 2019. Exploring the use of role-playing games in education. The MT Review. find
  25. Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. find
  26. Inioluwa Deborah Raji, Emily M. Bender, Amandalynne Paullada, Emily Denton, and Alex Hanna. 2021. Ai and the everything in the whole wide world benchmark. ↗
  27. Pranav Rajpurkar, Robin Jia, and Percy Liang. 2018. Know what you don’t know: Unanswerable questions for squad.
  28. Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2021. High-resolution image synthesis with latent diffusion models. ↗
  29. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. 2023. Llama: Open and efficient foundation language models. ↗
  30. Adam Trischler, Tong Wang, Xingdi Yuan, Justin Harris, Alessandro Sordoni, Philip Bachman, and Kaheer Suleman. 2016. NewsQA: A machine comprehension dataset. arXiv preprint arXiv:1611.09830. ↗
  31. Rianne Van den Berghe, Ora Oudgenoeg-Paz, Josje Verhagen, Susanne Brouwer, Mirjam De Haas, Jan De Wit, Bram Willemsen, Paul Vogt, Emiel Krahmer, and Paul Leseman. 2021. Individual differences in children’s (language) learning skills moderate effects of robot-assisted second language learning. Frontiers in Robotics and AI, page 259. ↗
  32. Elena Volodina, Lars Borin, and Ildikó Pilán, editors. 2014. Proceedings of the third workshop on NLP for computer-assisted language learning. LiU Electronic Press, Uppsala, Sweden. find
  33. Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, and Quoc V. Le. 2022. Finetuned language models are zero-shot learners. ↗
  34. Hongzhi Yang and Suna Kyun. 2022. The current research trend of artificial intelligence in language learning: A systematic empirical literature review from an activity theory perspective. Australasian Journal of Educational Technology, 38(5):180–210. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/rosa-2025-platform-generating-educational-activities-teach/