A diffusion-based model of language learning and interlingual distance

Understanding the process of language learning and quantifying interlingual relationships are central challenges in linguistics, cognitive science, and language education. In this paper, we propose a novel framework that models second language acquisition as a diffusion process within a structured, multidimensional space of languages. We introduce a formal measure of interlingual distance, grounded in linguistic features, to quantify structural and functional differences between languages. Building on Barenblatt-type nonlinear diffusion models, we represent language learning as a multicontinua diffusion process, where distinct components of language – such as phonetics, grammar, vocabulary, and pragmatics – are treated as separate, interacting continua. Each continuum evolves independently according to its own diffusion dynamics, capturing the heterogeneous difficulty and pace of learning across linguistic subsystems. The interaction between these continua reflects the coupling between linguistic competencies in real-world acquisition. We can validate this model with empirical data on second language learning rates across various language pairs, demonstrating that diffusion distances in each continuum correlate with observed learning difficulties in the corresponding language domain. This approach not only offers a new theoretical lens on language learning but also provides a predictive framework for curriculum design, learner modeling, and applications in multilingual NLP and AI systems.

1. Dörnyei Z. The Psychology of Second Language Acquisition. Oxford: Oxford University Press. 2009.

2. Ellis NC. Selective attention and transfer phenomena in SLA: Contingency, cue competition, salience, interference, overshadowing, blocking, and perceptual learning. Applied Linguistics. 2006;27(2):164-194. DOI: 10.1093/applin/aml015

3. Vabishchevich PN, Grigoriev AV. Numerical modeling of fluid flow in anisotropic fractured porous media. Numerical Analysis and Applications. 2016;9(1):45-56. DOI: 10.1134/S1995423916010055

4. Grigorev AV, Vabishchevich PN. Two-level approach for numerical modeling of blood flow in the liver lobule. Journal of Numerical Analysis, Industrial and Applied Mathematics. 2022;16(1-2):15-28.

5. Barenblatt GI, Zheltov IP, Kochina IN. Basic concepts in the theory of seepage of homogeneous liquids in fissured rocks. Journal of Applied Mathematics and Mechanics. 1960;24(5):1286-1303. DOI: 10.1016/0021-8928(60)90107-6

6. Rabinovich M, Ordan N, Wintner S. Found in translation: Reconstructing phylogenetic language trees from translated texts. Transactions of the Association for Computational Linguistics. 2017;5:169-182. DOI: 10.1162/tacl_a_00058

7. Devlin J, Chang MW, Lee K, Toutanova K. BERT: Pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT. 2019;4171-4186. DOI: 10.18653/v1/N19-1423

8. Artetxe M, Schwenk H. Massively multilingual sentence embeddings for zero-shot cross-lingual transfer and beyond. Transactions of the Association for Computational Linguistics. 2019;7:597-610. DOI: 10.1162/tacl_a_00288

9. Conneau A, Khandelwal K, Goyal N, et al. Unsupervised cross-lingual representation learning at scale. Proceedings of ACL. 2020;8440-8451. DOI: 10.18653/v1/2020.acl-main.747

10. Lake BM, Baroni M. Generalization without systematicity: On the compositional skills of sequence-to-sequence recurrent networks. Proceedings of ICML. 2018;80:2873-2882. Available at: https://proceedings.mlr.press/v80/lake18a. html