Elamur Research⌕
☰
home / papers / 2023 / Can Grammarly and ChatGPT accelerate language change? AI-powered technologies and their impact on the English language: wordiness vs. conciseness

Can Grammarly and ChatGPT accelerate language change? AI-powered technologies and their impact on the English language: wordiness vs. conciseness

Paper · 2023 · Karolina Rudnicka

Summary

The paper asks whether AI writing tools might speed up an existing trend in English toward shorter, less elaborate expression. Corpus studies already show this trend. In COHA, for example, average sentence length has fallen markedly since the nineteenth century, and long purpose subordinators like in order to have been losing ground to the plain to-infinitive. Rudnicka ran a case study on that subordinator. She took correct sentences written by native speakers from a corpus of contemporary English and submitted them to Grammarly and to ChatGPT, which was prompted to rewrite them "in a better way." Both tools removed nearly all occurrences of in order to. They did so no matter how short or formal the original sentence was. The edits made sentences shorter on average but did not improve readability. The author argues that, at the scale these tools are used, they could become a new external, higher-order driver of language change rather than just a mirror of it.

Key points

  • The study variable is wordiness vs. conciseness. The probe construction is the purpose subordinator in order to, which has a shorter to-infinitive equivalent.
  • COHA background trend: magazine sentences are about 10 words shorter than in the early 1800s, non-fiction about 8 words shorter, and newspapers almost 5 words shorter than in the mid-1800s.
  • Grammarly and ChatGPT removed 95–100% of in order to instances. The edits were not adjusted for sentence length or formality.
  • The modifications lowered mean sentence length but did not raise readability scores.
  • The scale argument rests on usage figures. Grammarly grew from 1M daily users in 2015 to 30M in 2020, ChatGPT reached about 100M monthly users within two months, and Grammarly ships a browser extension and a Text Editor SDK.

Interesting ideas

  • I1. The study audits a rewriting system by tracking one construction and measuring how often it survives, which turns a vague notion of tool style into a countable rate.
  • I2. The inputs are already-correct native-speaker sentences, so any edit shows the tool's preferences rather than real error correction. This acts as a built-in negative control.
  • I3. Edits are checked against the metric they supposedly serve: the study compares sentence length and readability before and after rewriting, and finds that shortening did not help readability.
  • I4. The author tests whether a rewrite depends on context, such as length or formality, and finds a blanket rule instead.
  • I5. In order to and the to-infinitive are treated as near-equivalent ways to express purpose, which is why tools can swap them freely.

For Mova

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

  • I1 (back): Run the text → graph → text compressor over a set of constructions (in order to, so as to, passives) and report survival rates, so silent stylistic drift in generation becomes visible.
  • I2 (learn): Add held-out gold sentences that need no repair to the evaluation of self-induced repair rules, and flag any rule that fires on them as overreaching.
  • I3 (learn): Whenever a teacher tier or repair step claims to improve some property, measure that property directly before and after instead of trusting the change.
  • I4 (back): Make register-sensitive choices like in order to vs. bare to an explicit, explainable parameter of generation rather than a fixed default.
  • I5 (frames): Map both forms to the same purpose role in the frame, so meaning is unchanged when surface wording varies, and keep the original form only as a style attribute.

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

🎓University of GdańskGdansk · Poland2 works · 4 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksHuman-Computer Interaction · 160 works

🧭 Topics 2Research subjects and languages the work is about.English · 2,379 worksHistorical linguistics & language change · 250 works

Cite

Cite as Karolina Rudnicka. 2023. Can Grammarly and ChatGPT accelerate language change? AI-powered technologies and their impact on the English language: wordiness vs. conciseness. arXiv:2502.04324 · Procesamiento del Lenguaje Natural, 2023, pp. 205-214.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.26342/2023-71-16 · arXiv 2502.04324

@misc{rudnicka2023can,
  title = {{Can Grammarly and ChatGPT accelerate language change? AI-powered technologies and their impact on the English language: wordiness vs. conciseness}},
  author = {Karolina Rudnicka},
  year = {2023},
  doi = {10.26342/2023-71-16},
  eprint = {2502.04324},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.26342/2023-71-16}
}

References

37 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Akimoto, M., Kytö, M., Scahill, J., and Tanabe, H. 2010. Language change and variation from Old English to Late Modern English: a festschrift for Minoji Akimoto: Vol. 114. find
  2. Peter Lang Barrot, J. S. 2020. Integrating Technology into ESL/EFL Writing through Grammarly. RELC Journal 3368822096663. ↗
  3. Biber, D. and Conrad, S. 2009. Register, Genre, and Style. Cambridge: Cambridge University Press. ↗
  4. Crystal, D. 2004. Making Sense of Grammar. Pearson Longman. Davies, M. 2008-. The Corpus of Contemporary American English (COCA): 520 million words, 1990-present. Available online at http://corpus.byu.edu/coca/ Davies, M. 2010-. The Corpus of Historical American English (COHA): 400 million words, 1810- 2009. Available online at https://corpus.byu.edu/coha/ ↗
  5. Doshi, R. H., Bajaj, S. S., and Krumholz, H. M. 2023. ChatGPT: Temptations of Progress. American Journal of Bioethics, 1–3. ↗
  6. Durkin, P. 2014. Borrowed Words: A History of Loanwords in English. Oxford: Oxford University Press. find
  7. Fahnestock, J. 2011. Rhetorical Style: The Uses Of Language In Persuasion. New York: Oxford University Press. find
  8. Filipan-Žignić, B., Legac, V., and Sobo, K. 2016. The influence of the language of new media on the literacy of young people in their school assignments and in leisure. Linguistics Beyond and Within (LingBaW), 2, 77–96. ↗
  9. Fowler, H. W. and Butterfield, J. (ed.). 2015. Fowler’s Dictionary of Modern English Usage. 4th ed. Oxford: Oxford University Press. find
  10. Fries, U. 2010. Sentence Length, Sentence Complexity and the Noun Phrase in 18th-Century News Publications. In Merja Kytö, John Scahill and Harumi Tanabe (eds.), Language Change and Variation from Old English to Late Modern English [Linguistic Insights 114], 21–33. Bern: Peter Lang. find
  11. Geertsema, S., C Hyman, C., and Van Deventer, C. 2011. Short message service (SMS) language and written language skills: educators' perspectives. South African Journal of Education, 31(4), 475–487. ↗
  12. Gordijn, B., and Have, H. T. 2023. ChatGPT: evolution or revolution? Medicine, Health Care, and Philosophy, 26(1), 1–2. ↗
  13. Gross, A. G., Harmon J. E. and Reidy M. S. 2002. Communicating science: the scientific article from the 17th century to the present. Oxford: Oxford University Press. find
  14. Hames, T. and Rae, N. C. 1996. Governing America: History, Culture, Institutions, Organisation, Policy. Manchester University Press. find
  15. Hickey, R. 2012. Internally and externally motivated language change. In The Handbook of Historical Sociolinguistics, Juan Manuel Hernández-Compoy and Juan Camilo Conde-Silvestre (eds.), 401–421. Malden. MA: Wiley-Blackwell. ↗
  16. Hilpert, M. 2013. Constructional change in English: developments in allomorphy, word formation, and syntax. Cambridge: Cambridge University Press. find
  17. Hiltunen, R. 1983. The decline of the prefixes and the beginnings of the English phrasal verb. The evidence from some Old and Early Middle English texts. Turku: Turun Yliopisto. find
  18. Hundt, M. 2014. The demise of the being to V construction. Transactions of the Philological Society, 112(2). 167–187. ↗
  19. Jin, T., and Lu, X. 2018. A Data-Driven Approach to Text Adaptation in Teaching Material Preparation: Design, Implementation, and Teacher Professional Development. TESOL Quarterly, 52(2): 457-467. ↗
  20. Kempf, L. 2021. German so-relatives: Lost in grammatical, typological, and sociolinguistic change. In Svenja Kranich and Tine Breban (eds.), Lost in change: Causes and processes in the loss of grammatical elements and constructions, 291–332. Amsterdam and Philadelphia: John Benjamins. find
  21. Klare, G. R. 1975. Readability and the behaviour of readers. In Shelley A., Harrison, Lawrence M. Stolurow, Improving Instructional Productivity in Higher Education, 235–240. New Jersey: Educational Technology Publications Englewood Cliffs. find
  22. Lewis, E. H. 1894. The History of the English Paragraph. The University of Chicago Press. find
  23. Looi, M-K. 2023. Sixty seconds on . . . ChatGPT. BMJ (Online), 380, 205–205. ↗
  24. Los, B. 2005. The Rise of the To-Infinitive. Oxford: Oxford University Press. ↗
  25. Mair, C. 2006. Twentieth-century English: History, variation and standardization. Cambridge: Cambridge University Press. find
  26. Mehrabi Boshrabadi A., and Bataghva Sarabi, S. 2016. Cyber-communic@tion etiquette: The interplay between social distance, gender and discursive features of student-faculty email interactions. Interactive Technology and Smart Education, 13(2), 86–106. find
  27. Naderi, B., Mohtaj, S., Ensikat, K., and Möller, S. 2019. Subjective Assessment of Text Complexity: A Dataset for German Language. ArXiv abs/1904.07733. ↗
  28. O'Neill, R. and Russell, A. 2019. Stop! Grammar time : University students' perceptions of the automated feedback program Grammarly. Australasian Journal of Educational Technology, 35(1), 42–56 Petré, P. 2010. The functions of weorðan and its loss in the past tense in Old and Middle English. English Language and Linguistics, (14), 457–484. find
  29. Rudnicka, K. 2018. Variation of sentence length across time and genre: influence on the syntactic usage in English. In Diachronic Corpora, Genre, and Language Change, Richard Jason Whitt (ed.), 219 – 240. Amsterdam and Philadelphia: John Benjamins. ↗
  30. Rudnicka, K. 2019. The Statistics of Obsolescence: Purpose Subordinators in Late Modern English. NIHIN Studies. Freiburg: Rombach. ↗
  31. Rudnicka, K. 2021a. The “negative end” of change in grammar: terminology, concepts and causes. Linguistics Vanguard 7(1), 20200091. find
  32. Rudnicka, K. 2021b. In order that – a data-driven study of symptoms and causes ofobsolescence. Linguistics Vanguard 7(1), 20200092. find
  33. Rudnicka, K. 2021c. So-adj-a construction as a case of obsolescence in progress. In Lost in Change. Causes and Processes in the Loss of Grammatical Elements and Constructions [Studies in Language Companion Series 218], Svenja Kranich and Tine Breban (eds.), 51 – 73. Amsterdam: John Benjamins. ↗
  34. Schmidtke-Bode, K. 2009. A Typology of Purpose Clauses. Amsterdam and Philadelphia: John Benjamins. ↗
  35. Tagg, C. 2015. Exploring Digital Communication: Language in Action. Routledge. find
  36. Van Dis, E., Bollen, J., Zuidema, W., van Rooij, R., and Bockting, C. L. 2023. ChatGPT: five priorities for research. Nature (London), 614 (7947), 224–226. ↗
  37. Westin, I. 2002. Language Change in English Newspaper Editorials. Amsterdam: Rodopi. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/rudnicka-2025-grammarly-chatgpt-accelerate-language-change/