Earticle

본 연구는 고차적 사고가 요구되는 영어 설득작문 과제에 스캐폴딩과정을 적용할 필요성 및 생태학적 타당도가 높은 스캐폴딩의 과정을 분석할 필요성에 근거하여, 영어작문수업 과정에서 이루어지는 교사와 학생, 학생과 학생간의 언어적 상호작용, 학생들의 작문전략 활용 및 작문수행의 결과를 관찰분석제시하여 보다 총체적이고 심층적으로 학습과제의 스캐폴딩 과정을 검토하였다. 본 연구는 K고등학교 학생 28명의 작문학습의 과정을 대상으로 질적 접근과 양적 접근을 병행하였다. 본 연구에서 분석된 학교의 사회문화적 맥락, 교사의 감수성과 교수법, 학습자의 수준별 과제접근 가능성 등은 스캐폴딩의 현장적용가능성을 보여주었다.

The purpose of this study is to analyze the process of scaffolding in English(EFL) writing tasks based on Vygotskian perspectives. Qualitative methods were mainly used to build the large picture of scaffolding process in English writing. Quantitative methods were secondarily used to supplement the qualitative results. A science high school in which gifted and quasi-gifted students attend was selected as a research site. An English teacher, 28 students and a native speaker(American) participated in the research. Eight focal students were observed and interviewed in depth. The findings of the study are as follows; First, macroscaffolds in English writing class were students' ownership, task appropriateness, the sequence of learning, collaborative interaction and internalization. Second, microscaffolds found in the process of dialogic scaffolding in English writing class differently worked according to student's level, writing strategy and rhetorical structure. Third, there were significant differences between the writing performance before scaffolding and the one after scaffolding in terms of the results of holistic and analytic scoring. Regarding differential effect, instructional scaffolding was more effective for beginning level students. The rhetorical elements such as concluding paragraph, overall cohesion and use of conjunctions in particular showed relatively low improvement. Fourth, in terms of research method, this study indicated that ecologically relevant scaffolding research was possible by holistically and deeply observing, recording and analyzing the whole process of a class for a semester.

Federated learning provides an efficient integrated model for distributed data, allowing the local training of different data. Meanwhile, the goal of multi-task learning is to simultaneously establish models for multiple related tasks, and to obtain the underlying main structure. However, traditional federated multi-task learning models not only have strict requirements for the data distribution, but also demand large amounts of calculation and have slow convergence, which hindered their promotion in many fields. In our work, we apply the rank constraint on weight vectors of the multi-task learning model to adaptively adjust the task's similarity learning, according to the distribution of federal node data. The proposed model has a general framework for solving optimal solutions, which can be used to deal with various data types. Experiments show that our model has achieved the best results in different dataset. Notably, our model can still obtain stable results in datasets with large distribution differences. In addition, compared with traditional federated multi-task learning models, our algorithm is able to converge on a local optimal solution within limited training iterations.

Despite the growing interest in using deep reinforcement learning (DRL) for drone control, several challenges remain to be addressed, including issues with generalization across task variations and agent training (which requires significant computational power and time). When the agent’s input changes owing to the drone’s sensors or mission variations, significant retraining overhead is required to handle the changes in the input data pattern and the neural network architecture to accommodate the input data. These difficulties severely limit their applicability in dynamic real-world environments. In this paper, we propose an efficient DRL method that leverages the knowledge of the source agent to accelerate the training of the target agent under task variations. The proposed method consists of three phases: collecting training data for the target agent using the source agent, supervised pre-training of the target agent, and DRL-based fine-tuning. Experimental validation demonstrated a remarkable reduction in the training time (up to 94.29%), suggesting a potential avenue for the successful and efficient application of DRL in drone control.

Purpose: We examined the effect of dual-task and single-task training on serial reaction time (SRT) task performance to determine whether SRT is based more on motor or perception in a dual-task. Methods: Forty healthy adults were divided into two groups: the dual-task group (mean age, $21.8{\pm}1.6$ years) and the single-task group (mean age, $21.7{\pm}1.6$ years). SRT task was conducted total 480 trial. The four figures were presented randomly 16 times. A unit was set as 1 block that would repeat 10 times. Thus, there were a total of 160 trials for each of the three color conditions. The dual-task group performed an SRT task while detecting the color of a specific shape. The end of the task, subjects answered the specific shape number; the single-task group only performed the SRT task. The study consisted of three parts: pre-measurement, task performance, and post-measurement. Results: Differences of pre and post reaction time between two group was higher for the dual-task group as compared to the single task group and there was a significant interaction between time and group (p<0.05). Conclusion: Our results indicate that. short term period SRT is not quiet effective under dual-task conditions, individuals need additional cognitive processes to successfully navigate a task This suggests that dual-task training might not be appropriate for motor learning enhancement, at least when the training is over a short period.

The purpose of this study is to examine the relationship between collaborative information seeking and users' learning style preferences and their experience of information systems. The study investigates the role of four different factors including learning style, task complexity, and user experience in collaborative information seeking in digital environments. Sixty participants (30 pairs) were randomly chosen from volunteer graduate students of Kharazmi University (Iran). Participants completed Kolb's learning style questionnaire and a user experience questionnaire and then performed two information seeking tasks (one simple and one difficult) in a lab setting. They could exchange information with their partners or a librarian using Skype. The sessions were recorded using Camtasia. The results showed that with an increase in task difficulty, collaborative information seeking activities increased and more interactions with partners and the librarian occurred. The number of executive help-seeking requests was higher than the number of instrumental help-seeking requests. This research confirms that learning style is related to the way users interact with the digital library and help seeking. The research showed that in difficult tasks, the differences among users with different learning styles become more evident, and that generally interactions increase in more difficult tasks. Among the learning styles, the accommodating style had the highest number of relationships with collaborative information seeking variables. Most of the statistically significant relationships between users' prior computer experience and collaborative information seeking variables were related to the time variable.

Large-scale pre-trained vision models such as ViT, CLIP, and SAM provide strong foundations for diverse vision tasks, motivating recent Mixture-of-Experts (MoE) approaches that combine multiple experts. However, existing methods often rely on static or implicit routing strategies, limiting adaptability to task semantics and input characteristics. We propose a task-adaptive vision expert routing framework based on competency learning guided by predictive uncertainty. We define expert competency as the relative reduction in predictive uncertainty induced by inter-expert interaction, and formulate expert routing as a learning problem driven by this signal. Our method uses task embeddings derived from textual descriptions to guide expert routing, refines expert features through cross-expert interaction, and aggregates them adaptively into a unified representation. By directly optimizing routing and feature composition using an uncertainty-based competency signal, the model learns how expert collaboration improves task-specific prediction reliability. Extensive experiments on diverse vision tasks demonstrate superior generalization performance and adaptive routing behavior aligned with task semantics.

In this paper, end devices are considered here as agent, which makes its decisions on whether the network will offload the computation tasks to the edge devices or not. To tackle the resource allocation and task offloading, paper formulated the computation resource allocation problems as a sum cost delay of this framework. An optimal binary computational offloading decision is proposed and then reinforcement learning is introduced to solve the problem. Simulation results demonstrate the effectiveness of this reinforcement learning based scheme to minimize the offloading cost derived as computation cost and delay cost in industrial internet of things scenarios.

Task offloading in multi-access edge computing (MEC) systems is critical for managing computational tasks in dynamic urban environments. Existing strategies face challenges such as high communication overheads and regional performance deviations including centralized and distributed methods. Clustering approaches have been explored to address these issues, yet they often rely on physical proximity to form clusters, overlooking the variability in task rate distributions across edges. To overcome these limitations, this paper proposes a graph-driven inter-cluster resource distribution (GIRD) clustering scheme that clusters edge nodes based on task request distribution and computing resource status, ensuring similar resource utilization across clusters. Building on this, a proximal policy optimization (PPO)-enabled intra-cluster task offloading algorithm (PITO) is introduced to determine one execution server for task offloading?either an edge server within a cluster or a cloud server?using various network state information. This dynamic decision-making process optimizes a multi-objective function that includes task processing delay, consumed energy, success rate, and cloud cost. Simulation results demonstrate the proposed GIRD-PITO framework achieves superior task success rates, reduced delays, and improved regional performance fairness, making it a promising solution for large-scale MEC systems.

Background: Deficiencies in the ability to maintain balance are common in elderly. Augmented feedback such as knowledge of results (KR) can accelerate learning and mastering a motor skill in older people. Objects: We designed this study to examine whether one session of Wii-Fit game with self-regulated KR is effective for elderly people, and to compare the effect of two different timings of self-regulated KR conditions. Methods: Thirty-nine community-dwelling elders, not living in hospice care or a nursing home, participated in this study. During acquisition, two groups of volunteers were trained in 10 blocks of a dynamic balancing task under the following 2 conditions, respectively: (a) a pre-trial self-regulated KR ($n_1=18$), or (b) a post-trial self-regulated KR ($n_2=21$). Immediate retention tests and delayed retention tests of balancing performance were administered in 15 minutes and 24 hours following acquisition period, respectively. Results: In both groups, significant improvements of balancing performances scores were observed during the acquisition period. Regardless of the group, mean of balancing performance scores on retention tests were well-maintained from the final session. There were no significant differences between groups in balancing performance scores during the acquisition period (p>.05); however, the post-trial self-regulated KR group exhibited significantly higher balancing performance scores in both the immediate retention test and delayed retention test than that of the pre-trial self-regulated KR group (p<.05). Conclusion: Therefore, subjects who regulated their feedback after a dynamic balancing task, during the acquisition period, experienced more efficient motor learning during the retention period than did subjects who regulated their feedback before a dynamic balancing task. Accordingly, in case of presenting the KR of motor learning in clinical settings to elders who reduced dynamic balance abilities, the requesting time of KR is imperative according to self-estimation processes as well as types of KR and practice.

This paper presents a real-time vehicle detection and tracking system using an unmanned aerial vehicle (UAV) to address challenges in dynamic urban environments. The system combines a convolutional neural network (CNN) for vehicle detection with a deep Q-network (DQN)-based navigation policy for continuous tracking. Input images are enhanced using contrast limited adaptive histogram equalization (CLAHE) and unsharp masking. The CNN jointly predicts vehicle center coordinates and probabilistic heatmaps, while a self-attention module captures long-range spatial dependencies to improve detection under clutter and occlusion. The DQN is trained on multi-step spatiotemporal states to learn optimal UAV movement strategies under diverse weather and structural conditions. Experiments conducted in a three-dimensional (3D) urban simulation environment using Unity’s machine learning agents (ML-Agents) show that the self-attention design reduced pixel-level localization error by about 7%, and the DQN-based tracking policy achieved stable convergence after approximately 2000?3000 episodes. These results demonstrate high tracking accuracy and system stability, highlighting the potential of the proposed approach for real-world UAV-based traffic monitoring applications.

Background: This pilot study aims to develop an automated, multi-task deep learning system as a proof of concept for both the classification of pregnancy status and gestational day prediction in Korean crossbred goats. Methods: Two 1-year-old goats were followed longitudinally until mid to late gestation. A complete data set containing 11,092 high-resolution image frames (6,647 pregnant and 4,445 non-pregnant) was created from the video sequences. The dataset was segmented at the frame level with 80% allocated to the training partition (n = 8,875), 10% to the validation partition (n = 1,110), and 10% to the test partition (n = 1,107). Multi-task Four convolutional neural networks (ResNet18, EfficientNet-B0, MobileNetV3-Large, and ConvNeXt-Tiny) were trained following a multi-task learning strategy with the AdamW optimizer, OneCycle scheduling, and a hybrid Binary Cross- Entropy and Huber loss function. Results: Accuracy was achieved for the four architectures, with more than 99.5%. The MobileNetV3-Large model performed the best in terms of frame-level classification accuracy, with 100% sensitivity and specificity. For the regression-based task for predicting gestational days, EfficientNet-B0 got the best performance with a framelevel Mean Absolute Error (MAE) of 1.3 days. The video-level MAE of 0.882 days and the competitive video MAE of 1.204 days were the outcomes for the aggregated video sequences using ConvNeXt-Tiny and EfficientNet-B0, respectively. Conclusions: The present pilot study lays the foundation for the methodological feasibility of small optimized deep learning models for accurate monitoring of fetal development and for the automation of reproductive diagnostics.

This study examines the effects of learning conditions (implicit learning condition and explicit learning condition) and task types (comprehension and production task) on the learning of an easy rule (subject-verb inversion in the sentence with location adverbial) and a hard rule (pseudo-cleft clause). One hundred thirty first-year university EFL students with low proficiency were divided into four groups based on the combination of learning conditions and task types. After that, pre-and post-test data on the grammaticality judgement test and controlled written test were analyzed by using the two-way Multiple Analysis of Covariance (MANCOVA). Results showed that the effects of learning conditions are not statistically different on the comprehension of easy and hard grammatical rules whereas the explicit learning condition has more effects on producing easy and hard rules than the implicit learning condition does. They also revealed that the effects of task types are statistically different on the learning of an easy and a hard rule. In addition, production task has more effects on the learning of both an easy and a hard rule than the comprehension task. It could be suggested that the results of this study could be applied in designing more effective instructional condition for the EFL adult learners with low proficiency.

This study aims to identify and investigate the role of task value (drawn from expectancy-value theory), surface/deep learning strategies, and critical thinking for predicting language learners’ selfregulation and ultimate language achievement in a single framework. To this end, 190 participants, majoring in English at bachelor’s level at a state university in Iran, were selected based upon convenience sampling. A structural equation model showed that both motivational and cognitive components were significant predictors of language achievement and self-regulation. The results reveal that task value is a positive predictor of deep/surface learning strategies and metacognitive selfregulation, which provides evidence that awareness of the importance and value of a task, directs learners to choose more appropriate self-regulation strategies to monitor their process of learning and structure their own cognitive manners. The findings demonstrated that critical thinking predicts improvement in language achievement. Theoretically speaking, it has been argued that the ability to critically analyze learning materials and evaluate information enables learners with advanced proficiency to actively take part in the learning process, be further dedicated to their tasks, and more effectively and efficiently engaged in learning strategies, therefore making more successful learners. Finally, the implication for teachers and policy makers is discussed.

With the emergence of sociolinguistic approach in second language learning, instructors have begun to acknowledge the native language (L1) as a legitimate tool with the potential to facilitate second language (L2) learning mainly in output-based tasks. This study focuses on how the use of L1 by L2 learners affects their task achievement. Two communicative tasks were used, and tasks were carried out by 32 high school students in L2 only, or in L1 and L2 during group test sessions held on different days. Tape-recorded transcripts of learners when performing the tasks, interviews, and questionnaires were collected and analysed to investigate effect on the use of L1. The findings suggest that during performing the tasks, learners used their L1 to complete their tasks for a variety of functions. Through the L1, they explained and negotiated the task each other, or checked their understanding or compared answers to the task against their peers. To allow the learners to use of their L1 was even more effective than to urge them to use L2 only in a foreign language class using task. Further investigation indicates that, if one of goals of output-based introduction is considered as the successful completion of the task, the learners’ use of L1 may be beneficial to attain the goal when performing completely in L2 is impossible or beyond the learners’ linguistic ability.

Note-taking tasks can be considered a process to achieve communication by helping learners organize linguistic information they strive to convey through cognitive activities. To successfully communicate, learners must develop various linguistic and cognitive skills, such as advanced listening proficiency and the reconstruction of intended stories. This study aims to discover the major types of errors affecting university-level learners’ note-taking activities and a teacher’s scaffolding guidance in order to foster note-taking skills and learning strategies. Seventy-nine university students enrolled in an audio visual English class took part in this study. They were required to routinely carry out note-taking activities by watching both CNN news reports and academic lectures dealing with social issues. This study found that the learners routinely missed a substantial number of content words despite repetitive note-taking tasks. In addition, important phrases were omitted due to variations in pronunciation, unfamiliar proper nouns, and complex numbers. It was also revealed that students appeared to struggle with several abbreviations and symbols when reconstructing the meaning of the information. In order to tackle these problems, various teaching strategies were introduced and implemented to increase note-taking effectiveness as well as maximize learners’ confidence and help them more accurately grasp the content. Based on these results, this study recommends providing structured scaffolding for learners to perform effective and efficient note-taking tasks. This will have a positive influence on learning strategies. It will also encourage students to actively participate in tasks and produce well-organized information on given subjects, which require linguistic information processing.

The purpose of this study is to examine whether the difference in involvement load induced by different vocabulary learning tasks affects Korean EFL adult learner’ vocabulary short-term learning and long-term retention. This study also investigates how different tasks (Reading, Gap-filling, Unscrambling) with different involvement loads affect vocabulary learning of learners at high proficiency level. For this study, a total of 35 EFL adult learners participated, they were randomly assigned to one of three groups to complete an assigned task. To measure the effectiveness of different task-induced involvement, a pre-test was administered to check both the participants’ pre-knowledge regarding the 10 target words and to prove homogeneity of three groups. Then, two post-tests were implemented, one immediately after the treatment and one two weeks after to measure their short-term learning and long-term retention on the target words. For the adult learners, the difference of involvement load was a significant factor in short-term learning, whereas it was partially supported for long-term retention, because Task 2 was more influential than Task 3. Therefore, the present study suggests that a more effective task should be considered for adult learners, who have different amount of time to prepare tests or study English, in teaching environments.

Task–based reading activities are of crucial value today, and consequently learners’ proficiency is more important than their abstract knowledge of language rules. It seems that learners’ familiarity with task-based reading activities may increase learners’ proficiency. Therefore, this study investigated the effect of task-based reading activities such as text completion and pupil generated questions on vocabulary learning and retention of Iranian intermediate EFL learners. To conduct the study, three intact classes of learners who had already finished Top Notch Fundamental A and B (Saslow & Ausher, 2011) in previous semesters in an English language institute were selected as the participants of the study. To ensure the homogeneity of the participants, those who got a score between 30-47 from the total score of 60 in OPT were selected as the intermediate level for main participants of the study (N=47). As the data were normally distributed, one way ANOVA and repeated measure ANOVA were employed for the statistical analyses of the study. The findings indicated that using task-based reading activities such as text completion and pupil-generated questions has significant and meaningful impacts on Iranian EFL learners’ vocabulary learning and retention. The implementations of the study are discussed.

The present study explored the effects of different task types on the autonomous vocabulary learning of EFL learners. To this end, two types of tasks with similar but slightly different levels of involvement loads were developed: a corpus-based task and a word map task. The participants of the study were 41 EFL students from a university in Seoul, Korea. They were divided into two groups: a corpus group (n=20) and a word map group (n=21). A total of 33 target words were selected and the experiment was administered during a period of five weeks. Major findings from the study were as follows: First, the corpus task proved to be more effective than the word map task in productive vocabulary learning, whereas the two tasks showed similar effectiveness in receptive vocabulary learning. Second, the two tasks varied in the effectiveness of vocabulary learning depending on the learners' vocabulary proficiency levels and the specific domains of vocabulary knowledge (receptive and productive). Lastly, the corpus group responded more positively than the word map group to their task and to the vocabulary learning. The findings indicate that the corpus task has advantages over the word map task and provide ‘residual’ empirical support for Hulstijn and Laufer’s (2001) involvement load hypothesis. Based on the findings, some pedagogical implications were made for the effective teaching and the autonomous learning of vocabulary.

This study aims to investigate the impact of topic familiarity and vocabulary learning tasks on vocabulary acquisition in an ESP (English for Specific Purposes) setting, focusing on 16 adult learners with varying proficiency levels (A1-B1), all training to be cooks. Participants worked with six cooking recipes, three on familiar topics and three on unfamiliar topics, with 30 target words assigned to one of three tasks: word list, flashcards, or fill-in-the-gaps. Immediate and delayed post-tests (three weeks later) were administered using the Vocabulary Knowledge Scale (VKS). Despite the small sample size, significant differences in retention are noted across proficiency levels and task types over time. Results from repeated-measures ANOVAs showed that topic familiarity alone did not significantly impact vocabulary acquisition. However, the interaction between topic familiarity and task type was significant, as were the effects of task types and proficiency levels individually. The interaction between task type and proficiency level also showed a significant effect on vocabulary learning. Participants generally performed better with the fill-in-the-gaps method, but these effects decreased over time. The study highlights the importance of considering both proficiency levels and task types for effective vocabulary instruction.

This study examined how three task activities with receptive (R) reading, cloze (CZ) and receptive-productive (RP) integration differently lead to L2 idiom recall and retention by Korean EFL adult learners when they are enhanced with bolding, L1 glossing and three repetitions. It further looks into what features of the enhanced input tasks influence L2 idiom gains. The theories of input enhancement, receptive and productive task input are the basis for the present study. The results revealed that overall the three tasks contributed to high score of L2 idiom gains in immediate receptive, productive recall and delayed receptive retention, except delayed productive retention. There was no significant difference in idiom recall and retention by the three tasks. The descriptive analysis showed R group performed better on immediate receptive recall, while CZ group led the highest score of immediate productive recall. In terms of retention gap between immediate and delayed tests, the three tasks yielded high percentage of receptive idiom retention. The CZ task yielded the highest percentage of receptive retention but relatively low percentage of productive retention, while the R group retained most memory of productive knowledge. The results suggest some implications for effective L2 idiom teaching materials using input enhanced tasks such as bolding, L1 glossing and repetition.