The importance of learning outcomes of online English for Specific Purpose courses in higher education and choosing the best bended model of language teaching has increased rapidly and a lot of studies have proved this issue [1,2]. However, a few researchers have been discovered that how blended model can affect learning out comes in higher education and accounting students in particular.

Shorey et al. [3] mentioned that by integrating blended learning within the course, communication skills and self-efficacy among accounting students were improved. In addition, Halasa et al.[4] claim that traditional model of learning is not the best teaching practices and blended model can make students taking the lead in the knowledge of accounting.

There for, this study is aim to provide accounting students perception about learning ESP based on blended learning model. 

Theretical background

Teachers should not necessarily control and manage new knowledge, skills, attitudes acquiring process through appropriate learning content and reflecting learner’s individual needs and preferences. In fact, blended learning is a student center learning technique that provide self-paced, self-administered pedagogy and learning activity along with best learning contents [5]. Both students and teachers can improve their skill and digital literacy by applying modern digital devices and applications. Also, Poon [7] mention that blended learning is a applicable not only to study individually but also in cramped one [8]. 

Research Problem, Question, Objective 

In higher education, the link between online and physical courses is the blended model. For preparing higher education accounting student, non-study has been done. Now, the question is whether the process of ESP learning for accounting professional can be improved and developed by blended model. Thus, the main goal of this pilot study is to determine and analyze accounting statements about comparison between blended learning and physical one as well as their learning out comes. 

Hypotheses

Six hypotheses were design to achieve goals: 

• H₁: There are significant paired differences between results of pretest and posttest  1 in group E statistically           

• H₁: There are significant paired differences between results of pretest and posttest 2in group E statistically. 

• H₁: There are significant paired differences between results of pretest and posttest 1in group C statistically.

• H₁: There are significant paired differences between results of pretest and posttest 2 in group C statistically.

• H₁: There are significant paired differences between results of pretest and posttest 1in group E and C statistically. 

• H₁: There are significant paired differences between results of posttest2 in group E and C statistically.

Due to two reasons based on Shapiro-Wilk or Kolmogorov-Smirnov tests, for getting p values, the exact or non-parametric tests should be applied [9]: (1) samples small number in data file (N= 46 E group, N= 40 C group), (2) zero values appearance in some of data columns that show of distribution. Wilcoxon paired exact test was applied that is used for processing the paired based comparisons and belongs to non-parametric tests. To the parametric paired T-test in case of the normality of data distribution, it works as an alternative [10].

A pair of a zero hypothesis H0 and an alternative hypothesis H1 for testing hypotheses were used. Refer to p value and to the significance level α, the zero hypothesis cannot be rejected (for p >α), or in favor of the alternative hypothesis, the zero hypothesis can be rejected (in case of p <α). The statistic results are supported by the effect size that approved the strength of rejecting the zero hypothesis, besides the p value.

Method, tool, sample

Because of data collected via didactic tests, the comparative method was applied. The entrance knowledge was not monitored but considered zero, due to all students had not studied ESP for accounting in the past. The instruction process was taken 16 weeks (one semester). The research was done at the non-public university in Iran, within the ESP course. Accounting students were split into two groups; group 1 (experimental group, E) include 46 students and the blended model was adapted. This group have physical and online classes and all contents along tasks were delivered via LMS Moodle. Group 2 (control group, C), include 40 students, participated in the physical class and LMS Moodle was not applied in this group. All materials which were available in LMS for group 1 were provided in the printed form to them. In the middle of the semester (i.e., after eight weeks of instruction in the blended, or face-to-face manner) both groups sat for posttest1 (posttest1 in E group, posttest1 in C group) which monitored students’ knowledge. After another eight-week period, i.e., at the end of semester, posttest2 was administered (posttest2 in E group, posttest2 in C group). Both the posttest1 and posttest2 consist of 70 tasks providing maximum score of 100 each. Active replies were required from the respondents; the tasks were not of multiple-choice type. Tests were assessed according to the following criteria: grade A: 100%–90% of maximum test score, B: 89%–79%, C: 78%–60%, F: 59%–0%.

Moreover, students of the experimental group reflected their opinions on the blended learning in the questionnaire. It included five areas of interest to provide the feedback, particularly focusing on students’ preference to face-to-face, or blended approach, other (which ones) subjects taught through blended learning would students appreciate, which type/s of study materials students consider the most useful, and which approach was more motivating for their learning were under the focus.

Following the research design, results of testing six hypotheses are presented first, followed by the results of questionnaire collecting the feedback from students of the experimental group. Using the techniques based on the quantitative research, the determined research problems, questions and objectives are confirmed in this contribution. Testing the hypotheses was generally and frequently used in this type of research complementing with descriptive statistics, e.g., boxplot. As one of the advantages of testing hypotheses, the guarantee of the statistical significance can be considered. Particularly, the paired comparisons of obtained results of the students’ improvement can be observed. For purposes of comparisons of appeared pretests and posttests, the paired tests are the appropriate option [10].

With regards to a normal probability distribution of data, parametrical or non- parametrical statistical paired test were exploited. In each testing the hypothesis or normality distribution of data, the obtained p value is compared to the declared statistical significance level. In the field of education, the applied significance level is 0.05. For purposes of the paired comparisons, the parametrical test Paired T-test should be used (for p greater than significance level) or the non-parametrical Wilcoxon paired test should be applied (for p lower than significance level) [10]. Data collected via didactic tests were processed by software PAST Statistics version 2.17 [11]and IBM SPSS Statistics, version 26.

Hypotheses 1H and 2H

The zero and alternative hypotheses were tested to compare the pretest and posttest1 scores and the pretest and posttest2 scores in the experimental group. Results of statistic testing are displayed in Table 1.

According to p = 3.51 ×10–9 <α = 0.05, the zero hypothesis 1H0 is rejected in favor of the alternative hypothesis 1H1. Therefore, there are the statistical significant paired differences between results of pretest and posttest1 in group E at the significance level a = 0.05. Due to effect size r = 1, the rejecting zero hypothesis was significantly proved.

According to p = 3.49× 10^-9 < α= 0.05, the zero hypothesis 2H₀ is rejected in favor of the alternative hypothesis 2H₁. Therefore, there are the statistical significant    paired   differences   between   results   of  the  pretest and posttest2 in group E at the significance level α= 0.05. Due to effect size r = 1, the rejecting the zero hypothesis was significantly proved.

Table 1: Results of testing hypothesis 1H – Paired Comparison between Pretest and Posttest1 in Group E and 2H – Paired Comparison between Pretest and Posttest2 in Group E

Table 2: Results of testing hypothesis 3H – Paired Comparison between Pretest and Posttest1 in Group C and 4H – Paired Comparison between Pretest and Posttest2 in Group C

Table 3: Results of testing hypothesis 5H – Paired Comparison between Posttests1 in Groups E and C and 6H – Paired Comparison between Posttests2 in Groups E and C.

Hypotheses 3H and 4H

The zero and alternative hypotheses were tested to compare the pretest and posttest1 scores and the pretest and posttest2 scores in the control group. Results of statistic testing are displayed in Table 2.

According to p = 3.49× 10^-9 < α= 0.05, the zero hypothesis 3H₀ is rejected in favor of the alternative hypothesis 3H₁. Therefore, there are statistically significant paired differences between results of the pretest and posttest1 in group C at the significance level α= 0.05. Due to effect size r = 1, the rejecting the zero hypothesis was significantly proved.

According to p = 3.49× 10^-9 < α= 0.05, the zero hypothesis 4H₀ is rejected in favor of the alternative hypothesis 4H₁. Therefore, there are the statistically significant paired differences between results of the pretest and posttest2 in group C at the significance level α= 0.05. Due to effect size r = 1, the rejecting the zero hypothesis was significantly proved.

Hypotheses 5H and 6H

The zero and alternative hypotheses were tested to compare the pretest and posttest1 scores and the pretest and posttest2 scores in the experimental and control group. Results of statistic testing are displayed in Table 3. According to p = 7.41 × 10^-9 < α= 0.05, the zero hypothesis 5H₀ is rejected in favor of the alternative hypothesis 5H₁. Therefore, there are the statistically significant paired differences between results of the posttests1 in groups E and C at the significance level α= 0.05. Due to effect size r = 0.95, the rejecting the zero hypothesis was significantly proved.

According to p = 3.18 × 10^-9 < α= 0.05, the zero hypothesis 6H₀ is rejected in favor of the alternative hypothesis 6H₁. Therefore, there are the statistically significant paired differences between results of the posttests2 in groups E and C on the significance level α= 0.05. Due to effect size r = 1, the rejecting zero hypothesis was significantly proved.

Students’ Feedback on Blended Learning

Students expressed their opinions on blended learning through the questionnaire. The results show that  96% of 46 students preferred the blended approach; 79% would appreciate to have blended learning in other subjects as well, particularly accounting history(9%), economy(4%), ethic(2%) and others; some types of study materials were considered very useful for learning: video recordings (21%), grammar exercises (18%), vocabulary exercises (14%), extra study materials providing texts with extra or more detailed information about the related topic, links to such materials, sometimes also extra exercises, labelling pictures, reading comprehension etc. (7%), topic revision in English language (6%), links to dictionaries (2%);– blended approach (62%) was more motivating to self-study for the students compared to the face-to-face learning (21%), both modes were also appreciated (17%).

According to p = 7.41 × 10–9 < α= 0.05, the zero hypothesis 5H₀ is rejected in favor of the alternative hypothesis 5H₁. Therefore, there are the statistically significant paired differences between results of the posttests1 in groups E and C at the significance level α= 0.05. Due to effect size r = 0.95, the rejecting zero hypothesis was significantly proved.

According to p = 3.18 × 10–9 < a = 0.05, the zero hypothesis 6H₀ is rejected in favor of the alternative hypothesis 6H₁. Therefore, there are the statistically significant paired differences between results of the posttests2 in groups E and C on the significance level α= 0.05. Due to effect size r = 1, the rejecting zero hypothesis was significantly proved.

Students’ Feedback on Blended Learning

Students expressed their opinions on blended learning through the questionnaire. The results show that. – 96% of 46 students preferred the blended approach;– 79% would appreciate to have blended learning in other subjects as well, particularly in accounting principle and auding (21%), banking(10%), accounting history(9%), economy(4%), ethic(2%) and others; some types of study materials were considered very useful for learning: video recordings (21%), grammar exercises (18%), vocabulary exercises (14%), extra study materials providing texts with extra or more detailed information about the related topic, links to such materials, sometimes also extra exercises, labelling pictures, reading comprehension etc. (7%), topic revision in English language (6%), links to dictionaries (2%);– blended approach (62%) was more motivating to self-study for the students compared to the face-to-face learning (21%), both modes were also appreciated (17%).

To sum up, students reported significantly higher preferences in the blended model of the course in comparison to the physical classes. After studying in the blended model, they expressed higher motivation and appreciated traditional strengths, i.e., the possibility of individual pace and independence of study. They also demonstrated better learning outcomes in the test scores, despite no significant differences between the groups were discovered. However, the students with lower level of English performed slightly better in the group, which was presented with the blended model.

Both course delivery approaches examined in this pilot study proved themselves efficient in today’s higher education. The authors believe that the traditional course delivery will not continue to offer benefits that can be fully obtained in online formats. Nevertheless, blended instruction does not appear to impair students’ performance and it enhance their appreciation of the concepts in some cases. Students in the blended learning group indicated they would definitely take another course using this method. Moreover, this delivery format offers increased flexibility to both students and teachers, who can operate from home not only in cases like the corona virus crisis, which we are experiencing right now.

Research results were analyzed using the quantitative research including the testing the hypotheses and descriptive boxplot (Fig. 1). The data show that starting from no knowledge at the beginning of the research period, they significantly improved in posttest1 which was administered in the middle of the semester. However, hardly any increase can be detected at the end of the period (+2.94). Nevertheless, the increase in pair scores in the experimental group compared to the control one in posttest1 is statistically significant (+7.06, Table 3), as well as in posttest2 (+7.541, Table 3).

The fact that students reach higher cognitive increase when the process of instruction is supported by information and communication technologies, i.e., in blended learning, is also proved by these research results.

Moreover, as clearly seen from the feedback questionnaire, test scores and numerous studied conducted in the past, higher medical students also appreciated traditional positive characteristics of blended learning, i.e., the face-to-face contact with teachers and at the same time the possibility to study any time, any place, any pace, as displayed in results of questionnaire. These features refer to learning general English and ESP. However, as ESP for higher medical staff does not appear frequently in the study programs, the contribution of this study is high. As the principles of ESP teaching/learning are identical, the discovered findings can be applied in engineering study programs.

1. Simonova, I. "Blended Approach to Learning and Practising English Grammar with Technical and Foreign Language University Students: Comparative Study." Journal of Computing in Higher Education, vol. 31, 2019, pp. 249–272. https://doi.org/10.1007/s12528-019-09219-w.

2. Kostolanyova, K., and I. Simonova. "Adaptive E-Learning Model for Learning English as a Second/Foreign Language." International Journal of Innovation and Learning, vol. 27, no. 3, 2020, p. 274. https://doi.org/10.1504/IJIL.2020.106811.

3. Shorey, S., et al. "Blended Learning Pedagogy Designed for Communication Module among Undergraduate Nursing Students: A Quasi-Experimental Study." Nurse Education Today, vol. 61, 2018, pp. 120–126.

4. Halasa, S., et al. "Comparing Student Achievement in Traditional Learning with a Combination of Blended and Flipped Learning." 2020. https://doi.org/10.1002/nop2.492.

5. Malach, A. Programované Vyučování [Programmed Learning]. 1st ed., Akademie, Praha, 1977.