Elamur Research⌕
☰
home / papers / 2020 / MT syntactic priming effects on L2 English speakers

MT syntactic priming effects on L2 English speakers

Paper · 2020 · Natália Resende, Benjamin Cowan and Andy Way

Summary

The paper asks whether reading machine translation output changes how second-language speakers later build their own English sentences. Nineteen Brazilian Portuguese speakers (20 were recruited) with intermediate or advanced English took part in a pretest-priming experiment adapted from psycholinguistics. The target was English possessive noun phrases. These have two forms: a prepositional version that mirrors Portuguese word order ("the handles of the cutlery") and a prenominal compound that does not ("the cutlery handles"). In the pretest, participants produced English descriptions of pictures without any MT exposure. In the priming phase, they first translated with Google Translate and then described pictures aloud. After seeing Google Translate output, participants used the compound form much more often and the prepositional form much less. Proficiency also mattered: stronger speakers showed a larger shift. The authors read this as evidence that MT output can prime L2 syntax and may help learners reach structures that clash with their native word order. They present it as the first test of syntactic alignment between an MT system and its human users.

Key points

  • Participants: 20 Brazilian Portuguese speakers living in Dublin were recruited and one was excluded, leaving 19 (14 women, mean age 33.7). Proficiency was measured with the 25-question online Cambridge General English test (mean score 13.8).
  • Materials: 104 images were used across the pretest and priming phases. Stimuli ran in PsychoPy, and spoken answers were recorded with a voice recorder.
  • Response shares (Table 1):
    • Pretest: PNP 38.42%, NP 33.42%, other 28.16%.
    • After Google Translate: PNP 10.50%, NP 55%, other 34%.
    • This is a rise of 21.58 points for the non-prepositional NP.
  • Model: a mixed-effects logit model predicting the produced structure from the English proficiency score, the test phase, and their interaction, with random intercepts and test-phase slopes by participant and random intercepts by item.
    • Test phase was significant (p < .001).
    • Proficiency score was significant (p < .001).
    • The Test × proficiency interaction was significant (p < .05), with more proficient speakers showing a larger shift toward NP.
  • Planned future work: more participants, other structures such as datives with and without prepositions, the role of system popularity, and whether the imitation is conscious.

Interesting ideas

  • I1. The study pits two English variants of the same meaning against each other, where one copies the speaker's native order and the other does not, so any shift toward the unfamiliar variant clearly reflects outside influence rather than L1 transfer.
  • I2. Each subject is measured on a baseline before exposure and again after it, so every person serves as their own control.
  • I3. Random intercepts for items and random slopes for subjects keep a few unusual people or pictures from driving the overall result.
  • I4. Answers that fit neither target structure are kept as a separate "other" category instead of being discarded, which shows how often the task fails to produce either form.
  • I5. The syntax people just read biases the structures they produce next, and the effect is stronger for less familiar forms.

For Mova

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

  • I1 (frames): Treat "office table" and "table of the office" as one possession frame, and add paired test sentences like these to check that the frame builder ignores the surface form.
  • I2 (learn): Run the same held-out items before and after adding a repair rule or teacher tier, alongside an untouched control set, so that real gains can be told apart from drift.
  • I3 (learn): Score parser or reader changes with per-sentence and per-genre random effects instead of one pooled accuracy, so that a handful of easy sentences cannot inflate a gain.
  • I4 (answer): Report "neither right nor wrong form" outputs as their own bucket in test scoring, so that abstentions and off-target answers stay visible.
  • I5 (back): Let the generator optionally mirror the structures in the user's input, or deliberately vary them, and make that choice visible in the explanation trace.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 9,086 worksArtificial Intelligence · 8,889 worksArts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 worksNeuroscience · 95 worksCognitive Neuroscience · 95 works

🧭 Topics 2Research subjects and languages the work is about.Machine translation · 958 worksEnglish · 1,373 works

Cite

Cite as Natália Resende, Benjamin Cowan and Andy Way. 2020. MT syntactic priming effects on L2 English speakers. Proceedings of the 22nd Annual Conference of the European Association for Machine Translation.

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

@inproceedings{resende2020mt,
  title = {{MT syntactic priming effects on L2 English speakers}},
  author = {Natália Resende, Benjamin Cowan and Andy Way},
  booktitle = {Proceedings of the 22nd Annual Conference of the European Association for Machine Translation},
  year = {2020},
  url = {https://aclanthology.org/2020.eamt-1.26/}
}

References

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

  1. Baayen, R.H., D.J. Davidson, and D.M. Bates. 2008. Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59(4):390 – 412. Special Issue: Emerging Data Analysis. ↗
  2. Barr, Dale J., Roger Levy, Christoph Scheepers, and Harry J. Tily. 2013. Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68(3):255 – 278. ↗
  3. Bates, Douglas, Martin Mächler, and Bin Dai. 2011. lme4: Linear Mixed-Effects Models Using S4 Classes, volume 0.999375-33. 01. find
  4. Bock, Kathryn and Zenzi M. Griffin. 2000. The persistence of structural priming: Transient activation or implicit learning?. Journal of Experimental Psychology: General, 129(2):177 – 192. ↗
  5. Bock, J. Kathryn and Antony S. Kroch. 1989. The isolability of syntactic processing. In M.K., Carlson G.N.and Tanenhaus, editor, Linguistic Structure in Language Processing. Studies in Theoretical Psycholinguistics, volume 7. Springer), Dordrecht. ↗
  6. Bock, Kathryn, Helga Loebell, and Randal Morey. 1992. From conceptual roles to structural relations: Bridging the syntactic cleft. Psychological Review, 99(1):150 – 171. ↗
  7. Bock, J.Kathryn. 1986. Syntactic persistence in language production. Cognitive Psychology, 18(3):355 – 387. ↗
  8. Branigan, Holly P., Martin J. Pickering, Jamie Pearson, Janet F. McLean, and Clifford I. Nass. 2003. Syntactic alignment between computers and people: the role of belief about mental states. In Alterman, Richard and David Kirsh, editors, Proceedings of the 25th Annual Conference of the Cognitive Science Society, July 31 - August 2 2003, Boston, Massachusetts, pages 186–191. Lawrence Erlbaum Associates. find
  9. Braningan, Holly P., Martin J. Pickering, Andrew J. Stewart, and Janet F. Mclean. 2000. Syntactic priming in spoken production: Linguistic and temporal interference. Memory and Cognition, 28(8):1297–1302. find
  10. Charny, Israel W. 1966. Integrated individual and family psychotherapy. Family Process, 5(2):179–198. ↗
  11. Chinnery, George. 2008. You’ve got some gall: Google-assisted language learning. Language Learning Technology, 12(1):3–11. ↗
  12. Clifford, Joan, Lisa Merschel, and Joan Munné. 2013. Surveying the landscape: What is the role of machine translation in language learning? Revista d’innovacio educativa, (10):108–121. find
  13. Cowan, Benjamin R., Holly P. Branigan, Mateo Obregón, Enas Bugis, and Russell Beale. 2015. Voice anthropomorphism, interlocutor modelling and alignment effects on syntactic choices in humancomputer dialogue. International Journal of Human-Computer Studies, 83:27 – 42. find
  14. Ferreira, Victor S. and Kathryn Bock. 2006. The functions of structural priming. Language and Cognitive Processes, 21(7-8):1011–1029. PMID: 17710210. ↗
  15. Garcia, Ignacio and Marı́a Isabel Pena. 2011. Machine translation-assisted language learning: writing for beginners. Computer Assisted Language Learning, 24(5):471–487. ↗
  16. Gaspari, Federico and John Hutchins. 2007. Online and free ! ten years of online machine translation : Origins , developments , current use and future prospects. find
  17. Gupta, Brij M. and S. M. Dhawan. 2019. Machine translation research a scientometric assessment of global publications output during 2007 16. DESI-DOC Journal of Library Information Technology, 39(1):31–38, 01. ↗
  18. Hartsuiker, Robert J. and Herman H.J. Kolk. 1998. Syntactic facilitation in agrammatic sentence production. Brain and Language, 62(2):221 – 254. ↗
  19. Heyselaar, Evelien, Peter Hagoort, and Katrien Segaert. 2017a. How social opinion influences syntactic processing—an investigation using virtual reality. PLOS ONE, 12(4):1–21, 04. ↗
  20. Heyselaar, Evelien, Peter Hagoort, and Katrien Segaert. 2017b. In dialogue with an avatar, language behavior is identical to dialogue with a human partner. Behavior Research Methods, (49):46–60. find
  21. Konopka, Agnieszka E. and Kathryn Bock. 2009. Lexical or syntactic control of sentence formulation? structural generalizations from idiom production. Cognitive Psychology, 58(1):68 – 101. ↗
  22. Levelt, Willem J.M and Stephanie Kelter. 1982. Surface form and memory in question answering. Cognitive Psychology, 14(1):78 – 106. ↗
  23. Marinis, Theodore, Leah Roberts, Claudia Felser, and Harald Clahsen. 2005. Gaps in second language sentence processing. Studies in Second Language Acquisition, 27(1):53–78. ↗
  24. McDonough, Kim. 2006. Interaction and syntactic priming: English l2 speakers’ production of dative constructions. Studies in Second Language Acquisition, 28(2):179–207. ↗
  25. Niño, Ana. 2006. Evaluating the suitability of using raw machine translation output as input for foreign language written production. Ph.D. thesis. find
  26. Oviatt, Sharon, Courtney Darves, and Rachel Coulston. 2004. Toward adaptive conversational interfaces: Modeling speech convergence with animated personas. ACM Trans. Comput.-Hum. Interact., 11(3):300–328, September. find
  27. Pickering, Martin J. and Holly P. Branigan. 1998. The representation of verbs: Evidence from syntactic priming in language production. Journal of Memory and Language, 39(4):633 – 651. ↗
  28. Pickering, Martin J. and Victor S. Ferreira. 2008. Structural priming: A critical review. Psychological Bulletin, 134(3):427 – 459. ↗
  29. Pickering, Martin J. and Simon Garrod. 2004. Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27(2):169–190. ↗
  30. R Core Development Team, R. 2011. R: A language and environment for statistical computing. Viena, Austria. find
  31. Shin, Jeong-Ah and Kiel Christianson. 2012. Structural priming and second language learning. Language Learning, 62(3):931–964. ↗
  32. Stoyanchev, Svetlana and Amanda Stent. 2009. Lexical and syntactic priming and their impact in deployed spoken dialog systems. In Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Short Papers, NAACL-Short ’09, page 189–192, USA. Association for Computational Linguistics. find
  33. Suzuki, N. and Y. Katagiri. 2007. Prosodic alignment in human-computer interaction. Connection Science, 19(2):131 – 141. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/resende-2020-mt-syntactic-priming-effects-l2/