Cooperative Learning and Academic Writing Skills: An Application of the Collective Working Memory Effect

Abstract

Cognitive Load Theory (CLT) has primarily been applied to individual learning, while research on collaborative learning under CLT remains inconclusive. This experimental study investigated the effect of collective working memory on cognitive load, writing performance, and learning efficiency among 150 Indonesian undergraduates. Participants were assigned to either an individual learning (control) or cooperative learning (experimental) condition. Baseline writing performance differed between groups, so gain scores were analyzed. Results showed that both groups improved significantly in writing performance, with no significant differences between conditions. Cognitive load increased slightly in the cooperative learning group, while learning efficiency showed a small positive trend. These findings suggest that cooperative learning may support writing performance and efficiency in complex tasks, although clear advantages over individual learning were not established. Further research is needed with balanced baseline groups and longer interventions.

1. Introduction

Human cognitive architecture, within the framework of Cognitive Load Theory (CLT), consists of working memory, long-term memory, and the connection between them (Sweller et al., 2019). Working memory has limited capacity, and it is typically able to process only three to five elements of new domain-specific information at once (Cowan, 2008, 2010). In contrast, long-term memory has unlimited capacity and duration, allowing information to be stored after being processed in working memory (Cowan, 2008; Sweller, 2003).

CLT, pioneered by Sweller et al. in the early 1980s, has been shown to be a useful tool for creating instructional design in relatively well-structured procedures and concepts (Sweller et al., 2019). It has also been successfully used to generate a variety of learning objectives. However, P. A. Kirschner et al. (2018) argued that CLT has typically always been associated with individual learning and rarely applied or adopted within the collaborative instructional design. In addition, F. Kirschner et al. (2009), who pioneered the concept of collaboration within the cognitive load framework, stated that there are quite a number of studies on collaboration, but the results are inconclusive. In some studies, highly structured or scripted collaboration could make learners more actively engaged in the learning activities, e.g., self-directed learning, verbalizing explanations, etc. However, there were also negative results such as social loafing, ineffective information sharing known only by individual group members, etc. Moreover, many studies focus on group performance rather than individual learning outcomes (F. Kirschner et al., 2009; Zambrano et al., 2019b).

Loyens et al. (2008) further point out that inquiry-based instructional approaches, integral to collaborative learning, encourage active engagement, and the development of transferable skills. However, the concept of collective working memory remains underexplored in collaborative learning contexts (F. Kirschner et al., 2011). The impact of collaboration is often determined by the task design and the structure of the collaboration. Highly structured tasks that promote active engagement, such as verbalizing explanations and self-reflection, have been shown to enhance learning efficiency (F. Kirschner et al., 2009; Sankaranarayanan et al., 2021). Conversely, ill-structured tasks with unrestrained communication can lead to inefficiencies in group learning (F. Kirschner et al., 2011).

The idea that groups can expand their cognitive capacity by sharing mental resources—the collective working memory effect (F. Kirschner et al., 2011), has gained attention in the research of collaborative learning within the framework of CLT in recent years (Zambrano et al., 2019a, 2019b, 2022; Lange et al., 2021; Du et al., 2022; Jiang & Kalyuga, 2022; Sankaranarayanan et al., 2021). The collective working memory effect works by distributing cognitive demands across members, thereby overcoming the limitations of individual working memory (F. Kirschner et al., 2009; Sweller et al., 2019). While this construct cannot be measured directly, it can be inferred indirectly through performance and cognitive load outcomes observed in collaborative versus individual learning conditions.

Group work or collaborative learning within the CLT framework showed more positive and consistent results with a better understanding of the instructional and task designs. When learners collaborate on high-complexity tasks, their combined cognitive resources can expand their working memory capacity, improving learning outcomes (Jiang & Kalyuga, 2022; Sweller et al., 2019). However, task complexity and the spatial distribution of information should be taken into consideration (Guzmán & Zambrano, 2024). The integration of collective working memory into CLT has shown promising results, particularly in collaborative tasks (Jiang & Kalyuga, 2022; Zambrano et al., 2019a). Collaborative learning environments can reduce cognitive load and improve learning outcomes, especially when task complexity and group interaction are optimized (Sankaranarayanan et al., 2021).

Cognitive load refers to the amount of working memory required to perform a learning task (Sweller, 2010) and can be categorized into intrinsic, extraneous, and germane load. Intrinsic load arises from the inherent complexity of the material being learned, including the number of elements and their interactions, and is influenced by prior knowledge (Sweller, 2010). For example, constructing grammatically correct sentences involves a greater intrinsic load compared to simply repeating words, due to the numerous elements involved in sentence construction (Klepsch et al., 2017). Extraneous load, on the other hand, refers to unnecessary cognitive effort induced by suboptimal instructional procedures (Sweller, 2010). For instance, a problem-solving task can create a higher extraneous load than learning from an example, as it requires learners to explore potential solutions independently. Finally, germane load arises from the engagement of learners in the process of learning, such as self-explanation or collaborative discussion, which contributes to schema building (Klepsch et al., 2017; Sweller et al., 2019). The aim of instructional design from a cognitive load perspective is, therefore, to manage intrinsic cognitive load, minimize extraneous load, and maximize germane cognitive load.

Cognitive load is commonly assessed through subjective and objective measures (Zavgorodniaia et al., 2020). Subjective measures often involve self-assessments of mental load and task difficulty, while objective measures utilize physiological or behavioral data (Paas et al., 2003). Furthermore, cognitive load is frequently measured alongside learning outcomes or performance to assess instructional effectiveness (Ouwehand et al., 2021). Combining mental effort with performance could give a clearer picture of the effectiveness of the learning instructions and the learning outcomes (Paas & van Merriënboer, 1993; van Gog & Paas, 2008). Moreover, efficiency in learning is also considered when learners achieve higher performance relative to their perceived mental load or use less effort than expected based on their performance (Jiang & Kalyuga, 2022; Paas & van Merriënboer, 1993).

As indicated, learners can benefit from expanding their working memory capacity by group work on high complexity tasks. Academic writing, particularly for second language learners, is, for instance, a domain fraught with cognitive challenges. Writing tasks require the integration of procedural, linguistic, and content knowledge, which can overwhelm an individual’s working memory (Al-Fattah, 2018; Jiang & Kalyuga, 2022). To et al. (2013) stated that IELTS (International English Language Testing System) writing tasks have high linguistic complexity, especially the lexical density. These IELTS writing tasks are known for their high cognitive demands due to their complexity and the need for learners to process and synthesize visual data into written discourse (Yu et al., 2010). Iwashita et al. (2021) reported that due to logical thinking, teachers thought IELTS Academic Writing Task 2 was harder than Task 1, while students thought Task 1 was harder than Task 2 because of the graph layouts. Even though each test-taker is unique, and no work will be to everyone’s taste, it was determined that IELTS Academic Writing Task 1 required a high level of cognitive processing. The different kinds of graphs have an impact on students’ cognitive processing of the visual data and their adherence to visual rules when reproducing their understanding of the graph in written English discourse (Yu et al., 2010).

Previous studies applying CLT to academic writing have generally focused on individual learning, with fewer examining collaborative writing (Chan, 2013; Chan et al., 2017; Clark & Yu, 2022; Jiang & Kalyuga, 2022; F. Kirschner et al., 2009; P. A. Kirschner et al., 2018; Révész et al., 2017). The present study aims to fill this gap by exploring the effect of collective working memory on academic writing skills within a cooperative learning framework. Cooperative learning methods, such as Cooperative Integrated Reading and Composition (CIRC), facilitates collaboration and supports the collective working memory effect, making them particularly suitable for academic writing tasks, which involve complex cognitive processing. CIRC promotes cognitive engagement by having learners collaborate to process and share information (Slavin, 2016). This approach has been shown to improve learning outcomes, as it fosters positive interdependence and individual accountability, key components of successful collaborative learning (Johnson & Johnson, 2002). CIRC’s structure, which integrates reading, comprehension, and writing tasks, aligns well with the demands of complex cognitive tasks like academic writing (Slavin, 2016).

The Present Study

This study applies the CIRC approach to facilitate collaborative learning activities (collective working memory) and to expand the working memory capacity of the learners in acquiring academic writing skills, i.e., IELTS Academic Writing Task 1. The goal of this study was to figure out the effect of the collective working memory effect on the cognitive load (mental effort), writing performance (learning outcomes), and learning efficiency of the experimental group (collaborative learning) compared to the control group (individual learning) on a high-complexity writing task.

It was hypothesized that the use of collective working memory, which is facilitated by cooperative learning (CIRC), would affect individual mental load, writing performance, and learning efficiency in IELTS academic writing Task 1. In addition, it is expected that the participants in the experimental group (collaborative learning) would perform differently from the participants in the control group (individual learning) in terms of mental load, writing performance, and learning efficiency after the experiment.

2. Methods

2.1. Participants

150 undergraduate students (95 females and 55 males) from the English Education Department at Universitas Sarjanawiyata Tamansiswa, Indonesia, participated and were randomly assigned in this study. The age ranged from 18 to 31 years (M = 22.03, SD = 2.27). All participants were at the same education level and program. None of the participants had previously received formal instruction on writing or summarizing charts or graphs, suggesting limited familiarity with the learning materials in this study. English is the second or third language for all participants. All participants were given the same learning materials, task duration, and experimental procedures, with the only difference lying in the learning condition (individual vs. cooperative learning). They consented to participate in this research.

2.2. Materials

The learning materials for this study were adapted from The Official Cambridge Guide to IELTS (Cullen et al., 2014), which is widely used for IELTS preparation. The IELTS Academic Writing Task 1 requires candidates to describe, summarize, or explain visual data, such as charts, graphs, or tables (Uysal, 2010). This study focused on the “describing charts and graphs” task. Three different charts and graphs from The Official Cambridge Guide to IELTS (Cullen et al., 2014) were used in the pre-test, learning task, and post-test. All assessments were conducted offline, using printed materials. In addition to the learning tasks, participants were given writing guidelines for peer review activity and completed a mental load questionnaire, both of which were provided in printed form. The mental-load questionnaire and IELTS rubric used in this study are provided in the Supplementary Materials.

2.3. Procedures

The study involved two groups: an experimental group and a control group. The experimental group participated in cooperative learning (CIRC), while the control group worked individually. Participants were randomly assigned to one of these groups, with 75 students in each group. The experiment lasted for 150 min for both groups, and it consisted of five stages: initial preparation (25 min), pre-test (30 min), learning task (50 min), break (15 min), and post-test (30 min). The experimental conditions differed only in the methods of learning (cooperative learning vs. individual learning).

To test the materials and procedures, a pilot study was conducted before the experiment. The participants in both groups were divided into smaller subgroups, with the experimental group working in triads and the control group working individually. The participants in the experimental group were divided into eight or nine triad groups at a single time and were invited to come to the lab. An instructor and the experimenter guided each time of experiments. The instructor guided the experiments and controlled the time duration at each stage.

2.3.1. Initial Preparation and Pre-Test

In the initial preparation stage, the instructors introduced the experiment and its procedures to ensure that all participants understood the purpose, the timeline, and their roles. Following this, participants signed a consent form and completed the pre-test, which measured their baseline writing skills. The pre-test involved a 20 min writing task, followed by a questionnaire (10 min) (Leppink et al., 2013), to measure their perceived mental load.

2.3.2. Learning Task

The learning task involved writing about a graph or chart, similar to the pre-test, but with additional instructions tailored to each group’s learning method.

Control Group (Individual Learning)

The control group received instructions for individual learning, with the following sequence:

Reading Task (20 min): Participants read the materials provided about IELTS Academic Writing Task 1, which included instructions, criteria, tips, and examples. They were asked to perform self-explanation to ensure understanding.

Writing Task (20 min): After reading, participants completed the writing task individually.

Reviewing and Revising (10 min): Participants were instructed to review and revise their writing using the provided peer review guidelines.

Experimental Group (Cooperative Learning)

The experimental group followed a similar sequence but with collaborative learning procedures:

Reading Task (20 min): Participants were divided into triads. Each member of the triad read different materials and explained their content to the others, fostering collaboration.

Writing Task (20 min): After the reading stage, participants collaboratively wrote the essay, with each member contributing to the task based on their assigned role (A, B, or C).

Reviewing and Revising (10 min): In the final stage, participants rotated roles and collaboratively reviewed their writing using the peer review guidelines.

Roles were rotated in such a way that each participant experienced all three roles during the learning task, ensuring equal involvement and contribution within the group. This collaborative structure aimed to reduce cognitive load by promoting shared mental load and enhancing engagement.

2.3.3. Post-Test

After completing the learning task, participants took the post-test, which was identical to the pre-test. Following the post-test, participants again completed the mental load questionnaire to figure out any changes in their perceived cognitive load.

2.4. Data Analysis

Data analysis was conducted using SPSS 29. The independent variable was the learning condition (individual learning vs. cooperative learning), with 75 participants in each group. The dependent variables included the following: cognitive mental load, writing performance, and learning efficiency. A two-tailed independent sample t-test was conducted to check differences in prior knowledge of writing skills between the control and experimental groups. Cohen’s d was to determine the effect size, i.e., small (0.20), medium (0.50), and large (0.80) (Cohen et al., 2018). Then, the researchers calculated the mean difference scores between the pre-test and post-test. These scores were then analyzed using one-way between subjects ANOVA to find out if there was a different effect of learning conditions on each group. Last, a one-way within subject ANOVA was separately conducted for analyzing each group. This was performed to figure out if there was a significant effect of learning conditions on each group. Partial eta-squared was to determine the effect size, i.e., small (0.01), medium (0.06), and large (0.14) (Cohen et al., 2018). G*Power(version 3.1) statistical software (Erdfelder et al., 1996) was used to calculate the statistical power associated with testing multiple hypotheses. A medium effect size of intervention was used, f = 0.25; α = 0.05. Study power up to 0.80. The estimated sample size of 128 participants was obtained from the statistical software. Thus, the current sample (N = 150) was sufficient.

2.4.1. Measurement of Mental Load

The mental load or cognitive load, that the students experienced during the experiment was measured by using a questionnaire. This questionnaire is a subjective rating scale measurement of cognitive mental load, which was developed by Leppink et al. (2013). In the questionnaire, there are ten items for the measurement of Intrinsic Load (Items 1, 2, and 3, e.g., “The topic covered in the activity was very complex”), Extraneous Load (Items 4, 5, and 6, e.g., “The instructions and/or explanations were full of unclear language”), and Germane Load (Items 7, 8, 9, and 10, e.g., “The activity really enhanced my knowledge and understanding of writing”). The rating scale is from 0 to 10. This questionnaire was reported highly reliable with Cronbach’s alpha from 0.76 to 0.95 (Leppink et al., 2014). In the present study, the reliability of the analysis showed quite similar results of internal consistency for Intrinsic Load (α = 0.86), Extraneous Load (α = 0.87), and Germane Load (α = 0.92), with an overall scale of α = 0.76. This mental load measurement was conducted using printed forms. Students gave their evaluations immediately after the pre-test and post-test to obtain the precise experience of students’ mental loads. In the questionnaire, the order of questions was jumbled from the pre-test and post-test stages to avoid similar patterned answers from the participants.

2.4.2. Measurement of Performance

Participants’ writing performance was assessed using the IELTS Academic Writing Task 1 rubric (Cambridge Assessment English, 2020). The tasks required participants to summarize or describe information from charts or graphs provided. The rubric evaluates four criteria: task achievement, coherence and cohesion, lexical resource, and grammatical range and accuracy. Each criterion was scored on a scale from 0 to 9, with scores for all four criteria averaged to yield a final score for each participant. Two trained raters, both certified IELTS instructors, independently scored the essays with a discussion process prior to and after scoring. Interrater agreement between two IELTS raters was measured using Weighted Kappa and intraclass correlation coefficients (ICC). The pre-test scores showed outstanding consistency, with Weighted Kappa = 0.91, p < 0.001, and ICC(2,k) = 0.99, 95% CI [0.98, 0.99]. The post-test scores showed strong interrater reliability (Weighted Kappa = 0.98, p < 0.001, and ICC(2,k) = 1.00, 95% CI [1.00, 1.00]). These findings indicate a high level of agreement among raters at both time points, with especially significant consistency on the post-test. The high interrater agreement strengthens confidence in the reliability of the performance assessments.

2.4.3. Measurement of Efficiency

Learning efficiency (E) was calculated based on the difference between participants’ performance and mental load scores in the pre-test and post-test phases. Performance and mental load scores were z-standardized across the full sample before computing the learning efficiency score. The formula for efficiency, adapted from Paas and van Merriënboer (1993), E = (zP − zR)/√2) where P represents performance, R represents mental load, and √2 is the constant derived from the standard deviation of both variables. The efficiency score reflects the effectiveness of learning, taking into account both cognitive load and task performance.

It should be noted that collective working memory was not measured directly, as no standardized instrument currently exists for this purpose. Consistent with prior research (e.g., F. Kirschner et al., 2009; Paas et al., 2003), we operationalize the effect of CWM indirectly through outcome measures (mental load, writing performance, and efficiency) under cooperative conditions. The cooperative intervention (CIRC) was designed to foster group-based distribution of cognitive load, thereby creating conditions under which CWM could be inferred.

3. Results

An independent sample t-test showed that the experimental group (M = 4.41, SD = 0.57) started with lower writing performance scores compared to control group (M = 4.61, SD = 0.55), t(148) = 2.25, p = 0.026. The effect size for this difference was medium, with Cohen’s d = 0.367, and 95% CI = [0.044, 0.689]. However, there were no significant differences for mental load and learning efficiency. Descriptive statistics (see Table 1) revealed mean scores and standard deviations for pre-test and post-test across groups, and the difference scores between pre-test and post-test.

Due to significant differences in the prior knowledge of participants in the control group and experimental group on writing skills, the researchers calculated the difference score between pre-test and post-test and conducted one-way between subject ANOVA to figure out if there was a different effect between groups or learning conditions (cooperative learning vs. individual learning). The results met the assumption of homogeneity of variance (p > 0.05) for all dependent variables. However, it showed that there were no significant differences between learning conditions (individual learning vs. cooperative learning) on mental effort F(1,148) = 1.469, p = 0.228, η² = 0.010, 95% CI = [0.000, 0.063], writing performance F(1,148) = 0.428, p = 0.514, η² = 0.003, 95% CI = [0.000, 0.043], and learning efficiency F(1,148) < 0.001, p = 0.998, η² < 0.001, 95% CI = [0.000, 0.000].

A one-way within subject ANOVA was separately conducted in each group to compare the effect of different learning conditions (cooperative vs. individual learning) on mental effort, writing performance, and learning efficiency. It was found that there was a significant effect of different learning conditions on writing performance for both groups or learning conditions as follows: the control group (individual learning) Wilks’ Lambda = 0.362, F(1,74) = 130.37, p < 0.001, η² = 0.638, 95% CI = [0.575, 0.818] and the experimental group (cooperative learning) Wilks’ Lambda = 0.510, F(1,74) = 71.14, p < 0.001, η² = 0.490, 95% CI = [0.484, 0.783]. For mental load and learning efficiency, there were not any significant effects found in both groups. The results for mental load in the control group (individual learning) were Wilks’ Lambda = 0.999, F(1,74) = 0.04, p = 0.842, η² = 0.001, 95% CI = [−0.275, 0.337], and in the experimental group (cooperative learning) were Wilks’ Lambda = 0.952, F(1,74) = 3.76, p = 0.056, η² = 0.048, 95% CI = [−0.008, 0.589]. While for learning efficiency, the results for the control group (individual learning) were Wilks’ Lambda = 1.000, F (1,74) < 0.001, p = 0.998, η² < 0.001, 95% CI = [−0.201, 0.201]. And for the experimental group (cooperative learning), was Wilks’ Lambda = 1.000, F (1,74) < 0.001, p = 0.999, η² < 0.001, 95% CI = [−0.243, 0.243].

4. Discussion

It was hypothesized that collective working memory, facilitated through Cooperative Integrated Reading and Composition (CIRC), would positively influence mental load, learning outcomes, and learning efficiency in IELTS academic writing Task 1. In addition, it was expected that participants in the experimental group, leveraging collective working memory, would outperform those in the control group across the dependent variables. Examination of the means indicated that both (control and experimental) groups have an increase in mental load and writing performance after the experiment. For learning efficiency, the experimental group experienced an increase while the control group was in contrast.

4.1. Mental Load

Participants in the experimental group generally reported lower perceived mental load compared to those in the control group, aligning with cognitive load theory’s premise that collaborative learning can distribute cognitive demands among group members, thereby reducing individual mental load (F. Kirschner et al., 2009). Nonetheless, the experimental group also showed a slightly higher rise in mental load from the pre-test to the post-test, which was probably an increase in germane load caused by the transactional costs and learning engagement (Klepsch et al., 2017; Sweller et al., 2019). These processes, while initially taxing, are integral to deep learning and have been shown to support the development of higher-order thinking skills (Loyens et al., 2008; van Merriënboer & Sweller, 2005).

Throughout the experiment, the average perceived mental load of the experimental group was consistently lower than that of the control group, even though the mental load increased in the post-test. The control group showed a modest rise in perceived mental load from pre-test to post-test. In contrast, the experimental group exhibited a slightly higher increase from pre-test to post-test (see Table 1). These findings suggest that although collaborative activities introduce additional germane load, they do not significantly hinder learning processes when appropriately structured. On the contrary, the collaborative group maintained its performance gains, underscoring the effectiveness of such tasks in balancing cognitive load while enhancing engagement and outcomes (Sweller et al., 2019).

4.2. Writing Performance

Both groups demonstrated a statistically significant improvement in writing performance from pre-test to post-test, confirming the overall effectiveness of the instructional interventions employed in this study. However, the control group consistently outperformed the experimental group in both pre-test and post-test assessments, reflecting a slightly higher baseline proficiency in writing. With unequal baselines of scores between groups, direct comparison between pre-test and post-test scores of each group may not fairly describe the benefit of collaboration. In this study, the growth scores between pre-test and post-test were analyzed to provide a clearer perspective on relative progress. The similar gains suggest that cooperative learning can foster meaningful improvements even when baseline inequality exists. It could also suggest that while collaborative learning fosters significant gains in writing performance, its advantages may be more pronounced in scenarios with equal baseline abilities. This finding is consistent with prior work (Slavin, 2016), which emphasizes that the benefits of collaborative learning are more pronounced when learners begin with comparable abilities.

4.3. Learning Efficiency

An examination of learning efficiency revealed divergent trends between the two groups. The experimental group demonstrated a slight improvement in efficiency, whereas the control group experienced a marginal decline. These findings align with previous research, which indicates that well-structured collaborative environments can balance cognitive demands and writing performance, thereby optimizing learning efficiency (Fanguy et al., 2021; Paas et al., 2003). Although the magnitude of the effect is not substantial, this directionality could suggest that cooperative learning promotes more efficient learning.

While collaboration inherently imposes additional cognitive demands due to transactional costs, these demands appear to facilitate meaningful and engaged learning. The experimental group’s slight efficiency gains highlight the potential of cooperative learning, when well-structured, to achieve a productive balance between mental load and writing performance. This observation reinforces the importance of instructional designs that not only enhance cognitive engagement but also promote efficient learning (Slavin, 2016).

Although collective working memory was not measured directly, the observed differences in mental load and efficiency between groups are consistent with the operation of CWM. These findings therefore provide indirect evidence supporting the theoretical claim that cooperative structures can expand cognitive capacity by distributing cognitive demands. In addition, it is important to take into account that the experimental group’s writing performance was lower than the control group at baseline. Thus, it might have reduced the observable benefits of cooperative learning in this situation and affected the relative improvements seen. This also highlighted the significance of assessing early proficiency levels when designing experiments. Gain scores, or the different scores between pre-test and post-test, offer a more profound understanding of the relative improvements made by each group, even though pre-test and post-test comparisons are still useful.

This study has several limitations that should be considered in interpreting the findings. First, the 150 min intervention period may have been insufficient for participants to fully experience and demonstrate the long-term benefits of cooperative learning, potentially underestimating its impact. The short duration of the study may have limited the full realization of collaborative learning benefits. Longitudinal studies are needed to investigate how sustained exposure to cooperative learning tasks influences learning outcomes, mental load, and efficiency over time (Slavin, 2016; Zambrano et al., 2019a). Second, the baseline inequality in writing performance between the control and experimental groups complicates the interpretation of gains, as initial proficiency differences could have biased the magnitude of improvement in each group. Although growth scores were used to address this limitation, future research should aim for balanced groups to achieve a clearer view of intervention effects. The role of prior knowledge and prior collaborative experience assures further exploration, as it may moderate the effectiveness of group interactions and task outcomes (Loyens et al., 2008; Zambrano et al., 2019a). Finally, the absence of a follow-up assessment restricts understanding of the retention and transfer of skills beyond the immediate post-test, limiting insight into the durability of collaborative learning effects. As this study only assessed learning outcomes immediately before and after the intervention. To better understand the long-term effects of collaborative learning methodologies on academic writing, follow-up assessments to measure knowledge transfer and retention need to be considered. To sum up, future research should incorporate longer intervention durations, ensure balanced baseline abilities, and include follow-up testing to better assess sustained learning outcomes.

5. Conclusions

This study examines the role of collective working memory in cooperative learning during a high-complexity writing task. Both individual and cooperative learning groups showed improvements in writing performance, and the cooperative learning group demonstrated a small positive trend in learning efficiency. However, no significant differences were found between groups. These results suggest that while cooperative learning may help manage cognitive demands and foster efficiency, its advantages over individual learning remain tentative. As both groups were not equal at the start, growth scores (the improvement of pre-test and post-test) were analyzed rather than direct group comparison to provide a clearer view of the intervention effect. Future research should use balanced baseline groups, longer interventions, and follow-up testing to better understand the potential long-term benefits of cooperative learning in academic writing.

Author Contributions

Conceptualization, methodology, original draft preparation, D.S. and R.R.; original draft preparation, R.R.; original draft preparation, materials preparation, D.S., I.H., and A.V.G.; formal analysis, original draft preparation, R.R., S.L., M.B.; supervision, project administration, review and editing, R.R., S.L., and M.B. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Universitas Sarjanawiyata Tamansiswa, Yogyakarta (115.B/UST/FKIP/Dkn/IP/IV/2024 and 19 April 2024).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors on request.

Acknowledgments

The authors would also like to thank the research writing assessors and instructors, Estri Oktarena, Victa Sari Dwi Kurniati, and Andhi Dwi Nugroho, who helped with data collection and assessment.

Conflicts of Interest

The authors declare no conflicts of interest. The funding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Table

Table 1

Descriptive Statistics.

Tests	Variable	Control Group (N = 75)		Experimental Group (N = 75)
		M	SD	M	SD
Pre-test	Mental load	5.13	1.60	4.80	1.44
	Writing	4.61	0.55	4.41	0.57
	Efficiency	−0.0005	1.04	0.0001	1.12
Post-test	Mental load	5.16	1.47	5.09	1.11
	Writing	5.31	0.57	5.04	0.74
	Efficiency	−0.0008	1.09	0.0003	1.11
Mean Difference	Mental load	0.03	1.33	0.29	1.30
	Writing	0.70	0.53	0.63	0.65
	Efficiency	−0.0003	0.87	0.0001	1.06

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/educsci15101392/s1, Questionnaire S1: Mental Load Questionnaire; Table S1: Writing Assignment Guidelines.

References

Al-Fattah, N. Cognitive load theory in the context of second language academic writing. Higher Education Pedagogies; 2018; 3, 1 pp. 385-402. [DOI: https://dx.doi.org/10.1080/23752696.2018.1513812]

Cambridge Assessment English. IELTS writing band descriptors (public version): Task 1 (Academic); 2020; Available online: https://www.ielts.org/-/media/pdfs/writing-band-descriptors-task-1.ashx (accessed on 11 October 2023).

Chan, S. Establishing the validity of reading-into-writing test tasks for the UK academic context. Doctoral dissertation; University of Bedfordshire: 2013.

Chan, S.; Bax, S.; Weir, C. Researching participants taking IELTS academic writing task 2 (AWT2) in paper mode and in computer mode in terms of score equivalence, cognitive validity and other factors (Technical Report); British Council, IDP: IELTS Australia, and Cambridge English: 2017.

Clark, T.; Yu, G. Test preparation pedagogy for international study: Relating teacher cognition, instructional models and academic writing skills. Language Teaching Research; 2022; 29, 2 pp. 726-750. [DOI: https://dx.doi.org/10.1177/13621688211072381]

Cohen, L.; Manion, L.; Morrison, K. Research methods in education; 8th ed. Routledge: 2018; [DOI: https://dx.doi.org/10.4324/9781315456539]

Cowan, N. Sossin, W. S.; Lacaille, J.-C.; Castellucci, V. F.; Belleville, S. Chapter 20 what are the differences between long-term, short-term, and working memory?. Progress in brain research; Elsevier: 2008; Vol. 169, pp. 323-338. [DOI: https://dx.doi.org/10.1016/S0079-6123(07)00020-9]

Cowan, N. The magical mystery four: How is working memory capacity limited, and why?. Current Directions in Psychological Science; 2010; 19, 1 pp. 51-57. [DOI: https://dx.doi.org/10.1177/0963721409359277] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20445769]

Cullen, P.; French, A.; Jakeman, V. The official Cambridge guide to IELTS: For academic & general training: Student’s book with answers; Cambridge University Press: 2014.

Du, X.; Chen, C.; Lin, H. The impact of working memory capacity on collaborative learning in elementary school students. Frontiers in Psychology; 2022; 13, 1027523. [DOI: https://dx.doi.org/10.3389/fpsyg.2022.1027523] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36532992]

Erdfelder, E.; Faul, F.; Buchner, A. GPOWER: A general power analysis program. Behavior Research Methods, Instruments, & Computers; 1996; 28, 1 pp. 1-11. [DOI: https://dx.doi.org/10.3758/BF03203630]

Fanguy, M.; Baldwin, M.; Shmeleva, E.; Lee, K.; Costley, J. How collaboration influences the effect of note-taking on writing performance and recall of contents. Interactive Learning Environments; 2021; 31, 7 pp. 4057-4071. [DOI: https://dx.doi.org/10.1080/10494820.2021.1950772]

Guzmán, J.; Zambrano, R. J. Effects of split-attention and task complexity on individual and collaborative learning. Education Sciences; 2024; 14, 9 1035. [DOI: https://dx.doi.org/10.3390/educsci14091035]

Iwashita, N.; Sasaki, M.; Stell, A.; Yucel, M. Japanese stakeholders’ perceptions of IELTS writing and Speaking tests and their impact on communication and achievement (Technical Report); British Council, IDP: IELTS Australia, and Cambridge English: 2021.

Jiang, D.; Kalyuga, S. Learning English as a foreign language writing skills in collaborative settings: A cognitive load perspective. Frontiers in Psychology; 2022; 13, 932291. [DOI: https://dx.doi.org/10.3389/fpsyg.2022.932291]

Johnson, D. W.; Johnson, R. T. Tindale, R. S.; Heath, L.; Edwards, J.; Posavac, E. J.; Bryant, F. B.; Suarez-Balcazar, Y.; Henderson-King, E.; Myers, J. Cooperative learning and social interdependence theory. Theory and research on small groups; Springer US: 2002; pp. 9-35. [DOI: https://dx.doi.org/10.1007/0-306-47144-2_2]

Kirschner, F.; Paas, F.; Kirschner, P. A. A cognitive load approach to collaborative learning: United brains for complex tasks. Educational Psychology Review; 2009; 21, 1 pp. 31-42. [DOI: https://dx.doi.org/10.1007/s10648-008-9095-2]

Kirschner, F.; Paas, F.; Kirschner, P. A. Task complexity as a driver for collaborative learning efficiency: The collective working-memory effect. Applied Cognitive Psychology; 2011; 25, 4 pp. 615-624. [DOI: https://dx.doi.org/10.1002/acp.1730]

Kirschner, P. A.; Sweller, J.; Kirschner, F.; Zambrano, R. J. From cognitive load theory to collaborative cognitive load theory. International Journal of Computer-Supported Collaborative Learning; 2018; 13, 2 pp. 213-233. [DOI: https://dx.doi.org/10.1007/s11412-018-9277-y]

Klepsch, M.; Schmitz, F.; Seufert, T. Development and validation of two instruments measuring intrinsic, extraneous, and germane cognitive load. Frontiers in Psychology; 2017; 8, 1997. [DOI: https://dx.doi.org/10.3389/fpsyg.2017.01997] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29201011]

Lange, C.; Costley, J.; Fanguy, M. Collaborative group work and the different types of cognitive load. Innovations in Education and Teaching International; 2021; 58, 4 pp. 377-386. [DOI: https://dx.doi.org/10.1080/14703297.2020.1788970]

Leppink, J.; Paas, F.; van der Vleuten, C.; van Gog, T.; van Merriënboer, J. J. G. Development of an instrument for measuring different types of cognitive load. Behavior Research Methods; 2013; 45, pp. 1058-1072. [DOI: https://dx.doi.org/10.3758/s13428-013-0334-1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23572251]

Leppink, J.; Paas, F.; van Gog, T.; van der Vleuten, C. P. M.; van Merriënboer, J. J. G. Effects of pairs of problems and examples on task performance and different types of cognitive load. Learning and Instruction; 2014; 30, pp. 32-42. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2013.12.001]

Loyens, S.; Magda, J.; Rikers, R. Self-directed learning in problem-based learning and its relationships with self-regulated learning. Educational Psychology Review; 2008; 20, pp. 411-427. [DOI: https://dx.doi.org/10.1007/s10648-008-9082-7]

Ouwehand, K.; van der Kroef, A.; Wong, J.; Paas, F. Measuring cognitive load: Are there more valid alternatives to Likert rating scales?. Frontiers in Education; 2021; 6, 702616. [DOI: https://dx.doi.org/10.3389/feduc.2021.702616]

Paas, F.; Tuovinen, J. E.; Tabbers, H.; van Gerven, P. W. M. Cognitive load measurement as a means to advance cognitive load theory. Educational Psychologist; 2003; 38, 1 pp. 63-71. [DOI: https://dx.doi.org/10.1207/S15326985EP3801_8]

Paas, F.; van Merriënboer, J. J. G. The efficiency of instructional conditions: An approach to combine mental effort and performance measures. Human Factors; 1993; 35, 4 pp. 737-743. [DOI: https://dx.doi.org/10.1177/001872089303500412]

Révész, A.; Michel, M.; Lee, M. Investigating IELTS academic writing task 2: Relationships between cognitive writing processes, text quality, and working memory (Technical report); British Council, IDP: IELTS Australia, and Cambridge English: 2017.

Sankaranarayanan, R.; Kwon, K.; Cho, Y. Exploring the differences between individuals and groups during the problem-solving process: The collective working-memory effect and the role of collaborative interactions. Journal of Interactive Learning Research; 2021; 32, pp. 43-66. Available online: https://psycnet.apa.org/record/2021-80316-002 (accessed on 5 October 2025). [DOI: https://dx.doi.org/10.70725/636732aueosi]

Slavin, R. E. Mayer, R. E.; Alexander, P. A. Instruction based on cooperative learning. Instruction based on cooperative learning. Handbook of research on learning and instruction; 2nd ed. Routledge: 2016; pp. 388-404. [DOI: https://dx.doi.org/10.4324/9781315736419]

Sweller, J. Ross, B. Evolution of human cognitive architecture. The psychology of learning and motivation; Academic Press: 2003; Vol. 43, pp. 215-266.

Sweller, J. Element interactivity and intrinsic, extraneous, and germane cognitive load. Educational Psychology Review; 2010; 22, 2 pp. 123-138. [DOI: https://dx.doi.org/10.1007/s10648-010-9128-5]

Sweller, J.; van Merriënboer, J. J. G.; Paas, F. Cognitive architecture and instructional design: 20 years later. Educational Psychology Review; 2019; 31, 2 pp. 261-292. [DOI: https://dx.doi.org/10.1007/s10648-019-09465-5]

To, V.; Le, T.; Le, Q. A comparative study of nominalisation in IELTS writing test papers. International Journal of Innovative Interdisciplinary Research; 2013; 1, 4 pp. 15-21.

Uysal, H. H. A critical review of the IELTS writing test. ELT Journal; 2010; 64, 3 pp. 314-320. [DOI: https://dx.doi.org/10.1093/elt/ccp026]

van Gog, T.; Paas, F. Instructional efficiency: Revisiting the original construct in educational research. Educational Psychologist; 2008; 43, 1 pp. 16-26. [DOI: https://dx.doi.org/10.1080/00461520701756248]

van Merriënboer, J. J. G.; Sweller, J. Cognitive load theory and complex learning: Recent developments and future directions. Educational Psychology Review; 2005; 17, pp. 147-177. [DOI: https://dx.doi.org/10.1007/s10648-005-3951-0]

Yu, G.; Rea-Dickins, P.; Kiely, R. The cognitive processes of taking IELTS academic writing task 1 (Technical report); British Council, IDP: IELTS Australia, and Cambridge English: 2010.

Zambrano, R. J.; Kirschner, F.; Sweller, J.; Kirschner, P. A. Effects of group experience and information distribution on collaborative learning. Instructional Science; 2019a; 47, 5 pp. 531-550. [DOI: https://dx.doi.org/10.1007/s11251-019-09495-0]

Zambrano, R. J.; Kirschner, F.; Sweller, J.; Kirschner, P. A. Effects of prior knowledge on collaborative and individual learning. Learning and Instruction; 2019b; 63, 101214. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2019.05.011]

Zambrano, R. J.; Kirschner, F.; Sweller, J.; Kirschner, P. A. Effect of task-based group experience on collaborative learning: Exploring the transaction activities. British Journal of Educational Psychology; 2022; 93, 4 pp. 879-902. [DOI: https://dx.doi.org/10.1111/bjep.12603] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37128656]

Zavgorodniaia, A.; Duran, R.; Hellas, A.; Seppala, O.; Sorva, J. Measuring the cognitive load of learning to program: A replication study. United Kingdom & Ireland computing education research conference (UKICER 2020); ACM: 2020; pp. 3-9. [DOI: https://dx.doi.org/10.1145/3416465.3416468]