Open research questions in Text Readability and Simplification
58 unresolved questions extracted from the limitations and future-work sections of 1,140 Text Readability and Simplification papers in our library. Each links back to the study that raised it.
What the literature leaves open
Feedback literacy—the understandings, capacities, and dispositions needed to make sense of feedback and use it While existing research has extensively examined AI-generated feedback in terms of writing product outcomes, the process of how learners develop feedback literacy through sustained interaction with AI remains insufficiently understood.
Developing EFL Learners’ Feedback Literacy through AI-Assisted Dialogic Interactions in English Academic Writing: A Mixed-Methods Study · 2026 · DOIFinally, the increasing demand for lecturers’ TOEFL ITP scores for key performance indicators and accreditation purposes means that this issue needs to be investigated further.
Moreover, future research should examine multiple dimensions of writing quality simultaneously and track changes over extended periods to better understand the longitudinal effects of AIGCF on writing development.
Assessing the integration of artificial intelligence-generated content feedback in English language writing learning · 2026 · DOIFuture research should examine real EFL student texts. Future studies could address this constraint by incorporating diverse reference datasets or multiple human annotators to better account for the fluid and subjective nature of language. Future research should consider adopting more clearly anchored rubrics or involving language learners themselves to assess practical comprehensi- bility, rather than relying exclusively on expert judgment. Given the sensitivity of LLM outputs to prompt design, future research should investigate how varying prompts influence readability, detail, comprehensibility, and accuracy (Singh et al. Furthermore, the assessment of explanatory feedback was based on only 20 sentences, which is insufficient for draw- ing firm conclusions.
Comparative Analysis of LLM-Based Writing Tools for Error Correction and Feedback: A Study on ChatGPT-3.5, ChatGPT-4, Gemini, and Claude 3 · 2026 · DOI” Instead, the findings suggest that excessive or poorly timed feedback streams can generate cognitive interference rather than consolidation, a nuance underexplored in previous hybrid-feedback research (Dong, 2024).
Learners, in turn, become more independent and empowered, with AI granting them agency to explore, test and produce language in ways previously limited by the constraints of print media.
ZeroGPT, however, displayed inconsistent results for human-written Croatian texts: One sample was detected as ‘Mixed signals, with some parts generated by AI/GPT’ with a significant AI percentage, another as ‘Likely human-written, may include parts generated by AI/GPT’ with a moderate AI percentage, and a third as ‘Most of your text is AI/GPT-generated’ with a high AI Information Research, Vol.
Several limitations must be considered. First, the examined sample is small, potentially affecting generalisability of results. Second, the study focuses exclusively on English language students and instructors in higher education; students with other specialisations and instructors from other majors are outside the scope of this research. Third, the impact of utilising AI technology may differ for students with limited English language proficiency. Finally, the study does not account for other AI assistant writing tools that may offer additional writing-related techniques not reported here.
Investigation of Students' Perceptions of the Use of Artificial Intelligence in Writing Assignments · 2026 · DOIknowledge acquisition and ■ APPENDIX A. CODING SCHEME This appendix operationalises the eight coding dimensions named in Section 2.5 into allowable codes and one-sentence definitions. The codebook was used for every retained record in Table 1, Tables 2-5, and Appendix F. ■ D1 – AI Technology Single dominant technology of the manipulated or assessed variable, drawn from the following closed set: LLM/GenAI (large language model or generative AI as the intervention; e.g., ChatGPT, GPT-4, GPT-4o, Claude, DeepSeek); ASR (automatic speech recognition coupled to a learner model); ITS (intelligent tutoring system with a learner-state model); Adaptive-Rec (adaptive recommendation of content); Chatbot/IPA (chatbot or intelligent personal assistant with natural-language understanding); Multimodal-AI (multimodal generation or recognition with model-level inference, e.g., DALL-E, Sora, voice-mode GenAI); DL-Assess (deep-learning-based assessment); Robotics (educational robotics with AI components); Multi (more than one of the above, with no single dominant technology). A record is assigned exactly one code; multi-technology studies receive Multi.
Artificial Intelligence in Teaching Chinese as A Target Second Foreign Language: A Scoping Review (2021–2026) · 2026 · DOI• Simplification Scope: Our simplification strategies focused primarily on lexical and syntactic complexity. We did not explicitly address more complex semantic simplifications, such as breaking down long concepts, adding explanations, or altering discourse structures. • Generalizability of Methods: The syntactic simplification rules and ESV strategies were developed for a specific dataset. Their effectiveness and general applicability to any arbitrary Punjabi text require further validation. • Our findings are therefore a baseline, and future work would benefit from applying more modern techniques with larger datasets.
Adopting a technology-enhanced approach to language assessment, rather than a technology-driven approach, we critically assessed the suitability of the state-of-the-art GEC system for assessing language accuracy in Korean, an understudied language in this regard.
Evaluating the applicability of the transformer-based grammatical error correction system for assessing language accuracy · 2025 · DOISecond, that it is imperative for educators to impart to our students the significant limitations of generative AI’s knowledge-production abilities, as algorithms trained on large language models (LLMs) reproduce historic inequalities.
Future research directions are provided based on the limitations of this study; these limitations include a small sample size, the employment of a single task type and L1 group, and the current GPT system’s inability to assess speaking domains beyond fluency and pronunciation.
Exploratory study on developing and evaluating a GenAI-based speech scoring system for L2 teaching practitioners · 2025 · DOIAcknowledging the history of writing with technologies and writing as technology, the development of GAI warrants attention to pedagogical and ethical implications in writing-intensive engineering classes.
A Pilot Study Inquiring into the Impact of ChatGPT on Lab Report Writing in Introductory Engineering Labs · 2024 · DOIIn the specific field of journalistic writing, its use for the moment is limited to assisting tasks in the writing process or suggesting topics, while the automatic generation of content is not widely accepted due to the ethical implications that it entails, especially those related to the originality of content.
Possibilities and challenges of Artificial Intelligence in the teaching and learning process of Journalism Writing. The experience in Spanish universities · 2024 · DOI” It did not, however, bring up the caveat that greets you when you open up ChatGPT itself: that it “may occasionally generate incorrect information,” that it “may occasionally produce harmful instructions or biased content,” or that it has “limited knowledge of the world and events after 2021.
Where does ChatGPT fit into the Framework for Information Literacy? The possibilities and problems of AI in library instruction · 2023 · DOIEducational researchers could benefit from the utility of ChatGPT in identifying gaps in the literature, generating new ideas and developing hypothesis, devising surveys or rating scales, conducting systematic reviews, eliminating human error in analysis of large datasets as well as drafting and editing scientific manuscripts.
Second, only one medical condition—testicular torsion—was evaluated; therefore, the findings may not be generalizable to other diseases or medical specialties.
Large Language Models in Patient Education: A Comparative Readability Analysis of Testicular Torsion Information Generated by ChatGPT-5 and Gemini 2.5 Pro · 2026 · DOIAlthough these challenges are not as well-known as others, such as hallucination and interpretability, they pose significant concerns in educational settings if they are not understood and mitigated before implementation.
Trust, Privacy, and Context Dependence of Large Language Models in Educational Contexts: Technical Challenges and Opportunities · 2026 · DOIMore than that, the false expectations could be easily built while using LLMs as they might seem to be ‘perfect’ in the eye of the user, but they are constantly evolving, adding new capabilities and the analyses are always limited to the present time.
We found mixed results: Whereas ChatGPT can mainly be used in surface syntax matters, it can produce hallucinations at the level of complex syntactic analyses.
Exploring the Relationship between ChatGPT Use and University Students’ Syntactic Argumentation · 2026 · DOIEvidence remains limited on purpose-built classroom systems that embed constrained, learner-initiated GenAI feedback within technology-mediated collaborative writing, align that feedback with explicitly taught forms, and assess outcomes beyond an immediate posttest.
Embedding constrained, learner-initiated GenAI feedback in collaborative L2 writing: Targeted grammar gains with a two-week follow-up · 2026 · DOIDespite growing interest in GenAI-assisted writing instruction, limited research has examined how visual enhancements to GenAI chatbot output might improve revision outcomes and learners’ emotional experiences.
Enhancing EFL writing with visualised GenAI feedback: A cognitive affective theory of learning perspective on revision quality, emotional response, and human-computer interaction · 2025 · DOIResults suggest that ChatGPT simplification positively influences reading comprehension and inferencing, but its impact on reading anxiety remains inconclusive.
Does AI Simplification of Authentic Blog Texts Improve Reading Comprehension, Inferencing, and Anxiety? A One-Shot Intervention in Turkish EFL Context · 2024 · DOIUsing two-minute trials, two participants improved their typing rate using word prediction software, one participant had mixed results, and the participant with the fastest pre-intervention typing speed had a decreased typing rate with word prediction.
Using Word Prediction Software to Increase Typing Fluency with Students with Physical Disabilities · 2004 · DOI
Most-cited papers in Text Readability and Simplification
- How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment · JMIR Medical Education · 2023 · 1,852 citations
- A SWOT analysis of ChatGPT: Implications for educational practice and research · Innovations in Education and Teaching International · 2023 · 863 citations
- AI-generated feedback on writing: insights into efficacy and ENL student preference · International Journal of Educational Technology in Higher Education · 2023 · 417 citations
- Comparing the quality of human and ChatGPT feedback of students’ writing · Learning and Instruction · 2024 · 371 citations
- Analyzing the role of ChatGPT as a writing assistant at higher education level: A systematic review of the literature · Contemporary Educational Technology · 2023 · 324 citations
- Exploring the opportunities and challenges of NLP models in higher education: is Chat GPT a blessing or a curse? · Frontiers in Education · 2023 · 276 citations
- Performance of GPT-3.5 and GPT-4 on the Japanese Medical Licensing Examination: Comparison Study · JMIR Medical Education · 2023 · 275 citations
- Evaluating Academic Answers Generated Using ChatGPT · Journal of Chemical Education · 2023 · 255 citations
- An Exploratory Study of EFL Learners’ Use of ChatGPT for Language Learning Tasks: Experience and Perceptions · Languages · 2023 · 216 citations
- Exploring Applications of ChatGPT to English Language Teaching: Opportunities, Challenges, and Recommendations · Teaching English as a Second or Foreign Language--TESL-EJ · 2023 · 180 citations
Most recent work
- The impact of generative AI on academic reading and writing: a synthesis of recent evidence (2023–2025) · Frontiers in Education · 2026
- Generative Artificial Intelligence for Automated Qualitative Feedback: A Cross-Comparison of Prompting Strategies · RELC Journal · 2026
- AI writing detectors are ineffective, unreliable and harmful · English Teaching Practice & Critique · 2026
- Clue before correction: ChatGPT-enhanced strategy for promoting autonomous and reflective language learning · Innovation in Language Learning and Teaching · 2026
- ChatGPT-4o as an automated scoring tool for writing assessment: Strengths and weaknesses · International Journal of Assessment Tools in Education · 2026
- Statistical and qualitative analysis of ChatGPT and human raters in preservice teachers’ writing assessment · International Journal of Assessment Tools in Education · 2026
- Evaluating Rater Effects of Large Language Models in Automated Essay Scoring: GPT, Claude, Gemini, and DeepSeek · Educational Measurement Issues and Practice · 2026
- The use of Copilot, Gemini and ChatGPT in the context of foreign language learning and teaching: An academic technology review · Contemporary Educational Technology · 2026
- Exploring AI-Driven Written English Assessment: Toward Improved Assessment Quality and Learner Outcomes · Journal of Language Teaching and Research · 2026
- An Exploratory Evaluation of GPT-4’s Consistency as an English Essay Rater: A Many-Facet Rasch Model Analysis of AI versus Human Rating Patterns · International Journal of TESOL Studies · 2026
Find a gap in your own Text Readability and Simplification sub-topic
This page shows what the Text Readability and Simplification literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.
Open the Research Gap Finder →