Elamur Research⌕
☰
home / papers / 2025 / Constructions All the Way Up: From Sensory Experiences to Construction Grammars

Constructions All the Way Up: From Sensory Experiences to Construction Grammars

Summary

The authors test a core tenet of construction grammar on a computational model. The tenet is that all linguistic knowledge, from concrete words to abstract schemas, consists of form–meaning pairings that humans acquire through situated communication. In the model, a learner agent builds a grammar directly from sensory data. It has no predefined lexicon, ontology or pre-specified categories. It learns within the language game paradigm by communicating with a tutor agent. The model combines two experiments that previously existed separately. In referential games, the learner associates words with perceptual concepts. In a question–answer game, it builds on these words to acquire partially schematic constructions whose meanings are specified by procedural semantics. Learning is organized as curriculum learning, with interactions that gradually increase in complexity. To connect the two parts, the authors added a component that autonomously infers a category hierarchy from the links between construction slots and the words that fill them. In the synthetic CLEVR environment, the learner answers questions from a subset of the data almost without errors. The work offers a mechanistic alternative to LLMs: here, abstract constructions emerge from experience rather than from next-token prediction.

Key points

  • The model stores each concept as a set of weighted Gaussian distributions over sensory features. For example, in the concept for "red", all the weight falls on the RGB channels. The distributions and weights are refined each time the tutor provides the correct answer.
  • The meaning of a question is represented as a program composed of primitive cognitive operations, such as OBSERVE-SCENE, FILTER and COUNT. The learner acquires new constructions through two mechanisms: intention reading (an abductive search for a program that yields the correct answer) and pattern finding.
  • The grammar contains both fully substantive lexical constructions and item-based constructions such as HOW-MANY-?X-ARE-THERE-CXN. Thanks to the categorial network, the learner can combine known constructions in combinations it has not seen before.
  • The new component infers a single-layer category hierarchy. It replaces the predefined hierarchy from the model of Nevens et al. (2022). The following hierarchy emerged on its own: shapes (separately in the singular and the plural), colors, materials and sizes.
  • De Vos et al. (2024), using a hand-written grammar, achieved 96% on the full CLEVR dataset. This model reached nearly 100%, but on a subset, so the authors consider a direct comparison invalid.

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: Linguistics & languages

Cite

Cite as Jérôme Botoko Ekila, Lara Verheyen, Katrien Beuls and Paul Van Eecke. 2025. Constructions All the Way Up: From Sensory Experiences to Construction Grammars. Proceedings of the Second International Workshop on Construction Grammars and NLP.

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

@inproceedings{ekila2025constructions,
  title = {{Constructions All the Way Up: From Sensory Experiences to Construction Grammars}},
  author = {Jérôme Botoko Ekila, Lara Verheyen, Katrien Beuls and Paul {Van Eecke}},
  booktitle = {Proceedings of the Second International Workshop on Construction Grammars and NLP},
  year = {2025},
  url = {https://aclanthology.org/2025.cxgsnlp-1.9/}
}

References

66 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Omri Abend, Tom Kwiatkowski, Nathaniel J. Smith, Sharon Goldwater, and Mark Steedman. 2017. Bootstrapping language acquisition. Cognition, 164:116– 143. find
  2. Afra Alishahi and Suzanne Stevenson. 2008. A computational model of early argument structure acquisition. Cognitive Science, 32(5):789–834. find
  3. Yoav Artzi and Luke Zettlemoyer. 2013. Weakly supervised learning of semantic parsers for mapping instructions to actions. Transactions of the Association for Computational Linguistics, 1:49–62. our card
  4. Barend Beekhuizen. 2015. Constructions Emerging. Ph.D. thesis, Universiteit Leiden. find
  5. Barend Beekhuizen and Rens Bod. 2014. Automating construction work: Data-oriented parsing and onstructivist accounts of language acquisition. In Ronny Boogaart, Timothy Colleman, and Gijsbert Rutten, editors, Extending the Scope of Construction Grammar, pages 47–74. De Gruyter Mouton, Berlin, German… find
  6. Heike Behrens. 2009. Usage-based and emergentist approaches to language acquisition. Linguistics, 47(2):383–411. find
  7. Liesbet De Vos, Jens Nevens, Paul Van Eecke, and Katrien Beuls. 2024. Construction grammar and procedural semantics for human-interpretable grounded language processing. Linguistics Vanguard, 10(2):565–574. find
  8. Katrien Beuls, Kateryna Gerasymova, and Remi van Trijp. 2010. Situated learning through the use of language games. In Proceedings of the 19th Annual Machine Learning Conference of Belgium and The Netherlands (BeNeLearn), pages 1–6. find
  9. Holger Diessel. 2017. Usage-based linguistics. In Mark Aronoff, editor, Oxford Research Encyclopedia of Linguistics. Oxford University Press, Oxford, United Kingdom. find
  10. Katrien Beuls and Sebastian Höfer. 2011. Simulating the emergence of grammatical agreement in multi-agent language games. In Proceedings of the Twenty-Second International Joint Conference on Artificial Intelligence, pages 61–66, Washington, D.C., USA. AAAI Press. find
  11. Peter Ford Dominey. 2005a. Emergence of grammatical constructions: Evidence from simulation and grounded agent experiments. Connection Science, 17(3-4):289–306. find
  12. Katrien Beuls and Paul Van Eecke. 2024. Humans learn language from situated communicative interactions. What about machines? Computational Linguistics, 50(4):1277–1311. find
  13. Peter Ford Dominey. 2005b. From sensorimotor sequence to grammatical construction: Evidence from simulation and neurophysiology. Adaptive Behavior, 13(4):347–361. find
  14. Katrien Beuls and Paul Van Eecke. 2025. Construction grammar and artificial intelligence. In Mirjam Fried and Kiki Nikiforidou, editors, The Cambridge Handbook of Construction Grammar, pages 543– 571. Cambridge University Press, Cambridge, United Kingdom. find
  15. Peter Ford Dominey and Jean-David Boucher. 2005. Learning to talk about events from narrated video in a construction grammar framework. Artificial Intelligence, 167(1):31–61. find
  16. Jonas Doumen, Katrien Beuls, and Paul Van Eecke. 2024. Modelling constructivist language acquisition through syntactico-semantic pattern finding. Royal Society Open Science, 11(7):231998. find
  17. Claire Bonial and Harish Tayyar Madabushi. 2024. Constructing understanding: on the constructional information encoded in large language models. Language Resources and Evaluation. find
  18. Jonas Doumen, Veronica J. Schmalz, Katrien Beuls, and Paul Van Eecke. 2025. The computational learning of construction grammars: State of the art and prospective roadmap. Constructions and Frames, 17(1):141–174. find
  19. Jérôme Botoko Ekila, Jens Nevens, Lara Verheyen, Katrien Beuls, and Paul Van Eecke. 2024. Decentralised emergence of robust and adaptive linguistic conventions in populations of autonomous agents grounded in continuous worlds. In Proceedings of the 23rd International Conference on Autonomous Agents… find
  20. Ewa D abrowska and Elena Lieven. 2005. Towards a lexically specific grammar of children’s question constructions. Cognitive Linguistics, 16(3):437–474. find
  21. Charles J. Fillmore. 1988. The mechanisms of “construction grammar”. In Annual Meeting of the Berkeley Linguistics Society, volume 14, pages 35–55. find
  22. Joan Bybee. 2010. Language, usage and cognition. Cambridge University Press, Cambridge, United Kingdom. find
  23. Emília Garcia Casademont and Luc Steels. 2015. Usage-based grammar learning as insight problem solving. In Proceedings of the EuroAsianPacific Joint Conference on Cognitive Science, pages 258–263. find
  24. Nancy Chang. 2008. Constructing grammar: A computational model of the emergence of early constructions. Ph.D. thesis, University of California, Berkeley, Berkeley, CA, USA. find
  25. Emília Garcia Casademont and Luc Steels. 2016. Insight grammar learning. Journal of Cognitive Science, 17(1):27–62. find
  26. David L. Chen and Raymond J. Mooney. 2008. Learning to sportscast: a test of grounded language acquisition. In Proceedings of the 25th International Conference on Machine learning, pages 128–135. find
  27. Judith Gaspers and Philipp Cimiano. 2012. A usage-based model for the online induction of constructions from phoneme sequences. In Proceedings of the 2012 IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL), pages 1–6. IEEE. find
  28. William Croft. 2001. Radical construction grammar: Syntactic theory in typological perspective. Oxford University Press, Oxford, United Kingdom. find
  29. Joachim De Beule and Benjamin K. Bergen. 2006. On the emergence of compositionality. In The Evolution of Language. Proceedings of the 6th International Conference (EVOLANG6), pages 35–42, Singapore, Singapore. World Scientific. find
  30. Judith Gaspers and Philipp Cimiano. 2014. A computational model for the item-based induction of construction networks. Cognitive Science, 38(3):439–488. find
  31. Judith Gaspers, Philipp Cimiano, Sascha S. Griffiths, and Britta Wrede. 2011. An unsupervised algorithm for the induction of constructions. In Proceedings of the 2011 IEEE International Conference on Development and Learning (ICDL), volume 2, pages 1–6. IEEE. find
  32. Tom Kwiatkowski, Luke Zettlemoyer, Sharon Goldwater, and Mark Steedman. 2011. Lexical generalization in CCG grammar induction for semantic parsing. In Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, pages 1512–1523. Association for Computational Linguistics. our card
  33. Judith Gaspers, Philipp Cimiano, Katharina Rohlfing, and Britta Wrede. 2017. Constructing a language from scratch: Combining bottom–up and top–down learning processes in a computational model of language acquisition. IEEE Transactions on Cognitive and Developmental Systems, 9(2):183–196. find
  34. George Lakoff. 1987. Women, fire, and dangerous things: What categories reveal about the mind. University of Chicago Press. find
  35. Ronald W. Langacker. 1987. Foundations of cognitive grammar: Theoretical prerequisites, volume 1. Stanford University Press, Stanford, CA, USA. find
  36. Kateryna Gerasymova and Michael Spranger. 2010. Acquisition of grammar in autonomous artificial systems. In Proceedings of the 19th European Conference on Artificial Intelligence (ECAI-2010), pages 923–928. IOS Press. find
  37. Elena Lieven. 2014. First language learning from a usage-based approach. In Thomas Herbst, Hans-Jörg Schmid, and Susen Faulhaber, editors, Constructions Collocations Patterns, pages 9–32. De Gruyter Mouton, Berlin, Germany. find
  38. Kateryna Gerasymova and Michael Spranger. 2012. An experiment in temporal language learning. In Luc Steels and Manfred Hild, editors, Language Grounding in Robots, pages 237–254. Springer, New York, NY, USA. find
  39. Martin Loetzsch. 2015. Lexicon formation in autonomous robots. Ph.D. thesis, Humboldt-Universität zu Berlin, Berlin, Germany. find
  40. Jens Nevens. 2022. Representing and learning linguistic structures on the conceptual, morphosyntactic, and semantic level. Ph.D. thesis, Vrije Universiteit Brussel, Brussels: VUB Press. find
  41. Adele E. Goldberg. 1995. Constructions: A construction grammar approach to argument structure. University of Chicago Press, Chicago, IL, USA. find
  42. Jens Nevens, Jonas Doumen, Paul Van Eecke, and Katrien Beuls. 2022. Language acquisition through intention reading and pattern finding. In Proceedings of the 29th International Conference on Computational Linguistics, pages 15–25, Gyeongju, Republic of Korea. find
  43. Adele E. Goldberg. 2003. Constructions: A new theoretical approach to language. Trends in Cognitive Sciences, 7(5):219–224. find
  44. Justin Johnson, Bharath Hariharan, Laurens van der Maaten, Li Fei-Fei, C. Lawrence Zitnick, and Ross Girshick. 2017. CLEVR: A diagnostic dataset for compositional language and elementary visual reasoning. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2901– 2910. our card
  45. Jens Nevens, Paul Van Eecke, and Katrien Beuls. 2020. From continuous observations to symbolic concepts: A discrimination-based strategy for grounded concept learning. Frontiers in Robotics and AI, 7(84). find
  46. Eleanor Rosch, Carolyn B. Mervis, Wayne D. Gray, David M. Johnson, and Penny Boyes-Braem. 1976. Basic objects in natural categories. Cognitive Psychology, 8(3):382–439. find
  47. Frédéric Kaplan, Luc Steels, and Angus McIntyre. 1998. An architecture for evolving robust shared communication systems in noisy environments. Technical report, Sony CSL Paris. find
  48. Joshua Rozner, Leonie Weissweiler, Kyle Mahowald, and Cory Shain. 2025. Constructions are revealed in word distributions. arXiv preprint arXiv:2503.06048. find
  49. Tom Kwiatkowski, Sharon Goldwater, Luke Zettlemoyer, and Mark Steedman. 2012. A probabilistic model of syntactic and semantic acquisition from child-directed utterances and their meanings. In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics,… our card
  50. Gerard Salton and Michael J. McGill. 1983. Introduction to Modern Information Retrieval. McGraw Hill. find
  51. Michael Spranger. 2015. Incremental grounded language learning in robot-robot interactions: Examples from spatial language. In 2015 Joint IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob), pages 196–201. IEEE. find
  52. Tom Kwiatkowski, Luke Zettlemoyer, Sharon Goldwater, and Mark Steedman. 2010. Inducing probabilistic CCG grammars from logical form with higher-order unification. In Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing, pages 1223–1233. Association for Computational… find
  53. Michael Spranger. 2017. Usage-based grounded construction learning: A computational model. In The 2017 AAAI Spring Symposium Series, pages 245–250, Washington, D.C., USA. AAAI Press. find
  54. Michael Spranger and Luc Steels. 2015. Co-acquisition of syntax and semantics: an investigation in spatial language. In Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence, pages 1909–1915, Washington, D.C., USA. AAAI Press. find
  55. Luc Steels. 1995. A self-organizing spatial vocabulary. Artificial Life, 2(3):319–332. find
  56. Luc Steels. 1999. The Talking Heads experiment: Volume I. Words and Meanings. Best of Publishing, Brussels, Belgium. find
  57. Luc Steels. 2004. Constructivist development of grounded construction grammar. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics (ACL-04), pages 9–16. Association for Computational Linguistics. our card
  58. Luc Steels and Emília Garcia Casademont. 2015. Ambiguity and the origins of syntax. The Linguistic Review, 32(1):37–60. find
  59. Luc Steels, Paul Van Eecke, and Katrien Beuls. 2022. Usage-based learning of grammatical categories. arXiv preprint arXiv:2204.10201. — [arXiv:2204.10201](https://arxiv.org/abs/2204.10201) find
  60. Michael Tomasello. 2003. Constructing a Language: A Usage-Based Theory of Language Acquisition. Harvard University Press, Harvard, MA, USA. find
  61. Paul Van Eecke. 2018. Generalisation and specialisation operators for computational construction grammar and their application in evolutionary linguistics Research. Ph.D. thesis, Vrije Universiteit Brussel, Brussels: VUB Press. find
  62. Remi van Trijp. 2008. The emergence of semantic roles in fluid construction grammar. In The Evolution of Language, pages 346–353. World Scientific. find
  63. Leonie Weissweiler, Valentin Hofmann, Abdullatif Köksal, and Hinrich Schütze. 2022. The better your syntax, the better your semantics? Probing pretrained language models for the English comparative correlative. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing… find
  64. Terry Winograd. 1972. Understanding natural language. Cognitive Psychology, 3(1):1–191. find
  65. William A. Woods. 1968. Procedural semantics for a question-answering machine. In Proceedings of the December 9-11, 1968, Fall Joint Computer Conference, Part I, pages 457–471, New York, NY, USA. find
  66. Shijia Zhou, Leonie Weissweiler, Taiqi He, Hinrich Schütze, David R. Mortensen, and Lori Levin. 2024. Constructions are so difficult that even large language models get them right for the wrong reasons. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language… find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ekila-2025-constructions-all-way-up-sensory/