This research focuses on creating and evaluating the efficacy of Languafrasa aimed at decreasing plagiarism rates in student essays. Beyond lowering similarity scores, the study positions Languafrasa as an educational writing companion that helps students compare alternative expressions, preserve meaning, and practise responsible paraphrasing in Indonesian academic writing. Languafrasa is designed to incorporate AI-powered paraphrasing functionalities in Indonesian and 102 other languages. The study followed an experimental design involving 30 students who were divided into two groups. One group used conventional essay writing methods, while the other utilized the Languafrasa application. The experimental outcomes revealed a significant decrease in plagiarism levels among students using the Languafrasa application. Additionally, students in the experimental group demonstrated higher efficiency than those in the control group. Feedback from the questionnaire administered to all participants corroborated these results, with most students favoring the application over traditional writing methods. Users found Languafrasa enjoyable, user-friendly, and an effective aid in essay-writing instruction. These findings suggest that Languafrasa contributes not only to textual originality but also to students' confidence and engagement in learning how to paraphrase.
Digital reading environments make students' reading processes more visible because they capture annotations, discussion posts, comprehension attempts, and activity logs. However, these traces are often reduced to participation totals rather than used to inform the design of reading pedagogy. This study developed and evaluated a dialogic digital reading learning environment for Indonesian literacy. Using a design-based mixed-methods approach, 100 undergraduate students engaged with genre-based Indonesian texts over six weeks through interactive annotation, threaded discussion, audio support, and comprehension practice. The dataset comprised 1,842 annotations, 1,126 discussion comments, pretest and posttest reading comprehension scores, platform activity logs, and 20 follow-up interviews. The learning environment was organized around five dialogic engagement dimensions: text-grounded noticing, inferential linking, evaluative stance, dialogic uptake, and metacognitive reflection. Coding showed strong interrater reliability (Cohen's kappa = .86). Reading comprehension increased from pretest (M = 69.84, SD = 9.18) to posttest (M = 83.72, SD = 7.64), t(99) = 15.99, p < .001, with a repeated-measures effect size of dz = 1.60. Dialogic uptake and inferential linking were the strongest predictors of comprehension gain, and the regression model explained 65% of the variance. Interview findings indicated that annotations helped students locate textual difficulty, discussion threads supported peer-mediated interpretation, and comprehension tasks prompted students to return to the text rather than respond superficially. The study offers a pedagogical model for designing Indonesian digital reading environments as social, reflective, and evidence-informed learning spaces.
Kata Kunci:
digital readingdialogic engagementlearning environmentsocial annotationIndonesian literacy
BAHASTRA, 46 (2), 1-14.Abstrak:Kata Kunci:artificial intelligenceparaphrasingplagiarismacademic writingLanguafrasamultilingual
Pedagogy Review, 5 (3), 301-317.Abstrak:Kata Kunci:digital readingdialogic engagementlearning environmentsocial annotationIndonesian literacy