Elamur Research⌕
☰
home / papers / 2021 / A quantitative investigation of English adnominal modifiers

A quantitative investigation of English adnominal modifiers

Paper · 2021 · Tsy Yih and Haitao Liu

Summary

The paper asks how often English noun heads actually take modifiers. Theory has described the layered order of adnominals (determiner, quantifier, adjective, noun modifier, head), but there has been little counting. The authors apply dependency-based valency theory to nouns. They use the UD version of the GUM treebank and count seven adnominal relations (det, amod, compound, nmod, nummod, acl, acl:relcl) on three classes of heads: common nouns, proper nouns and pronouns. They compute Mean Valency per class. They also introduce the Mean Valency Pattern (MVP), which is built on frequency of dependencies per head (FDPH), because the older Probabilistic Valency Pattern (PVP) gives within-class proportions that cannot be compared across classes of different sizes.

The main results:

  • Common nouns have significantly higher valency than proper nouns, and proper nouns have higher valency than pronouns.
  • Most modifiers on proper nouns are really pieces of multiword names.
  • Pronouns are almost never modified.

The work offers a quantitative baseline for studying noun-phrase structure.

Key points

  • The data is the UD version of GUM, a balanced spoken and written corpus: 7,409 sentences, 116,748 words excluding punctuation, 11 genres from academic writing to vlogs and how-to guides.
  • Only seven UD adnominal relations are counted. Noise such as cc is excluded. UD was chosen over SUD because its content-word heads make noun-phrase modifiers easy to find.
  • Determiners are the most frequent modifier but appear with fewer than half of common nouns. The authors attribute this to bare plurals and generic reference. After det come amod, then nmod and compound, with nummod rare.
  • Proper nouns attract slightly more compound than common nouns. Much of their non-zero valency is due to complex names, for example "University of Cambridge" and "the United Kingdom".
  • Pronoun modification is essentially limited to nmod and relative clauses, as in "those who …". The study ignores genre differences, which the authors leave for future work.

Interesting ideas

  • I1. Normalizing relation counts per head rather than as a share of all relations within a class (MVP vs PVP) makes classes of very different sizes comparable.
  • I2. Most modifiers on proper nouns turn out to be internal parts of multiword names, not restrictive modifiers.
  • I3. Pronouns get modified only when they carry too little information alone, as in "those who". The relative clause then becomes essential to identify the referent.
  • I4. Profiling which of a fixed set of adnominal relations each head class normally takes gives an expected dependent inventory per class.
  • I5. The frequent absence of determiners is explained by bare plurals and generic reference.

For Mova

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

  • I1 (learn): compute per-head rates of each relation by UPOS for gparse and perceptron output, compare them to gold UD as a cheap sanity check, and flag drift when silver data shifts these rates.
  • I2 (world): when building entities, collapse det, compound and nmod inside a PROPN span into a single name, so that "University of Cambridge" is not read as a university restricted by a location.
  • I3 (frames): in coreference, treat a pronoun with an attached acl:relcl or nmod as a restricted description to resolve through its clause, not as an ordinary anaphor.
  • I4 (tree): turn these profiles into plausibility checks or self-induced repair triggers, for example flagging amod or det attached to a PRON as a likely attachment error.
  • I5 (world): mark a bare plural with no det as a candidate kind-level (generic) reading, so it states a property of a type rather than introducing new individuals into the text world.

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, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,379 works

Cite

Cite as Tsy Yih and Haitao Liu. 2021. A quantitative investigation of English adnominal modifiers. Proceedings of the 35th Pacific Asia Conference on Language, Information and Computation.

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

@inproceedings{yih2021quantitative,
  title = {{A quantitative investigation of English adnominal modifiers}},
  author = {Tsy Yih and Haitao Liu},
  booktitle = {Proceedings of the 35th Pacific Asia Conference on Language, Information and Computation},
  year = {2021},
  url = {https://aclanthology.org/2021.paclic-1.50/}
}

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. Abeillé, A. (Ed.). (2003). Treebanks: Building and using parsed corpora (Vol. 20). Springer Science & Business Media. find
  2. Bache, C. (2000). Essentials of mastering English. De Gruyter Mouton. ↗
  3. Davidse, K., & Breban, T. (2019). A cognitive-functional approach to the order of adjectives in the English noun phrase. Linguistics, 57(2), 327-371. ↗
  4. Dixon, R. M. (1977). Where have all the adjectives gone?. Studies in Language, 1(1), 19-80. ↗
  5. Čech, R., Pajas, P., & Mačutek, J. (2010). Full valency. Verb valency without distinguishing complements and adjuncts. Journal of quantitative linguistics, 17(4), 291-302. find
  6. Croft, W. (1990). A conceptual framework for grammatical categories (or: a taxonomy of Propositional Acts). Journal of Semantics, 7(3), 245-279. ↗
  7. Du Bois, J. W. (1980). Beyond definiteness: The trace of identity in discourse. The pear stories: Cognitive, cultural, and linguistic aspects of narrative production, 3, 203-274. find
  8. Gao, J., & Liu, H. (2019). Valency and English learners’ thesauri. International Journal of Lexicography, 32(3), 326-361. ↗
  9. Gao, J., & Liu, H. (2020). Valency Dictionaries and Chinese Vocabulary Acquisition for Foreign Learners. Lexikos, 30, 1-32. ↗
  10. Gao, S., Zhang, H., & Liu, H. (2014). Synergetic properties of Chinese verb valency. Journal of Quantitative Linguistics, 21(1), 1-21. ↗
  11. Halliday, M. A. K. (1985). An Introduction to Functional Grammar. London: Edward. find
  12. Hopper, P. J., & Thompson, S. A. (1984). The discourse basis for lexical categories in universal grammar. Language, 60(4), 703-752. ↗
  13. Hudson, R. A. (1984). Word grammar. Oxford: Blackwell. find
  14. Hudson, R. (1994). About 37% of word-tokens are nouns. Language, 70(2), 331-339. ↗
  15. Keizer, E. (2007). The English noun phrase: The nature of linguistic categorization. Cambridge University Press. find
  16. Liu, H. (2009a). Dependency grammar: From theory to practice. Science Press. find
  17. Liu, H. (2009b). Probability distribution of dependencies based on a Chinese dependency treebank. Journal of Quantitative Linguistics, 16(3), 256-273. ↗
  18. Liu, H., & Feng, Z. (2007). Probabilistic Valency Pattern Theory for Natural Language Processing. Linguistic Sciences, 6(3), 32–41. find
  19. Lu, B. (1993). The head-oriented grammar. Shanghai Educational Publishing House. find
  20. Lu, Q., Lin, Y., & Liu, H. (2018). Dynamic Valency and Dependency Distance. Quantitative Analysis of Dependency Structures, 72, 145. ↗
  21. Mel'cuk, I. A. (1988). Dependency syntax: theory and practice. SUNY press. find
  22. Nivre, J., De Marneffe, M. C., Ginter, F., Goldberg, Y., Hajic, J., Manning, C. D., ... & Zeman, D. (2016, May). Universal dependencies v1: A multilingual treebank collection. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) (pp. 1659-1666).
  23. Pan, X., & Liu, H. (2014). Adnominal Constructions in Modern Chinese and their Distribution Properties. Glottometrics, 29, 1-30. find
  24. Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1985). A comprehensive English grammar. London and New York: Longman. find
  25. Rijkhoff, J. (2002). The noun phrase. Oxford University Press. ↗
  26. Tesnière, L. (1959). Eléments de la structurale. Paris: Klincksieck. syntaxe Vincze, V. (2014). Valency frames in a Hungarian corpus. Journal of quantitative linguistics, 21(2), 153-176. find
  27. Yan, J. & Liu, H. (2021). Quantitative Analysis of Chinese and English Verb Valencies Based on the Probabilistic Valency Pattern Theory. CLSW2021. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yih-2021-quantitative-investigation-english-adnominal-modifiers/