Elamur Research⌕
☰
home / papers / 2016 / Mental Lexicon Growth Modelling Reveals the Multiplexity of the English Language

Mental Lexicon Growth Modelling Reveals the Multiplexity of the English Language

Paper · 2016 · Massimo Stella and Markus Brede

Summary

The authors ask whether the order in which children acquire English words is better explained by one kind of word relation or by several kinds together. They model the mental lexicon as a three-layer multiplex network. The same 4731 English words appear on every layer, and each layer uses a different link type: free associations from the Edinburgh Associative Thesaurus, synonyms from WordNet 3.0, and phonological similarity computed from IPA transcriptions that were checked by hand.

A minimal growth model builds the lexicon one word at a time.

  • Candidate words are proposed in some order, such as word length, frequency, semantic degree or multiplex features.
  • A candidate with no phonological link to the words already present is rejected with probability f.

The model's assembly times are compared with empirical age-of-acquisition data using interquartile-range overlap. The orderings that use both phonological and semantic degree match real acquisition best. Orderings based on frequency alone or on single layers fall short, especially when f is high. This supports the view that word learning draws on several levels of linguistic structure at once.

Key points

  • The multiplex has N = 4731 words in three unweighted, undirected layers: EAT free associations, WordNet 3.0 synonyms, and phonological neighbours that differ by one phoneme.
  • Representativeness was checked against Opensubtitles frequencies, which cover more than 1.4·10^8 tokens.
    • The sample over-represents frequent words, and its frequency tail follows a power law with exponent about 1.83.
    • An average word has 4.78 phonemes and 5.39 letters.
  • The layers have different degree distributions:
    • the association layer has a heavy tail with many hubs;
    • the phonological layer cuts off near k ≈ 30;
    • the synonym layer is roughly exponential.
  • All layers are small-world compared with configuration models. The phonological diameter is almost three times its mean path length.
  • The rejection probability f separates two regimes.
    • At low f, frequency or semantic-degree orderings fit acquisition data reasonably well. The authors read this as lexical learning.
    • At high f, only orderings that combine phonological and semantic information match. The authors read this as phonological learning.

Interesting ideas

  • I1. Store several relation types (sound, synonymy, association) over one shared node set and score a node by summing its degree across layers; this shows centrality that no single graph reveals.
  • I2. Compare each real network with a configuration model that has the same degree distribution, so that structure is not credited to degree alone.
  • I3. Judge a model by how much its interquartile range overlaps the empirical one rather than by mean error; this suits small, skewed distributions.
  • I4. Before trusting a lexical resource, check its word-length and frequency profile against a large corpus of everyday usage.
  • I5. Grow a structure item by item under different insertion orders, and compare the resulting timelines with real acquisition data to see which features drive learning.

For Mova

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

  • I1 (world): The WordNet/FrameNet/Wikidata knowledge graph could be kept as aligned layers over shared lemmas, with cross-layer degree used to rank which senses or entities are central.
  • I2 (learn): This is a ready-made negative control for any graph statistic Mova reports, such as frame co-occurrence or coreference chain structure.
  • I3 (learn): Evaluation of silver data or self-play outputs could test whether the distributions of parse depth or answer timing stay within the reference IQR, not just whether averages match.
  • I4 (tokens): The lemma lexicon in crate en could be audited against subtitle-style frequencies to catch gaps in common words before tagger or morphology training.
  • I5 (learn): Training curricula could be ordered by combined frequency and graph degree, then compared against alternative orderings to find which order best supports learning.

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

🎓University of SouthamptonSouthampton · United Kingdom4 works · 2,409 citations

📚 Shelves 8The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 worksSocial Sciences · 744 worksPsychology · 798 worksDevelopmental and Educational Psychology · 260 worksPhysics and Astronomy · 25 worksStatistical and Nonlinear Physics · 22 worksCultural Studies · 10 works

🧭 Topics 1Research subjects and languages the work is about.English · 1,373 works

Cite

Cite as Massimo Stella and Markus Brede. 2016. Mental Lexicon Growth Modelling Reveals the Multiplexity of the English Language. arXiv:1604.01243.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.1007/978-3-319-30569-1_20 · arXiv 1604.01243

@misc{stella2016mental,
  title = {{Mental Lexicon Growth Modelling Reveals the Multiplexity of the English Language}},
  author = {Massimo Stella and Markus Brede},
  year = {2016},
  doi = {10.1007/978-3-319-30569-1\_20},
  eprint = {1604.01243},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.1007/978-3-319-30569-1_20}
}

References

27 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jean Aitchison. Words in the mind: An introduction to the mental lexicon. Wiley, 2012. find
  2. Adrien Barbaresi. Language-classified Open Subtitles (LACLOS). PhD thesis, BBAW, 2014. find
  3. Andrea Baronchelli, Ramon Ferrer-i Cancho, Romualdo Pastor-Satorras, Nick Chater, and Morten H Christiansen. Networks in cognitive science. Trends in cognitive sciences, 17(7):348–360, 2013. ↗
  4. Federico Battiston, Vincenzo Nicosia, and Vito Latora. Structural measures for multiplex networks. Physical Review E, 89(3):032804, 2014. ↗
  5. N. M Beckage, A Aguilar, and E Colunga. Modeling lexical acquisition through networks. In Proceedings of CogSci2015, 2015. find
  6. Nicole M Beckage and Eliana Colunga. Language networks as models of cognition: Understanding cognition through language. In A Mehler, P Blanchard, B Job, and S Banisch, editors, Towards a Theoretical Framework of Analyzing Complex Linguistic Networks. Springer, 2015. ↗
  7. Stefano Boccaletti, G Bianconi, R Criado, Charo I Del Genio, J Gómez-Gardeñes, M Romance, I Sendina-Nadal, Z Wang, and M Zanin. The structure and dynamics of multilayer networks. Physics Reports, 544(1):1–122, 2014. ↗
  8. Tom Brondsted. http://tom.brondsted.dk/text2phoneme/, 2008. ↗
  9. Allan M Collins and Elizabeth F Loftus. A spreading-activation theory of semantic processing. Psychological review, 82(6):407, 1975. ↗
  10. Simon De Deyne and Gert Storms. Word associations: Network and semantic properties. Behavior Research Methods, 40(1):213–231, 2008. ↗
  11. Charles A Ferguson and Carol B Farwell. Words and sounds in early language acquisition. Language, pages 419–439, 1975. ↗
  12. Erika Hoff, Cynthia Core, and Kelly Bridges. Non-word repetition assesses phonological memory and is related to vocabulary development in 20-to 24-month-olds. Journal of Child Language, 35(04):903–916, 2008. ↗
  13. GR Kiss, Christine A Armstrong, and Robert Milroy. An associative thesaurus of english. Medical Research Council, 1972. find
  14. Victor Kuperman, Hans Stadthagen-Gonzalez, and Marc Brysbaert. Age-of-acquisition ratings for 30,000 english words. Behavior Research Methods, 44(4):978–990, 2012. ↗
  15. George Miller, Christiane Fellbaum, Randee Tengi, P Wakefield, H Langone, and BR Haskell. WordNet. MIT Press Cambridge, 1998. find
  16. David S Moore and George P McCabe. Introduction to the Practice of Statistics. WH Freeman/Times Books/Henry Holt & Co, 1989. find
  17. Adilson E Motter, Alessandro PS de Moura, Ying-Cheng Lai, and Partha Dasgupta. Topology of the conceptual network of language. Physical Review E, 65(6):065102, 2002. ↗
  18. Mark Newman. Networks: an introduction. Oxford University Press, 2010. ↗
  19. Richard G Schwartz and Laurence B Leonard. Do children pick and choose? an examination of phonological selection and avoidance in early lexical acquisition. Journal of child language, 9(02):319–336, 1982. ↗
  20. Mariano Sigman and Guillermo A Cecchi. Global organization of the wordnet lexicon. PNAS, 99(3):1742–1747, 2002. ↗
  21. Massimo Stella and Markus Brede. Investigating the phonetic organisation of the English language via phonological networks, percolation and Markov models. Accepted in Lecture Notes in Computer Science, 2015. ↗
  22. Massimo Stella and Markus Brede. Patterns in the english language: Phonological networks, percolation and assembly models. Journal of Statistical Mechanics, 2015:P05006, 2015. ↗
  23. Mark Steyvers and Joshua B Tenenbaum. The large-scale structure of semantic networks: Statistical analyses and a model of semantic growth. Cognitive science, 29(1):41–78, 2005. ↗
  24. Carol Stoel-Gammon and Judith A Cooper. Patterns of early lexical and phonological development. Journal of child language, 11(02):247–271, 1984. ↗
  25. Michael S Vitevitch. What can graph theory tell us about word learning and lexical retrieval? Journal of Speech, Language, and Hearing Research, 51(2):408– 422, 2008. ↗
  26. Michael S Vitevitch, Kit Ying Chan, and Rutherford Goldstein. Insights into failed lexical retrieval from network science. Cognitive psychology, 68:1–32, 2014. ↗
  27. Fernanda Marafiga Wiethan, Letı́cia Arruda Nóro, and Helena Bolli Mota. Early lexical and phonological acquisition and its relationships. In CoDAS, number 4, pages 260–264. SciELO Brasil, 2014. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/stella-2016-mental-lexicon-growth-modelling-reveals/