A Language-agnostic Model of Child Language Acquisition

What it is about

The authors port Abend et al.'s semantic-bootstrapping learner from English to Hebrew. The learner sees one child-directed utterance at a time, paired with a logical form, and learns a CCG lexicon with EM over Dirichlet-process counts. The Hebrew data are from the CHILDES Hagar corpus (3,295 utterances), converted from UD annotation. The English Adam corpus (5,320 utterances) is the control. The learner gets word order right in both languages: SVO is favoured in both, faster and more clearly in English. Word meanings are right for the top 50 words in both (100%), while syntactic category agrees for 76% in English and 46% in Hebrew. With two distractor logical forms, Hebrew fails to settle on SVO. The gap is explained by richer morphology (more unseen word forms), and by copula-less Hebrew sentences: 73.9% of Hagar utterances are under three words, against 15.2% in Adam.

Ideas

• I1. A word-order preference measure over UD trees: for each sentence with a root verb, subject and object, count the six orders of S, V and O, normalise, and plot the preference over a corpus, as the paper does for its six scores. Run it on treebanks/he/ and on an English treebank to compare the two.
• I2. A learning order that uses only evidence from sentences whose words are all already known (the "critical examples" of the paper). Use it as a curriculum in world: learn a clause-order rule only from sentences with no unseen words. Measure how many examples qualify in each language.
• I3. A flag for copula-less clauses: count clauses with no verb and no copula in each treebank, and mark them as non-evidence for word order before a verb-based rule runs.

For Mova

I1 is a measure in 7. Syntax and in world. I2 is a learning-order rule for world. I3 goes into 7. Syntax (copula-less predication) and Nouns, adjectives, pronouns and numerals for the nominal clause.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. arXiv 2408.12254

@misc{mahon2024language,
  title = {{A Language-agnostic Model of Child Language Acquisition}},
  author = {Mahon, Louis and Abend, Omri and Berger, Uri and Demuth, Katherine and Johnson, Mark and Steedman, Mark},
  year = {2024},
  eprint = {2408.12254},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2408.12254}
}

References

23 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Abend, O., Kwiatkowski, T., Smith, N.J., Goldwater, S., Steedman, M., 2017. Bootstrapping language acquisition. Cognition 164, 116–143. find
2. Ambridge, B., 2020. Against stored abstractions: A radical exemplar model of language acquisition. First Language 40, 509–559. find
3. Bowerman, M., 1973. Structural relationships in children’s utterances: Syntactic or semantic?, in: Moore, T. (Ed.), Cognitive Development and the Acquisition of Language. Academic Press, pp. 197–213. find
4. Brown, R., 1973. A First Language: the Early Stages. Harvard University Press, Cambridge, MA. find
5. Chater, N., Christiansen, M.H., 2018. Language acquisition as skill learning. Current opinion in behavioral sciences 21, 205–208. find
6. Chomsky, N., 1981. Lectures on Government and Binding. Foris, Dordrecht. find
7. Doron, E., 2000. Word order in hebrew, in: Lecarme, J., Lowenstamm, J., Shlonsky, U. (Eds.), Research in Afroasiatic Grammar. John Benjamins, pp. 41–56. find
8. Goldberg, Y., Elhadad, M., 2013. Word segmentation, unknown-word resolution, and morphological agreement in a Hebrew parsing system. Computational Linguistics 39, 121–160. find our card
9. Grimshaw, J., 1981. Form, function, and the language acquisition device. the logical problem of language acquisition, ed. by cl baker and john j. mccarthy, 165–182. find
10. Hyams, N., 1986. Language Acquisition and the Theory of Parameters. Reidel, Dordrecht. find
11. Kwiatkowski, T., Sharon, G., Zettlemoyer, L., Steedman, M., 2012. A probabilistic model of syntactic and semantic acquisition from child-directed utterances and their meanings, in: EACL. find
12. Lavi-Rotbain, O., Arnon, I., 2023. Zipfian distributions in child-directed speech. Open Mind 7, 1–30. find
13. Li, P., Xu, Q., 2022. Computational modeling of bilingual language learning: current models and future directions. Language Learning . find
14. MacWhinney, B., 1998. The childes system. Handbook of child language acquisition , 457–494. find
15. Mattys, S., Jusczyk, P., Luce, P., Morgan, J., 1999. Phonotactic and prosodic effects on word segmentation in infants. Cognitive Psychology 38, 465–494. find
16. Pinker, S., 1979. Formal models of language learning. Cognition 7, 217–283. find
17. Räsänen, O., Khorrami, K., 2019. A computational model of early language acquisition from audiovisual experiences of young infants. arXiv preprint arXiv:1906.09832 . ↗
18. Regier, T., 2005. The emergence of words: Attentional learning in form and meaning. Cognitive science 29, 819–865. find
19. Schlesinger, I., 1971. Production of utterances and language acquisition, in: Slobin, D. (Ed.), The Ontogenesis of Grammar. Academic Press, New York, pp. 63–101. find
20. Steedman, M., 2001. The syntactic process. MIT press. find
21. Szubert, I., Abend, O., Schneider, N., Gibbon, S., Mahon, L., Goldwater, S., Steedman, M., 2024. Cross-linguistically consistent semantic and syntactic annotation of child-directed speech. Language Resources and Evaluation , 1–50. find our card
22. Warstadt, A., Mueller, A., Choshen, L., Wilcox, E., Zhuang, C., Ciro, J., Mosquera, R., Paranjabe, B., Williams, A., Linzen, T., Cotterell, R. (Eds.), 2023. Proceedings of the BabyLM Challenge at the 27th Conference on Computational Natural Language Learning, Association for Computational Linguistics, Singapore. URL: https://aclanthology.org/2023.conll-babylm.0. ↗
23. Yang, C.D., et al., 2002. Knowledge and learning in natural language. Oxford University Press on Demand. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/langstream-2024-mahon-language-agnostic-model-child-language/