Modeling dichotomous technology use among university EFL teachers in China: The roles of TPACK, affective and evaluative attitudes towards technology

Abstract Despite a consensus that technologies facilitate English as a Foreign Language (EFL) teaching, Chinese EFL teachers are not using technologies at the optimal level of expectation. To address the problem of ineffective technology use, this study purports to delineate the interactions among three teacher internal variables (i.e., Technological Pedagogical Content Knowledge (TPACK), affective and evaluative attitudes towards technology) and two technology usage variables (i.e., technology use for face-to-face and online instruction). Data were collected from 261 EFL teachers at 17 universities in China via a self-reported questionnaire and analyzed using structural equation modeling. Results revealed that teachers’ TPACK, which related to their evaluative attitudes, positively influenced their actual technology use for both face-to-face and entire online instruction. Their evaluative attitudes also positively affected technology use for face-to-face instruction. In contrast, affective attitudes influenced neither type of technology use. The main theoretical values of this study were conceptualizing technology usage as a dichotomous variable to better fit the blended learning context and introducing a new dyadic perspective on attitudes towards technology. It also provided practical implications for computer-assist language learning policymakers and EFL faculty professional development in China.


PUBLIC INTEREST STATEMENT
Drive by the national curriculum policy and the outbreak of COVID-19, the online-and-offline blended model for studying English as a foreign language has been highly promoted at Chinese universities. In order to successfully teach with blended model, the first issue university English teachers must address is to effectively use information and communication technology tools for pedagogical purposes. Studies have shown that teachers' professional knowledge and attitude toward technology are the two biggest internal factors affecting their use of technology. Therefore, we investigated 261 Chinese university English teachers' technology using frequencies, professional knowledge level, attitudes toward technology, and further analyzed the relationships among the three matters. The result revealed more fine-grained predictors of university English teachers' technology using behaviors, particularly in the context of blended learning. The research finding contributes to a better understanding of teachers' internal factors affecting technology use as well as provides practical ways to enhance technology using efficiency.

Introduction
In China, the exploration of information and communication technology (ICT) in education began in the 1990s (Li, 2014;Li & Ni, 2012). Currently, the use of ICT in teaching and learning as a policydriven strategy for its educational transformation and modernization has become increasingly prevalent (Ma et al., 2019;Teo et al., 2018;N. Wang et al., 2021). The importance of integrating ICT into English as a Foreign Language (EFL) teaching was even more emphasized by Chinese policymakers, not only because today's global communication in English mainly took place in ICTmediated environments; more importantly, it greatly fitted non-English speaking teachers and learners who were unable to learn English in a natural English-speaking environment through facilitating the creation of authentic language-rich environments , promoting interactive language teaching and learning activities (Golonka et al., 2014) and broadening the students' international perspectives (S. C. Yang & Chen, 2007).
To realize the many potential benefits of ICT for EFL study, the relevant ICT tools must be effectively adapted and used for pedagogical purposes by teachers (Liu et al., 2018;Long et al., 2017;Sabzian & Gilakjani, 2013). Aiming at improving the university EFL teachers' effectiveness in technology use, the Chinese Ministry of Education (MOE) recently issued a College English curriculum document titled "The Guidelines on College English Teaching (2020) "(Ministry of Education of the People's Republic of China, 2020). This national document explicitly put forward that with the increasing integration of modern ICT tools such as multimedia, big data, virtual reality, and artificial intelligence into EFL teaching and learning, the university EFL teachers should adapt themselves to up-to-date ICT environment, upgrade their skills in cutting-edge technology use, fully utilize technological approaches to explore broader teaching resources, create a diverse language learning environment and implement blended instructions that benefit students' autonomous and personalized learning.
Nevertheless, notwithstanding the stipulated standards and requirements on EFL teachers' professional competencies, many Chinese EFL teachers were found not fully prepared for teaching with modern ICT tools (Z. Hu & McGrath, 2011;Liu et al., 2017;Mei et al., 2018). The paradox between the pervasive promotion of technology integration into English teaching and ineffective technology use was common (Gao, 2012;Teo et al., 2018). To name a few, ICT was often employed for relatively superficial teaching tasks rather than deep ones (Fraillon et al., 2014); technology use has remained a teaching or presentation tool for one-way knowledge dissemination (Li, 2014), which often led to a loss of interactive learning opportunities (Xu, 2010). It seems urgent to tap into Chinese university EFL teachers' use of technology in teaching particularly considering they can exert an impact on such a large population of EFL learners. 1 Effective technology use in teaching has long been a sophisticated problem linking to multifaceted factors (Zhao & Frank, 2003). Multiple internal variables, such as teachers' technology knowledge and competency (Fathi & Yousefifard, 2019;Lai et al., 2018), pedagogical beliefs (Ding et al., 2019;Liu et al., 2017;Tondeur et al., 2017), and attitudes toward technology (Sabzian & Gilakjani, 2013;Sointu et al., 2019; can significantly affect how teachers use technology in their teaching practice. Among the various internal factors, teachers' knowledge and attitudes were the two most frequently cited barriers to their technology integration (P. A. Ertmer et al., 2012;Hew & Brush, 2007). This assertion was supported by many empirical evidences that Chinese EFL teachers' technological pedagogical content knowledge (TPACK, hereafter) and their attitudes towards technology were both significant predictors of their intention to use technology (Hsu, 2016;Liu et al., 2017;Mei et al., 2018). Albeit teachers' TPACK and attitudes towards technology play similar critical roles in technology use, the subtle relations among the three constructs are not fully understood (Baturay et al., 2017;Scherer et al., 2018). Lacking knowledge of the interwoven interaction among teacher knowledge, attitudes, and technology using behaviors might threaten the potentials of EFL teachers and make it challenging to overcome the barriers posed by the two internal factors.
As such, research contributing to understanding the interaction between teachers' knowledge and attitudes in the context of educational technology integration were strongly called for by prior studies (e.g., Joo et al., 2018;Scherer et al., 2018;. Additionally, given teaching contexts are constantly changing, particularly where emerging technology is concerned (P. A. Ertmer et al., 2012), along with a long notice that language teachers and learners have problematic interaction with technologies (Liu et al., 2017;Ma et al., 2019;Mei et al., 2018), it is of great importance to revisit the phenomenon of EFL teachers' technology integration. To address the above-mentioned knowledge gaps, the main research question guiding this study reads: To what extent do Chinese EFL teachers' TPACK, attitudes towards technology and technology use relate to one another?

Contemporary ICT tools used by EFL teachers in China
The ongoing round of College English Reform begun in 2002 is the biggest scaled and most influential College English curriculum reform that has ever taken place in China (Chen, 2010). It is noteworthy that using ICT as a tool in college English teaching and learning has been acclaimed as a catalyst for this reform. The recently issued national curriculum document GCET (2020) emphasized that the university EFL teachers should "incorporate high-quality online resources to optimize teaching contents, create video lectures, combine Massive Open Online Courses (MOOCs) with traditional classroom courses, and implement blended learning in order to promote students' proactive, autonomous and personalized learning" (Ministry of Education of the People's Republic of China, 2020). With this stateinitiated curriculum reform, transforming teacher-dominated instruction to student-centered learning, offering online and blended courses with the support of new ICT tools has already become a growing trend of foreign language teaching in Chinese higher education (N. Wang et al., 2021).
As Chinese education is highly centralized, many Chinese universities are positively responding to the top-down advocacy in virtues of building online courses delivered through MOOC platforms, creating video lectures, designing EFL blended instructional models, and so forth. Meanwhile, a variety of emerging technologies, especially Web 2.0 technologies, are embraced by Chinese technology-savvy teachers to facilitate EFL classrooms. For instance, in order to create an immersive, supportive, constructive, and participatory learning environment for EFL learners, there is a rapid growth of blended college English courses that incorporated MOOCs provided on the four leading MOOC platforms in China: XuetangX, iCourse, CNMOOC and Chinese MOOCs (N. Wang et al., 2021). In addition to MOOC platforms, indigenous smartphone applications, such as WeChat and QQ with multiple functions (e.g., hold-to-talk messaging, broadcast messaging, video conferencing, file sharing , and location sharing) have been frequently used as mobile learning platforms (Mei et al., 2018). Social networking, QQ Zone and Sina Microblog, which accounted for 30% of Internet users in China, have also been introduced to higher education classrooms for improving Chinese EFL learners' collaborative learning capacities and English writing achievements (Li, 2016;Mei et al., 2018;Y. Wang et al., 2014).

The dichotomy of technology use for blended EFL learning
Driven by both national curriculum policy and the outbreak of COVID-19, blended instruction has become the mainstream approach to learn English at Chinese universities today. Mediated by a variety of ICT tools, ways to design and implement blended EFL learning are either asynchronous or synchronous (Hastie et al., 2010;Szeto, 2015). The blended EFL learning in China was set in a most ubiquitous blended learning mode in which face-to-face (f2f, hereafter) instruction and online learning were combined through using Web platforms and tools (e.g., Peng & Fu, 2021;Sun & Qiu, 2017;N. Wang et al., 2021). This type of blended learning mainly involved an instructional model incorporating asynchronous videos or synchronous online instruction that allowed teachers and students to participate in both physical and virtual classrooms. To successfully implement a set of blended teaching tasks, the university EFL teachers need to use basic technological infrastructure consisting of a regular computer, network facilities, and real-object projectors in general university teaching venues, as well as utilize video recording devices such as digital cameras, smartphones and tablets, video editing software like iMovie and videoconferencing platforms such as Tencent Meeting and Zoom to teach online.
In f2f with computer-mediated classrooms, EFL teachers often take advantage of relatively wellestablished technological tools to support knowledge transmission and interactive activities. Teachers use computer and other demonstration devices (e.g., laptop) as well as presentation and multimedia software (e.g., PowerPoint, Adobe Premier) for classroom teaching. Additionally, outside the f2f class, teachers would use online tools and computer software such as learning management system, Web searching engines, and chat tools to plan lessons, manage courses, and deal with the students' affairs. Compared with f2f teaching, entire online instruction is a kind of newer instructional mode that calls for typical technological tools to serve remote teaching and learning activities. According to GCET (2020), online instruction consists of two elements: asynchronous online content supplementary to f2f courses and synchronous lectures and communications delivered entirely online (Ministry of Education of the People's Republic of China, 2020). Specifically, asynchronous online contents are often manifested as recorded microlessons. Such microlessons are often created by video recording devices like digital camera and video editing software (e.g., iMovie). When teachers implement synchronous lectures and communications, they mainly rely on synchronous communication tools such as videoconferencing (e.g., Tencent Meeting, Zoom) and MOOC platform (e.g., CQOOC).
It should be noted that the use of some technological tools can be interchangeable across different teaching contexts. For example, the use of learning management system can help teachers better manage teaching affairs in both f2f and virtual classrooms. However, for different teaching purpose in different contexts, teachers would consider the necessity of specific technologies and have distinct frequencies of using them. For instance, computers with projection systems are the primary equipment in f2f classrooms but not a necessity for entire online teaching, while videoconferencing platforms and video editing software are often employed for creating and delivering synchronous video lessons other than f2f lectures. Therefore, based on the dichotomy of instructional context as either f2f or entire online, the frequently and necessarily used technologies by Chinese EFL teachers are classified as f2f instruction technology (FT), and entire online instruction technology (OT) in the present study (see Table 1).  Mishra and Koehler (2006) attempted to identify the types of knowledge required for this purpose. Building on Shulman's (1986) theory of pedagogical content knowledge (PCK), Mishra and Koehler (M. J. Koehler et al., 2008;M. Koehler & Mishra, 2009;Mishra & Koehler, 2006) proposed the TPACK framework, which included the new arena of educational technology. This framework offered the first theoretical conceptualization of integrating technology into instruction (Rosenberg & Koehler, 2015). It has since been widely used to guide research on technology education and teacher professional development (Chai et al., 2013;Voogt et al., 2013).

The role of TPACK in technology use
As shown in Figure 1, according to M. Koehler and Mishra (2009), the TPACK framework comprises seven distinct components: CK (knowledge of teaching content); PK (knowledge of teaching methods); TK (knowledge of technology usage); TCK (knowledge of how technology enables new Table 1

Technological tools used for Example Technology Application/Tools
Face-to-face instruction mediated by computer Computers with projection systems.
Video recording equipment (e.g., digital camera, camcorder, tablets. expressions of specific content); TPK (knowledge of how the techniques of teaching and learning can change when particular technologies are used); PCK (knowledge of the transformation of subject matter for teaching); and TPACK (integrating knowledge of content, pedagogy, and technology to achieve truly meaningful and deeply skilled teaching with technology).
By separating these interwoven forms of knowledge and skills and considering the interactions between them, the TPACK framework facilitates further analysis of the complex phenomenon of technology integration in teaching. Through the TPACK framework, teachers' thought processes and actions associated with technology integration have been observed and described in dynamic classroom settings (M. Koehler & Mishra, 2009).
In the last decade, many researchers have sought to determine whether specific TPACK construct affects teachers' technology usage intentions and behavior in the real world. For example, Pamuk (2012) found PCK to be the most significant factor predicting the overall technology integration efforts of 78 Turkish preservice teachers. Similarly, Mei et al. (2018) revealed that TPACK directly influenced 295 Chinese EFL preservice teachers' intention to use Web 2.0 technologies. Hsu (2016) confirmed TPACK was critical to 158 in-service Taiwanese English teachers' adoption of mobile-assisted language learning. Although these studies found a positive predictive relationship between TPACK and technology integration, what they have claimed to account for explaining technology integration was teachers' intention to use technology, while they failed to distinguish one's intention to perform a behavior from their actual behavior (Liu et al., 2017). The behavioral intention may lack efficacy in explaining long-term future use of technology (Bergeron et al., 1995). Besides, teachers' intention to use technology was not aligned with their actual technology integration practice (H.-d. Yang & Yoo, 2004;. Therefore, exploring the relationship between teachers' TPACK knowledge and their actual technology using behavior is necessary for enriching the existing theoretical models such as the Technology Acceptance Model, Theory of Reasoned Action, and Theory of Planned Behavior (Liu et al., 2017). Consequently, more empirical research is needed to address the relationship between TPACK and teachers' actual use of technology in teaching (Avidov-Ungar & Eshet-Alkalai, 2011;Joo et al., 2018). Based on both theories and prior study findings, this study hypothesizes that Chinese EFL teachers' TPACK positively influences their actual technology uses. Combining the conceptualization of dichotomous technology use stated in earlier section, the first two hypotheses read: H 1: The Chinese university EFL teachers' TPACK will positively influence their use of FT; H 2: The Chinese university EFL teachers' TPACK will positively influence their use of OT;

The role of attitudes towards technology in technology use
Teacher knowledge and expertise play essential roles in effective technology integration. However, they are neither useful nor effective unless teachers are convinced of the value of technology or the technology integration is consistent with their attitudes towards technology (Ding et al., 2019;Ertmer & Ottenbreit-Leftwich, 2010;Yildirim, 2000). Indeed, teacher knowledge, attitudes, and beliefs are intertwined (Baturay et al., 2017;Scherer et al., 2018) and consequently conceived as inherent parts of teacher agency (Ertmer & Ottenbreit-Leftwich, 2010;Tondeur et al., 2017). When teachers are required to integrate technology to support teaching, they must also update their knowledge of instructional practices, strategies, methods, and approaches and change their beliefs, attitudes, and pedagogical ideologies (Ertmer & Ottenbreit-Leftwich, 2010). There has been a longstanding tradition of studies on examining teachers' attitudes towards technology since the emergence of educational technology (Scherer et al., 2018). People's attitudes towards a new technology are a critical factor affecting its diffusion (Albirini, 2006;Rogers, 2010). Researchers (e.g., Bullock, 2004;Kersaint, 2003;Woodrow, 1992) have widely accepted that it is necessary to take teachers' attitudes into consideration and foster teachers' positive attitudes towards technology for effective technology integration. Despite the significance of attitudes in technology using behaviors, the concept of attitudes has not always been a focus in various competing models in research areas of technology acceptance (H.-d. Yang & Yoo, 2004;Zhang et al., 2008). The existing technology acceptance models could only explain limited variance in attitudes, which however, should not be simply interpreted as "perceived usefulness" or/and "perceived ease" (Li, 2014).
As suggested by social psychology literature, attitudes were a complex variable with both cognitive and affective facets. The two facets of attitudes were assumed to influence technology integration behavior through different psychological mechanisms (H.-d. Yang & Yoo, 2004). Nevertheless, given that Davis et al. (1989) indicated that attitudes had little value in explaining technology adoption, the distinction between affective and cognitive facets of attitudes was often overlooked in technology attitude research (Goodhue, 1988;H.-d. Yang & Yoo, 2004;Swanson, 1982). H.-d. Yang and Yoo (2004) pointed out the weak association between attitudes and technology use might be attributed to the composite measure of the attitude construct. They further contended there was a significant difference between the cognitive and affective facets of attitudes in mediating users' technology use. Thus, they recommended that future studies should pay substantial attention to the distinction among the cognitive and affective dimensions of attitude regarding technology acceptance and re-examine the explanative power of affective attitude, taking it as a dependent variable instead of a mediator.
Different dimensions of attitudes in previous studies were conceptualized as, for example, attitudes towards the general use of technology/attitudes towards the use of technology in teaching (Scherer et al., 2018), behavior-oriented ICT attitudes/object-oriented ICT attitudes (Zhang et al., 2008), perceived ease/usefulness of technology (Davis et al., 1989). However, these dimensions of attitudes were often highly associated with one another and thus hard to be distinguished (Scherer et al., 2018). There then exists a common methodological problem and erroneous interpretations caused by high correlations among several dimensions of attitudes towards technology in previous studies (Scherer et al., 2018). Based on Wiggins (1989, 1991) experimental results, affect and evaluation were two distinctive facets of attitudes. In order to gain more distinctive facets of attitudes towards technology, this present study will incorporate both affect and evaluation as two distinct facets consisting of attitudes towards technology: affective (AAT) and evaluative attitudes towards technology (EAT). Breckler and Wiggins (1989, p. 253) defined affect as "emotional responses and feelings engendered by attitude object", and evaluation as "thoughts, beliefs, and judgements about an attitude object". In the field of educational technology research, Venkatesh (2000) defined the role of emotion in the nomological net of TAM, which mainly referred to the engendered feelings (i.e., enjoyment, anxiety, apprehension, fear, etc.) when one was faced with the possibility of using technologies in teaching. Combining Breckler and Wiggins (1989) definition on affect and that of Venkatesh (2000), AAT should primarily point to emotions and feelings, both favoring and disfavoring, which were directly or/and indirectly aroused or associated with the use or tendency to use technology for educational purposes. On the other hand, according to a group of researchers who treated attitudes towards technology as evaluative judgments and emphasized the users' perceived value and relevancy of technology used in their job (e.g., Gibbone et al., 2010;Park & Ertmer, 2008), EAT was then defined to describe teachers' negative and positive thoughts and judgments on educational technologies' value and relevancy for teaching. Since Wiggins (1989, 1991) provided empirical evidence that both affect and evaluation significantly affected subsequent behaviors, the two components of attitudes towards technology (i.e., AAT, EAT) are assumed to impose significant influences on teachers' technology uses as well. Consequently, the next four hypotheses in present study read, H 3: The Chinese university EFL teachers' AAT will positively influence their use of FT; H 4: The Chinese university EFL teachers' AAT will positively influence their use of OT; H 5: The Chinese university EFL teachers' EAT will positively influence their use of FT; H 6: The Chinese university EFL teachers' EAT will positively influence their use of OT;

TPACK-Attitudes relations
Given that knowledge about TPACK-Attitudes relations may be helpful for expanding theoretical models of technology acceptance as well as scaffold teacher education in TPACK enhancement (Scherer et al., 2018), studies concerning relationships between TPACK and attitudes towards technology have drawn growing attentions over recent years (e.g., Avidov-Ungar & Eshet-Alkalai, ; Baturay et al., 2017;Scherer et al., 2018). Scherer et al. (2018) pointed out there was a scarcity of research that had fully elaborated the relationship between TPACK and attitudes towards technology. Several previous studies focusing on this knowledge gap indicated positive associations between TPACK constructs and attitudes towards technology to a different extent. For example, Baturay et al. (2017) found a low-level correlation between Turkish K12-level teachers' TPACK and attitudes towards computer-assisted education. Incorporating TPACK as an external variable to the TAM, Mei et al. (2018) and Hsu (2016) also confirmed positive relations between Chinese preservice EFL teachers' overall TPACK scores and the two attitudinal factors, perceived usefulness and ease of use. Based on a sample of 688 Finnish preservice teachers, Scherer et al. (2018) provided insights to how specific TPACK dimensions related to three attitudes dimensions (i.e., general attitudes toward ICT, attitudes toward the educational use of ICT, and perceived ease of ICT use), that the preservice teachers' positive attitudes toward ICT were correlated with higher self-beliefs in TPACK, TCK, TPK and TK.
However, considering most of the studies were conducted on the preservice samples (e.g., Joo et al., 2018;Mei et al., 2018;Scherer et al., 2018), investigation on TPACK-attitudes relation among in-service teachers is still lacking. In-service teachers might be differentiated from preservice teachers because they could generate mastery experiences, actual pressure and different views on ICT use in their classrooms (Mei et al., 2018;Scherer et al., 2018;Teo, 2015). Additionally, Scherer et al. (2018) highlighted there is a continuous need for a more fine-grained exploration incorporating broader and more distinct facets of technology attitudes or beliefs to expand the view on the TPACK-attitudes relations. Thus, it is timely to explore the TPACK-attitudes relations, particularly with account for the dyadic view of attitudes towards technology as well as using inservice teacher samples. Prior empirical findings have laid the basis for the present study to hypothesize a positive correctional relationship existing between TPACK, affective and evaluative attitudes towards technology among Chinese EFL teachers. The last two hypotheses will then read, H 7 : The Chinese university EFL teachers' TPACK will be positively correlated with AAT; H 8: The Chinese university EFL teachers' TPACK will be positively correlated with EAT.

Research model
Based on the aforementioned theoretical and empirical literature, this present study formulated eight research hypotheses that needed to be examined. Specifically, a research model containing four pairs of hypothesized relationships was proposed and demonstrated in Figure 2.

Participants and procedure
A total of 261 EFL teachers from 17 universities across the eastern, central, and western parts of Mainland China via non-probability volunteer sampling participated in this study. Over half (54.3%) of the participants were between 30 to 40 years old, followed by those who were above 40 years old (38.4%) and under 30 years old (12.5%). A majority (66.3%) of the participants had over 10year teaching experience and had received technology training before (69.4%). Although the proportion of female participants was substantially high (80%), this was basically in line with the current gender ratio of English teachers in Chinese universities. The majority (93.0%) of the participants have master's and higher degrees.
Considering the different developing level of the regions where the participants come from may have possible socio-economical influence on the statistical results, we ensured that the participants were all from provincial capital cities, and the levels of information technology facilities in their universities were similar.
An online invitation including an URL to the electronic questionnaire was sent to the participants through an online survey platform (www.wjx.com). All participants were well informed of research goals and their rights to withdraw participation at any stage of this study. Consent was given by participants who indicated that they agreed to a set of statements listed in the front page of the survey questionnaire. The research met the required ethical standards and was approved by the research team's university ethics committee. As shown on the survey platform, each participant took no more than 20 minutes to complete the questionnaire, which took place entirely anonymously. Altogether, 261 valid questionnaires were achieved.

Measures
A multiple-item questionnaire with three subscales was designed to collect quantitative data on five study constructs: TPACK, AAT, EAT, FT, and OT. The 19 scaled items can be found in the Appendix. In addition, demographic information items (e.g., age group, gender, location) were also included in this questionnaire.

Technology Uses Scale (TUS)
TUS was an originally developed scale aiming to measure the participants' frequency in using FT and OT during the recent semester. The initial item pool was generated from the prior interview data, Faculty Technology Survey (Vannatta & O'Bannon, 2002), and shortened version of Faculty Technology Survey (Vannatta & Fordham, 2004). TUS consisted of two constructs, FT and OT with three items, respectively. The content and face validity of this 6-item scale was established by two experts in this field of English education and two university EFL teachers' review. It was answered using a 5-point scale: (1) none, (2) rarely, (3) moderate, (4) often, (5) high.  teachers learning to teach English as a foreign language". It was answered on a 5-point Likert Scale ranging from "Strongly disagree" to "Strongly agree." 3.2.3. Technology Attitudes Scale (TAS; Gibbone et al., 2010) TAS was comprised of two constructs, AAT and EAT with three and six items for each. These nine items were adapted from the survey of The Secondary Physical Educators' Attitudes and Technology Practices Inventory (SPEATPI) (Gibbone et al., 2010) which has been well tested for validity and reliability. Given this present study conducted among EFL teachers, items related to PE subjects in SPEATPI were changed into "English subjects." It was answered on a 5-point Likert Scale ranging from "Strongly disagree" to "Strongly agree."

TPACK-EFL Scale
All scales were presented in both published English versions and translated Chinese versions. To ensure the face and content validity of the score, each item underwent the process of translation and back-translation (Beaton et al., 2000) by the authors and a university EFL teacher together.

Data analysis
IBM SPSS 24 was employed for descriptive statistics calculation. To analyze the collected data and examine the hypothesized relationships among different variables, structural equation modeling (SEM) was conducted by using AMOS 24.0. The first step was to estimate the measurement model, which shows how well the observed variables measure the underlying constructs. The second step, the structural model that describes hypothesized relationships among the exogenous and endogenous variables (as shown in Figure 2), was tested. Relevant results would be interpreted based on model fit criteria and indices to provide evidence for supporting research hypotheses.

Descriptive statistics
The means, standard deviations, skewness, and kurtosis for all measured variables were analyzed to confirm the multivariate normal distribution (as shown in Table 2). Most mean scores were above the midpoint of 3.00 (except AAT1, AAT2, and OT3 which were 2.96, 2.61, and 2.48). Moreover, the standard deviations ranged from .74 to 1.35. According to Kline's (2010) recommendations that the skew and kurtosis indices should be within |3| and |10| respectively, the data in this study met the assumption of a multivariate normal distribution as the values of skewness ranged from −2.34 to .53 and kurtosis values ranged from −1.23 to 5.96.

Testing the measurement model
As the first step of SEM analysis to confirm the construct validity of data in this study, the measurement model was estimated by conducting confirmatory factor analysis (CFA) with maximum likelihood estimation. While the method of maximum likelihood estimation is a robust procedure of SEM, this method can be applied if data meet the assumption of multivariate normality (Schumacker & Lomax, 2010). In this study, the multivariate normality of collected data was examined by calculating the Mardia's normalized multivariate kurtosis value (Mardia, 1970). According to Bollen (1989), if Mardia's coefficient is less than p (p + 2) (p is the number of observed variables), then the data shows multivariate normality. In this study, the Mardia's coefficient for data was 112.80, which was lower than 399 calculated by the formula p (p + 2) (p = 19). Therefore, a multivariate normal distribution of the data was confirmed and the maximum likelihood estimation method can be used in the SEM.
To examine the reliability and validity of observed variables in the model, composite reliability (CR) and average variance extraction (AVE) were used. Both indicators were judged to be adequate when their values equal or exceed .50 (Fornell & Larcker, 1981). Additionally, to assess the validity of questionnaire items, the standardized estimates of items were also estimated, with values higher than .50indicating the items explain their latent variables well (Hair et al., 2010). Moreover, the Cronbach's alpha coefficients of all subscales were tested to explore their internal Note. N = 261. TPACK = Technological pedagogical content knowledge; AAT = Affective attitudes towards technology; EAT = Evaluative attitudes towards technology; FT = face-to-face instruction technology uses; OT = online instruction technology uses.
reliability. As shown in Table 3, the CRs, AVEs, and standardized estimates of "TPACK", "EAT" and "AAT" were at the acceptable ranges, while the AVEs of "FT" and "OT" were not adequate. However, CRs and standardized estimates of these two variables supported the reliability and validity of their items. The Cronbach's α coefficients of all subscales ranged from .66 to .96 (see Table 3). Loewenthal and Lewis (2018) argued that scales with Cronbach's alpha values above .60 can be acceptable when they have fewer than ten items. Therefore, all subscales had acceptable α coefficients, which suggested their internal consistency.
To test the fit of the measurement model, five indices were used: the ratio of minimum fit function chi-square to its degree of freedom (CMIN/DF) was used with a range of no more than 3.0, indicating an acceptable fit between the hypothesized model and the data (Carmines & McIver, 1981), two incremental fit indices as Tucker-Lewis Fit Index (TLI) and Comparative Fit Index (CFI) with values that are equal to or greater than .90 indicating an adequate model fit (Hair et al., 2010), the root mean square error of approximation (RMSEA) with a value ranges from .05 to .08 representing a fair fit (Browne & Cudeck, 1993), and a standardized root mean square residual (SRMR) with a value less than .08 was suggested by L. T. Hu and Bentler (1999) for the good models. Results from CFA of the sample data revealed that the measurement model had acceptable fit, as CMIN = 321.59, CMIN/DF = 2.27, TLI = .93, CFI = .94, RMSEA = .07, and SRMR = .05.

Testing the structural modeling and the hypotheses
The structural model had a good model fit (CMIN = 327.90, CMIN/DF = 2.29, TLI = .93, CFI = .94, RMSEA = .07, and SRMR = .05). As demonstrated in Table 4 and Figure 3, among the eight research hypotheses, four of them were supported by the data in this study (i.e., H 1 , H 2 , H 5 , and H 8 , all p < .05), while 4 of them were not supported (i.e., H 3 , H 4 , H 6 , and H 7 ). The results revealed that FT and OT were both significantly influenced by TPACK (H 1 , H 2 ), and FT was also influenced by EAT (H 5 ) but not by AAT (H 3 ). Besides, TPACK was found to be significantly correlated with EAT (H 8 ) rather than AAT (H 7 ). These results suggested that TPACK, which was correlated with EAT, was an important construct that directly influences teachers' frequencies of using technology to teach across f2f classrooms and online settings. Overall, two endogenous variables, FT and OT had respectively 17.6 % and 20.0% of total variance explained by three exogenous variables, including TPACK, EAT, and AAT.

Discussion
This study investigated five study variables, namely, TPACK, affective and evaluative attitudes towards technology, the technology uses for f2f and entire online instruction, as well as their potential predictive and correlational relationships based on a sample of 261 Chinese EFL teachers from 17 Chinese universities. Overall, our finding supported half of the research hypotheses: teachers' TPACK would positively influence their use of FT and OT; EAT would affect the use of FT; TPACK is correlated with EAT.

Supported structural relationships in the hypothesized model
First, Chinese EFL teachers' TPACK would directly and positively predict their use of ICT tools for f2f and entire online instruction. It implies that EFL teachers who possess a higher level of TPACK are more likely to increase their actual use of technology for EFL blended teaching practice. Although a substantial body of research have affirmed a linkage between teachers' TPACK and technology use, the bulk of the study still focused on teachers' intention or willingness to use technology rather than actual use of technology in teaching. For instance, Mei et al. (2018) and Hsu (2016) revealed that the TPACK of Chinese EFL teachers can positively affect their intentions to use Web 2.0 and mobile technologies. Comparatively, this present finding clarified the connection between teacher knowledge and their actual use behavior to boot. As articulated by M. Koehler and Mishra (2009), teachers' thought processes and knowledge (i.e., TPACK) and teachers' action and the observable effects (i.e., actual technology uses) were two key domains of technology integration. This study contributed to TPACK research body by offering a better understanding of how the two key domains associated with one another. It is in accordance with but also a step forward from previous studies (e.g., Mei et al., 2018; Pamuk, 2012;. For another aspect, since the conceptualization of technology use in this current study differed from that in previous studies, it also shed light on EFL teachers' dichotomous use of technology for blended learning in the higher education. That was, the pivotal role of teachers' skills and expertise in the integration of pedagogy, content, and technology should be continuously attached importance in the new blended learning environment. Either teaching with mature technological infrastructure for f2f instruction or newer emerging tools for entire online instruction, successful technology integration primarily requires EFL teachers to prepare themselves with personal proficiency in the English language, the ability to teach English as well as skills in combining technological resources. Second, EAT was found to have a positive influence on Chinese EFL teachers' use of FT. In line with prior findings that teachers' perception of relevance or importance of technology in curricula has been found to positively predict their technology use (e.g., Gibbone et al., 2010;Kanaya et al., 2005), it was not unexpected that Chinese university EFL teachers' perception of technology  affordance in providing high-quality language teaching and learning also triggered their actual uses of technology. Haines (2015) asserted that language teachers must well recognize the value and usefulness of technology before they could actually use them in teaching. The technological tools for f2f instruction, including presentation software, whiteboard, and digital devices, have been well-established for years and comparatively familiar to foreign language teachers (Golonka et al., 2014). Knowing more about the usability and functions of those classroom-based technologies, the Chinese EFL teachers could posit higher frequency in using them.
Third, Chinese EFL teachers' TPACK was significantly correlated with their EAT. This finding suggested that the more TPACK knowledge the teachers believed they have, the more likely they found the technology helpful in teaching English. In other words, Chinese EFL teachers' tendency to gain knowledge in technology, pedagogy, and content is partially related to their perceived usability and performance of the technology system displayed to them. This might be explained by P. Ertmer et al.'s (2015) two related belief systems: one's belief about technology is linked to the belief about his or her own competencies for using technology. According to Scherer et al. (2018), the belief about the capabilities within the TPACK knowledge domains might be a source of their attitudes towards technology. The success in processing technology-related tasks would lead to increased confidence in TPACK, which in turn positively affected teachers' attitudes towards technology (Tschannen-Moran & Johnson, 2011). This result reminded us to view the belief system about the usefulness of technology and one's competencies for using technology as a whole so as to formulate more precise teacher-education pedagogy in the improvement of TPACK.

Unsupported structural relationships in the hypothesized model
AAT was not a significant predictor of technology uses, regardless of the type of technology. This result directly echoed H.-d. Yang and Yoo's (2004) and Teo et al.'s (2018) study findings. H.d. Yang and Yoo's (2004) study showed the cognitive attitude rather than affective attitudes significantly influenced 211 American undergraduates' technology usage behavior. They attributed why affective attitudes lacked significance to explain technology acceptance as affective attitudes should be regarded as an independent variable rather than a mediator in the research model. Alternatively, the affective attitude might be more related to another exogenous variable, the user satisfaction, not technology acceptance. However, the current study results overturned their hypotheses since affective attitudes did not directly influence teachers' technology use behavior even if it was treated as a dependent variable, not a mediator. This result also supported Teo et al.'s (2018) finding that Chinese university EFL teachers' attitudes towards technology, namely, the degree they like or dislike technology, would not affect their use of technology in teaching English. According to Teo et al. (2018), this might be attributed to the Chinese cultural differences. Contrary to EAT, the lack of support for AAT→FT and AAT→OT meant that as conscientious teachers, Chinese EFL teachers' technology using behavior was more affected by the effectiveness and efficiency they have perceived rather than their personal feelings about using technology. Affected by Confucian-heritage culture, which emphasized group interest and conformity in behavior (Hofstede, 2001), the Chinese EFL teachers' technology uses would be more likely to be influenced by compulsory mandates and significant others, rather than their individual preferences.
In addition, although EAT was found as a predictor of Chinese EFL teachers' classroom technology use, it would not predict their use of OT. Due to the relative unfamiliarity and less frequency in using newer technological tools such as video editing software and conferencing platforms, it was unlikely for Chinese EFL teachers to be accurately informed of the affordance of those newer technologies. Therefore, their attitudes, particularly concerning value judgments, would not capture their actual use of OT in teaching.
Lastly, TPACK was not significantly correlated with AAT. Although one's affective reaction, particularly anxiety and apprehension caused by technology usage has long and widely been accepted as significant components of technology attitudes (e.g., Howard & Smith, 1986;Igbaria & Chakrabarti, 1990), Venkatesh (2000) argued that with the increasing pervasiveness of ICT in the workplace and homes, the emotion of anxiety exhibited by users might not be still relevant. In addition, Mei et al. (2018) articulated the widespread of digital devices such as smartphones, laptops, tablets, as well as the big progress in Web 2.0 technologies in China, made users' interaction with technology became much easier and more intuitive today. Thus, it is reasonable that the need for high-level TPACK knowledge and skills would not much relate to the decreasing difficulty perceived by teachers for using technology.

Limitations and recommendations for future study
There were several limitations to this study. First, self-reported data, though is commonly used in studies of technological integration, may still generate response bias and thus inflate the genuine associations among variables (Spector, 1994). Alternative data sources to capture broader phenomena on similar topics in future studies are recommended. Second, this study considered affective and evaluative attitudes as two parallel independent variables that separately influence technology uses, while it overlooked the possible mutual influences generated within the two dimensions of attitudes. Given the contagious and malleable nature of attitudes, the technology users' affective and evaluative attitudes were postulated to affect one another and even transform overtime under certain conditions. Thus, future studies were recommended to employ longitudinal study design and build reciprocal models in which the evaluative and affective attitudes are expected to influence technology use at different time points. Such models would better reflect the dynamic and malleable character of attitudes in technology acceptance.
Third, due to the differences in culture and discipline, this result may not represent teachers with different cultural or discipline backgrounds, which would impact the ability to generalize this result. Besides, unlike EAT, the survey items of AAT may stand in need of improvement as the three items in the instrument only captured minimal emotion types. Measuring a broader range of emotions in future studies is expected to sharpen the results. In addition, it is also crucial for future studies to consider the interaction between other domains of the TPACK framework (i.e., TK, CK, PK, TPK, TCK, PCK) and teachers' technology uses to expand the validity of the TPACK framework's explanatory power.

Implications and conclusions
From a theoretical perspective, this study built a conceptual model to examine the hypothesized relationships among technology actual usage and two most frequently cited teachers' internal factors: teacher knowledge and attitudes about technology use. Contrary to previous TAM studies, this study provided a new dyadic perspective on attitudes towards technology, treating affective and evaluative attitudes as the independent variables, and validated the explanative power of TPACK on technology usage. This study directly contributed to the extant knowledge on technology integration and also enriched the current TPACK research by focusing on a domain-specific construct (i.e., EFL) as well as enhanced its applicability in Asian cultures.
From a practical perspective, the study shed light on EFL stakeholders (i.e., EFL teachers, school leaders, educational policymakers, educational software/application designers, etc.) in China. For a long time, the teachers' lackluster responses to using ICT resources in classrooms have discouraged school leaders and educational policymakers who made continuous efforts in building hightech learning environments. They usually got stuck with teachers' inappropriate attitudes and/or pedagogical beliefs and felt powerless to change the teachers' attitudes which were inherent parts of the teachers' psychology. This study highlighted the pivotal role of Chinese EFL teachers' knowledge and competence regarding technology integration that not only impacted technology usage behaviors directly but also linked to their attitudes toward technology. Thus, the school leaders and teacher educators could address ineffective technology use by focusing more on the EFL teachers' limited knowledge, capacities, or skills for technology integration. Providing targeted professional training in TPACK for in-service teachers was possible to raise their actual use of ICT in EFL teaching. Overall, to fulfill the promise of ICT-supported EFL education in China, policy, infrastructure building, hardware and software should be combined with fostering both TPACK and positive attitudes in EFL teachers.

Funding
The research project has received approval from the Panel on Research Ethics of University of Macau (the authors' institution). The reference number of the ethics report is SSHRE20-APP019-FED. For further information regarding research ethics, please contact the corresponding author of University of Macau, TEL:853-88228833