The incorporation of digital technology into the classroom has vastly altered how teachers teach and how students learn. By embracing the use of digital technology, it’s possible to surpass the physical confines of the in-person classroom; therefore, increasing the medium through which teachers can connect with students and students among each other (Nelson, 2013; O’Connor, 2020). The use of mobile technology, which provides access to VLEs through various educational applications, has continued to show an increase in use and Perceived usefulness by both teachers and students (Anatolyevna et al., 2018; Davison & Lazaros, 2015). Surani and Hamidah (2019) state that the utilization of online learning allows students to easily connect and provide students with an opportunity to share and utilize subject matter from a variety of sources that would help increase learning efficiency. Their study also finds that students are willing to continue using online learning methods citing convenience as a factor.

As the world enters the 4th Industrial Revolution, using technology in the classroom to teach real-life skills becomes a facet of all subjects, including EFL (Alakrash & Razak, 2020; Hashim & Yunus, 2019). Jones et al. (2019) found that Korean EFL students generally had positive feelings towards technology in the 4^th Industrial Revolution but were not as positive towards their use in learning English. Teacher readiness was one reason that was given. Wen and Kim Hua (2020) found that a teacher’s self-perceived ICT competence, accessibility of infrastructure and online resources, and suitability of the work environment were all positive predictors of teacher behavioral intention related to using 4^th Industrial related skills in the classroom. However, it and Ganapathy (2019) found that there could be difficulties for teachers making this adjustment, including lack of awareness, lack of time, lack of ICT skills, lack of training, and lack of infrastructure.

With the outbreak of Coronavirus in early 2020, schools had to turn to the use of 4th Industrial Revolution skills to teach all subjects (Wen & Kim Hua, 2020). According to UNESCO, COVID-19 has been the cause of schools worldwide temporarily shutting down, affecting approximately 1.2 billion students in 186 countries since the beginning of 2020. As a result, schools have been compelled to utilize online platforms or transition to blended or hybrid learning practices. Graham (2006) defines the method of blended learning as follows: “Blended learning systems combine face-to-face instruction with computer- mediated instruction” (p. 5). Essentially, education has been forced to accept technology to connect teachers and students to maintain effective class instruction. In response to the COVID-19 pandemic, online-based virtual learning environments have been used by the vast majority of universities worldwide (O’Connor, 2020) as a replacement for classes previously taught in person. This pandemic has become the impetus to push technology forward in an unprecedented manner that can highlight its advantages and flaws. Although technology’s fundamental role in education began as a tool to aid in learning, this unforeseen shift has propelled technology to the forefront, making it the primary vehicle for how teachers and students connect across the globe.

Writing is a production-based skill that requires its creators to record their thoughts in the written word. Traditionally this skill was used with a scribe on a functional surface, but as technology progressed, new tools and mediums allowed creators to record their writing in more innovative ways. printing presses and later typewriters allowed writing to be created more easily and disseminated to a broader audience. It is no surprise that the computer and the digital landscape in which it thrives have given rise to a new form of writing and further expanded the number of users and audiences that can be reached.

The subject of writing has been closely tied with the use of online resources to promote learning. Shepherd (2018) points out that digital writing is practiced regularly in many facets of modern life and should be acknowledged as an essential element of current writing classes. The concept of utilizing online learning platforms to foster and promote writing skills is a trend that has existed for some time (Townsend et al., 2013). The benefits provided by online-based writing tools have often been implemented in conjunction with in-person classes. Online resources allow for the potential for faster feedback and more in-depth explanations that enable students to recognize their errors and make corrections promptly. Ni (2013) also echoes that the quality and quantity of interaction between students and teachers may be increased through using online platforms for teaching writing.

While it is clear that technology not only has a place in the writing classroom, it will likely remain an integral part of teaching methodology indefinitely. Much research focuses on how this technology is used in conjunction with face-to-face learning, such as flipped classrooms or blended learning. Yet, as has been illustrated by the Coronavirus pandemic, there is likely a chance that using online platforms to conduct classes may become the sole teaching method. As Griffin and Minter (2013) point out, much research has been undertaken on developing effective in-person teaching techniques and methodologies. However, that same attention must also be given to learners working in an online environment, showing an existing gap in addressing teaching in an online environment.

Technology’s use in the classroom remains an ever- growing and dynamic element of teaching and learning. As a result, how the use of this technology is perceived and its acceptance is also in constant flux. The Technological Acceptance Model (TAM) introduced by Davis (1986) and further developed as the TAM2 (Venkatesh & Davis, 2000) can be used to help understand how its users accept this technology.

The theory behind TAM centers on the individual’s intention to use a given system determined by two primary beliefs; perceived usefulness and perceived ease of use (Davis, 1989; Davis et al., 1989). The first belief is the user’s perceived usefulness, which reflects how a system can be used to enhance their job, or in education, to successfully learn and complete the task at hand. The second belief, perceived ease of use, refers to how a user believes they are capable of using a particular system. Based on users’ perceived usefulness and ease of use in a system, their intent to use or continue using this system will be affected. The TAM2 further elaborates on Davis’s initial TAM model through the addition of factoring in social influence processes (Venkatesh & Davis, 2000).

For the survey, the TAM was broken into three stages. The first stage uses the subjective norm (SN), output quality (OQ), results demonstrability (RD), and perceived ease of use (EoU) to predict the students’ perceived usefulness (PU) of the online writing system. The second stage of the model explores how well SN, PU, and EoU can be used as predictors for the student’s intention to use (IU) the technology. The final stage delves into the relationship between IU and the online writing system’s usage behavior (UB) (See [Figure 1]).

Subjective norm refers to the social influences exerted on the users of the online system for writing. Amjad et al. (2020) defined SN as “the social influence or the influence of colleagues, friends, and teachers on learners’ acceptance” (p. 155) of a technology.

Output quality refers to the job relevance of the technology being studied (Venkatesh & Davis, 2000). In the case of an online writing system, the quality of the feedback would be necessary for student writing improvement.

Results demonstrability refers to how well a user can explain how the technological system aids their development. As Venkatesh and Davis (2000) explained, “Even effective systems can fail to garner user acceptance if people have difficulty attributing gains in their job performance specifically to their use of the system” (p. 192).

In accordance with the TAM model, this study will seek answers to the following questions:

A total of 97 students enrolled in a required English writing course at a mid-sized Korean university were given a survey about their perception of using an online writing system based on the Technology Acceptance Model (TAM). The survey was performed across two terms which met in the Spring term 2020 and 2021. The students were predominately in their first year at university, with 76 students reporting they were in the first year. Six students said that they were second-year students. A total of five students were in their third year, and eight were in their fourth year. Two students did not respond. The students were enrolled in a range of majors. The largest number of students were enrolled in engineering: 23 students. Eighteen business majors answered the survey, along with eight students who studied the humanities. There were seven students each enrolled in architecture, computer sciences, and education. Six students studied the social sciences. There were five arts and communication majors, and one student stated they studied law. Finally, ten students did not report their major.

The class was a three-credit-hour class that was taught over two 90-minute class periods. The classes met a total of 30 times over the 15-week semester as laid down by the university. The online writing classes consisted of videos based on the course textbook, followed by homework assignments. The homework was broken into assignments aimed at gauging a student’s participation with the video and writing tasks submitted to and then graded using an online system. Writing tasks were assessed using a rubric and a checklist by both teachers. The examinations for the class were also held online due to the safety issues surrounding COVID-19.

The students were given a link to the survey, which was posted on the class website at the end of the semester. There was a notice that participation was voluntary and that no identifying information would be collected. Students were asked to respond to 20 items broken into six factors of usage behavior based on Venkatesh and Davis’ (2000) TAM structure (see <Table 1>). Each question is on a four-point Likert-type scale with a four equaling strongly agree, and one equated to disagree strongly. The survey was administered through Google Forms.

The results were imported to IBM SpSS Statistics (Version 26) for statistical analysis. The descriptive statistics were first determined. The following test was to check if the data could be broken into planned factors by checking internal reliability and the feasibility of a factor reduction. The factors were then analyzed using a linear reduction to determine the way the variables completed the model. If all of the questions did not have a response, that response was removed listwise for analysis.

The items on the survey were broken into seven factors. The first factor of usage behavior (UB) contained a single item with a mean of 3.58. The second factor, intention to use (IU), had items ranging from 3.30 to 3.13. All of these scores were between strongly agree and agree. The third factor, perceived usefulness (PU), had four items with an average response ranging from 3.46 to 3.08. The next factor was perceived ease of use (EoU) which had four items with a range of average scores between 3.46 and 3.04. Subjective norm (SN) was the fifth factor with two questions that had two of the three lowest average responses at 3.02 and 2.77. SN also had the largest standard deviation, implying a more extensive range of differences in the responses. Output quality (OQ) was the sixth factor, which had three items, with a range of responses from 3.22 to 3.01. Finally, result demonstrability (RD) also had three items and had average responses between 3.23 and 3.12 (see <Table 1>).

All but one of the response averages was between strongly agree and agree, except for Item 14: people who are important to me think that I should use an online platform to improve my learning experience was the lowest item and the only one between agreeing and disagreeing at 2.77. Item 1: Currently, I am required to use an online platform to submit my writing homework had the highest score of any item on the survey at 3.58. This result is understandable because all classes were required to submit writing homework through an online system. The standard deviations ranged from .599 to .887.

To determine if the sample size was appropriate for factorial analysis, Beavers et al. (2013) stated two statistical tests should be applied: the Kaiser-Meyer-Olkin value (KMO) and Bartlett’s Test of Sphericity (BTS). The KMO was found to be .844, which is considered meritorious according to Dziuban and Shirkey (1974). And the BTS was found to be significant at an approximate χ² (171) = 1053.693, p = < .004. Both tests were passed, allowing for factorial analysis.

All items were included in the scale reduction and factorial analysis based on load factor and the percent of the total variance explained. peterson (2000) determined the threshold for each as 56.6% variance explained and an average absolute factor loading above .32 in a meta-analysis of 401 factor analysis studies. The standard of larger than one eigenvalue was used to determine the total variance explained for the survey at 67.3%. The twenty items in the survey had factor loadings ranging from .53 to .88 (see <Table 1>). Therefore, all of the items were suitable for use in the analysis.

The Cronbach’s alpha was used to determine internal consistency within the scale items and the entire questionnaire. The whole survey had a Cronbach’s alpha of .919. The first scale item, Intention to Use (IU), had a Cronbach’s alpha of .722. Perceived Usefulness (PU) had a .859, and Perceived Ease of Use (EoU) had .781. The Cronbach’s alpha for scale item four, Subjective Norm (SN), scale item five, Output Quality (OQ), and item six, Result Demonstrability (RD), were .673, .788, and .729 respectively (see <Table 1>). Tavakol and Dennick (2011) reported that the overall Cronbach alpha should be between .700 and .950. One of the factors is slightly lower than the level, but Tavakol and Dennick (2011) stated that a low number of questions could be the reason.

A standard multiple regression was performed to determine how different elements in the TAM for online writing homework predicted usage behavior within the theoretical model. According to the model, PU consisted of four possible parts: EoU, SN, OQ, and RD. The item to total correlations was statistically significant to PU at the p < .004 level. All of the correlations with PU had a large significant effect size of over .50 (See <Table 2>). The interitem relationships were all also significant, with only one being slightly below the medium effect size for the Pearson’s correlation as defined by Cohen (Prajapati et al., 2010). The inter-item and item to total correlations indicate a proper sample size for statistical testing (Amjad et al., 2020).

The overall regression utilizing the four predictors was statistically significant at an adjusted R² = .59, F (4, 85) = 32.83, p ≤ .004, indicating that the model can explain 59% of the variance of the regression. Three of the four predictors were statistically significant, with a positive slope indicating that the perceived usefulness also increased as the predictor increased. EoU had a standardized coefficient β = .24, t (89) = 2.60, p = .011, OQ had a standardized coefficient β = .43, t (89) = 4.30, p ≤ .004, and RD had a standardized coefficient β = .25, t (89) = 2.93, p = .004. SN did not statistically affect the perceived usefulness (See <Table 2>).

The second stage of the model looks at how the predictors PU, EoU, and SN affect the students’ Intention to Use (IU). The item to total (range of .65 to .83) and inter-item correlations (between .46 and .63) were both in the range that made the test valid with medium to large effect sizes. This section of the model had an adjusted R² = .74, F (3, 87) = 86.88, p ≤ .004, indicating that the model can explain 74% of the variance of the regression. All three predictors for IU were statistically significant. PU had the highest standardized coefficient β = .58, t (90) = 7.66, p ≤ .004, EoU had a standardized coefficient β = .18, t (90) = 2.61, p = .011, and SN was at β = .24, t (90) = 3.57, p ≤ .004. Since all beta coefficients were positive, IU increases as PU, EoU, and SN increase (see <Table 3>).

The final standardized regression was run to determine how IU affected a students’ Usage Behavior (UB). This section of the model had an adjusted R² = .05, F (1, 94) = 6.32, p = .014, indicating that only 5% of the variance of the regression can be explained by the model. The standardized coefficient β = .25, t (95) = 2.51, p = .014.