Malang, Indonesia, October 17, 2026 — The growing use of generative artificial intelligence (AI) in academic writing is prompting lecturers to reconsider how students’ writing skills should be assessed. A study by Fitrotul Maulidiyah, Rendi Yoga Saputra, Adzikirani, Rizka Rahmawati, and Rachmat Arnanda found that although many English as a Foreign Language (EFL) lecturers support assessing the writing process rather than focusing solely on final submissions, relatively few have implemented process-based assessment in their teaching.
The study, published in the Journal of Language Development and Linguistics (JLDL) in September 2026, highlights a gap between lecturers’ awareness of AI-related challenges and the assessment practices currently used in higher education. The findings also point to the need for clearer institutional policies, professional training, and assessment methods that recognize students’ independent thinking and learning processes.
AI Offers Benefits but Raises Assessment Concerns
Generative AI tools can help students improve grammatical accuracy, organize ideas, and develop written assignments. However, their increasing availability also creates challenges for lecturers seeking to determine whether submitted work accurately reflects students’ own abilities.
The researchers found that lecturers considered the possibility of AI making plagiarism easier and more difficult to detect the most significant concern, with an average rating of 4.56 out of 5. Concerns about AI potentially reducing students’ critical thinking skills received an average rating of 4.33, while the risk of inaccurate information generated by AI received a rating of 4.17.
At the same time, lecturers recognized the potential benefits of these tools. AI assistance with grammatical accuracy received an average rating of 4.21, while its ability to help students organize ideas received a rating of 4.08.
These findings suggest that AI presents both opportunities and risks for academic writing. Rather than treating the technology solely as a threat, assessment practices may need to distinguish between appropriate AI assistance and the use of AI in ways that undermine independent learning.
Research Examines Lecturers’ Experiences
The study used a convergent parallel mixed-methods approach to examine EFL lecturers’ perceptions and assessment practices in Indonesian vocational higher education.
The researchers collected questionnaire responses from 84 lecturers, supplemented by open-ended responses and follow-up interviews. Participants had between two and 25 years of teaching experience, providing perspectives from lecturers at different stages of their professional careers.
The research explored lecturers’ views on generative AI, their current assessment practices, and the institutional support available to them. By combining quantitative and qualitative information, the study examined both the extent of lecturers’ concerns and the practical challenges they face when assessing academic writing.
A Gap Between Awareness and Practice
One of the study’s most notable findings was the difference between lecturers’ preferences and their actual assessment practices.
Approximately 62% of respondents supported placing greater emphasis on the writing process rather than evaluating only the final product. However, only 29% reported implementing process-focused assessment in their teaching.
The findings also revealed limited institutional support. Only 17% of respondents reported that their institutions had explicit guidelines governing the use of AI in academic writing. Meanwhile, 74% expressed a need for formal policies, training, and resources to help them respond to the challenges associated with generative AI.
Lecturers also questioned whether existing assessment rubrics remained adequate. The statement that current rubrics sufficiently capture student learning in the context of AI use received a relatively low average rating of 2.38 out of 5.
Together, these results indicate that recognizing the importance of fair and meaningful assessment does not automatically translate into classroom practice. Time constraints, large classes, heavy workloads, and unclear institutional policies can make it difficult for lecturers to introduce more comprehensive assessment methods.
Assessing the Writing Process, Not Just the Final Product
To address these challenges, the researchers proposed a process-based assessment framework that evaluates not only the quality of the final text but also the steps students take to develop it.
The proposed rubric allocates assessment weight across five components:
- Idea development and argumentation — 30%: Evaluates the quality of students’ ideas, reasoning, and arguments.
- Process documentation and reflection — 20%: Examines drafts, revision records, and students’ reflections on how their writing develops.
- Language accuracy and register — 20%: Assesses grammatical accuracy and the appropriateness of language for the academic context.
- Critical engagement with sources — 20%: Evaluates how students interpret, assess, and integrate information from relevant sources.
- Ethical disclosure and referencing — 10%: Considers appropriate acknowledgment of sources and transparent disclosure of AI assistance.
This approach could help lecturers gain a clearer understanding of students’ learning and decision-making processes. Instead of relying exclusively on a polished final submission, lecturers could also examine successive drafts, revision logs, reflective commentary, and students’ explanations of their arguments.
Oral explanations may provide another way to assess whether students understand the ideas and evidence presented in their work. Disclosure of AI use could also help establish transparency and clarify how students have incorporated AI-generated suggestions into their writing.
Institutional Policies and AI Literacy Are Essential
The researchers’ findings highlight that improving academic writing assessment requires more than changing grading rubrics. Lecturers also need institutional guidance and practical support to implement new approaches effectively.
Formal AI-use policies could clarify acceptable and unacceptable forms of assistance, while professional development could help lecturers design assessments suited to AI-supported learning environments. Training in AI literacy may also help students evaluate generated information critically, recognize inaccuracies, and understand their responsibilities regarding academic integrity.
The researchers also identified linguistic equity as an important consideration. Students may have different levels of language proficiency and access to digital tools, so assessment practices should avoid assuming that all students have the same resources or technological skills.
By combining process documentation, critical thinking, transparent AI disclosure, and clear institutional guidance, higher education institutions may be better equipped to preserve academic integrity while recognizing the potential benefits of generative AI.
Implications for Higher Education
The study suggests that the future of academic writing assessment will depend on how effectively institutions balance technological assistance with independent learning. AI can support students’ writing development, but the presence of fluent and well-organized text alone may not demonstrate a student’s understanding or writing ability.
Process-based assessment offers one possible response by making students’ reasoning, revisions, source evaluation, and learning decisions more visible. However, implementing this approach requires realistic workload planning, suitable assessment designs, and consistent institutional support.
The researchers noted limitations in the study, including the use of purposive and snowball sampling in Indonesian higher education and reliance on self-reported perceptions and practices. The findings therefore reflect the experiences of the participating lecturers and do not establish how assessment practices change over time through direct, long-term observation.
Nevertheless, the results provide useful insight into the challenges lecturers face as generative AI becomes more common in academic writing. They also underscore the importance of developing assessment systems that evaluate not only what students produce, but how they think, learn, revise, and use technology responsibly.
Author Information
- Fitrotul Maulidiyah — penulis korespondensi, Politeknik Negeri Malang, Indonesia.
- Rendi Yoga Saputra — The University of Nottingham, Nottingham, Inggris.
- Adzikirani — Politeknik Negeri Malang, Indonesia.
- Rizka Rahmawati — Politeknik Negeri Malang, Indonesia.
- Rachmat Arnanda — Politeknik Negeri Jakarta, Indonesia.
Source
Maulidiyah, F., Saputra, R. Y., Adzikirani, Rahmawati, R., & Arnanda, R. (2026). Rethinking Academic Writing Assessment in the Generative AI Era: EFL Lecturers’ Perceptions and the Shift Towards Process-Based Assessment. Journal of Language Development and Linguistics, 5(2), 299–316.
DOI: https://doi.org/10.55927/jldl.v5i2.19018
Journal website: https://journal.formosapublisher.org/index.php/jldl
.jpg)
0 Komentar