Because the study used intact classes rather than random assignment, the findings should be interpreted as quasi-experimental rather than fully causal. Qualitative interviews were also conducted with a subsample of participants to understand their perceptions and the processes involved in engaging with each feedback source. The quantitative results showed significant gains for both the AI and peer feedback groups, but with distinct patterns.

Students who received AI-generated feedback showed a particular advantage in acknowledging different feedback sources. The peer feedback group improved across a broader set of dimensions, including appreciating feedback, acknowledging different feedback sources, and making judgments about the comments they received. In contrast, the teacher feedback group showed limited improvement, with no significant changes in any of the measured dimensions of writing feedback literacy.

The authors note that most previous research on feedback literacy has focused mainly on teacher or peer feedback, leaving open the question of how feedback from different sources may differentially shape students' development in this area. The qualitative interviews helped explain these quantitative patterns. Students working with AI and peer feedback reported greater autonomy and more critical engagement when interpreting and using feedback, whereas teacher feedback did not foster these qualities to the same degree.

The authors suggest that these differentiated effects across feedback sources point to the value of integrating multiple feedback channels in L2 writing instruction, rather than relying on a single source, to optimize how students develop as receivers and users of feedback. Within the scope of this study, combining AI, peer, and teacher feedback appears promising for cultivating different components of feedback literacy. The study's scope carries limitations worth noting.

It involved a specific population of Chinese university students over a single 10-week intervention, and the use of intact classes means pre-existing differences between groups cannot be fully ruled out. The published version is an accepted manuscript shared early to provide faster access, and it remains subject to further edits before the final Version of Record replaces it. The findings therefore describe measurable changes in this particular context rather than universal outcomes, and the authors frame their results as highlighting how different feedback sources can be combined to support writing instruction.

Subject of Research: Effects of AI, peer, and teacher feedback on university students' L2 writing feedback literacy

Article Title: Exploring the impact of AI, peer, and teacher feedback on university students’ L2 writing feedback literacy

Article References: Yan, J., & Meng, Y. (2026). Exploring the impact of AI, peer, and teacher feedback on university students’ L2 writing feedback literacy. Humanities and Social Sciences Communications. https://doi.org/10.1057/s41599-026-09383-2

Image Credits: AI Generated

DOI: 10.1057/s41599-026-09383-2

Keywords: feedback literacy, second-language writing, AI feedback, peer feedback, teacher feedback, quasi-experimental study, Chinese university students, L2 education, writing instruction, mixed methods, Exploring, impact

The study, led by Neda Naqdi Babaee, adopted a nonrandomized pretest-posttest design. Twenty-five learners received ChatGPT-generated feedback while 21 received teacher-written comments from the same instructor, who taught both groups within an identical instructional sequence. Across two complete feedback-revision cycles, students in each condition wrote argumentative and expository paragraphs of roughly 200 words and revised them after receiving comments. Writing was scored by two trained raters, blind to condition and testing phase, using adapted IELTS-based analytic descriptors covering task achievement, coherence and cohesion, lexical resource, and grammatical accuracy, with the four domain scores averaged into an overall composite.

The headline result was clear-cut. After adjusting posttest scores for corresponding pretest performance, the ChatGPT condition showed a robust advantage in overall writing, F(1, 43) = 16.86, p < .001, with a partial eta squared of 0.282. None of the baseline differences between groups was statistically significant, and tests confirmed that the relationship between pretest and posttest scores was comparable across conditions, supporting the validity of the adjustment approach.

The domain-level results demanded more caution. Unadjusted tests favored the ChatGPT condition for task achievement, coherence and cohesion, and lexical resource, but not for grammatical accuracy. Because five related outcomes were examined, the researchers applied a Holm adjustment across the four domain tests as a familywise sensitivity analysis. Only task achievement survived this correction, with an adjusted p value of .031. Lexical resource, at an adjusted p of .057, and coherence and cohesion, at .081, fell just above the threshold and were interpreted as suggestive rather than confirmatory. Robustness checks told a similar story: raw posttest comparisons, nonparametric tests, and Hedges g values ranging from 0.69 to 1.27 supported the ChatGPT advantage for several outcomes, but gain-score analyses reached the conventional significance boundary only for overall writing.

Perhaps the most practically important null finding concerned grammar. Both conditions directed attention to grammatical accuracy, and both groups improved descriptively over the four weeks, yet neither the adjusted models nor any of the sensitivity analyses detected a reliable difference between them. The authors suggest that short interventions may be insufficient to distinguish grammatical development, which typically requires repeated practice and sustained attention to recurring error patterns. The result, they stress, does not establish complementary roles for the two feedback sources; it indicates only that the implemented ChatGPT condition offered no clear grammatical advantage over teacher comments.

The qualitative strand of the study, based on open-ended post-intervention responses analyzed through reflexive thematic analysis, revealed two broad themes: actionable revision support, and trust, control, and feedback authority. Students in the ChatGPT condition repeatedly described the AI comments as immediate and elaborated. One learner explained that the feedback did not simply flag a weak paragraph but showed that an example was insufficient and suggested how to explain it more clearly, helping revise content rather than only grammar. Another said the most useful part was organization, because the system identified where connections between sentences were unclear, allowing new linking words or reordering of ideas — a shift, the student noted, from a previous habit of only checking grammar.

Vocabulary emerged as a particularly revealing area of learner experience. Students reported that ChatGPT offered alternatives for repeated or imprecise wording, expanding their options for academic expression, but they also emphasized the responsibility to judge appropriateness. One participant noted that some suggested words were too formal for their level and had to be checked against the intended sentence. This combination of availability and selection — a wider menu of choices paired with the learner’s own evaluative filtering — recurred throughout the accounts and is consistent with the favorable but statistically inconclusive lexical pattern in the quantitative results.

Trust worked differently across the two conditions. ChatGPT users valued speed and detail but did not treat the output as uniformly authoritative; several reported that suggestions sometimes altered their intended meaning or produced grammatical advice that made sense only in context, so they read carefully and rejected comments that distorted their ideas. Some linked immediacy to momentum, saying that instant access let them revise while their intended meaning was still fresh, and one student reported feeling more independent and confident. Teacher feedback recipients, by contrast, trusted comments because the instructor knew the course, the task, and the assessment expectations — but that authority could narrow revision. Several said they corrected only the parts the teacher marked and did not explore alternative versions, appreciating the prioritization while sometimes wanting more explanation or examples for content expansion and lexical refinement.

The authors are explicit that these findings concern implemented feedback conditions, not feedback source in isolation. The ChatGPT condition bundled AI generation with rapid access, extensive comments, and revision alternatives, while the teacher-written condition was more selective and contextually situated, so the estimated difference may reflect differences in amount, timing, specificity, and elaboration rather than the mere presence of AI. The study also cannot claim complementarity, because it included no combined ChatGPT-plus-teacher condition; a sequenced approach remains a pedagogical proposition for future testing rather than an empirical conclusion. Additional limitations include the nonrandomized allocation, the modest single-university sample, the four-week duration without a delayed posttest, and the fact that the qualitative strand captured perceptions rather than observed draft-to-draft uptake.

For educators navigating the rapid arrival of generative AI in language classrooms, the study offers a measured message. AI-generated feedback appears most useful when learners need elaborated options for task fulfillment, organization, or lexical revision, provided instruction requires them to evaluate suggestions against meaning and task criteria. The null grammatical result warns against assuming such tools can replace explicit grammar instruction or sustained corrective-feedback practice, and teacher feedback retains a distinct role in prioritizing problems and aligning revision with course goals. What the research most clearly demonstrates is not that ChatGPT is a superior teacher, but that the way feedback is delivered — how much, how fast, and how elaborated — may matter as much as who or what produces it.

Subject of Research: Comparing ChatGPT-generated and teacher-written feedback effects on EFL university students' writing development

Article Title: Comparing ChatGPT and teacher feedback for EFL writing development

Article References: Babaee, N. N. (2026). Comparing ChatGPT and teacher feedback for EFL writing development. Discover Education, 5(1), Article 1211. https://doi.org/10.1007/s44217-026-02329-5

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02329-5

Keywords: ChatGPT, generative AI, EFL writing, written corrective feedback, second language writing, teacher feedback, quasi-experimental study, ANCOVA, learner perceptions, task achievement, lexical resource, grammatical accuracy

The episode comes from a new study published in Discover Education by Eshetie Kasie, an Ethiopian TEFL practitioner and researcher at Bahir Dar University. The research followed ten undergraduate English as a foreign language (EFL) students across one full academic semester as they used generative AI, mostly through mobile phones and often through patchy internet connections, to manage writer’s block. The findings extend an influential framework for critical AI literacy with two new layers, one emotional and one structural, and they offer a rare Global South perspective on a conversation that has been dominated by classrooms in wealthy, well-connected countries.

Writer’s block is not a trivial complaint. Researchers describe it as a tangle of cognitive overload, emotional strain, linguistic difficulty, and conflicts of identity that surface during composition, and these factors intensify in EFL contexts where students face limited proficiency, cultural expectations, and high-stakes assessment. A student who struggles to retrieve vocabulary in a second language often experiences that struggle as anxiety rather than as a purely linguistic gap, which then compounds the original difficulty. In Ethiopia, where English is the medium of instruction but Amharic and other local languages dominate everyday life, the blank page can feel especially unforgiving. Generative AI promised relief, but the study shows that the relief was conditional, arriving only when students learned to interrogate the tool rather than simply obey it.

The study’s theoretical anchor is the Affective Process and Strategic Engagement (APSE) framework, developed by researchers from first-year second-language writing classrooms in the United States. APSE identifies four interrelated dimensions of critical AI literacy: awareness of what AI can and cannot do, positionality toward its use, strategies for interacting with it, and evaluation of its output. But the framework was built in a setting that assumes stable internet access and Anglocentric rhetorical norms, assumptions that collapse quickly in a university where power outages are routine and connectivity is mobile-only. Kasie’s extension, which he calls APSE-Extended for Block, keeps the four core dimensions and wraps them in an inner affective layer addressing anxiety reduction, confidence building, and identity resilience, and an outer equity layer addressing connectivity constraints, linguistic hierarchies, and resistance to Anglocentric assumptions embedded in AI outputs.

The methodology was deliberately multimodal. Kasie purposefully selected ten undergraduates, five women and five men spanning first to third year, all of whom screened as experiencing moderate or high writer’s block severity. Over fourteen to sixteen weeks, he gathered semi-structured interviews before, during, and after the semester, weekly reflective journals, anonymized ChatGPT conversation logs, successive writing drafts, and classroom observations. Amharic portions of journals and interviews were professionally translated and checked against the originals. To quantify how critically students engaged with the technology, Kasie compared every AI-generated output in the conversation logs against the corresponding segment of each student’s revised draft, coding outputs as accepted, substantially revised, or rejected, and then corroborating those judgments in stimulated recall sessions where students explained their decisions.

The developmental pattern that emerged was strikingly consistent. Most participants began the semester trusting ChatGPT almost unconditionally for English writing tasks, then encountered inaccuracy, cultural misrepresentation, and Anglocentric bias the moment they applied it to Ethiopian content. One student asked for help drafting a reflection on Timkat, the Ethiopian Orthodox celebration of Epiphany, and received a generic description of a religious festival that omitted the procession of the tabot and the ritual blessing of water entirely. The student responded by researching and rewriting the section from memory and family accounts. By semester’s end, eight of the ten participants treated AI outputs as provisional starting points requiring cultural filtering rather than finished material.

Positionality, the framework’s term for how writers situate themselves relative to the tool, proved equally consequential. Eight participants actively reclaimed authorship and resisted perceived Western or native-speaker bias in AI outputs. A third-year student named Selam, given a pseudonym in the study, described how the AI consistently framed gender issues through a Western feminist lens, presenting women’s oppression as uniform across cultures. That framing, she said, did not reflect her own story, and she had to decide for herself what was true for Ethiopian women. Students revised or rejected content that flattened communal and relational framings common in Amharic argumentative writing into individualistic Western argument structures. A second-year student, Kidist, rejected outright an AI characterization of Ethiopian schools as underdeveloped and backward, rewriting the passage to say instead that her schools faced limited resources but also showed innovation and resilience.

The quantitative picture reinforces the qualitative one. Based on the draft-log comparisons, nine participants rejected an estimated 60 to 80 percent of AI-generated content overall, particularly when it appeared biased, generic, or culturally inappropriate, and they did not simply delete unwanted material but rewrote it to reflect their own framing and tone. Nine students also developed structured prompting strategies that functioned as affective scaffolding for block relief. A first-year student, Dawit, described asking ChatGPT for three short ideas about Ethiopian youth unemployment without full sentences, then using those fragments to begin his own writing. Others requested a single counterargument to test against their thesis, or an outline skeleton they filled in entirely with their own wording. All ten participants reported that careful, limited use of generative AI reduced anxiety and increased confidence, with one student describing AI as a helper, not a boss.

Yet the equity layer of the extended framework tells a sobering story. All ten participants faced persistent connectivity challenges and linguistic mismatch. Dawit explained that when the internet cuts off, his ideas disappear and he feels blocked again, a description of how an infrastructure gap translates directly into a personal interruption of the writing process. Students often drafted offline and revised only once connectivity returned. The study connects this lived constraint to broader scholarship: a systematic review of AI use in Global South EFL classrooms found that AI holds real potential for student autonomy only when implementation is context-sensitive and equity-driven, while research on South African higher education cautions against treating AI literacy as a purely technical skill separable from local conditions. Notably, one first-year student with high block severity copied full paragraphs from AI outputs without revision all semester, suggesting that block severity, not just year of study, shapes how readily students resist algorithmic norms.

The practical payoff is a four-phase classroom cycle that educators in similar contexts can adapt: an awareness phase in which students experiment with AI on local topics and discuss cultural mismatches as they arise; a positionality phase in which they revise AI content to incorporate Amharic rhetorical patterns and local perspectives; a strategies phase introducing spark prompting for ideas only, preserving ownership of the text; and an evaluation phase in which students apply a personal checklist judging whether an output respects Ethiopian realities and sounds like their own voice. Kasie is careful about the limits of his findings, which remain bound to ten students at a single university over one semester, and he calls for longitudinal and comparative research across urban and rural institutions. But the core insight travels well beyond Bahir Dar. Critical AI literacy, the study suggests, is not just a technical skill of writing better prompts. It is an emotional and political practice in which learners decide what the machine may say about them, and in which a coffee ceremony, a religious festival, or a school described as backward can become the moment a student stops outsourcing judgment and starts writing as themselves.

Subject of Research: Critical AI literacy and writer's block among Ethiopian EFL undergraduates using generative AI in low-resource settings

Article Title: Extending the affective process and strategic engagement framework to address writer’s block in Ethiopian English as a foreign language contexts through critical artificial intelligence literacy and ethical generative artificial intelligence in low-resource settings

Article References: Kasie, E. (2026). Extending the affective process and strategic engagement framework to address writer’s block in Ethiopian English as a foreign language contexts through critical artificial intelligence literacy and ethical generative artificial intelligence in low-resource settings. Discover Education, 5(1), Article 1133. https://doi.org/10.1007/s44217-026-02049-w

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02049-w

Keywords: critical AI literacy, generative AI, writer's block, EFL writing, Ethiopia, Bahir Dar University, APSE framework, Global South, equity, ChatGPT, second language writing, digital divide