Innovation in Language Learning

The proposed work continues a previous study in which we defined the distinctive characteristics of a literary work as its top deviations from a reference corpus in terms of a large selection of measurable textual characteristics, including readability characteristics, grammatical constructions, and literary devices [8]. Our current step consists of applying the conclusions reached in that study to improve existing literary adaptations for EFL learners. Our hypothesis is that, as adapted works are made to more closely match the style of their original counterparts, readers’ engagement and comprehension will improve. 

In the experimental setting, we will use one chapter as present in two adaptations (each at a different CEFR level) of each of two canonical novels (The Picture of Dorian Gray and Moby Dick). Each of these four texts will be presented in three alternative versions: 1) the baseline (the adaptation’s unaltered text), 2) an enriched baseline (the text after undergoing modifications based on the conclusions of our statistical analysis), and 3) a text derived through the fully automatic transfer of the original work’s style onto its adapted counterpart [6]. To produce the enriched baseline, we will follow a defined modification pipeline. We will increase the presence of each work’s most prominent characteristics – in proportions corresponding to their degree of prominence and an extent calibrated to the extract’s length. Similarly, we will increase the presence of the vocabulary items that are most overrepresented in the original work when compared against the associated corpus. The derived text’s CEFR level will be automatically measured at this point [1]. If it has not increased relative to the baseline text, prominent sentences (defined through the proxy of high surprisal [3]) will be progressively incorporated, stopping immediately before the level increases. 

We will compare the texts produced in the three scenarios through human evaluation, in which readers will be provided with one of the 12 examined chapter versions and asked to answer several comprehension questions, as well as a set of open-ended and Likert-scale questions concerning their perception of the reading experience (e.g. “Did the story catch your attention?”; “What do you think will happen next?”). The participants will be EFL learners matched in proficiency level to the corresponding texts. 

 

Keywords: adapted literature, literary style, corpus analysis

 

REFERENCES

 

[1] Arase, Y., Uchida, S., & Kajiwara, T. (2022). CEFR-Based Sentence Difficulty Annotation and Assessment. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (pp. 6206–6219). Abu Dhabi, United Arab Emirates: Association for Computational Linguistics.

https://aclanthology.org/2022.emnlp-main.416/

[2] Biber, D. (1988). Variation Across Speech and Writing. Cambridge: Cambridge University Press.

[3] Danescu-Niculescu-Mizil, C., Cheng, J., Kleinberg, J., & Lee, L. (2012). You Had Me at Hello: How Phrasing Affects Memorability. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (pp. 892–901). Jeju Island, Korea: Association for Computational Linguistics. https://aclanthology.org/P12-1094/

[4] DuBay, W. H. (2004). The Principles of Readability. Costa Mesa, CA: Impact Information.

[5] Hatzfeld, H. A. (1953). A Critical Bibliography of the New Stylistics Applied to the Romance Literatures, 1900–1952. Chapel Hill: University of North Carolina Press. (University of North Carolina Studies in Comparative Literature, No. 5.)

[6] Horvitz, Z., Patel, A., Singh, K., Callison-Burch, C., McKeown, K., & Yu, Z. (2024). TinyStyler: Efficient Few-Shot Text Style Transfer with Authorship Embeddings. In Findings of the Association for Computational Linguistics: EMNLP 2024 (pp. 13376–13390). Miami, Florida: Association for Computational Linguistics. https://aclanthology.org/2024.findings-emnlp.781/

[7] Hutcheon, L. (2006). A Theory of Adaptation. New York: Routledge.

[8] Nikolova-Stoupak, I., Schaeffer-Lacroix, E., & Lejeune, G. (2026). Quantifying Literary Style: A Corpus-Based Study. In Proceedings of the Seventh International Conference on Computational Linguistics in Bulgaria. Sofia, Bulgaria.

[9] Riley, P., Constant, N., Guo, M., Kumar, G., Uthus, D., & Parekh, Z. (2021). TextSETTR: Few-Shot Text Style Extraction and Tunable Targeted Restyling. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (pp. 3786–3800). Association for Computational Linguistics. https://aclanthology.org/2021.acl-long.293/

[10] Spitzer, L. (1948). Linguistics and Literary History: Essays in Stylistics. Princeton: Princeton University Press.

[11] Thiel, E. (2013). Downsizing Dickens: Adaptations of Oliver Twist for the Child Reader. In A. Müller (Ed.), Adapting Canonical Texts in Children’s Literature (pp. 143–162). London: Bloomsbury Academic.

https://aclanthology.org/2022.flp-1.7/

[12] Wellek, R., & Warren, A. (1949). Theory of Literature. New York: Harcourt, Brace and Company.

[13] Wu, Y., & Deng, X. (2025). Implementing Long Text Style Transfer with LLMs through Dual-Layered Sentence and Paragraph Structure Extraction and Mapping. arXiv. https://arxiv.org/abs/2505.07888