ETR&D

This study examined the usefulness and impact of AI-enabled automated essay scoring (AES) in providing formative feedback on student writing drafts. Despite the well-documented importance of writing instruction and feedback, educators face significant challenges in delivering timely, actionable feedback to students due to time constraints, large class sizes, and workload pressures. These barriers often limit students’ opportunities to practice writing and receive meaningful feedback on their writing. To address this problem, this research investigated WriteGrader, an AI-powered AES tool designed to score student essays and provide accurate and timely feedback. Results demonstrated strong agreement between WriteGrader scoring and expert human ratings, indicating reliable scoring accuracy. Teachers evaluated the AI-generated feedback as highly accurate and actionable, particularly for criteria related to organization, mechanics, and evidence use. Most significantly, students who received formative AI feedback on drafts showed measurable improvements in their final submissions compared to initial drafts. These findings suggest that AI-enabled AES tools can serve as effective teaching assistants, providing students with immediate, consistent feedback while alleviating teacher workload. The study contributes to understanding how generative AI can be integrated into writing instruction to support student learning and teacher effectiveness.

Multimedia reading media has become a vital tool in children's education. This study investigates the influence of augmented reality (AR) picture books and electronic picture books on children's popular science reading. Sixty-two children aged 5–6 were randomly assigned to either an AR or an electronic groups. Over 2 weeks, both groups completed four rounds of popular science picture books reading over 2 weeks, with consistent content in each round. Tests were conducted after the first and fourth rounds, while the middle two rounds involved reading only. The results revealed a divergence in performance between the two groups. The electronic group demonstrated stronger outcomes in learning target popular science vocabulary, explicit reading comprehension, and physical engagement, In contrast, the AR group exhibited more pronounced strengths in implicit reading comprehension, intrinsic reading motivation, verbal engagement and emotional engagement. Additionally, physical and verbal engagement played mediating roles in the process through which multimedia reading media influenced popular science reading comprehension. The divergence in outcomes can be explained by established pedagogical theories. For the electronic group, their stronger performance in vocabulary and explicit comprehension aligns with cognitive load theory. The linear presentation and immediate feedback of electronic books likely reduced extraneous cognitive load, allowing young learners to allocate more cognitive resources to foundational knowledge acquisition. In contrast, the AR group’s advantages in implicit comprehension and intrinsic motivation are well-supported by self-determination theory and principles of multimedia learning. By providing immersive, three-dimensional visualizations and interactive scenarios, AR books fulfilled children’s psychological needs for autonomy and competence, thus boosting intrinsic motivation. This enriched, embodied interaction also fostered a deeper, more integrated implicit understanding of the popular science concepts. This study, therefore, provides empirical evidence for selecting developmentally appropriate media and offers important implications for fostering children's academic language development.