Investigating the efficacy of blended learning in ESL classrooms

Abstract The purpose of this research is to check the efficacy of blended mode of learning in ESL (English as a Second Language) classrooms in Pakistan under the theoretical construction of Transactional Distance Theory. For this purpose, the data was collected for pre-test, post-test and self-administered questionnaire from 499 ESL learners of private universities in Pakistan which were offering the blended learning programs. The post-test results indicated significant improvement in student’s English language proficiency. In the same vein, the Partial Least Square (PLS)-Structural Equation Modelling (SEM) techniques results confirmed that most of the factors namely ease of use, computer literacy, need for face-to-face interaction, pre-built learning mechanism and resistance to change have a significant relationship with blended learning except the two factors namely, technical difficulty during the attempt and student assistance in lab.


Introduction
The demand of neo-millennial learners and the overarching interplay of technology have bound the educators to direct their focus towards the digitization of pedagogies especially from the last two decades (Dede, 2004). The current scenario is further affected by the pandemic; hence, online learning is no more an option but a compulsion (Cabauatan et al., 2021;. Most of the contemporary educational discourse focuses on the implementation, possibilities, challenges, efficacy and outcomes of varying forms of online teaching and learning (Tate & Warschauer, 2022;Ismatullayeva, 2021;Shahzad et. al., 2021;Di Vaio et al., 2020a, Zhai et al., 2020. This implies that the need for innovation is the key to match the pace in the field and ESL (English as a Second Language) classrooms in Pakistan need such innovations where the digitized pedagogies should be implemented for bringing betterment (Afridi, 2021;Ahmad & Rao, 2013;Dar & Khan, 2015) considering the multiple reasons discussed in the upcoming section.
Referring to the importance of English language in Pakistan, English is a medium of instruction. Additionally, it is a status symbol and the language of the privileged class, hence an individual possessing command over English enjoys privilege over any other language user in Pakistan (Ashraf, 2022;Begum, 2022;Shamim, 2008, Rahman, 2002. However, student's mass performance reveals contrary results as majority of the students possess below average level of proficiency (Ahmad & Rao, 2013;Dar & Khan, 2015;Haidar & Fang, 2019). There are a number of issues reported to be the cause of low language proficiency of the learners such as anxiety, outdated syllabus, conventional teaching practices, rotten examination system, lack of resources, poor student-teacher relationship (Panhwar & Bell, 2022;Ahmad & Rao, 2013;Memon et al.,2018, Andrabi, Das, & Khwaja, 2006. In such scenario, global research on modern ESL classrooms identifies blended learning as an effective method of teaching for improving the language skills of the students (Ramalingam, Yunus & Hashim, 2022;Heilporn et al., 2021;Jnr et al.., 2020, Ramalingham). Blended learning or asynchronous learning is one of the forms of online learning.) Hrastinski (2019) defines it as a mix or blend of online and face-to-face activities. The students have an access to online portal where they attempt multiple exercises, watch relevant videos or lectures, read handouts and this is then synchronized with face-to-face classroom discussions, activities and lectures (Wong, Tatnall & Burgess, 2014). Asynchronous learning refers to the use of such online modes where the students can access the uploaded course materials in different forms according to their ease. Such activities usually have long due dates and learners can work on their own pace. The course content can take multiple forms like audios, videos, multiple-choice questions, handouts, articles, power-point presentations, etc. (Phanphech et al., 2022). These are accessible to the students through learning management system (LMS). LMS can be described as online classroom lodging to the content (Garrison & Kanuka, 2004).
However, the smooth implementation of blended learning requires proper infrastructure, trained staff, digital literacy and student motivation (Goos et al., 2020). Along with the discussed factors, blended mode in ESL classrooms needs a review on teaching and learning resources, and content as well. There is a disparate literature referring to the successful implementation of blended mode of learning in a marginalized context. For instance, a study in Vietnam reports major drawbacks in adoption of blended learning in English language classrooms (Le, Allen, Johnson, 2022, highlighting the lack of knowledge and trained staff. On the contrary, a study conducted in a Thai university reports a significant improvement in reading skills of the undergraduate students after the adoption of blended mode by the department (Yudhana, 2021). Similarly, Taiwanese study indicates positive impact upon student's language learning after taking up the blended mode . More interestingly, an empirical analysis discussing the schools in many developing countries reports varying results in different contexts of different sectors of one country highlighting teaching pedagogies, students' computer literacy and access, and availability of resources as the major causes of difference in outcomes (Teo & Divakar, 2021).
Thus, the quoted literature indicates both effective and ineffective outcomes of implementation of blended mode in ESL classrooms in a developing context. Hence, there is a dire need of evaluation of efficacy of blended mode of learning in Pakistani context. Although there has been voluminous work done globally in the field of blended learning (Jnr et al.., 2020), the efficacy of blended mode with reference to its implementation in technical terms and behavioral response in a compromised context still needs further research. In addition, the students' perception of blended learning further needs to be explored in ESL context (Gaol & Hutagalung, 2020). Therefore, one of the aims of the study is to check the efficacy of blended learning in ESL classrooms in the Pakistani context. Another aim is to review the students' perception of blended learning in ESL classrooms. The evaluation is done through the lens of Moore's Transactional Distance Theory (1993). This study aims to answer: What factors including easy access to computers, computer literacy, face-to-face interaction, pre-built learning mechanism, resistance to change, students' assistance in lab and technical difficulty during the attempt, influence the online language learning? Does the application of blended learning mode bring any improvement in the English Language proficiency level of the students at the tertiary level? And it aims to provide the recommendations pertaining to the implementation of asynchronous language learning in a developing context.

Transactional distance theory
The underpinning theory supporting the conceptual framework of the study is Transactional Distance Theory (TDT). This theory was originated by Moore in 1993. According to this theory, the psychological and communication distances are the hurdles to be crossed for achieving effective outcomes from a distance learning programme. Shearer, Gregg, Joo and Graham, (2014) explain there are three main components of Moore's model namely the dialogue, the structure and the autonomy. The interplay of these three components presents a framework for a distance learning programme.
The dialogue refers to the chances of interactions made possible for the exchange of ideas through the meaningful development of the structure of the programme. This implies that the more the interaction, the less will be the psychological and communication distance. On the contrary, there will be more psychological and communication gap in a more structured form of mediated education (Shearer, 2009). The autonomy is the third factor which can be defined as the self-directedness of the learners. This implies that the incorporation of tightly structured activities with no or less interaction raises the autonomy or self-directedness among learners. It can also be explained with reference to the autonomous learners who need no or less interaction (Shearer et.al., 2014 In a nutshell, the theory explains the three significant dimensions of the distance learning; where the structure includes the teaching text, assignments, activities, assessments and the other relevant content, the dialogue refers to the level of interaction and the autonomy defines the learner's control over his/her own learning (Ariza & Hancock, 2003).
The TD theory is used in this study as it provides an effective framework for the online learning and has been adopted in the number of educational researches (Ariza & Hancock, 2003). The theory will serve as a rationale to comprehend the supportive and unsupportive factors in the proposed asynchronous programme. This research categorizes the influential factors identified from the literature review into two domains of the Transactional Distance Theory. The first category, the dialogue, includes the computer literacy, face-to-face interaction, pre-built learning mechanism and preparedness or resistance to change. The second category, i.e. the structure in the given context includes easy access to computers, technological assistance in lab and technical difficulty. The autonomy is reflected through the findings from both the categories of the dialogue and the structure.

Hypothesis development
Blended learning has been identified as an effective innovation but any university in a developing country cannot ignore the obstacles in incorporation of such innovations in academics (Gaol & Hutagalung, 2020). Literature about the context presents a number of influential factors in the implementation of blended learning mode which are discussed in the upcoming section. Along with the theoretical review in the previous section, following review represents the empirical relationship of the identified independent variables on the dependent variable, i.e. English language learning in asynchronous mode.

Easy access to computers and language learning
Computer assisted language learning (CALL) is an old concept proved to be effective in language learning (Chapelle, 2009). However, this study is conducted in a developing country where the lack of resources is identified as one of the obstacles in ESL learning . Hence, the programs have provided the access to computers in labs of the universities where the students may attempt the online work. It is assumed to have a positive relationship with the language learning in the given context. H1: There is a significant relationship between easy access to computers and English language learning.

Computer literacy and language learning
Computer literacy refers to the students' knowledge and practical skills to use the computers. It is stated as one of the common issues in implementing blended Learning (Folorunso et al., 2006, p. 496). The studies conducted by Li & Zhu, (2013) in China also confirm that computer literacy has positive relationship with language proficiency of the students.
H2: There is a significant relationship between computer literacy and English language learning.

Need for face-to-face interaction and language learning
Blended learning is a mixed mode of learning and teaching where the learners attempt online exercises along with face-to-face classes, hence there should be a balanced proportion of the virtual and face-to-face interaction (Heilporn et al., 2021). This implies that the students do need face-toface interaction along with the online activities as explained by TD theory (Moore, 2013) in the previous section. Thus, this hypothesis is relevant to the face-to-face classes instead of opting complete online model. The students tend to have a positive influence on their learning if they feel connected. Wang, Guo, He, &Wu, 2019 andVaughan (2008), confirm that a moderate inclusion of face-to-face interaction is compulsory for getting positive outcomes from blended learning.
H3: There is a significant relationship between the face-to-face interaction and English language learning.

Pre-built learning mechanism and language learning
Pre-Built learning mechanism refers to the students' learning background and existing knowledge. It explains how the students have been learning English throughout or to what kind of teaching practices have they been exposed to earlier. Many studies such as Grigoryan (2020); (Teo & Divakar, 2021)) show a positive relationship between the pre-built learning mechanism and the language learning. It confirms that the students with prepared mindset are more likely to achieve improvement as compared to those who lack the preparedness in terms of knowledge about the practices in blended mode of learning language.
H4: There is a significant relationship between pre-built learning mechanism and English language proficiency.

Resistance to change and language learning
Resistance to change refers to the students' lack of interest or low acceptance level in adopting new technology for learning. It is identified as one of the issues in the developing countries (Folorunso et al., 2006). This is explained as preparedness towards the use of blended learning and it is confirmed to have positive relationship with learning or resistance to change has negative relationship with the learning (Cabauatan et al., 2021). In this study resistance to change is termed as preparedness.
H5: There is a significant relationship between preparedness and English language learning.

Student assistance and language learning
Students' assistance here refers to the technical support provided to the students while their work in labs. The management of both the universities provided an easy access to the computers in lab for the students during the programs discussed in this study. Diversified and seamless learning platform is one of the recommendations in setting up a blended learning environment and it is proved to have a positive relationship with language learning (Yousufi, 2020, P. 95). This implies that the learners should be provided with a supportive system while they attempt the blended learning course. Cabauatan et al., (2021) and Farid et al., (2015) state that there is a significant relationship between the students' learning and the technological assistance during the blended learning.
H6: There is a significant relationship between students' (technical) assistance and English language learning.

Technical difficulties during the attempt and language learning
There have been number of obstacles identified by the researchers in implementation of the blended learning program and technical issues in VLE or lab (Hehir et al., 2021;Hermida, 2020). This refers to the establishment of a setup where the students do not face any technical issue while they take up the blended learning course. However, Bing & Zheng, 2020June) confirm no significant relationship between the technical difficulty and the language learning of the students. It implies that there is no connection between any technical obstacle and the expected improvement in learning of the students.
H7: There is significant relationship between technical difficulty and English language learning.

Methodology
The objective of the study is to investigate the efficacy of blended learning in ESL classrooms of Pakistani private sector universities. The proposed hypotheses have been tested using quantitative research approach. The quantitative approach is deemed best to test statistical data and generalize the results when it comes to teaching pedagogies (Eyisi, 2016). Likewise, Ma (2015) and Apuke (2017) opine that the qualitative research approach is suitable for small sample size and inductive method whereas the study under consideration is established using a theory with a purpose to investigate the application on a wider scale.
Both cross-sectional and longitudinal research design have been employed. Longitudinal research design is used for conducting English proficiency test as it is appropriate when the data is collected at different point in times (Caruana et al., 2015;Kehr & Kowatsch, 2015;Menard, 2007). Moreover, the data was collected from the same respondents in two phases pre-test and post-test as suggested by Shek and Sun (2012). Whereas, the cross-sectional research design was used for self-administered survey questionnaire because data was collected at one time and this research design is appropriate for self-administered questionnaire (Lou et al., 2010;Sritharan et al., 2021).

Data collection methods and procedures
Reviewing the record of the students' performance in the entry test, the admission interview, and the classroom results of the previous students, the universities' management proposed a remedial blended learning program. The students were able to take online and face-to-face sessions focusing on the basic components of English language i.e. reading, listening, grammar, vocabulary and speaking. The content from varying authentic sources was modified for meeting the contextual needs. The intervention was established with an aim to bring improvement in the language proficiency of the students.
The target population of the current study was the ESL learners of the private universities who were following the blended learning programs. The core reason for selecting the private universities was that these universities had a major contribution in the student development through the use of technology. The technique applied here is convenient sampling. It is described as a sampling technique where the researcher collects data from easily accessible part of the population (Rahi, 2017). Thus, a sample of 499 ESL freshmen was used for evaluating the results of pre-& post language proficiency test and feedback survey.

Research instrument
At the initial stage of investigation, the English language proficiency test was conducted on campus. The purpose of the test was to diagnose the language proficiency level of the freshmen enrolled in the university. The test was pitched at B2 according to Common European Framework of Reference (CEFR). This computer-based test comprised of four sections namely "reading, grammar, vocabulary and listening". The test was devised by the distinguished faculty members and was blind reviewed by three PhD faculty members for ensuring the validity and reliability of the content of the test. The pre-test English language proficiency test conducted in the beginning was repeated in the end of the program for evaluating the efficacy of the intervention. This test was similar to the diagnostic test for getting the clear picture of progress in terms of English language of the students. On the next stage of inquiry, student's feedback about the program in form of the questionnaire was collected. The variables in the questionnaire focused on the issues in implementation of a blended learning program as highlighted in the theoretical review.
Evaluation or feedback in the form of survey from the university students is a common and effective technique used worldwide; where the students fill a self-administered questionnaire for giving their responses about the teachers, course content or other components of the program (Douglas et al., 2009, p. 306). Thus, a feedback was taken from the students in form of a survey. The questionnaire for the survey was adapted from adapted from Qureshi, Ilyas, Yasmin, & Whitty, 2012. The variables in the questionnaire including easy access to computers, computer literacy, face-to-face interaction, pre-built learning mechanism, preparedness were measured by three items, whereas the technical difficulty during the attempt, and students' assistance in lab were measured by two items. Table 1 shows the demographics of the students who responded to the survey and took part in the pre & post language proficiency test: The table shows that almost 67% of the students from the sample are males, Likewise, 85% i.e. the majority is from matriculation background. However, 98% of the students range between 18 to 25 years old.

Data analysis
Pre & post-test was analysed using descriptive statistics where t-test was run on PLS to compare the results. This provided the clear picture of difference between the language proficiency of the students before and after the intervention. Self-administered questionnaire from the students was analysed to test the regression results between the study variables, we used Smart PLS (SEM). The results from the collected data have been analysed using Smart-PLS version 3. The results discuss the factor analysis and the relationship between the variables. PLS-SEM method is used for testing the model. The initial step in the procedure was to achieve the standard reliability and validity. Next, Smart-PLS is used for the assessment of the measurement model as recommended in Ali et al., (2018). Measures comprise items' and individual item's reliability in PLS assessed by loadings on their latent as prescribed in (Hulland, 1999). The model consists of first-order and second-order construct (higher order construct). The lower loadings indicate the model explanation to be small whereas the Higher loadings interpret that the shared variance between construct and measurement is also higher as compared to error variance. Thus, assessed parameters relating the constructs decease in case the model is explained to be small (Hulland, 1999). Furthermore, the indicators are closely knitted and interchangeable in case of the reflective measurement model. Consequently, the reliability and validity of such indicators should be assessed and reported in detail. Hence, composite reliability (CR) is utilized as a criterion for the measurement model assessment (MMA). The CR ensures that each item maintains its individual standing and is loaded only in the factor it represents. In addition, composite reliability of the construct is explained through Cronbach's alpha's value (Hair et al., 2016). However, validity is assessed through convergent and discriminant validity criteria. Thus, CFA is been employed for Internal consistency (CR), convergent validity (average variance extracted (AVE) and discriminant validity (cross loadings and Fornell-Larcker criterion).

Paired sample t-test
The results from the paired t-test show a significant difference between the two means i.e. the results of the Pre-test and the Post-Test (see, Table 2). The probability value is less than 0.05 which indicates the rejection of the Null Hypothesis (there is no significant difference between the value of the two means), thus the students' language proficiency has significantly improved due to the intervention of the focused blended learning course.
The findings from the paired t-test report a positive impact of teaching innovation upon the students' learning. This supports the need for change in the educational set up and specifically in language teaching methodologies. The reported results align with the TD theory and number of researches in the educational field which support the innovation in learning and teaching methods to suit the millennial needs (Yousufi, 2020;Keppell, 2014;Berg, Joffe, & Porto, 2016). This highlights that a major improvement in language proficiency can be achieved even through a minor input and effort. The focused sample had only been provided with scaffolding for eight weeks yet the results are significant. Hence, any intervention for a longer duration can bring long lasting impact even in a developing country like Pakistan. This finding is also consistent with the international

Reliability and validity results
To sum up, factor loadings represent the evidence for the individual items and their significance in the model, Cronbach alpha confirms the overall variable, CR is for the construct, Average Variance Extracted (AVE) is for convergent validity, and cross loadings are for discriminant validity. Additionally, discriminant validity is explored through Fornell-Larcker and Heterotrait-Monotrait Ratio of the correlation (HTMT). It is important to assess the reliability before proceeding thus, the figure 1 below displays measurement model assessment.

Individual item reliability
Individual item reliability as widely addressed in the literature (Hair et al., 2016) emphasizes that each item under measurement model provides a separate factor loading. Findings in the Table 3 provides the information about all the items with the factor loadings of above the threshold point of 0.5. 0.3 is the minimum requirement for internal consistency of the model whereas loading with 0.40 is acceptable, however, equal to or more than 0.50 is considered as significant Hair et al. (2016), or equal is considered as significant while going for the SEM analyses. In addition, the items with low factor loading should ideally be deleted (Hair et al., 2016).

Internal consistency & Reliability
The internal consistency and reliability of the construct, composite reliability or CR as explained in Bijttebier et al. (2000) is the extent to which the same concept is measured by all the items added in the scale of the study. This is assessed through, Cronbach alpha and CR coefficients (Hair et al., 2016;Peterson & Kim, 2013 In addition, the value of Cronbach alpha (α) if is greater than 0.90 is excellent, greater than 0.8 is good whereas greater than 0.7 is acceptable Mallery and George (2003) for justification of internal consistency of the construct. Moreover, the value of composite reliability should be more than 0.7 (Hair, Black, Babin, & Anderson, 2010). The findings reported in table 0.0

Figure 1. Measurement Model
show that CR values for the variables satisfy the criterion of 0.7. In addition, the values for Cronbach Alpha is also ranged from 0.705 to 0.918. Hence, all the variables depict higher levels of reliability.

Convergent validity
Convergent validity of the construct confirms that all the parameters used to measure the variable are interconnected. Average variance extracted is employed to confirm the convergent validity of the construct. The value of the convergent validity (AVE) must be greater than 0.5, which means that more than half of the variance is explained by the latent construct in comparison to items (Chew & Chan, 2008). As per the findings reported in Table 2 all the AVE values are greater than 0.5 and ranged from 0.768 to 0.755, hence it is concluded that convergent validity is established.

Discriminant Validity
It assesses the independence of the construct from other constructs used in the model. It is assessed through application of Fornell and Larcker (1981) criterion and Heterotrait-Monotrait Ratios. Fornell & Larcker Criterion: According to criterion the squared root of average variance extract should be greater than each of the correlation of the construct, i.e., the diagonal values should be higher in their rows and columns (Hair et al., 2016). Table 4 is in line with the mentioned criterion thus proving the construct used to be discriminant.

Heterotrait-monotrait ratio (HTMT)
Heterotrait-Monotrait Ratio (HTMT) is relatively new method to assess the discriminant validity proposed by Henseler et al. (2015). This approach acknowledges the efficacy of Fornell & Larcker, however it states that Fornell & Larcker and Cross Loadings do not specify the lack of discriminant validity (Henseler et al., 2015). In addition, it is based on "multitrait-multimethod matrix" for the assessment of discriminant validity, moreover it is recommended in number of studies (Alshehhi et al., 2018;Haider et al., 2018;Henseler et al., 2015). Two methods have been proposed for using HTMT; first one is criterion whereas the second one is statistical test (Henseler et al., 2015). As per the criterion method the HTMT should be less than 0.85 (Clark & Watson, 1995) and it should not exceed 0.90 (Gold et al., 2001). Table 5 show the values for HTMT ratios for all the variables and all the ratios for the variables are less than 0.85. Hence, these findings strongly indicate the discriminant validity.

Path analysis results
Our results show that out of seven hypotheses, five have been supported while two of them have not been supported (see, Table 6). The easy access to computers has a positive significant relationship with English language learning (β = .266, p < .001), confirmed H1. Second hypothesis  (H2), i.e. computer literacy also shows a significant relationship with language learning (β = .219, p < .001), also confirmed H2. Third hypothesis (H3) i.e. preference for face-to-face interaction a significant impact on English language learning (β = .127, p < .001), supporting H3. In addition, pre-built learning mechanism also significant impact on English language learning (β = .266, p < .001), hence, H4 supported. Moreover, resistance to change (preparedness) also presents a significant relationship with English language learning of the students (β = .266, p < .05), which supports H5. However, students' assistance (in technical domain or in labs) does not have any significant impact on the English language learning of the students (β = −.005, p > .05), rejecting H6. Similarly, technical difficulty also does not show any significant impact upon language learning of the students (β = −.031, p > .05), therefore, H7 is also rejected.

Discussion
The purpose of the study was to review the improvemnet of performance of freshmen students at a private university in an English language proficiency test with refrence to the application of the blended mode of learning. In addition, it envisioned to analyze the efficacy of a blended learning program through identification of obstacles in the context. The main idea was the to evaluate the possibility of improvement through incorporation of remedial plug-in. The findings show that easy access to computers has a positive significant relationship with English language learning. It explains that marginalized students if provided with easy access to online learning material are expected to achieve improvement in language. Again, it can be viewed in relation with the purpose of implementing blended learning program in the context. Furthermore, this finding is aligned with the results in  and Li and Zhu (2013)'s study. The findings further show that computer literacy also has a significant relationship with language learning. This justifies that students with computer literacy have more chances to acquire greater language competency as compared to those who lack in computer skills. This implies that the implementers are well aware of the need of time and they have taken due steps for bringing a change. Moreover, the result is similar to (Yudhana, 2021;Li & Zhu, 2013;Folorunso et al., 2006;Keppell,) who predicts seamless learning as the future of learning and teaching environment. Further, preference for face-to-face interaction a significant impact on English language learning. This implies that implementation of blended learning program is the contextual need as affirmed in the study by Shahzad et al. 2016)) who claims that application of blended learning program yields progressive results. This finding is also similar to Wang et. al. (2019) and Garrison and Vaughan (2008). This means that a moderate level of interaction according to the need of the context is helpful in bringing improvement in the language proficiency of the students. In addition, the findings show that pre-built learning mechanism also significant impact on English language learning. This means PBLM has a positive relationship with language learning. It is in alignment with the highlighted obstacles in the literature review. Therefore, it simply supports the idea that learning becomes easier if the students' pre-built learning style matches the prevailing classroom teaching methodologies as stated in Qureshi, Ilyas, Yasmin, Whitty, 2012, p.315. However, it does not mean that those teaching styles are always more efficient, therefore the students may take time in adapting innovation but it can produce better results once they become used to of new methods. The findings about pre-built learning mechanism are consistent with Grigoryan (2020)(), as well. Moreover, the findings exhibit that resistance to change(preparedness) also presents a significant relationship with English language learning of the students. This proves that adaptability plays a vital role in learning of the students. The acceptance of innovative teaching and learning methods in classroom ensures the improvement in language proficiency of the students as confirmed in the research by (Cabauatan et al., 2021;. This affirms the conclusion that there is a need for change in teaching and learning methods suiting to the millennial needs of the time. In contrast to the previous literature, the findings show that students' assistance (in technical domain or in labs) does not have any significant impact on the English language learning of the students. It is understandable in the given context since majority of the millennials possess gadgets and it is not difficult for them to understand simple instructions. However, the findings are opposite to (Cabauatan et al., 2021;Farid et al., 2015) who report that the students' technical assistance plays a role in their learning. Similarly, technical difficulty also does not show any significant impact upon language learning of the students. This can be understood through report on provision of ample resources by the university which is an evidence of support for any technical issue arising in learning of students. Thus, students may not have possibly encountered any noticeable hurdle while taking classes which is contrary to Khan, Nawaz and Khan (2011) findings who indicate the presence of technical difficulty as one of the major obstacles. However, the study by Bing & Zheng, 2020, also states that there is no significant impact of technical difficulty upon learning. A study by Teo & Devakar (2021) also reports varying results in different context. Moreover, the results from the t-test also validate that significant improvement is observed among the students' proficiency after taking the classes in the suggested remedial program. Thus, over all findings are aligned with the general researches in the field that there exists an inevitable gap between the prescribed standards and the real practices in the field of English language teaching in Pakistan.

Recommendations
The study offers number of suggestive measures which should be undertaken in English language teaching and learning pedagogy. Firstly, the study emphasizes on implementation of blended learning as one of the practical possibilities. In addition, it rejects the assumed hurdles highlighting the improvement it can bring. The whole practice validates the value of A-synchronous learning and it highlights the possibility of implementation in the Pakistani context. It suggests that integration of technology in ESL classrooms is inevitably a beneficial step despite the prevailing limitations of resources. The findings further imply that neo-millennials being tech-savvy are comparatively easier to be trained through the use of computers instead of conventional rotelearning methods.
The other significant recommendation on behalf of the findings related to the resistance to change, is the application of counselling. The students must be counselled and motivated for anxiety is among the major findings as one of the reasons for low level of language proficiency and demotivated attitude. Furthermore, it is not only at the tertiary level that the teaching pedagogy is to be revised rather the students at the tertiary level would not even require any remedial course if they have been provided with standard teaching at their primary levels as confirmed through the results on pre-built learning mechanism during the survey.
Another recommendation is the extension in the program since the t-tests have confirmed improvement but they do not indicate the presence of C2 (highest level of proficiency) which means that the students do need further support in form of such a program. Thus, it can be inferred in the light of the deduced findings that the educationists can bring a huge improvement even through little efforts, since if the incorporation of mere eight classes can bring a change, then a well-structured establishment of blended learning program can aid a promising influence (Dede, 2004;Keppell, 2014;van den Berg, Joffe, & Porto, 2016).

Future research recommendations
The future researches in the field of blended learning may include the analysis of the content incorporated since it plays a major role in prospecting the extent to which any program has the potential to bring improvement among the students. Furthermore, an action research on classroom practices in face-to-face sessions can also be conducted for reviewing the teaching methodologies adopted by the teachers. This may help in reflection upon the core weaknesses and strengths in classroom practices as to how the incorporation of blended learning may support teachers in the inculcation of self-directed learning among the students specially at the tertiary level. Further research in the area would also break the resistance to change among the novice educators in the field of blended learning. Finally, for the current study, the sample consisted of students from private education institutions of Pakistan only, future studies could test this model by taking sample from public or public and private both, and compare the results.