Elamur Research⌕
☰
home / papers / 2007 / Network statistics on early English Syntax: Structural criteria

Network statistics on early English Syntax: Structural criteria

Paper · 2007 · Bernat Corominas-Murtra

Summary

The paper sets out an experimental design for studying how English syntax emerges in a child's speech using complex-network tools. Adult notions of grammaticality would discard most early utterances, even though many of them clearly have structure. The author therefore proposes descriptive annotation criteria for deciding what counts as structured in early child grammar. These criteria assume a constituency hierarchy and a grammar built on lexico-thematic relations. Each accepted utterance is analysed into headed constituents, and those constituents are projected onto directed word-to-word links. Pooling the links across utterances yields a syntactic network whose nodes are lexical items. The work uses the PETER corpus from CHILDES, annotated with the DGA-Annotator. It also argues a theoretical point: such networks show the global usage patterns of a grammar, not its internal rules. The resulting annotation supports two analyses: statistics on the average size of structures, ⟨S⟩, and standard complexity measures on the network across successive acquisition stages. A small worked example shows the full pipeline.

Key points

  • Data come from the PETER corpus of the CHILDES database. The child's lines are pulled out with a PERL routine, wrapped in XML, and annotated in the DGA-Annotator.
  • Arcs are built by a fixed procedure:
    • an arc runs from each complement to its phrase head;
    • when the complement is itself a phrase, the arc runs from the inner head to the outer head, which preserves merge order;
    • the head is the semantically most relevant item;
    • finite verbs head the sentence.
  • The criteria cover both rejected and accepted productions. Accepted cases include missing arguments, "to be", infra-specified lexical items, first functional particles, duplicated function words, non-structural items and negation.
  • Unintelligible tokens ("xxx", "0") are dropped. In the example, four structures of sizes 4, 2, 3 and 3 give ⟨S⟩ = 3, with 9 lexical nodes and 7 edges.
  • Network measures include degree assortativity ρ. A negative ρ means the network is disassortative and a positive ρ means it is assortative.

Interesting ideas

  • I1. The author replaces adult grammaticality with lenient, explicit criteria for when a fragment still counts as structured, so partial utterances are kept rather than thrown away.
  • I2. A headed binary constituency tree is turned into directed head-ward arcs, with inner heads linking to outer heads so the order of merging can be recovered.
  • I3. Statistics over the aggregated network describe how a grammar is used, not how it works internally, so global measures should not be read as evidence about specific rules.
  • I4. Average structure size measured per time slice gives a single, simple indicator of growing syntactic capacity.
  • I5. Unintelligible or empty tokens are filtered out explicitly before annotation, so placeholder symbols never become graph nodes.

For Mova

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

  • I1 (tree): The same approach could define which partial trees from noisy or fragmentary input are still worth keeping, instead of discarding sentences the parser cannot fully attach.
  • I2 (tree): This is a compact, explainable rule set for converting constituency-style treebanks into dependency arcs during UD dialect conversion.
  • I3 (learn): Corpus-level statistics of silver trees should be read as usage profiles, and parser rules should be checked with targeted negative controls instead.
  • I4 (learn): A per-text or per-checkpoint mean constituent size could serve as a cheap progress signal during self-play, or for grading the difficulty of training texts.
  • I5 (tokens): A dedicated placeholder or noise token class in the tokenizer would stop such markers from leaking into lemmas and trees.

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

🎓Universitat Pompeu FabraBarcelona · Spain47 works · 11,723 citations🤝Institució Catalana de Recerca i Estudis AvançatsBarcelona · Spain6 works · 453 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksDevelopmental and Educational Psychology · 499 worksPhysics and Astronomy · 44 worksStatistical and Nonlinear Physics · 40 works

🧭 Topics 2Research subjects and languages the work is about.English · 2,263 worksSyntax & grammar formalisms · 214 works

Cite

Cite as Bernat Corominas-Murtra. 2007. Network statistics on early English Syntax: Structural criteria. arXiv:0704.3708.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.0704.3708 · arXiv 0704.3708

@misc{corominasmurtra2007network,
  title = {{Network statistics on early English Syntax: Structural criteria}},
  author = {Bernat Corominas-Murtra},
  year = {2007},
  doi = {10.48550/arxiv.0704.3708},
  eprint = {0704.3708},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.0704.3708}
}

References

23 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Melçuck (1989) Dependency Syntax: Theory and Practice. University of New York, New York find
  2. Hudson, R. (2006). Language networks. The new word grammar. New York: Oxford University Press. find
  3. MacWhinney, B. (2000). The CHILDES project: Tools for analyzing talk. Third Edition. Mahwah, NJ: Lawrence Erlbaum Associates. find
  4. Popescu, M. (2003) Dependency Grammar Annotator, Building Awareness in Language Technology (editors: Florentina Hristea & Marius Popescu). Editura Universităţii din Bucureşti, 2003, p.17-34. find
  5. Chomsky, N. (1965) Aspects of the theory of syntax. MIT Press. Cambridge, MA. ↗
  6. Chomsky, N. (1995) The minimalist program. MIT Press. Cambridge, MA. find
  7. Chomsky N. (1963) Formal properties of grammars. In Handbook of Mathematical Psychology Vol II. Wiley Publishing Company, New York (13) (14) (15) Network statistics on early English Syntax: Structural Criteria find
  8. Chaitin, G. J. (1987) Algorithmic information Theory. Cambridge University Press. Cambridge, MA. ↗
  9. Chomsky N. (1963) Formal properties of natural languages. In Handbook of Mathematical Psychology Vol II. Wiley Publishing Company, New York find
  10. Zipf, G. K. (1949) Human Behaviour and the Principle of Least-Effort Addison-Wesley. Cambridge MA. find
  11. Bollobàs, B. (1998) Modern Graph Theory. Springer. New York. ↗
  12. Chomsky, N. & Miller, G. (1963) Finitary models of language users. In Handbook of Mathematical Psychology Vol II. Wiley Publishing Company, New York find
  13. Ferrrer i Cancho, R., Solé, R. V. & Kohler, R. (2004) Patterns in syntactic dependency networks. Phys. Rev. E. 70 056135. find
  14. Ferrer i Cancho, R. (2005) The structure of syntactic dependency networks: insights from recent advances in network theory. In The problems of quantitative linguistics, Altmann, G., Levickij, V. & Perebyinis, V. (eds.). Chernivtsi: Ruta. pp. 60-75. find
  15. Radford, A. (1990) Syntactic Theory and the Acquisition of English Syntax: the nature of early child grammars of English Blackwell, Oxford. find
  16. Bloom, L.; Hood, L. & Lightbown, P. (1974) Imitation in language development: If, when and why. Cognitive Psychology, 6, 380-420. ↗
  17. Bloom, L.; Hood, L. & Hood, L. (1975) Structure and variation in child Language. Monographs of the society for Research in Child Development, 40, (serial No. 160) ↗
  18. Veneziano, E. & Sinclair, H. (2000) The challenging status of filler syllabes on the way to grammatical morphemes. find
  19. Bloom, L. (1970) Language development: form and function in emerging grammars (Research Monograph No. 59) The MIT Press, Journal of Child Language, 27, 461-500. find
  20. Hyams, N. (1997) Language Acquisition and the theory of Parameters. Dordrecht: Reidel find
  21. Braine, M. D. S. On two types of models of the internalization of grammars. In D. I.. Slobin (ed.), The ontogenesis of grammar: a theoretical symposium. Academic Press (New York)(1971) pp. 153-186 find
  22. Dorogovtsev, S. N. & Mendes, J. F. F. (2003) Evolution of Networks: From Biological Nets to the Internet and WWW, Oxford University Press, New York. Network statistics on early English Syntax: Structural Criteria ↗
  23. Newman, M. E. J.(2002) Assortative mixing in Networks. Phys. Rev. Lett. 89, 208701 find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/corominas-murtra-2007-network-statistics-early-english-syntax/