Effects of Blended Instruction on students’ paragraph writing performances: the case of first year Health science pharmacy students at Pharma College Hawassa Campus, Ethiopia

Abstract

This study focused on paragraph writing abilities as it is a fundamental skill for academic and professional success. Improving paragraph writing lays the groundwork for addressing advanced writing tasks and is vital for effective global communication in English. The objective of this study was, therefore, to examine effects of blended instruction on students’ paragraph writing performance at Pharma College Hawassa Campus, Ethiopia. A quasi-experimental design was used to attain this objective. A sample of 54 first-year pharmacy department (26 control and 28 experimental) students was taken purposefully. The control group attended conventional, face-to-face approach; whereas experimental group received instruction using blended learning approach via Google Classroom, which incorporated collaborative writing that allowed for peer interaction and scaffolding, extensive online resources for independent learning, face-to-face support, and feedback. Pre-test and post-test paragraph writing were conducted to test effects of blended instruction, and effect size was computed to measure the magnitude of intervention. Furthermore, students’ questionnaire was used to collect data on participants’ backgrounds and experiences. Results revealed improvements in both groups; however, overall, the experimental group in post-test demonstrated significantly better paragraph writing with a large effect size. Regression analysis highlighted flexibility, collaborative tasks, and instant feedback as key contributors to improvement, whereas impact of authentic materials and face-to-face interactions was comparatively less. Finally, the researchers recommended EFL teachers should benefit from applying blended instruction, which may in turn improve students’ paragraph writing performance in English language in the Ethiopian context.

Keywords:

• Blended instruction
• paragraph writing
• conventional approach
• online learning
• Google Classroom

Reviewing Editor:

• Lin Zhong, Workforce Education and Development, Southern Illinois University Carbondale, USA

SUBJECTS:

• Information & Communication Technology (ICT)
• Teachers & Teacher Education
• Higher Education
• Open & Distance Education and eLearning
• Language Teaching & Learning

Introduction

In the contemporary period of globalization, science and technology are developing rapidly, accompanied by major developments in many aspects of life, including political, cultural, social, educational, and economic situations (Nishanthi, 2018). Globalization has not only brought nations closer but also heightened the role of English as a primary language for international discourse. The enhanced status of the English language directly stems from the necessity of a unified medium to support the vast array of interactions that come with an increasingly connected world through technology. As an international language for communication, English serves as a common medium for diverse cultures and nations to communicate, collaborate, and exchange ideas, reinforcing its status as a crucial skill in the global job market. Consequently, learning English language has become vital for communication on a global scale, making it one of the prerequisites for mastering the subject of international communication (Nishanthi, 2018; Yang & Chen, 2007).

The rapid advancements in information and communication technologies, which have made English evolve as a key international language, have also greatly affected education right across the world. Such advancements have necessitated a direct link between technology and education, leading to an expansion of higher education and an increase in student enrollments in both developed and developing countries (Bonk & Graham, 2012). Education, therefore, has emerged as one of the key sectors influenced by the use of advanced technologies for teaching and learning (Allen & Velden, 2012). The interplay between technology and education has underscored the need for reforms to cater to the evolving landscape of higher education, which now demands a curriculum that is as dynamic as the technological advancements shaping it.

Similarly, realizing the importance of the English language, the Ethiopian Ministry of Education (1994) has taken significant steps to reform the Teaching of English as a Foreign Language (TEFL) in Ethiopian schools, colleges, and universities. As a result, in 1994, a new language-in-education policy was implemented in Ethiopia, stipulating that English would be the medium of instruction (MoI) in Ethiopian education from grade nine up to higher institutions and is a compulsory subject from grade one up to higher education. However, in most instances, English language courses in higher education did not always align with the needs of the students (MoE, 2018), highlighting a gap that needs to be addressed through educational reforms. These reforms are essential to ensure that students are not only proficient in English but also equipped with the necessary skills to thrive in a globalized, technologically advanced society.

Meanwhile, the MoE introduced ICT in schools in 1994 after evaluating the standard and applicability of the existing curriculum and the methods used by teachers (MoE, 1994). For instance, it implemented several Educational Reform Plans and policy documents during the 1990s to address the demands of the current technological period and to suit the various needs of all students. Hence, to lay the foundation for education, the ministry implemented ‘The Comprehensive Plan for Education in the Information Age’ between 1997 and 2000. In this regard, it included an ICT-equipped computer laboratory and classrooms to provide schools with advanced educational settings. To ensure the effective integration of ICT in English language instruction, specific initiatives such as computer-assisted language learning labs, digital teaching materials developed specifically for the English language curriculum, and ICT training for English language teachers were included in the plan. These initiatives enhanced students’ engagement and interaction with the English language through technology-driven methods. The outcomes of these efforts were reflected in increased access to the Internet, better proficiency levels in English among students, as evidenced by improved performance in digital literacy (MoE, 2018), national exit exam assessments, and a greater ability to engage in English-mediated communication.

Other policy programs such as the Education Sector Development Programme (ESDP V) document (MoE, 2009), the Curriculum Framework for Ethiopian Education (MoE, 2009), and the Education Road Map (MoE, 2018) emphasized the importance of incorporating technology into the Ethiopian educational system, particularly with regard to English language skills. Additional programs such as Strengthening Teachers English Proficiency (STEP), English and Digital Skills for Girls (EDS), Schools Connect program, Teaching English Africa: English Connects supported by British Council Ethiopia (MoE, 2018), English Language Improvement Program (ELTIP), and Teach English for Life Learning (TELL) training also highlighted the integration of technology in education. Therefore, English language proficiency and ICT-computer literacy are essential communication elements required for higher education (MoE, 2009, 2018).

However, the researchers in higher education have observed that EFL students who use English as a medium of instruction face significant challenges, particularly in writing skills. The primary goal of English writing instruction at educational institutions is to provide students with the written communication skills required to deal with challenges with English language writing on a local and global level (Yang & Chen, 2007). According to Brown (2001), writing is ‘the graphic representation of spoken language, and that written performance is much like an oral performance, the only difference lying in a graphic instead of auditory signals’ (p. 337). Hedge (2005) supports and expands on Brown’s (2001) idea, stating that writing involves producing correct and complete sentences and phrases.

Therefore, writing is about assisting students in producing complete pieces of communication, linking and developing ideas, arguments, or facts for a specific reader or set of readers. It is a process that requires time, practice, and occasionally proper instruction (Dastjerdi & Samian, 2011). According to Richards and Renandya (2002) and Oshima and Hogue (2007), there are four major stages of the writing process: planning, drafting, revising, and editing. These stages involve activities like pre-writing, drafting, revising, and editing, which are important for enhancing students’ writing skills (Hyland, 2003). Hedge (2005) further outlines five factors to consider when writing a paragraph: content, organization, vocabulary, language use, and mechanics. Therefore, it is essential to focus on improving writing skills, with paragraph writing serving as the foundational step before mastering other writing skills (Maulida et al., 2022).

Although writing is a crucial language skill, it is not a simple task that students can pick up quickly and easily somewhere else; instead, it requires a carefully thought-out procedural approach. Cole and Feng (2015) provided evidence for this claim by pointing out that although writing is one of the most important English language skills, it is also thought to be the most complex and challenging task. Mitchell (2018) provided additional support for the concept by stating that writing skill is an extremely difficult cognitive task that requires writers to be knowledgeable in a variety of fields, from the learner’s academic background and personal interests to specific linguistic, psychological, and cognitive phenomena. The writing process in the conventional classroom is more teacher-centered; teachers stand in front of students and deliver lectures, and students become passive and receive little feedback and support from their teacher due to the limited amount of classroom time (Svinicki et al., 2014).

However, in the context of higher education, the rapid development of technology in the past two decades has ushered in a new dimension of teaching known as online learning. The widespread usage of online courses and programs in recent decades can be seen as a trend towardthe use of new learning environments with new media and methods such as Distance Learning (DL), and E-Learning (EL). E-learning, according to Chou and Chou (2011), is the process of learning and teaching outside of a traditional classroom setting using digital technology. According to Strauss and Howe (2000), e-learning has become increasingly important in higher education and for students of current generations: centennials (born after 1997) and Millennials (born between 1981 and 1996). These generations grew up in the era of digital technology, where computers, smartphones, and the internet were commonplace. Hence, E-learning is considered as the modern learning methods and has some advantages including increased flexibility for learners, access to a wider range of resources and experts, and the ability to learn at one’s own pace.

Nevertheless, there are also drawbacks associated with e-learning, such as a lack of face-to-face interaction between learners and instructors, technical difficulties, and a lack of structure and motivation for some students. Moreover, some practical tasks cannot be completed online due to students’ varied learning styles, (Epignosis LLC, 2015). According to Sadeghi (2019), students find it more challenging to get in touch with teachers when learning online. As a result, a new mindset; Blended Learning (BL) has emerged that compromises online and face-to-face learning.

Blended learning and its features

Blended learning is an approach that combines the benefits of e-learning and face-to-face methods (Graham & Robinson, 2007). According to contemporary research on educational technology, blended learning is defined as the deliberate ‘blending’ of in-person and virtual learning activities to promote and assist learning (Boelens, et al., 2015). The basic idea behind blended instruction is to create a learning environment that leverages the strengths of both online and traditional teaching methods. However, the concept of combining face-to-face and online education is not new (Garrison & Kanuka, 2004).

According to Drysdale, et al. (2013), teaching and learning using blended learning has been adopted and studied repeatedly since the development of ICT in education. Blended learning, according to Garrison and Vaughan (2008), involves ‘restructuring and replacing traditional class contact hours’ rather than ‘adding online learning to face-to-face learning’. The concept emerged as educational institutions recognized the potential benefits of integrating technology into learning processes while still maintaining the advantages of face-to-face interactions with the preferences and capabilities of all generations (Centennials and Millennials). As a result, blended instruction is becoming more popular in higher education due to its efficiency and flexibility and has the potential to become the predominant method of course delivery, particularly in the post-COVID-19 pandemic (Agarwal, 2020; Dziuban et al., 2018).

Blended learning has several benefits in the writing context that make it crucial for this study and for enhancing paragraph writing. In this regard, Shand and Glassett Farrelly (2018) clearly state that blended learning benefits from the flexibility and abundance of tools such as grammar checkers, and online learning resources any place steadfastly. Blended instruction also offers the benefits of instant feedback on students’ paragraph writing skills, discussion forums, and collaborative writing. According to Klimova and Kacetl (2015), the use of Google Docs as a feature of Google Classroom, for instance, facilitates real-time feedback, discussion forums, and collaboration, allowing students to engage in the writing process more actively and receive immediate guidance from instructors. This exposure to various resources and tools can expand students’ knowledge and understanding of paragraph structure, including how to use topic sentences, supporting details, organization, and coherence, leading to improved writing performance. Students can also benefit from aspects of traditional education like getting input from instructors, getting clarification on tasks, and asking instructors and other students’ questions in real time (Shand & Glassett Farrelly, 2018).

However, it is essential not to forget that blended learning also has drawbacks that could hinder students’ ability to write paragraphs: lack of access to technology or unstable internet connectivity. According to Rasheed et al. (2020), blended learning has been examined for its effectiveness from two perspectives. On one hand, students may face challenges in possessing the necessary self-regulation skills to effectively navigate the demands of blended learning. On the other hand, instructors may encounter limitations in their access to appropriate tools and resources for designing and developing the essential course materials required for successful blended learning environments. Hence, this digital divide may lead to unequal access to online resources and affect how much students participate in blended learning (Maulida et al., 2022). In this context, the Digital Divide Network theory is worth mentioning as it draws attention to the differences in access to digital resources and technology. Therefore, it is imperative to close the digital gap and make sure every student has equal access to possibilities (Warschauer, 2003). In addition, students may require additional support and guidance to navigate online platforms effectively and maximize their learning experience. These studies recommended that additional research be conducted to explore and expand the current understanding of blended learning.

Theoretical Framework

The present study is grounded on two prominent theoretical frameworks: social constructivist and connectivist theories. The social constructivist and connectivist theories provide evidence-based guidance for creating an environment that enhances student engagement, collaborative learning, and critical thinking, which are fundamental skills for effective paragraph writing. Social constructivist theory suggests that learning can occur through social interaction and collaboration (Vygotsky, 1978). Social constructivism, according to Lalima and Dangwal (2017), forms the basis of blended learning. In this setting, social constructivism stresses collaborative learning, scaffolding, and peer feedback where students actively engage with their peers to construct knowledge together (Lin, 2014). Social constructivist theory includes scaffolding and ZPD (Zone of Proximal Development). The first essential idea of Vygotsky is scaffolding. According to Finnegan and Ginty (2019), the social constructivist theory and the concept of scaffolding are strongly related. Scaffolding allows students can correct their errors on their own by looking for answers or content on the Internet and online resources like writing templates or examples (Reinking, et al., 2009). Scaffolding will also assist students in deciding what to consider about learning and direct them to key concepts. Thus, with the help of scaffolding, students can better understand paragraph writing patterns, plan their ideas, and improve their paragraph writing abilities.

Conversely, ZPD (Zone of Proximal Development) is a way to improve students’ skills by correcting their errors by using the position of teachers or other students. According to Polly and Byker (2020), Vygotsky developed the idea that every person has a unique ZPD within the framework of social constructivism. They add that these tools enable people to execute tasks that are beyond what they can do on their own. Tools such as interactive writing platforms, online discussion forums, and virtual collaborative writing activities can foster student engagement and interaction in a supportive learning environment. In other words, ZPD can boost students’ abilities by correcting their errors. Blended Learning has given input or feedback on its benefits in teaching and learning. Feedback has the power to fix mistakes made by students and advance their paragraph-writing abilities. To put it another way, blended learning and the ZPD feedback concept are related.

The other suitable theoretical framework for this study is the connectivist theory. According to Siemens (2004), connectivism theory emphasizes the value of interactions and collaboration in learning through social networks that are mediated by technology. By incorporating technologies such as learning management systems, social media, and the internet into blended instruction, educators can facilitate student interactions and connections with experts and peers beyond the physical location of the classroom. This can improve access, increase the diversity of perspectives, and enhance students’ exposure to different modes of writing, advancing their paragraph writing performance.

In developing nations like Ethiopia, where the availability of skilled manpower, classroom facilities, and printed materials are so scarce and costly, the application of blended learning using appropriate educational theories in the teaching in general and learning process plays a significant role in reducing costs and improving the quality of education. Hence, taking the advantage of the 21st century technology seems reasonable to implement BL in teaching the English language to develop students’ ability to use the language in multi-dimensional aspects (Burns, 2010; Brown, 2015). This is particularly important for developing countries like Ethiopia, where the education system is struggling. Consequently, the current study evolved into a significant effort to improve students’ proficiency in paragraph writing in the Ethiopian context using blended instruction as a technology strategy. Therefore, based on the justification given in the background of this study, the researchers were motivated to carry out a study using a quasi-experimental design to examine the effects of using BL on students’ English paragraph writing performance at Pharma College, Hawassa Campus in southern Ethiopia.

Statement of the problem

The impact of modern information and communication technology has changed that how people learn languages, particularly English, and how they learn both online and offline (face-to-face and online) around the world. Currently, language learners have many opportunities to improve their language learning using different technologies (Dudeney et al., 2007). This approach has impacted instruction in general and the four English language skills specifically. Hence, one of the four basic skills in English that is emphasized in the education system is writing, which appears to be the most crucial and difficult skill to be acquired by native speakers and foreign/second language learners (Hamp & Heasley, 2006).

The problems of writing, particularly paragraph writing, involve understanding how to build topic sentences, develop ideas, establish cohesion and coherence, use appropriate vocabulary, and apply the right grammatical rules. There were also cognitive problems such as punctuation, capitalization, spelling, content, and organizational difficulties. Addressing these challenges within a limited amount of time in higher education writing classes is a challenging task. Moreover, traditional teacher-centered approaches to teaching writing often fail to cater to the diverse learning needs and limitations of students, particularly in today’s rapidly digitizing world. As a result, blended instruction, which combines face-to-face instruction with online learning possibilities, has been shown to improve students’ academic attainment in a variety of educational contexts (James, 2016; Liu, 2013). By incorporating technology-based tools, such as online writing resources, virtual discussion platforms, and interactive exercises, blended instructions can provide students with additional support and practice opportunities to improve their paragraph writing skills.

Previous studies

Several studies at various educational levels have been conducted on how blended learning is used in writing courses. For example, Liu (2013) investigated blended instruction in an academic writing course at a university in Beijing. The findings revealed an increase in student-student and student-teacher interactions, a gain in academic skills, a decrease in communication anxiety, and a growth in learning autonomy. Sabarun (2019) examined the paragraph writing instruction demands of 20 pupils. The results showed that students primarily encountered grammar issues when writing paragraphs and that they tended to consult the internet rather than other sources to find solutions.

A recent study on Google Docs supports and confirms blended learning effectiveness. For instance, according to Ebadi and Rahimi’s (2017) research, students who used Google Docs for peer editing outperformed those in face-to-face classes and saw a considerable improvement in their academic writing abilities. Additionally, they noted that when utilizing Google Docs, students worked more closely with their peers to revise and give feedback on written tasks than they did when doing them in person.

Although scant, some local research endeavors have explored the topic of blended learning within the Ethiopian educational context. An illustrative example is the study conducted by Geta and Olango (2016), who explored the impact of blended learning on students’ writing proficiency, alongside the perceptions of English language educators and students at Hawassa University. The study’s findings revealed that, within the writing skills course, students in the experimental group outperformed their counterparts in the control group.

Another recent local study that investigated the application of blended instructional approaches to enhance students’ writing skills was conducted by Mamo (2021). This study incorporated a blend of writing strategies, including product, process, and genre approaches, without the use of e-learning or technology. The results indicated that the use of these blended writing strategies had a positive impact on the EFL writing abilities of the students in the experimental group.

Studies on blended learning utilizing other emerging online platforms are still rare, and hence, this study is based on what the ‘second generation (web 2.0, centennials) technology’ has brought into educational institutions: ‘Google classroom platform’. The investigation of ‘Google classroom platform’ in writing courses has remained more under-researched than its other Web 2.0 counterparts (i.e. Moodle, bogs or wikis) (Ebadi & Rahimi, 2017).

The majority of the above studies have focused on the effects of online learning on improving models, methodologies, and instructors’ attitude toward blended learning. However, it can be said that few studies have focused on the impacts of on-line and conventional approaches on students’ performance in paragraph writing among Ethiopian EFL learners. In contrast, this study investigated both online and face-to-face activities in the paragraph writing course. Thus, in order to fill this gap, the present study sought to test the effects of blended learning in improving students’ paragraph writing performances on first-year pharmacy students at Pharma College, Hawassa Campus, Ethiopia. To this end, the following research questions were developed.

Research questions

This study aimed at examining the effect of using BL on students’ paragraph writing skills at Pharma College, Hawassa Campus. The following questions were also answered in this study.

1. Is there a statistically significant difference in mean scores on pre-test results between students who are taught writing through blended learning and students who are taught writing through conventional methods in paragraph writing?

2. What is the significant difference in mean scores on post-test results between students who are taught writing through blended learning and students who are taught writing through conventional methods in paragraph writing?

3. Which specific features of blended learning (if any) assisted students in writing paragraphs?

Hypotheses

The following null and alternative hypotheses are formulated.

H0: There will be no statistically significant difference in mean scores on pre-test results between students who are taught writing through blended learning and those who are taught writing through conventional methods in their paragraph writing.

H1: There will be a statistically significant difference in mean scores on pre-test results between students who are taught writing through blended learning and those who are taught writing through conventional methods in their paragraph writing.

H0: There will be no statistically significant differences in mean scores on post-test results between students who are taught writing through blended learning and the students who are taught writing through conventional methods in their paragraph writing.

H1: There will be statistically significant differences in mean scores on post-test results between students who are taught writing through blended learning and those who are taught writing through conventional methods in their paragraph writing.

Methods and materials

In this study, the researchers used a quasi-experimental design because it is suited to the nature of the study. According to Morris (2009), this type of design is suitable when it is not feasible or practical to randomly assign participants to different groups. This is mainly because, in the context of Pharma College, the students were already organized into existing sections based on the college’s enrollment and class allocation schedule. Disrupting these established groups to create random assignments could have been impractical (Creswell, 2008). Moreover, random assignment might raise ethical concerns, such as ensuring equal access to potentially beneficial educational interventions for all students (Dornyei, 2007). Consequently, to maintain effective comparison without changing the makeup of the groups, the researchers chose a quasi-experimental design. As a result, two intact classes of Pharmacy Department students who had already signed up for Communicative English Skills II were specifically taken for the study by the researchers. Hence, two non-equivalent groups were used to study the effects of the independent variable (the blended learning method) on the dependent variable (paragraph writing ability).

A pre-test and a post-test were conducted at the beginning and end of the course (Communicative English Skills II), respectively. Pre and post-test measurements are appropriate for research that seeks to evaluate the effectiveness of a treatment (Hancioglu et al., 2019). In this study, the participants in the experimental group learned through blended learning methods using Google Classroom in the online session in addition to face-to-face sessions. In contrast, the participants in the control group learned through only face-to-face or conventional methods to examine the effects of using BL on students’ paragraph writing performance at Pharma College, Hawassa Campus.

Research setting

This study was conducted at Pharma College, Hawassa Campus, a private Institution in Hawassa City, the capital of the Sidama Regional State of Ethiopia. Hawassa City lies approximately 275 kilometers to the south of Addis Ababa, and the Pharma College Hawassa Campus is situated in the south of the city. The study was conducted from October to December 2022.

Participants of the study

The targeted populations for the study were all 267 undergraduate first-year Pharmacy students at Pharma College, who were registered in the 2nd semester of the academic year 2022 for the course Communicative English Language Skills II (Enla1012).

Sampling techniques and sample size

Different sampling techniques were used to select the setting and study participants. Accordingly, Pharma College Hawassa Campus and participants of the study were selected using a convenient sampling technique due to the geographical proximity and familiarity of the context with the researchers. According to Dornyei (2007), convenient sampling is a non-probability sampling technique that involves the selection of participants based on their availability, accessibility, and willingness to participate in the study. The main reason the researchers chose Pharma College Hawassa for their study was that the college offered them access to a fully functional, contemporary ICT lab with internet connectivity, which they used for training and data gathering.

Furthermore, from the pharmacy department at the college, two intact sections of first-year students were chosen out of six available sections. The selection of students for the study was based on their availability, willingness to participate, and a certain level of technological proficiency, as outlined by Dornyei (2007). Subsequently, these selected students were randomly divided into two groups: the experimental group (N = 28) and the control group (N = 26). In total, the study involved 54 students. The control group received conventional paragraph writing instruction, whereas the experimental group was exposed to a blended learning approach to paragraph writing. Both groups received instruction from the first investigator over the course of eight weeks.

Data gathering instruments

For this study, the researchers developed two data gathering instruments: paragraph writing tests (Pre-Test and Post-Test) and a student questionnaire. Both tests were the same in content and numbers. The first part of the tests included written prompts that encouraged students to provide a response to a specific question. This type of task evaluates students’ ability to generate coherent and well-organized sentences. Another component of the tests was re-writing erroneous paragraphs, providing students with the opportunity to identify and correct errors in grammar, structure, or content. The test also involved writing paragraphs based on the given information in a table. In this task, students were required to write paragraphs that incorporated a topic sentence, supporting sentences, and a concluding sentence. This task assessed the students’ ability to structure paragraphs effectively.

The researchers administered pre-test and post-test before and after the intervention, respectively. The tests were administered to both experimental and control groups to measure the changes or improvements in students’ writing skills resulting from the intervention. The pre-and post-test writing were partly adopted and constructed from the Ministry of Education Module (MoE, 2019). Additionally, the researchers used concepts presented by authors such as Hughes (2003) and Heaton (1988) who have expertise in writing English language tests. The tests were also constructed on the basis of the minimum learning competency and components of the Communicative English Language Skills Course, ensuring alignment with the curriculum and desired learning outcomes. Two raters, selected as the senior most instructors, played a vital role in correcting the students’ papers. They used a rubric that encompassed multiple criteria, including structure, content, vocabulary, organization, grammar, and mechanics (Wahyuni, 2018). The rubric provides clear guidelines and objective standards by which the raters evaluated and scored the students’ work, ensuring consistency and fairness.

The second instrument utilized in this study was a student questionnaire designed to assess the computer literacy of EFL students and gather data on their personal and academic backgrounds. Furthermore, at the end of the intervention, a modified version of Arias’s (2015) five-point Likert scale questionnaire, tailored to evaluate blended learning components, was administered to the 28 students in the experimental group. This study aimed to measure their comprehension of blended learning features and the impact on their paragraph writing skills.

Validity

Validity denotes the extent to which a scale measures the construct it is supposed to. It is analogous to the scale’s accuracy (Agala et al., 2020). The instrument’s validity was examined using construct validity. This was because several types of validity fall under the umbrella of construct validity which is a psychological and emotional concept that cannot be confirmed through statistical measures; it was believed that it might be evaluated at different stages of the data collection process. Therefore, construct validity is upheld if the data-collecting tool prompts participants to include the required information in their paragraph writing. Additionally, to verify the validity of the test items and intervention materials, the researchers, and two Ph.D. ELT candidates and one English language expert reviewed and evaluated the tools before the study was conducted. These language experts offered their opinions on the tasks’ and materials’ applicability to the goals and objectives of the research as well as the way the questions were arranged. Finally, the necessary improvements were made after taking their suggestions and criticisms into account.

Inter-Rater reliability

In this research, the reliability of this instrument was determined by using inter-rater reliability. According to Creswell (2008), Inter-Rater Reliability is the consistency between 2 or more raters when evaluating or scoring the same set of responses or test papers with little or no difference in scores. In this study, 54 test papers were corrected by the two raters. In doing so, the Pearson Correlation coefficient was computed to assess the correlation of the scores marked.

Therefore, the inter-rater correlation of experimental group rater 1 against rater 2 in pre-test was r = .977. This implies that the two raters had a very positive correlation. Similarly, for the control group in the pre-test, the correlation of rater 1 against rater 2 was .989. This indicates an even higher level of agreement between the two raters as compared to the experimental group. For the posttest conditions, the inter-rater correlation of experimental group rater 1 against rater was, 0. 974. Likewise, the correlation coefficient of rater 1 against rater 2 for the control group in the post- test correlates at.989. This depicts a strong Positive Correlation and demonstrates a high level of consistency. However, for the experimental group in the post-test, the correlation between raters 1 and rater 2 was slightly lower at 0.974. Although this value is slightly lower, it still suggests a significant positive correlation between the two raters. This was found at a very high level and can be considered as acceptable inter-rater reliability coefficients.

Procedure of the study

In this study, the researchers followed a careful and well-planned set of procedures to ensure its success and validity. They first designed and prepared training materials, schedules, and instructions for both the Experimental and Control groups. Training manuals, guidelines, and schedules were prepared based on the guidelines of UNESCO and the Federal Ministry of Education of Ethiopia (FDRE Ministry of Education, 2009). To ensure that the study was ethical they had to receive a letter of approval from Hawassa University. First, the researchers obtained a letter of approval for the study from Hawassa University, which was written in September 2022, to collect data. Upon receiving the required letter of approval, the first investigator went to the Pharma College to submit the letter and introduced the study to the participants. This step was taken to inform the college administration about the study and to seek permission to conduct it in their institution.

Next, after obtaining permission, the researchers discussed the purpose of the study with the Dean and Head of the Pharmacy School and made an appointment to discuss it with the ICT technicians. Then, the first researcher contacted pharmacy students, discussed the research objectives and informed them about their role during the intervention. For this purpose, they signed an informed consent form to participate in the study voluntarily. After getting the official permission from Pharma College to gain access to the ICT center and classrooms for the intervention, the researcher started data collection process. One of the classes was assigned as the experimental group, while the other class was assumed to be the control group. The pre-test was administered to both the experimental and control groups to establish a baseline of the students’ paragraph writing abilities. This initial assessment occurred one week before the start of the instructional intervention, ensuring that students’ performance could be accurately measured before any treatment was applied.

Before starting the intervention, the researchers organized a half-day training sessions for the students. This training solved any technical problems related to information and communication technology (ICT) that might hinder the smooth implementation of the study. The experimental group students were asked to provide their e-mails to the researchers for registration on the Google Classroom platform to have access to the platform. Then, the first researcher created a class on Google classroom and sent the code to the students to join the class. Then, participants in the experimental group were taught how to create a Google account, join a class, and send messages, posts, and assignments. The training helped familiarize the students with the online learning platform (Google Classroom), troubleshoot any issues, submit assignments, communicate with the instructor and peers, and ensure their readiness for the intervention.

The instructional intervention began immediately after the pre-test and took a total of eight weeks, with two sessions per week for both the experimental and control groups between October and December 2022. The experimental group received blended learning, while Control group received conventional methods or face-to-face type of teaching. The same teacher (the researcher) taught paragraph writing in both groups.

During the intervention period, the researcher monitored the progress of both groups and provided feedback to the students to help them improve their writing skills. Throughout the intervention, the researcher developed several exercises to enhance various writing-related sub-skills. Participants of both experimental and control groups were asked to write sentences, paragraphs, and work on the material that was prepared for the intervention. The experimental group wrote their paragraphs and tasks on Google Docs; they were also asked to comment on their classmates’ posts and on discussion forum. They submitted their tasks to the instructor for feedback through Google Classroom. The researcher checked students’ posts and gave feedback.

Upon completion of the eight-week instructional intervention, the post-test was administered to both groups simultaneously on the same day, exactly one week after the final intervention session. Then, the post-test data were collected from the participants. Following the post-test, the researchers administered questionnaire for experimental group students. The questionnaire aimed to capture the students’ perceptions of the blended learning approach, to identify the features of blended learning on the Google platform that helped them improve their paragraph writing, their engagement with the online components, and the overall effectiveness of the intervention in enhancing their paragraph writing skills.

Finally, two raters assessed the post-test using a scoring rubric designed for writing tasks as they had done the pretest. Then, and the researchers analyzed the data using statistical methods to determine the impact of blended instructions on the paragraph writing performance of students.

Methods of data analysis

Data from participants’ paragraph writing was analyzed using SPSS (Statistical Package for the Social Science, version 25) to determine if there were any statistically significant effects on the students’ paragraph writing performance following the intervention. First, descriptive statistics for the pre-test scores of both groups were computed to determine their initial writing performance levels for each group. To compare the mean difference between the pre-test and post-test scores of each group, a paired sample t-test was conducted. In addition, an independent sample t-test was used to determine the mean difference between the pre-test and post-test scores for the two groups. Moreover, the researchers calculated Cohen’s d effect size to determine the magnitude of the differences observed in addition to P-value. As a result, Cohen’s (1988) benchmarks were used to interpret the results. Cohen categorizes the effect size into three categories: small effect size, medium effect size and large effect size; between 0.10 and .29, between 0.30 and .49, and 0.50 and above, respectively. Lastly, a multiple regression analysis was carried out to determine the essential elements of blended learning that have a substantial impact on paragraph writing performance.

Results of the study

Based on first-semester results and pre-test writing scores, an independent sample T-test was used to determine whether the computer abilities and English language ability of the experimental and control groups were similar or not. This section also presents the results of the regression analysis used to identify the blended learning features that seemed significant.

Students’ first semester English language performance

Table 1 examined the academic performance of students in Communicative English Skills I. The results revealed that the experimental group had a higher average score (M = 74.70, SD = 11.74) than the control group (M = 69.69, SD = 9.48). However, the independent Sample T-Test ($t_{54}=-1.738,p-\mathit{\text{value}}=0.088),$ suggested that the two groups were not different in terms of general English performance level before the treatment.

Table 1. Sample Independent T-test results between control and experimental groups.

Items	Group	N	Mean	SD	t	df	p-value
Students’ first semester Communicative English Skills I academic status	Control	26	69.69	9.48	−1.738	54	.088
	Experimental	28	74.70	11.74

Students’ general digital literacy skills

Table 2 assessed digital literacy skills across various tasks. In examining digital literacy among students, the control group displayed a higher mean in computer typing (M = 5.15, SD = 1.16) compared with the experimental group (M = 4.6, SD = 1.00), yet with no statistically significant difference (p = 0.06), suggesting similar typing proficiency. Similarly, both groups displayed comparable abilities in using language learning software, with the control group having a higher mean (M = 5.08, SD = 1.06) than the experimental group (M = 4.87, SD = 1.14), but with no significant difference (p = 0.479). On the other hand, the experimental group reported a higher mean score (M = 6.87, SD = 9.12) for using language learning websites compared with the control group (M = 4.65, SD = 1.5), yet the high variability within the experimental group suggests inconsistent skill levels.

Table 2. Literacy skills between the control and experimental groups.

Items	Group	N	Mean	SD	t	p-value
Typing English texts (on computer)	Control	26	5.15	1.16	1.92	0.06
	Experimental	28	4.60	1.00
Language Learning software	Control	26	5.08	1.06	0.714	0.479
	Experimental	28	4.87	1.14
Language Learning website	Control	26	4.65	1.50	−1.221	0.227
	Experimental	28	6.87	9.12
Create/ send / receive phone text messages.	Control	26	5.31	1.01	−2.374	0.021
	Experimental	28	5.80	0.48
Search engines e.g. Facebook, Google, Yahoo, online dictionary, etc	Control	26	4.92	1.47	−1.37	0.176
	Experimental	28	5.37	0.93
Online services e.g. registration, Assessment-Grade, payment fees etc	Control	26	4.27	1.56	−2.285	0.026
	Experimental	28	5.03	0.89
Use Email, or Telegram to write and send assignments	Control	26	4.85	1.54	−1.736	0.088
	Experimental	28	5.40	0.77
Summary	Control	26	4.88	1.36	−1.557	0.126
	Experimental	28	5.38	1.96

In mobile texting, the experimental group outperformed the control group (M = 5.8, SD = 0.48 vs. M = 5.31, SD = 1.01), a difference that was statistically significant (p = 0.021), indicating stronger skills in this domain. The experimental group also had higher means in employing search engines and online dictionaries (M = 5.37, SD = 0.93) compared with the control group (M = 4.92, 1.47), but without statistical significance (p = 0.176). When navigating online services for tasks such as registration and fee payments, the experimental group excelled (M = 5.03, SD = 0.89) over the control group (M = 4.27, SD = 1.56) with a significant difference (p = 0.026). When using email or telegram for assignments, the experimental group again had a higher mean (M = 5.4, SD = 0.77) versus the control group (M = 4.85, SD = 1.54), yet this was not significant (p = 0.088). However, the overall summary scores indicated that both groups had relatively similar digital literacy skills, with no significant difference in their overall abilities (t = -1.557, p = 0.126). The following table presents the pre-test results of the experimental and control groups.

Pre-test results of the experimental and control groups

Before the intervention, the students’ performance in writing paragraphs was examined to determine if the 2 groups are comparable or not. An independent sample t-test was employed (see Table 3).

Table 3. Pre-test scores for the experimental and control groups.

Dimension	Group	N	Mean	SD	t	df	Sig. (2-tailed)
Organization	Control	26	11.15	3.42	−1.117	52	0.269
	Experimental	28	12.30	4.08
Content	Control	26	10.94	3.20	−0.720	52	0.475
	Experimental	28	10.59	3.39
Grammar	Control	26	9.40	2.97	−0.415	52	0.680
	Experimental	28	9.73	2.85
Mechanics	Control	26	10.62	3.45	−0.574	52	0.568
	Experimental	28	9.09	2.58
Style	Control	26	9.37	3.01	−1.233	52	0.223
	Experimental	28	8.39	3.10
Total	Control	26	51.48	15.09	−0.893	52	0.376
	Experimental	28	50.11	14.73

Table 3 presents the Pre-Test scores of an independent sample t-test for the experimental and control groups. The t-test for the independent sample was used to determine whether or not there was significant difference. Accordingly, the overall findings revealed that the mean test score of the pre-test for the control group was (M = 51.48, SD = 15.09); whereas, it was (M = 50.11, SD = 14.73) for the pre-test of the experimental group. The average pre-test mean scores of the control group was somewhat higher than those of the experimental group. The mean difference of the pre-test scores of the two groups was (MD= -0.893). However, according to the overall results of the independent sample t-test, (t52=-0.893, p = 0.376), the pre-test scores of the two groups are not statistically significant (p=.376) in terms of their paragraph writing performance. Thus, both groups were comparable because their performance was similar before the experiment.

Pre-Test and post-test results of the control group

A paired sample t-test was conducted to see if there was a statistically significant difference at the p< .05 level between the Control group’s pre- and post-test results.

According to Table 4, a decrease in the average results of the content component was observed as students scored (M = 10.94, SD = 3.20) and (M = 9.06, SD = 3.13) in the pre-and post-tests, respectively. When we see dimension-wise analysis, students’ pre-content test scores are better than their post-test results. For the rest of the test scores, no significant difference was observed. However, when compared to the pre-test, the overall paired sample t-test findings among the control group revealed that students had shown some improvements on paragraph writing performance in the post-test, as it is natural to improve after learning a course.

Table 4. Paired sample t-test for comparing pre-test and post-test results for the control group.

Dimension	Test	Mean	SD	t	Df	Sig.(2-tailed)
Organization	Pre-test	11.15	3.42	3.281	25	0.003
	Post-test	12.75	2.89
Content	Pre-test	10.94	3.20	1.690	25	0.104
	Post-test	9.06	3.13
Grammar	Pre-test	9.40	2.97	2.618	25	0.015
	Post-test	10.17	2.90
Mechanics	Pre-test	10.62	3.45	−0.549	25	0.588
	Post-test	11.08	2.56
Style	Pre-test	9.37	3.01	−2.024	25	0.054
	Post-test	10.17	2.90
Total	Pre-test	51.48	15.09	−1.643	25	0.113
	Post-test	53.23	14.74

Pre-Test and post-test results of the experimental group

A paired sample t-test was used to determine whether there was a statistically significant difference at p< .05 level between the pre- and post-test results of each writing component of the experimental group.

The findings of the experimental group’s paired sample t-test are shown in Table 5. The pre-test’s mean score (M = 50.11, SD = 14.73) and that of the post-test is (M = 76.75, SD = 10.30). This shows that the average score of the Pre-test is lower than that of the post-test. The experimental group’s pre- and post-test scores have a mean difference in (MD= -10.48). The significance level for comparing the pre- and post-test average scores is p=.000. Thus, the p-values of the statistical tests for all five components are all statistically significant (p < 0.05).

Table 5. Paired sample t-test results of the experimental group.

Dimension	Test	Mean	SD	t	df	Sig.(2-tailed)
Organization	Pre-test	12.30	4.08	−5.808	27	0.000
	Post-test	16.50	2.09
Content	Pre-test	10.59	3.39	−8.968	27	0.000
	Post-test	15.77	2.30
Grammar	Pre-test	9.73	2.85	−8.231	27	0.000
	Post-test	14.48	2.15
Mechanics	Pre-test	9.09	2.58	−12.943	27	0.000
	Post-test	15.21	2.14
Style	Pre-test	8.39	3.10	−12.110	27	0.000
	Post-test	14.79	2.44
Total	Pre-test	50.11	14.73	−10.488	27	0.000
	Post-test	76.75	10.30

Post-test results of the experimental and control groups

The experimental and control groups’ post-test mean gain scores were compared using an independent-sample t-test at the p<.05 level to see if there was a statistically significant difference between the 2 groups.

Table 6 shows that the post-test mean scores of the experimental group on each writing component are higher than their counterparts in the control group. As a result, there were significant differences between the post-experimental and post-control groups in organization ($t_{52}=6.972;p<0.05),$ content ($t_{52}=5.956,p<0.05),$ Grammar ($t_{52}=6.852,p<0.05),$ mechanics ($t_{52}=6.852,p<0.05)$ and style ($t_{52}=9.094,p<0.05).$ The total scores between the post-experimental group (M = 76.75; SD = 10.30) and the post-control group (M = 53.23; SD = 14.74) were statistically significant differences ($t_{852}=7.935,p<0.05).$ Thus, the difference in the post-test scores of the experimental and control groups was statistically significant, and the p-values of the statistical tests for all five components were all statistically significant.

Table 6. Post-test results of the experimental and control groups.

Dimension	Group	N	Mean	SD	T	Df	Sig. (2-tailed)
Organization	Experimental	28	16.50	2.09	6.972	52	.000
	Control	26	12.75	3.46
Content	Experimental	28	15.77	2.30	5.956	52	.000
	Control	26	9.05	3.10
Grammar	Experimental	28	14.48	2.15	6.852	52	.000
	Control	26	10.17	2.89
Mechanics	Experimental	28	15.21	2.14	8.505	52	.000
	Control	26	11.08	3.13
Style	Experimental	28	14.79	2.44	9.094	52	.000
	Control	26	10.17	2.90
Paragraph Writing skills	Experimental	28	76.75	10.30	7.935	52	.000
	Control	26	53.23	14.74

However, effect size was calculated to evaluate whether the magnitude of the differences was attributable to blended instruction or not. This is used to determine whether the treatment has a small, medium, or large effect size on the study sample. In this study, the effect size was calculated using eta squared, was r = 0.548 for the overall paragraph writing performance. According to Cohen (1988), the obtained value was classified as a large effect size, revealing that blended learning had a significant effect on students’ paragraph writing performance as measured before and after the treatment (Table 7). It can be summarized that the null hypothesis is rejected as these groups do not have similar effects on students’ writing ability, and there is a meaningful difference between the two groups.

Table 7. Results of tests of between-subjects effects for writing skills.

Tests of Between-Subjects Effects
Dependent Variable: Score
Source	Sum of Squares	df	Mean Square	F	Sig
Corrected Model	10039.122^a	1	10039.122	62.966	.000
Intercept	214751.270	1	214751.270	1346.937	.000
Group	10039.122	1	10039.122	62.966	.000
Error	8290.712	52	159.437
Total	236834.000	54
Corrected Total	18329.833	53

a. R Squared = .548 (Adjusted R Squared = .539)

Tests of between-subjects effects

 

Regression coefficients’ model summary

As shown in Table 8, the regression model summary indicated that the combination of five independent variables-face-to-face, collaborative writing, authenticity, flexibility, and instant feedback accounts for 60.3% (R² Square = 0.603) of the variance in the dependent variable (students’ paragraph writing skill). This implies that a substantial portion of the variation in students’ paragraph writing test scores can be explained by these predictors, with only 39.70% attributed to other factors not included in this study. The low standard error suggests a high level of accuracy in the model’s predictions, indicating minimal first-order linear autocorrelations in the data.

Table 8. Regression Coefficient of Blended Learning design features.

Model	R	R Square	Adjusted R Square	Std. Error of the Estimate
1	0.776^a	0.603	0.531	.10925

^a Predictors: (Constant), face-to-face, collaborative writing, authenticity, flexibility, instant feedback.

^b Dependent variable: student writing skills exam score.

Table 9. Results of analysis of variance.

Model		Sum of Squares	df	Mean Square	F	Sig
1	Regression	2.377	5	0.4754	6.691	.000^b
	Residual	1.563	22	0.07105
	Total	3.94	27

^a Dependent variable: score.

^b Predictors: (Constant), face-to-face, collaborative writing, authenticity, flexibility, instant feedback.

Table 10. Results of multiple linear regression model.

Model	Unstandardized Coefficients		Standardized Coefficients	t	Sig	95.0% Confidence Interval for B
	B	Std. Error	Beta			Lower Bound	Upper Bound
(Constant)	.204	.084		.720	.479	.384	.793
Flexibility	.404	.120	.382	3.364	.003	.155	.653
Instant feedback	.256	.097	.301	2.648	.015	.055	.456
collaborative writing tasks	.223	.061	.313	3.648	.001	.096	.350
Authentic Language	.176	.081	.185	2.181	.040	.009	.343
Face-to-face	.073	.083	.070	.872	.393	−.100	.245

^a Dependent variable: paragraph writing skill exam score.

Results of ANOVA

The F-ratio in the ANOVA table tests whether the overall regression model is a good fit for the data (see Table 9). The table shows that the independent variables statistically significantly predict the dependent variable, F (5, 22) = 6.691, p < .0005 (i.e. the regression model is a good fit of the data), indicating that the independent variables collectively provide strong predictive power for the students’ paragraph writing performance.

Multiple linear regression model

The equation for a multiple linear regression is: $Y_{\mathit{\text{exam score}}}=0.204+.382\mathrm{*}$ flexibility +.301* instant feedback+ .313*collaborative writing tasks+ .185*authentic+ .070* face-to-face.

The Multiple Linear Regression Modeltable presents the influence of various instructional strategies on students’ paragraph writing skills, as indicated by their test scores (see Table 10). From the regression model, the standardized beta coefficients allow us to compare the relative impact of each predictor variable. The higher the beta value, the stronger its contribution becomes. Accordingly, the variable 'flexibility’ shows a positive and significant relationship with paragraph writing skill scores, as indicated by a beta coefficient of (.382) and a significant value (p = .003). This suggests that greater flexibility in the learning environment is associated with higher writing test scores. This was followed by collaborative writing tasks (Beta = 0.313) and instant feedback (Beta = 0.301), both of which were significant contributors to enhancing students’ paragraph writing skills.

The 'authentic materials ' variable, representing the use of real-world or meaningful writing tasks, has a smaller but still positive (Beta = 0.185) and significant value of (p = .040), indicating that authenticity in writing assignments can contribute to better writing performance. Lastly, 'face-to-face interaction’ variable has a positive but not statistically significant beta value of .070 (p = .393), suggesting that while traditional classroom interactions may contribute to writing skill development; their impact is not as pronounced within this model as the other variables.

Discussion

This study explored the effect of blended learning approach on the paragraph writing performance of pharmacy students at Pharma College, Hawassa, Ethiopia. Additionally, the study sought to identify specific features within blended learning in Google classroom that could enhance students’ performances in writing paragraphs. The first research question of this study was to determine whether a statistically significant difference exists in the mean pre-test scores for paragraph writing between students instructed in writing through blended learning and those instructed through traditional methods in paragraph writing.

Therefore, this section first discusses the students’ academic status, prior writing experience and digital knowledge. The results showed that the students’ first semester Communicative English Skills I academic status did not show a significant difference between the control and experimental groups, ensuring that both groups were at a comparable level before the intervention. Accordingly, the two groups were not different in terms of the general English language performance level before the treatment.

The evaluation of digital literacy skills conducted on the control and experimental groups also revealed no significant difference in the average scores of the two groups, indicating that both groups were equally proficient in tasks like typing English texts on computers and using language learning software. According to Hadad (2007), computer abilities are a prerequisite for the effective use of blended learning. This study showed that students appeared to have no difficulties or very little trouble with computer technology skills. This is most likely a result of their prior experience using digital technologies for various educational purposes, such as Facebook, Telegram, Tick Talk, and e-books for reading and sending emails. Furthermore, through other courses such as emerging technologies, the college may try to enhance students’ digital literacy abilities. This foundation of digital literacy was crucial for the successful implementation of blended learning, as it required students to engage with online platforms and resources. This finding is consistent with research by Picciano and Seaman (2007) and Hadad (2007), who pointed out that competency with computers and the internet is critical for successful blended learning. Likewise, this research aligns with the findings of Othman and Musa (2012), emphasizing the significance of internet access and speed in enhancing learner satisfaction within the context of blended learning.

Moreover, data analysis from students’ pre-test paragraph writing test further revealed that the average pre-test scores of the control group were somewhat higher than that of the experimental group. However, there was no significant difference between the two groups in the pre-test. This indicated that the null hypothesis was accepted, and the alternative hypothesis was rejected. Thus, it can be implied that the participants in both groups were almost at the same level of writing performance before the intervention.

The second research question was to check if there was a significant difference in the mean scores on post-test results between the control and experimental group students. To address this, a paired sample t-test was carried out for both the experimental and control groups. The aim was to determine the potential impact of blended learning on students’ paragraph writing skills, as assessed through pre-test and post-test evaluations. According to the findings, the means scores of the control group was increased to some extent between the pre-tests (M = 51.48) and post-tests (M = 53.23) results. However, the overall paired sample t-test result indicated that there was no significant difference between the results of the pre-and post-tests of paragraph writing skill.

On the other hand, the results of the experimental group, in the pre-test and post-test were statistically significant. It is also evident from Table 5 that the experimental group mean scores (M = 76.75) on the post-test were higher than those on the pre-tests (M = 50.11) of the same group. This suggests there were significant differences in scores from the pre-test to the post-test in each dimension of paragraph writing (see Table 5). These differences between the experimental group’s mean pre-test and post-test results might be attributable to the effects of the experimental treatment, which were exemplified by the usage of Google Classroom for instruction.

Similarly, the results of the post-test between experimental and control group, as indicated in Table 6, showed that both groups’ writing abilities had increased in terms of their overall writing performances. The blended form of instruction, on the other hand, improved students’ paragraph writing skills and led to improvements in each writing-related rating scale item on the post-test. This improvement was observed more for the experimental group students who used blended instruction. In this study, the experimental group students made significant development in terms of vocabulary, mechanics, punctuation, and grammar. The findings of this study are in agreement with those of Adas and Bakir (2013); Challob, Bakar, and Latif (2016), Ghahari and Ameri-Golestan (2013); Ebadi & Rahimi (2019). These researchers emphasized that students can enhance their writing mechanics and word choice by employing the accessibility features and add-ons provided by the Google platform, such as grammar checkers and vocabulary augmentation tools.

The overall results indicated that the experimental group, which received blended instruction, showed significant improvements with a large effect size in their paragraph writing skills compared with the control group. However, this study diverges from the other findings of Woodrich and Fan (2017), who observed no significant difference in effectiveness between conventional sessions and those assisted by Google Docs. This disparity might stem from variations in implementation, design, utilization, and student interaction with technology.

The third research question of this study was to investigate which particular features of blended learning, if any, contributed to assist students in paragraph writing. The findings revealed that, first, the benefits of online learning coupled with the teacher’s instructional approach aided students in composing paragraphs. Specifically, the flexibility inherent in blended learning emerged as the primary factor with a significant p-value of 0.003 and a standardized beta coefficient of 0.382 contributing to the enhanced paragraph writing skills observed in the experimental group students. Through Google Classroom and an online discussion board, the BL participants had more possibilities to solve problems. Therefore, it was discovered that ‘Google Docs’ gave students flexibility so they could obtain knowledge both online and from the teacher in person (offline). Because of the features of blended learning, including convenience, flexibility, and more time management allowances for each individual, students were able to learn at their own pace by using self-learning methods in an engaging fashion that was not possible with traditional teaching methods. Additionally, the idea of being free in time and space to learn something new worked well in this study. Students worked and submitted their homework and assignments using their cell phone via Gmail account from their home at any time.

This study aligns with Li’s (2018) assertion that Google Docs provides students with flexibility features that allow for independence in terms of time and space. Shand and Glassett Farrelly (2018) similarly argued that blended learning benefits from flexibility and the availability of online learning resources from any location. These findings resonate with those of other scholars in the field, such as Agarwal (2020) and Dziuban et al. (2018), who have highlighted the efficiency and flexibility of blended learning, contributing to its effectiveness and growing popularity in higher education. Consequently, this study supports the claim that Google Docs’ features, which offer time and location independence, indeed provide students with valuable flexibility. Thus, Google Docs provided a solution to this problem as the teacher could highlight specific errors one time for the students and that would save time.

The second feature is the use of collaborative writing tasks in Google classroom, with a beta of 0.313 and a p-value of 0.001. It was discovered that the experience of collaborative writing in Google Docs significantly improved the students’ writing skills. Students also improved their ability to share and organize ideas. Furthermore, the current study’s findings are consistent with previous research (Pae, 2011; Kessler et al., 2012; Ebadi & Rahimi, 2019) all of whom discovered that writing collaboratively using Google Docs increased students’ writing skills. Students who wrote online were more innovative and creative. The use of ‘Google Docs’ also helped students overcome their fears about making mistakes when developing their ideas. As a result, it was simple for the students to identify the errors they had made as well as the modifications, edits, and suggestions offered by the course instructor at various levels.

The findings of this study are similar to previous studies (Means et al., 2013; Bernard et al., 2014) that investigated the impact of blended learning features such as flexibility, collaborative activities, and timely feedback on improving learning outcomes. Additionally, since everything was electronic, it was simple for them to collaborate by reading each other’s writing and comparing and contrasting several revisions of the same writing assignment. This finding supports Hyland’s (2003) assertion that specific steps, including pre-writing, drafting, revising, and editing, contribute to enhanced writing skills. In contrast, traditional instruction often focuses on lectures and subsequent discussions, allocating less time for active collaboration among students. Students who engaged in online writing tended to exhibit greater innovation and creativity. This collaborative dimension aligns with Semin’s (2004) connectivist theory, underscoring the significance of networks and connections in the learning process.

Another significant aspect of blended learning is instant feedback, which was found to be a strong predictor in this study with a beta coefficient of 0.301 and a p-value of 0.015. The enhancement in students’ paragraph writing can be attributed to the immediate feedback facilitated by blended learning components, including face-to-face interactions and online engagement via Google Classroom. Thus, the Google platform used in this study allowed for the option of quick feedback, which not only helped teaching and learning but also inspired the students to work harder. The feedback primarily came from peers and teachers, consistent with the findings of Ebadi and Rahimi (2017) and Klimova and Kacetl (2015) who observed that students using Google Docs for peer feedback outperformed those in face-to-face classroom and their academic writing skills significantly developed. Meanwhile, because of all assignments were turned in and graded using the same platform, the students were able to get quick feedback on their work. Blended Learning, thus, has given input or feedback on its benefits in teaching and learning. This immediate feedback aligns with the principles of the social constructivist theory, which emphasizes learning through social interaction and collaboration (Vygotsky, 1978; Lalima & Dangwal, 2017).

The authentic language interaction gained through integrated learning was the other predicting component. The authenticity of writing tasks positively affects writing skills (beta = 0.185, p = 0.040), but it has less impact on writing paragraphs. Still, when writing assignments have a purpose and are connected to real-world situations, students do better. Thus, the improvement in writing abilities could also be ascribed to increased availability and engagement with digital resources and activities (Adas & Bakir, 2013).

According to this finding, it may be argued that online learning and off-line teaching using real materials were responsible for the experimental group’s gain in mean score. This outcome was in line with that of Mulyadi et al. (2021), who discovered that experimental students’ online communication increased when real materials, audio, video, MP3, YouTube, wiki, blog, Power Points and course content were made available online. This might be the reason why post-test scores for the experimental group significantly outperformed the pre-test results. Li (2022), in addition, emphasized the benefit of real-world language use in blended learning for students’ ability to write English paragraphs. Therefore, it is possible to argue that during online learning, the EFL students might have had access to online learning materials or uploaded course materials and teachers’ explanations of the key points even after the class on face-to-face/offline instruction. Hence, the interaction, feedback and discussion that occurred in Google Classroom might have helped to construct knowledge for improving paragraph writing performance (Laili & Muflihah, 2020).

The least significant predictor in this study was face-to-face interaction, which had a non-significant p-value of 0.393 and a beta of 0.070. Its effect is not statistically significant in this study, although having a positive beta value suggests some amount of contribution to the development of writing skills. This is most likely because students and teachers only have a limited amount of time in the classroom together, and teachers may not have enough time to give comments, feedback, and evaluations to every student at once. In blended learning, students do assignments at their own pace and receive online feedback from classmates, teachers, and online resources. This study indicates a lesser impact of face-to-face engagement, despite other studies (Warschauer, 2003) emphasizing the large influence of face-to-face contacts. This disparity may result from the way blended learning settings are developing, where digital resources are being incorporated more, which could reduce the weight assigned to in-person components.

Therefore, it is vital to mention that in the 21st century, blended learning has emerged as a transformative approach that aligns with multiple learning theories and addresses the challenges posed by the digital divide. In the teaching of paragraph writing, it is crucial to implement effective learning theories such as constructivist, connectivist, and social network theory to enhance learning experiences for students in Ethiopian higher education. These theories provide a foundation for student-centered and active learning methodologies, which have proven effective in enhancing paragraph writing skills for students.

In this study, blended learning was aligned with social constructivist theory of learning in improving students’ paragraph writing performance at Pharma College Hawassa campus, Ethiopia. Social Constructivist theory emphasizes active student engagement and knowledge construction, which can be facilitated through blended learning approaches (Vygotsky, 1978). By incorporating collaborative activities, hands-on projects, and interactive online resources, blended learning encourages students to develop their understanding of paragraph writing (Lalima & Dangwal, 2017). That is the better performance of experimental group could be attributed to the blended learning features and interactions with peers and their teacher.

Similarly, connectivist theory is pertinent to blended learning in this context. According to Siemens (2004), connectivism recognizes the importance of networks and connections in learning, and blended learning provides opportunities for students to engage in networked learning. Through online platforms and digital resources, students can connect with peers, experts, and online communities that enhance their understanding of paragraph writing. By engaging in collaborative discussions, sharing ideas, and accessing several perspectives and resources, students can deepen their knowledge and improve their paragraph writing skills.

In relation to the above theories, the two generations: centennials and Millennials are important to mention in this study since they are well suited to blended learning instruction as they are part of digital natives. In the context of this study and in Ethiopia in general, the centennials and Millennials are accustomed to using digital tools and have grown up in a digital age. However, it is important to consider and address any potential gaps in access to technology and internet connectivity. Therefore, it is crucial to acknowledge the Digital Divide Network theory, which is worth mentioning as it draws attention to the differences in access to digital resources and technology. This theory emphasizes equal access to technology and digital resources among students (Warschauer, 2003). In the context of Ethiopian higher education, where infrastructure and internet connectivity may be limited, it is essential to bridge the digital divide and ensure equal opportunities for all students. This study contributes to the field by showing that even in settings where access to technology may be limited, blended learning can significantly enhance students’ writing skills when digital literacy is sufficiently supported.

Conclusions

The study concluded that blended learning has a positive impact on the paragraph writing performance of pharmacy students at Pharma College, Hawassa campus, Ethiopia. The study utilized a quasi-experimental design and employed two main instruments for data collection: researcher-made writing tests (pre-test and post-test) and student questionnaires. The writing tests assessed students’ paragraph writing abilities before and after the intervention. The questionnaires gathered data on students’ backgrounds, experiences, and perceptions of blended learning components.

The experimental group, which received blended instruction through Google Classroom, demonstrated significant improvements in their paragraph writing skills compared to the control group, which was taught through traditional face-to-face methods. The research identified flexibility, collaborative writing activities, and instant feedback as the key features of blended learning that contributed to the enhancement of students’ writing abilities. However, the study found that face-to-face interaction had the least impact on paragraph writing skills within the blended learning model.

The study’s results align with previous research that highlights the benefits of collaborative writing and immediate feedback in blended learning environments for enhancing writing skills. The effectiveness of Google Docs in providing these benefits is particularly noted, supporting the idea that technology can facilitate peer collaboration and timely feedback, which are crucial for writing development. However, the study also notes divergence ideas from some previous findings where the impact of technology-assisted instruction was not as pronounced, indicating that the success of blended learning may be influenced by the specific context and implementation strategies.

In light of these findings, the study suggests that educators should incorporate blended learning strategies to support the development of writing skills, ensuring that students have the digital literacy capabilities to fully engage with blended learning tools. The recommendation is for educators to adapt their teaching methods to include blended learning approaches that leverage the strengths of both traditional and digital instructional methods.

Limitations

The present study identified a number of inherent constraints that will be addressed in future research. First, the duration of the study was relatively short, spanning only eight weeks, which may not have been sufficient to capture the full extent of students’ writing development or the long-term effects of blended learning. Second, the sample size was relatively small, with only 54 students participating, which limited the ability to generalize the results to a larger population.

Another limitation was the focus on descriptive paragraph writing, which means that the findings may not be applicable to other types of writing or comprehensive writing skills. This study was also limited to a single educational institution, which may have specific characteristics that do not represent other settings. Future research could address these limitations by conducting studies over a longer period, with larger and more diverse samples, and by exploring various types of writing to provide a more comprehensive understanding of blended learning’s impact on writing skills.

Authors’ contributions

All authors participated in the conception and design of this study. Teshome Bekele, the first researcher, was responsible for material preparation, data collection, and analysis. Teshome Bekele also took the lead in writing the initial draft of the manuscript. All authors were involved in commenting, revising, and enhancing the final manuscript for submission.

Ethical approval

Data collection for this study adhered to the ethical principles specified in the Hawassa University Ethical Guidelines for Human Subjects. Therefore, approval for the study protocol was granted by the Hawassa University’s College Ethics Review Committee (CRERC) under reference number [CSSH/101/2022]. Informed consent was obtained from all participants, and they were provided with clear information regarding the nature, purpose, and potential implications of the study. Confidentiality and anonymity of participants have been strictly maintained throughout the research process, and any identifying information has been appropriately safeguarded. Finally, the participants were informed that they had the freedom to decide whether or not to participate in the study, and they had the option to withdraw from the intervention at any point if they wished.

Informed consent

All subjects who participated in this study provided their informed consent.

Acknowledgments

The researchers would like to express their heartfelt appreciation to all individuals and organizations whose contributions were integral to the successful completion of this study. Special gratitude is extended to the English Language and Literature Department at Hawassa University, Hawassa College of Teacher Education, and Pharma College. The researchers would also like to thank Ato Seyoum Kebede and Dr. Wongelawit Seyoum, the owners of Pharma College, for their financial and material support, as well as Mr. Anteneh Fikre, Research and Community Service Coordinator, for organizing the second seminar in the College.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data supporting the results of this study can be accessed upon reasonable request from the corresponding author.

Additional information

Funding

This work was supported by Hawassa University and Pharma College Hawassa Campus, Ethiopia.

Notes on contributors

Teshome Bekele Sime

Teshome Bekele Sime is currently a Ph.D. candidate at Hawassa University and a Lecturer in Teaching English as a Foreign Language (TEFL) at Hawassa College of Teachers Education, Ethiopia. He earned his Bachelor’s degree in English from Jimma University and Master’s Degree in TEFL from Addis Ababa University. His research interests are educational technology and the role of the Blended Learning Approach in English language skills.

Mulu Geta Gencha

Mulu Geta Gencha (PhD.) is an Associate Professor in TEFL at Hawassa University, Ethiopia. He also serves as the Head for the Quality Assurance and Enhancement at Hawassa University in the College of Social Sciences and Humanities. He received a Bachelor’s Degree in English Language and Literature, as well as an MA and PhD in TEFL from Addis Ababa University. His research focuses on Technology Assisted English as a Foreign Language Writing and Blended Learning, termed as 'BLANDOPEDIA'. He has published a number of articles in reputable journals.

Taye Gebremariam Olamo

Taye Gebremariam Olamo (PhD.) is also an Associate Professor in TEFL at Hawassa University, Ethiopia. Additionally, he serves as the Dean of the School of Graduate Studies at Hawassa University. He earned his BA, MA, and PhD degrees in TEFL from Addis Ababa University. His research interests include both qualitative and quantitative research in English language teaching. He has also contributed to the field by publishing a number of articles in reputable journals.