Elamur Research⌕
☰
home / papers / 2018 / Probabilistic grammar and constructional predictability: Bayesian generalized additive models of <i>help</i> + (to) Infinitive in varieties of web-based English

Probabilistic grammar and constructional predictability: Bayesian generalized additive models of <i>help</i> + (to) Infinitive in varieties of web-based English

Paper · 2018 · Natalia Levshina

Summary

The paper asks what governs the optional to after help, as in "helped John (to) install". It also asks whether the factors behind this choice are the same across English varieties. Levshina draws on GloWbE web data from seven countries (the USA, Great Britain, Jamaica, India, Hong Kong, Ghana and Australia) and fits Bayesian generalized additive mixed-effects regression models. The models control for known factors: horror aequi, cognitive complexity, register, the morphological form of help, and whether a Helpee is present. On top of these, they test constructional predictability, measured in two directions: the infinitive given help and help given the infinitive.

The to-infinitive turns up more often when information content is high. The clearest effect comes from the surprisal of help given the verb, not from the reverse direction. These effects are non-linear, and their strength and shape differ by country. The work shows that efficiency-based marking can be driven by how unexpected the construction is, not only the word. It also argues that claims about universal principles need checking across language varieties.

Key points

  • Data come from the Global Web-based English corpus (GloWbE, Davies 2013), covering seven national varieties.
  • The two predictability measures mirror Schmid's (2000) Attraction (word given construction) and Reliance (construction given word). They are modelled with smooth, non-linear terms.
  • Some verbs make help highly surprising, among them be, have, do, say, ask, try, use. In the top 5% of such scores, to is more frequent than average in every variety. The effect holds even when auxiliaries be/have/do are removed.
  • Horror aequi and cognitive complexity behave very similarly across varieties.
  • Other factors are less uniform:
    • Average word length, used as a formality proxy, mostly favours to, but shows a reverse effect in Indian English.
    • Transitive infinitives favour to credibly only in the Hong Kong data.
    • The US model shows weak effects overall, plausibly because to is nearly extinct there.

Interesting ideas

  • I1. Measure association in both directions separately (word given construction vs. construction given word) instead of one symmetric score, since only one direction may carry the signal
  • I2. Optional function words tend to appear exactly where the structure is hard to recognize, so their presence is itself a cue about the writer's expectations
  • I3. Model predictors with smooth non-linear curves rather than straight lines, because effects can be flat over most of the range and appear only at the extremes
  • I4. Fit the same model per variety and compare, rather than pooling, to see which effects are truly shared
  • I5. Rerun the analysis without suspicious special cases (here auxiliaries) to confirm the effect is not driven by a few items

For Mova

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

  • I1 (tree): keep separate head→dependent and dependent→head statistics for xcomp/infinitival attachments, which helps when a frequent verb like be rarely signals a help-type complement.
  • I2 (back): when generating, insert optional to/that when the complement verb is weakly tied to its governor, making output easier to reparse in the text → graph → text loop.
  • I3 (learn): when analysing which features drive parser or repair-rule errors, bin or smooth continuous features instead of assuming linear impact.
  • I4 (learn): report evaluation per genre/variety slice (fables, school tests, web text) so a pooled score does not hide opposite-direction effects.
  • I5 (learn): use such exclusion runs as negative controls when validating silver data or induced rules, e.g. drop the most frequent lemmas and check that gains persist.

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

🎓Leipzig UniversityLeipzig · Germany26 works · 8,622 citations

 Fields, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Natalia Levshina. 2018. Probabilistic grammar and constructional predictability: Bayesian generalized additive models of <i>help</i> + (to) Infinitive in varieties of web-based English. Glossa: a journal of general linguistics.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.5334/gjgl.294

@article{levshina2018probabilistic,
  title = {{Probabilistic grammar and constructional predictability: Bayesian generalized additive models of <i>help</i> + (to) Infinitive in varieties of web-based English}},
  author = {Natalia Levshina},
  journal = {Glossa: a journal of general linguistics},
  year = {2018},
  doi = {10.5334/gjgl.294},
  url = {https://doi.org/10.5334/gjgl.294}
}

References

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

  1. Aylett, Matthew & Alice Turk. 2004. The smooth signal redundancy hypothesis: A functional explanation for relationships between redundancy, prosodic prominence, and duration in spontaneous speech. Language and Speech 47(1). 31–56. DOI: https://doi.org/10.1177/00238309040470010201 Bell, Alan, Jason Brenier, Michelle Gregory, Cynthia Girand & Dan Jurafsky. 2009. Predictability effects on durations of content and function words in conversational English. Journal of Memory and Language 60(1). 92–111. DOI: https://doi.org/10.1016/j.jml.2008.06.003 ↗
  2. Biber, Douglas. 1988. Variation across speech and writing. Cambridge: Cambridge University Press. DOI: https://doi.org/10.1017/CBO9780511621024 ↗
  3. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman grammar of spoken and written English. Harlow: Longman. find
  4. Bürkner, Paul-Christian. 2017. brms: An R Package for Bayesian Multilevel Models using Stan. Journal of Statistical Software 80(1). DOI: https://doi.org/10.18637/jss.v080.i01 ↗
  5. Bybee, Joan L. 1985. Morphology. A study of the relation between meaning and form. Amsterdam: John Benjamins. DOI: https://doi.org/10.1075/tsl.9 ↗
  6. Davies, Mark. 2013. Corpus of Global Web-Based English: 1.9 billion words from speakers in 20 countries. http://corpus.byu.edu/glowbe/. ↗
  7. Dixon, R.M.W. 1991. A new approach to English grammar, on semantic principles. Oxford: Clarendon Press. ↗
  8. Duffley, Patrick J. 1992. The English infinitive. London: Longman. find
  9. Givón, Talmy. 1990. Syntax: A functional-typological introduction 2. Amsterdam: John Benjamins. find
  10. Haiman, John. 1983. Iconic and economic motivation. Language 59(4). 781–819. DOI: https://doi.org/10.2307/413373 ↗
  11. Haspelmath, Martin. 2008. Creating economical morphosyntactic patterns in language change. In Jeff Good (ed.), Linguistic universals and language change, 185–214. Oxford: Oxford University Press. DOI: https://doi.org/10.1093/acprof: oso/9780199298495.003.0008 Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press. DOI: https://doi. org/10.1017/9781316423530 Jurafsky, Daniel, Alan Bell, Michelle L. Gregory & William D. Raymond. 2001. Probabilistic relations between words: Evidence from reduction in lexical production. In Joan L. Bybee & Paul J. Hopper (eds.), Frequency and the emergence of linguistic structure (Typological Studies in Language 45), 229–254. Amsterdam: John Benjamins. ↗
  12. Klein, Dan & Christopher D. Manning. 2003. Accurate Unlexicalized Parsing. In Proceedings of the 41th annual meeting of the Association for Computational Linguistics, 423–430. http://nlp.stanford.edu/manning/papers/unlexicalized-parsing.pdf. ↗
  13. Kruschke, John. 2011. Doing Bayesian data analysis. A tutorial with R and BUGS. Amsterdam: Elsevier. find
  14. Kurumada, Chigusa & T. Florian Jaeger. 2015. Communicative efficiency in language production: Optional case-marking in Japanese. Journal of Memory and Language 83. 152–178. DOI: https://doi.org/10.1016/j.jml.2015.03.003 ↗
  15. Leben, William. 1973. Suprasegmental phonology. Cambridge, MA: The Massachusetts Institute of Technology dissertation. find
  16. Levshina, Natalia. 2016. When variables align: A Bayesian multinomial mixed-effects model of English permissive constructions. Cognitive Linguistics 27(2). 235–268. DOI: https://doi.org/10.1515/cog-2015-0054 ↗
  17. Levy, Roger & T. Florian Jaeger. 2007. Speakers optimize information density through syntactic reduction. In Bernhard Schlökopf, John Platt & Thomas Hoffman (eds.), Advances in Neural Information Processing Systems (NIPS) 19. 849–856. Cambridge, MA: MIT Press. ↗
  18. Lind, Age. 1983. The variant forms of help to/help Ø. English Studies 64. 263–275. DOI: https://doi.org/10.1080/00138388308598255 ↗
  19. Lohmann, Arne 2011. Help vs. help to – a multifactorial, mixed-effects account of infinitive marker omission. English Language and Linguistics 15(3). 499–521. DOI: https://doi.org/10.1017/S1360674311000141 Mahowald, Kyle, Evelina Fedorenko, Steven T. Piantadosi & Edward Gibson. 2013. Info/ information theory: Speakers choose shorter words in predictive contexts. Cognition 126. 313–318. DOI: https://doi.org/10.1016/j.cognition.2012.09.010 ↗
  20. Mair, Christian. 2002. Three changing patterns of verb complementation in Late Modern English: a real-time study based on matching text corpora. English Language and Linguistics 6(1). 105–131. DOI: https://doi.org/10.1017/S1360674302001065 ↗
  21. Mair, Christian. 2015. Responses to Davies and Fuchs. English World-Wide 36(1). 29–33. DOI: https://doi.org/10.1075/eww.36.1.02mai ↗
  22. McEnery, Anthony & Zhonghua Xiao. 2005. HELP or HELP to: What do corpora have to say? English Studies 86(2). 161–187. DOI: https://doi.org/10.1080/0013838042000339880 ↗
  23. Norcliffe, Elisabeth & T. Florian Jaeger. 2016. Predicting head-marking variability in Yucatec Maya relative clause production. Language and Cognition 8(2). 167–205. DOI: https://doi.org/10.1017/langcog.2014.39 ↗
  24. Piantadosi, Steven T., Harry Tily & Edward Gibson. 2011. Word lengths are optimized for efficient communication. Proceedings of the National Academy of Sciences 108(9). 3526. DOI: https://doi.org/10.1073/pnas.1012551108 ↗
  25. R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. ↗
  26. Rohdenburg, Günther. 1996. Cognitive complexity and increased grammatical explicitness in English. Cognitive Linguistics 7(2). 149–182. DOI: https://doi.org/10.1515/ cogl.1996.7.2.149 Rohdenburg, Günther. 2003. Horror aequi and cognitive complexity as factors determining the use of interrogative clause linkers. In Günther Rohdenburg & Britta Mondorf (eds.), Determinants of grammatical variation in English, 205–250. Berlin: Mouton de Gruyter. DOI: https://doi.org/10.1515/9783110900019.205 ↗
  27. Rohdenburg, Günther. 2009. Grammatical divergence between British and American English in the nineteenth and early twentieth centuries. In Ingrid Tieken-Boon van Ostade & Wim van der Wurff (eds.), Current issues in Late Modern English (Linguistic Insights 77), 301–330. Bern: Peter Lang. find
  28. Schmid, Hans-Jörg. 2000. English abstract nouns as conceptual shells. From corpus to cognition. Berlin: Mouton de Gruyter. DOI: https://doi.org/10.1515/9783110808704 ↗
  29. Schmid, Hans-Jörg & Helmut Küchenhoff. 2013. Collostructional analysis and other ways of measuring lexicogrammatical attraction: Theoretical premises, practical problems and cognitive underpinnings. Cognitive Linguistics 24(3). 531–577. DOI: https://doi.org/10.1515/cog-2013-0018 Shannon, Claude E. 1948. A mathematical theory of communication. Bell System Technical Journal 27. 379–423 & 623–656. ↗
  30. Stan Development Team. 2015. Stan: A C++ library for probability and sampling, version 2.8.0. http://mc-stan.org/. ↗
  31. Stefanowitsch, Anatol & Stefan Th. Gries. 2003. Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8(2). 209– 243. DOI: https://doi.org/10.1075/ijcl.8.2.03ste ↗
  32. Szmrecsanyi, Benedikt. 2004. On operationalizing syntactic complexity. In Gérard Purnelle, Cédrick Fairon & Anne Dister (eds.), Le poids des mots. Proceedings of the 7th International Conference on Textual Data Statistical Analysis 2. 1032–1039. Louvain-la-Neuve: Presses universitaires de Louvain. find
  33. Walter, Mary Ann, & T. Florian Jaeger. 2008. Constraints on optional that: A strong word form OCP effect. In Rodney L. Edwards, Patrick J. Midtlyng, Colin L. Sprague & Kjersti G. Stensrud (eds.), Proceedings from the Annual Meeting of the Chicago Linguistic Society, 505–519. Chicago, IL: CLS. find
  34. Wasow, Thomas, Roger Levy, Robin Melnick, Hanzhi Zhu & Tom Juzek. 2015. Processing, prosody, and optional to. In Lyn Frazier & Edward Gibson (eds.), Explicit and implicit prosody in sentence processing, 133–158. New York: Springer. DOI: https://doi.org/10.1007/978-3-319-12961-7_8 Wasow, Thomas, T. Florian Jaeger & David M. Orr. 2011. Lexical variation in relativizer frequency. In Horst J. Simon & Heike Wiese (eds.), Expecting the unexpected: Exceptions in grammar, 175–195. Berlin: De Gruyter Mouton. DOI: https://doi. org/10.1515/9783110219098.175 Wood, Simon N. 2006. Generalized additive models: An introduction with R. Boca Raton, FL: Chapman and Hall/CRC. ↗
  35. Zipf, George. 1935. The psychobiology of language: An introduction to dynamic philology. Cambridge, MA: MIT Press. find
  36. How to cite this article: Levshina, Natalia. 2018. Probabilistic grammar and constructional predictability: Bayesian generalized additive models of help + (to) Infinitive in varieties of web-based English. Glossa: a journal of general linguistics 3(1): 55. 1–22, DOI: https://doi.org/10.5334/gjgl.294 ↗
  37. Submitted: 09 November 2016 find
  38. Accepted: 26 January 2018 find
  39. Copyright: © 2018 The Author(s). This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC-BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. See http://creativecommons.org/licenses/by/4.0/. Glossa: a journal of general linguistics is a peer-reviewed open access journal published by Ubiquity Press. OPEN ACCESS ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/levshina-2018-probabilistic-grammar-constructional-predictability-bayesian/