The influence of teachers' anxiety-reducing strategies on learners' foreign language anxiety

This quasi-experimental study investigated the effects on learner anxiety of anxiety-reducing strategies utilized by English as a foreign language (EFL) teachers in Saudi Arabia. The study was conducted in two stages. In the first stage, sources of foreign language (FL) anxiety for Saudi learners of English (N = 596) were identified using The Foreign Language Classroom Anxiety Scale (FLCAS). In the second stage, 465 learners were divided almost equally into two groups (experimental vs. control) and 12 teachers were recruited. Anxiety-reducing strategies were implemented exclusively in the treatment group for approximately eight weeks. These strategies targeted the sources of FL anxiety identified at the first stage of the study like fear of negative evaluation, communication apprehension, and the negative attitudes toward English language class. At the second stage, a classroom observation scale was used to evaluate teacher's anxiety-reducing practices. FLCAS was used to assess learners' FL anxiety levels before and after treatment. Statistical analyses (e.g. ANOVA and ANCOVA) were used to evaluate the study findings. These findings revealed that the intervention led to significantly decreased levels of FL anxiety for learners in the experimental group compared with increased levels of anxiety for those in the control group.

classroom as a scary place. They established that these pre-understandings can intensify the level of anxiety and therefore impact FL learning. Nicolson and Adams (2010) identified facilitative and inhibiting factors (e.g., anxiety) that can make the classroom either a place of comfort or an obstacle course. They established that students who had difficulty in minimizing their emotions and fears in the classroom found the experience of classroom speaking an inevitable and uncomfortable obstacle. Based on the classroom observations they conducted for over a decade, Adams and Margaret (2012, 1) concluded that 'discomfort [in the classroom] arose because of what students were asked to do, how they were asked to do it, or who they had to do it with.' Nicolson and Adams (2008) examined the interplay between learner identity and methods used in the management of speaking activities in the language classroom. They discovered that communicative activities that rely on historic conventions to capture a person's identity (e.g., personal questions about age, marital status, job, etc.) were perceived as intrusive and giving rise to anxiety by many students. Nicolson and Adams added that communicative-, group-, and student-centered methods may also challenge students' view of the teacher as an authority figure and make them re-evaluate their own identity as a learner. According to them, in such activities, students may participate less, feel uncomfortable, out of control, or worried that peers are leading rather than the teacher. These behaviors and feelings can easily evoke learners' anxiety in the classroom. Stroud and Wee (2006) highlighted the importance of the concept of learner's identitybased anxiety in the language classroom. They maintained that beyond the general assumption that student anxiety in the language classroom is basically competence-based (i.e., students become anxious, because they feel insecure about their language abilities, which evokes their anxiety in class), there is also identity-based anxiety, in which an individual may be more concerned with maintaining his or her relationship with particular groups (e.g., the desire for recognition, affiliation, and security and safety) than with his or her language abilities.

Anxiety and language learning
Given the vital role of anxiety in the success (or lack thereof) in attaining competence in an FL, this factor has been heavily researched since the late 1970s. Among the seminal works that notably influenced the field is the work of Scovel (1978), who acknowledged that early perspectives on anxiety generated inconsistent results concerning the relationship between anxiety and second language achievement (Zheng 2008). Another influential work was that of Horwitz, Horwitz, and Cope (1986), who identified FL anxiety as a situation-specific anxiety construct that is largely independent of other types of anxiety and who also contributed to developing a valid and reliable measurement of learner anxiety in the FL classroom, known as The Foreign Language Classroom Anxiety Scale (FLCAS). Subsequently, research increased considerably from the 1990s with the works of Aida (1994), Al-Shboul et al. (2013), Burden (2004), Cao (2011), Horwitz (2001, Horwitz and Young (1991), MacIntyre (1998), MacIntyre and Gardner (e.g., 1991Gardner (e.g., , 1994, Young (1994), and Zheng (2008), among many others.
FL anxiety is a complex, multidimensional phenomenon (Young 1991) that has various causes/sources. The primary sources of language anxiety explicated by Horwitz, Horwitz, and Cope (1986) are communication apprehension (fear of communicating with people), fear of negative evaluation (apprehension about others' evaluations and avoidance of evaluative situations), and test anxiety (performance anxiety stemming from the fear of failure). Young (1991Young ( , 1994 identified six interrelated potential sources of language anxiety that arise from three aspects: the learner (personal and interpersonal anxiety as well as learner beliefs about the language), the teacher (instructor's beliefs about language teaching and instructor-learner interactions), and the instructional practice (classroom procedures and language testing). Palacios (1998) recognized the level of perceived teacher support -'the help and friendship the teacher shows toward students; how much the teacher talks openly with students, trusts them, and is interested in their ideas'as a factor typically associated with higher (and lower) levels of learner anxiety.
MacIntyre and Gardner (1991, 112) described the anxious FL learner as 'an individual who perceives the L2 as an uncomfortable experience, who withdraws from voluntary participation, who feels social pressures not to make mistakes and who is less willing to try uncertain or novel linguistic forms.' Other empirical research (e.g., Aida 1994) has described anxious FL students as those who appear less willing to participate in learning activities and who have lower performance than nonanxious students.
FL anxiety negatively influences the quality of various aspects of language learning. A negative correlation between anxiety and achievement in an FL is well established in the literature (see Abu-Ghararah 1999;Aida 1994;Anyadubalu 2010;Atasheneh and Izadi 2012;Batumlu and Erden 2007;Gardner, Tremblay, and Masgoret 1997;Hewitt and Stephenson 2011;Horwitz 2001;Horwitz, Horwitz, and Cope 1986;MacIntyre 1998;Gardner 1991, 1994;Mahmood and Iqbal 2010;Wilson 2006;Young 1991). According to MacIntyre and Gardner (1991) and MacIntyre (1998), the effects of anxiety on second/FL learning and performance are projected onto social, cognitive, and personal aspects in addition to academic achievement.

EFL teaching and learning in Saudi Arabia
English was introduced in Saudi schools in the late 1950s (Al-Shammary 1984). It was initially taught only in the intermediate and secondary schools (Years 7-12). Only very recently has English been introduced in the last year of elementary schools (Year 6). The rationale for the late learning of English is that its acquisition might interfere with the acquisition of the mother language (Arabic) if taught at an early age. English is still considered a FL in Saudi Arabia (e.g., Al-Maini 2006) because of the dominance of native Arabic in most daily activities. This dominance results in relatively few opportunities to hear, speak, and write English inside or outside of the classroom (Al-Otaibi 2004). Rahman and Alhaisoni (2013) stated that the Saudi government has recognized the importance of English language by making it a mandatory subject in schools and universities. However, the level of achievement in learning EFL is far below expectations: After years of study, most Saudi students are unable to communicate in English and have only the most basic reading and writing skills. This phenomenon may be attributed to a number of cultural, social, religious, and political barriers that undermine the value of learning English in the Saudi context. Al-Zubeiry (2012) assumed that the low proficiency of Arab students in English stems from the fact that they have ethnocentric and authoritarian attitudes toward the English language and its culture. Pennycook (1994) noted some cultural misconceptions usually associated with teaching and learning English by a majority of the Saudi population, claiming that many Saudis feel that English is connected to forms of knowledge and culture that are oppositional or even threatening to their Islamic way of life. Alresheed (2012) investigated Saudi English teachers' beliefs and attitudes toward English language and its effects on their teaching practice. He found that religion and culture were the major factors affecting teachers' attitudes and their teaching styles. According to Alresheed, teachers who were highly influenced by conservative religion and culture had negative attitudes toward English and demonstrated old-fashioned and un-innovative teaching styles. Their teaching was restricted to completing the syllabus and preparing for exams, and students in their classes were discouraged from actively using English outside the class. In contrast, teachers who were less likely to be influenced by religion and culture had positive attitudes toward English. They actively encouraged students to use English in their daily lives and adopted a learner-friendly atmosphere in their classrooms. Khan (2011) elaborated that teaching of EFL in Saudi Arabia always demands expertise in the pedagogy for many sociocultural, economic, psychological, philosophical, educational, and pedagogic reasons. He clarified that the objective of teaching and learning English is unclear to a majority of the Saudi population, since English, to a greater extent, is seen as the main avenue through which Western (i.e. non-Islamic) culture enters the country. Khan concluded that when it comes to a place where English serves a very limited purpose like in Saudi Arabia, it becomes more crucial and painstaking to teach and learn.
In addition to religious and cultural restrictions, Al-Shehri (2004) highlighted some of the challenges to the EFL program in the Saudi context, asserting that English classroom instruction in Saudi Arabia tends to be teacher-centered rather than student-centered. Furthermore, English classes are typically overcrowded, with more than 40 students per class (Al-Mohanna 2010). In the Saudi context, an English language teacher is typically perceived as a presenter of knowledge, rather than as a facilitator of learning. This perceived role may be attributable to the common belief that the teacher is an authority figure and thus is the main source of knowledge and the controller of the class (Alrabai 2011). Consequently, learners in this context are typically considered passive observers.
Despite all these learner's anxiety-provoking situations, there are very few experimental studies on FL anxiety in the Saudi setting. Abu-Ghararah (1999) investigated the effect of FL anxiety on English achievement using the FLCAS and the grades of 240 secondary and university students. The study found a negative correlation between FL anxiety levels and student's achievement within the study sample. Al-Saraj (2013) used a case study design to examine the experiences of female college students learning EFL in Saudi Arabia. A variety of instruments, including questionnaires, classroom observations, and individual and group interviews, were used to gather information from ten participants regarding their perspectives and behavior. Participants reported experiencing moderate to high anxiety. Teacher-student interaction and teacher behavior were the key anxiety-provoking factors among all participants. Aljafen (2013) investigated FL writing anxiety among 296 Saudi university science students at Qassim University in Saudi Arabia. The participants were chosen from three science colleges: preparatory year, pharmacy, and engineering. The students completed the English Writing Apprehension/ Attitude Test (EWAT) scale survey to evaluate their writing anxiety. The findings of Aljafen's research showed that the three groups of learners shared almost the same moderate feeling of English writing anxiety. The weak previous English education of the learners was the main cause of this phenomenon.

Anxiety-reducing strategies
In view of the undeniable negative effects that anxiety typically exerts on language learning outcomes, the need for practical means to help learners diminish or at least cope with the anxiety that they experience in the course of FL learning does not appear to require justification. Numerous scholarly sources proffer theoretical suggestions to help learners diminish, reduce, or cope with their feelings of anxiety in the language classroom (e.g., Kao and Craigie 2013;Horwitz 1990;Horwitz, Horwitz, and Cope 1986;Kondo and Ling 2004;Nagahashi 2007;Onwuegbuzie, Bailey, and Daley 1999;Oxford 1990;Tallon 2008;Tsiplakides and Keramida 2009;Young 1991Young , 1999Zhang 2010). These sources suggest practical means of addressing anxiety sources that typically stem from learner characteristics, learner beliefs about learning an FL, teacher characteristics, language testing, classroom atmosphere, learning procedures, and other sources. Zheng (2008) emphasized that studies of this nature provide insights into how educators can develop appropriate interventions to decrease language anxiety among second/FL learners. Nearly all of these studies regard the FL teacher as the cornerstone of the implementation of these strategies. Horwitz, Horwitz, and Cope (1986) argued that the teacher holds the major responsibility for reducing learner anxiety by either focusing on the students' anxiety-provoking situation or by establishing a relaxed and less stressful learning atmosphere in the language classroom. These strategies may be referred to as 'foreign/second language anxiety-reducing strategies' and in the context of this paper, this term refers to 'the techniques used by the foreign/second language teacher to help language learners diminish, reduce, or at least cope with the feelings of anxiety they experience when learning a foreign/second language.' Despite the valuable theoretical assumptions of earlier studies regarding the usefulness of anxiety-reducing strategies, the empirical practicality of such strategies is yet to be established, as none of these studies have experimentally tested the effectiveness of these strategies in reducing learner anxiety in FL classes. Even with the appearance of recent attempts to test these strategies (e.g., Nagahashi 2007;Tsiplakides and Keramida 2009), these studies appeared to be more concerned with the general practice of teachers, rather than with utilizing specific strategies to reduce learner anxiety in the classroom. Because such studies also lacked comparison (control) groups, the results of the experimental intervention in the treatment classes could not be compared with those of the traditional methods of instruction followed among the control groups. This study primarily aims to shift from theory to practice, by utilizing anxiety-reducing strategies in the FL classroom and by evaluating their effectiveness in reducing learners' FL anxiety.

Study methodology
This study was conducted in two stages. The objective of the first stage was to empirically identify the sources of FL anxiety that Saudi learners typically experience when learning English. The objective of the second stage was to establish whether implementing anxietyreducing strategies in FL classrooms would induce positive changes in FL learner anxiety levels.

Participants
In the first stage, 596 volunteer male and female EFL learners were recruited. Demographic information regarding the learners participating at this stage is presented in Table 1.
This type of intervention study is rare because of the difficulty in gaining access to cooperative institutions; thus, it is notable that this study gained access to four institutions from three regions of Saudi Arabia during the second stage and garnered the cooperation of 12 teachers who were willing to undergo evaluation of their teaching practices. The reason for this success appears to be that the research project was generally valued by the teachers and learners who were approached for the study.
The 12 participating teachers were volunteer male EFL instructors. These participants represented a different range of ages, qualifications, teaching experiences, and regional backgrounds, as shown in Table 2.
The students participating in the second stage were 468 Saudi EFL learners who spoke Arabic as their first language and who studied at different levels, ranging from beginner to advanced. Social demographic information pertaining to the participating learners at this stage is provided in Table 3.
Given that the protocol of this research required real-time classroom observations to monitor and record teachers' and learners' in-class behaviors, we were unable to include female participants in the second stage of the study, because there are no coeducational settings in Saudi Arabia due to religious, social, and cultural restrictions. Female students and teachers in Saudi Arabia study and teach in schools and colleges that are completely separated from those of males. For this reason, this sort of participants was not approached in the second stage of the study, because it would be impossible for a male researcher to have an access to their institutions to conduct the classroom observations. We believe that the same restrictions affected the participants' gender representation at the first stage of the study, with females being underrepresented (only 41.78% of the sample population).

Instruments
The FLCAS, a 33-item instrument developed by Horwitz, Horwitz, and Cope (1986), was employed in the two stages of this study. In the first stage, this instrument was used to identify the sources and levels of FL anxiety for Saudi EFL learners. In the second stage, it was used to assess the degree to which students felt anxious during language class before and after treatment. The scale items were rated on a five-point Likert scale, with 1 representing 'strongly disagree' and 5 representing 'strongly agree.' The total scale scores ranged from 33 to 165, with high scores indicating high levels of FL anxiety. Nine items (2, 5, 8, 11, 14, 18, 22, 28, and 32) were negatively worded, and their scores were reversely computed. The scale demonstrated internal reliability, achieving an alpha coefficient of .93, with all items producing significant corrected item-total scale correlations. Test-retest reliability yielded r = .83 (p < .001) (Horwitz, Horwitz, and Cope 1986). The English version of this instrument is presented in Appendix 1.
In the second stage, a classroom observation instrument was also utilized to assess teacher practices based on the anxiety-reducing strategies that were implemented during the treatment period. This instrument focused on assessments of teacher behavior in terms of demonstrating proper behavior to students, reducing student language-learning anxiety, and building student self-confidence. The evaluation of each of these variables involved the evaluation of its subcomponents. For example, assessing the teacher's ability to reduce learners' anxiety involved the assessment of his ability to reduce their fear of negative evaluation, their communication apprehension, and their negative attitudes toward the English class. The variables on this scale were designed particularly for the purpose of the current study and were evaluated on a five-point frequency scale, with 1 assigned to 'Never' and 5 assigned to 'Very Often.' Higher scores on this scale indicate a greater ability of the teacher to reduce learner anxiety. This instrument underwent piloting procedures in three classes for approximately one month during the semester preceding the experimental treatment. The teacher of each class was interviewed immediately after his class was observed and was consulted about the consistency and accuracy of the recording of the instructional events that occurred during the class. Based on insights from the teachers, modifications and amendments were made to the classroom observation scale. To avoid subjective judgments, the items in this instrument were rated by two raters (the researcher and another expert in the research area). An interrater reliability analysis using the Kappa statistic was performed to determine consistency among these raters, who monitored and evaluated the teachers' behaviors in the three classes visited during piloting. The interrater reliability for the scale was found to be Kappa = .787 (p < .001), displaying a substantial level of agreement between raters. The instrument can be found in Appendix 2. Design and matching procedures The four participating institutions in the second stage were initially randomly selected by the researcher and then subjected to the following three conditions for study participation: the formal permission of the principals of these institutions, the approval of at least one English teacher from each institution for himself and his students to be involved in the study, and the consent of those students to participate in the study. The participating teachers were allocated to either the experimental or control group based on their expressed preference for involvement in a particular group. The two cohorts of teachers were carefully matched for qualifications, experience, and age to minimize the effects of any preexisting differences between the groups. Furthermore, each pair of teachers in the experimental and comparison (control) groups had taught the same course content to learners at similar study levels and of similar ages. This matching procedure aimed to ensure that the implementation of the anxiety-reducing strategies was the single key difference between the two groups of teachers.
The participating learners were almost evenly divided between the experimental and control groups (236 and 232 learners, respectively). The allocation of learners to the different groups naturally followed the allocation of the teachers: Learners were assigned to their own teachers.
Effective matching along these categorical and ordinal variables was tested statistically through a set of chi-squared tests performed on the teacher and student demographic data; the tests revealed that the conditions had a null effect on all variables (p > .05).

Procedures
Pre-treatment training Based on the results from the first stage of the study, an advisory guide was constructed to assist the teachers in the experimental groups with strategy implementation. The guide included a range of specific techniques to translate the pre-selected anxietyreducing strategies into a classroom context. Details regarding these strategies as well as excerpts and examples of the specific techniques used to operationalize the strategies are presented in the experimental treatment section.
Prior to the implementation of the experimental treatment, teachers in the experimental group underwent a two-week intensive training course on the use of the advisory guide conducted by the researcher. This training course was intended as an awareness-raising program to facilitate the teachers' thinking and practices in relation to FL anxiety.

Experimental treatment
The experimental treatment was conducted over an eight-week period during the 2012-2013 winter term of the Saudi academic year. During the experimental intervention period, the teachers in the experimental group implemented some strategies that had been recognized in the literature on anxiety and educational psychology (e.g., Dörnyei 2001a; Kao and Craigie 2013;Horwitz and Young 1991;Tallon 2008;Tsiplakides and Keramida 2009;Young 1991Young , 1999Zhang 2010) as being potentially effective in reducing learners' FL anxiety; moreover, the applied strategies were found to have the potential for positive effects on the sources of FL anxiety detected during the first stage of the study. In addition, they seemed to be transferable across diverse cultural and ethno-linguistic contexts (e.g., Japan, Taiwan, China, Greece) and might fit in the Saudi EFL context. Ease of implementation in the language classroom was also taken into account in the selection of these strategies. The following strategies were employed: (1) Demonstrate proper teaching behavior to your students.
(3) Reduce the fear of negative evaluation in learners. (4) Reduce the fear of language testing in learners. (5) Properly address learner anxiety-provoking beliefs and misconceptions. (6) Help students establish specific and realistic goals for learning English. (7) Increase students' self-confidence.
The implementation guide designated specific techniques and subtechniques with which to operationalize each of these strategies. Demonstrating proper teacher behavior to learners includes behaviors such as showing passion and enthusiasm for teaching the FL, demonstrating commitment toward students' progress, and developing positive personal relationships with students (i.e. showing respect and warmth to students, being friendly, caring, and patient).
Reducing learner communication apprehension involves practices such as providing students with opportunities for oral practice in the FL, allowing students to practice selftalk before they talk in real situations in the classroom, and paying attention to certain anxiety-provoking situations, such as speaking in front of the class and writing work on the board.
Reducing learners' fear of negative evaluation involves making students concentrate on learning tasks rather than on the evaluation of their performance, processing feedback sessions to address learners' errors, avoiding putting individual students 'on the spot' when correcting their errors, encouraging students to learn from their errors, and correcting learner errors using modeling.
Reducing learners' fear of language testing involves utilizing techniques such as giving students practice tests before the real test, avoiding making difficult tests or tests that do not match what was taught in class, making test instructions clear and familiar to students, clarifying the criteria that will be used for grading, and giving students sufficient time to finish exams.
Addressing learner anxiety-provoking beliefs and misconceptions involves helping learners to interpret anxiety-provoking situations in more realistic ways and to find approaches to cope with such situations, talking openly with students about the nature of language anxiety, addressing learners' beliefs that may evoke feelings of anxiety (e.g., mastering a FL is an overwhelming task, speech in an FL should be grammatically perfect, pronunciation must sound native or native-like, learners should understand every word of what they read or hear, etc.) Helping students establish specific and realistic goals for learning English includes techniques such as identifying students' individual goals for learning English; ensuring that students' goals for learning English are clear, realistic, and attainable; linking students' individual learning goals with the curriculum goals; and suggesting ways to overcome the obstacles that students may face in achieving their learning goals.
Increasing students' self-confidence can be achieved by reinforcing students' ability for success (e.g., setting high expectations for students, showing faith in students' abilities, enabling learners to achieve success) and recognizing students' efforts and achievements (i.e., providing them with positive feedback, appropriate praise, recognition, and rewards).
To implement these strategies in practice, the teachers were asked to expose the learners in their classes to anxiety-reducing and self-confidence-building moments during each lesson. Examples of self-confidence-building moments that occurred during a class recorded by the researcher are provided below.
Example. In the first visit to a class, while students were working on a given learning task, the teacher moved around the classroom and targeted a student who was sitting silently in the far back of the class with clear anxiety signs on his face. The teacher continued to offer practical encouragements and positive reinforcements to this student, while he was working on the assigned task: In the second visit to that class, the same student was sitting in the front row and was one of the most active students in class.

Data collection
The data for this study were collected over two stages. In the first stage, the FLCAS was administered to 596 volunteer male and female EFL learners to empirically identify the sources of FL anxiety that Saudi learners typically experience when learning English. Only those learners who were granted institutional approval to participate in the study and who provided their own consent to participate were included. Because a major goal of this study was to recruit as many participants as possible at this stage, the researcher used 'snowball sampling,' which, according to Dörnyei (2001b), involves a 'chain reaction,' whereby the researcher identifies several people who meet the criteria of the particular study and then asks these participants to identify additional members of the target population.
In the second stage, 468 Saudi EFL learners and 12 teachers were recruited. At this stage, 30-minute classroom observations to assess teachers' anxiety-reducing practices were conducted during two phases, at the beginning (Time 1, or T1) and end of the treatment (Time 2, or T2). During each phase, three observation sessions were conducted. In each session, the frequency of three main variables was observed and recorded: the teacher's behavior, his utilization of anxiety-reducing strategies, and his enhancement of learners' self-confidence. During each session, these variables were recorded by two raters for nine minutes in a consecutive manner, with a one-minute gap between each observation session. A substantial level of agreement between the raters was found for both phases (Kappa = .792 for the first phase, .796 for the second phase, p < .001 for both phases), confirming the reliability of this instrument.
The FLCAS was administered twice to learners at this stage: at the beginning and end of the treatment. The participating learners received ample reassurance of the anonymity and confidentiality of their responses. The questionnaire was administered in the subjects' mother tongue of Arabic to ensure that inadequate knowledge of English would not be a factor in the participants' presentation of their responses. To ensure the honesty of the learners' responses, neither the teachers nor any other representatives of the participating institution were allowed in the classroom during the administration of the questionnaire. The learners required 35-45 minutes to complete the entire questionnaire at each time point.

Data analysis
In the first stage of the study, descriptive statistics were generated to identify the levels of FL anxiety for Saudi learners of English. The FLCAS was also subjected to factor analysis with varimax rotation to reveal its underlying components and thereby identify FL classroom anxiety sources for Saudi EFL learners.
In the second stage, a two-condition (experimental vs. control) between-subject analysis of variance (ANOVA) was performed on all T1 aggregate indices of teachers' classroom observations and learners' self-report questionnaire responses to check for any preexisting differences between teachers and learners in the experimental and control groups with regard to the key anxiety variables. To identify the variables that changed over time in the experimental (vs. control) condition, a two-time (T1, T2) repeatedmeasure ANOVA was performed separately among the experimental and control teachers and learners. Finally, a condition between-subject analysis of covariance (ANCOVA) was performed on the T2 data collected at the end of treatment (with T1 data as covariates) to control for preexisting group differences that were likely to affect the study's key findings. The ANCOVA was used to reveal the significant changes between the study groups that occurred as a result of our experimental intervention. Dörnyei (2007, 222) elaborated on a significant feature of ANCOVA by stating that 'a special case of ANCOVA [test] in quasi-experimental designs occurs when we compare the post-test scores of the control and the treatment groups while reducing for the pre-test scores as the covariate.' He added that in such a case, if we find any significant difference between the posttest scores, then these differences will be related to the events that occurred after the pretest (the experimental intervention in this study), because the preexisting differences between the two groups would have been removed by controlling for the pretest scores.
Although some of our analyses could have been conducted in the t-test domain, we performed them in the ANOVA/ANCOVA domain to ensure that we could always control for effect size (strength of association). The partial eta squared (g 2 p ) was used as an estimator of the effect size of the treatment that was manipulated in this study. We drew on the findings of previous studies (e.g., Cohen 1988) in interpreting the effect size of the intervention based on the partial eta squared values: .01 = small effect, .06 = moderate effect, and .14 and above = large effect.

Results of the first stage of the study
Based on the classifications provided by earlier studies (e.g., Linh 2011), the mean scores were used to determine the levels of anxiety for Saudi EFL learners. The overall mean score for the FLCAS in the current study was 101.05/165. This result indicates that the majority of the study participants had a moderately high level of anxiety based on the values in Table 4. This finding is in line with that of Al-Saraj (2013). The reasons behind the high level of anxiety held by Saudi learners of EFL are diverse and complex. This phenomenon could be attributed to several causes, including the religious, social, and cultural misconceptions usually linked to learning of EFL in Saudi Arabia. Another contributing factor could be the nature of EFL learning/teaching in Saudi Arabia, where the teacher is usually seen as an autocratic figure rather than democratic figure (Alrabai 2011); a threatening atmosphere in which learners' errors are not tolerated (Tanveer 2007); the lack of the learners' involvement in class discussion and decision-making; a competitive atmosphere for learning, where learners work against each other instead of cooperating; overcrowded EFL classes that generate a real barrier to the proper involvement of learners; a ready-made EFL curriculum that often cares for the quantity rather than the quality of the content, forcing EFL teachers to work under pressure to cover the entire content on time; the assessment procedures that primarily depend on written tests to evaluate the learners' progress in the English language (Tanveer 2007); the strict classroom rules imposed by schools in Saudi Arabia, which do not allow the learners to experience sufficient freedom and therefore evoke feelings of anxiety. All of these obstacles generally limit the teachers' opportunities to expose their students to anxiety-reducing strategies in their classes.
A four-factor solution emerged as a result of the factor analysis conducted in the first stage. Ten items (3, 7, 10, 13, 19, 20, 21, 23, 31, and 33) indexed the first FLCAS component (fear of negative evaluation). This component accounted for 38.15% of the total variance. Items reflective of language test anxiety (e.g., 10,21) loaded strongly on this factor. It clearly appears that the strict and formal EFL classroom settings in Saudi Arabia generally create highly evaluative situations, where the learners' performance is constantly monitored and evaluated. This constant assessment means that the fear of a negative evaluation from both their teacher and peers is the biggest concern and the main source of anxiety for learners in their English language classes.
Eight items (1, 4, 9, 15, 24, 27, 29, and 30) reflected the second component of the FLCAS, communication apprehension, and accounted for 17.18% of the total variance. Although communication apprehension is considered to be a significant source of FL anxiety according to the results of previous research (e.g., Price 1991; Zhang 2010), it did not emerge as the primary source of anxiety for the Saudi EFL learners recruited in this study. This result might be attributable to the limited communication opportunities involving the learners in their language classes, which precluded communication from becoming the biggest concern of learners and, consequently, not the primary source of their EFL classroom anxiety.
Six items (6, 12, 16, 17, 25, and 26) constituted the third FLCAS component, negative attitudes toward the English class (9.35% variance). The fourth component obtained loadings from numerous items (2, 5, 8, 11, 14, 18, 22, 28, and 32) and accounted for 5.32% of the total variance. All of the items in the fourth component were originally negatively worded items, indicating a lack of anxiety in the FL learner. This component was found to be independent of the other three components although it was significantly correlated with them. Because this factor reflected a learner's confident approach to FL class, it was referred to as 'self-confidence' and was thus treated as an indicator of a lack of anxiety in FL class (i.e., high scores on this factor indicated low levels of FL classroom anxiety). The fourth component that was extracted in this study is similar to that found in the study of Burden (2004), who labeled it 'foreign language classroom non-anxiety.' In sum, the findings of this analysis revealed that there are three sources of anxiety for Saudi EFL learners: fear of negative evaluation, communication apprehension, and negative attitudes toward English class. The correlations among these sources in addition to the fourth emerging component (self-confidence) are reported in Table 5.

Results of the second stage of the study
Were there any preexisting differences between the study groups before experimental treatment?
To identify any preexisting differences between teachers in the experimental and control groups with regard to the key variables, a two-condition (experimental vs. control) between-subject ANOVA was performed on all T1 aggregate indices. The findings reported in Table 6 reveal that the condition factor had a null effect on all teacher variables at T1. This result indicates that there were no significant preexisting differences between the teachers in the two groups in their ability to reduce learners' FL anxiety and to promote their self-confidence prior to experimental treatment. This set of null differences between the experimental and control teacher groups indicates that our matching procedure was indeed effective in the creation of two groups of teachers who were comparable in most key social dimensions. The findings of the two-condition ANOVA test performed on the T1 data and reported in Table 7 reveal that the condition factor had a null effect on all but one of the learner variables at the beginning of treatment. This result indicates that there were no significant preexisting differences between learners in the two groups in terms of anxiety components and in their overall levels of anxiety toward learning EFL prior to the experimental intervention. The only significant difference was in learners' self-confidence (F(1, 466) = 15.53, p < .001, g 2 p = .03, M = 3.12 for the experimental group, M = 3.30 for the control group). The insignificant differences in most variables between the experimental and control learner groups are another indicator of the successful and effective matching procedure that this study employed.

Changes in teacher behaviors and practices over time
To identify teacher variables that changed over time in the experimental (vs. control) condition, a two-condition (experimental vs. control) × two-time (T1, T2) mixed-model ANOVA was performed using time as the repeated measure factor for each of the teacher variables. As shown in Table 8, all variables changed significantly over time (p < .05) as a result of the time × condition interaction factor. A two-time (T1, T2) repeated-measure ANOVA was also conducted. The figures reported in Table 8 indicate that the anxiety-reducing practices of the teachers changed significantly over time exclusively among teachers in the experimental group and that no significant changes occurred over time in the control group for any of the variables. The mean scores of these variables displayed a positive change over time (T1, T2) in the treatment group (e.g., reducing anxiety MT1 = 3.39, MT2 = 4.27) compared with a decline for most of these variables in the control group (e.g., reducing anxiety MT1 = 3.21, MT2 = 3.19). The variable with the largest increase over time in the experimental group was demonstrating proper teacher behavior (F(1, 5) = 159, p < .001, g 2 p = .97, MT1 = 3.72, MT2 = 4.77). This finding indicates that among all of the variables, teacher Note: Sig. = significant differences were detected between students in the experimental and control groups in this variable at T1. Note: Exp. = experimental group; Cont. = control group; M = mean; SD = standard deviation; F = variance of the group means; p = significance value; g 2 p = partial eta squared; No Sig. = no significant differences were detected between teachers in the experimental and control groups in this variable.

Innovation in Language Learning and Teaching
behavior was the most positively affected by the implementation of anxiety-reducing strategies in the treatment classes.

Changes in learner anxiety over time
A two-time (T1, T2) repeated-measure ANOVA was performed on the experimental and control learners' data to identify the variables that changed over time in the experimental (vs. control) condition. As shown in Table 9, the time × condition interaction factor had a significant effect on all learners' variables over time (p < .001). A two-time (T1, T2) repeated-measure ANOVA was also performed separately among the experimental and control learners to reveal whether any significant differences occurred over time and whether these changes were larger in the experimental vs. control learner groups. The relevant F statistics and p values for the effects of time separated by conditions, which are reported in Table 9, indicate that learner anxiety (in terms of fear of negativeevaluation, communication apprehension, negative attitudes toward English class, and overall anxiety) decreased significantly over time in the experimental learner group only. In addition to this decline in learner anxiety over time in the experimental group, there was a significant increase in learners' self-confidence (F(1, 235) = 10, p < .01, g 2 p = .04, MT1 = 3.12, MT2 = 3.25). All of these variables significantly changed in the opposite direction over time in the control group, with increased levels of anxiety and decreased self-confidence (see Table 9). These findings indicate that the experimental treatment incrementally affected these variables beyond the effects of standard teaching practices and basic maturation. The component of FL classroom anxiety that exhibited the largest change in the experimental group over time was the fear of negative evaluation variable (F(1, 235) = 93, p < .001, g 2 p = 0.28, MT1 = 3.13, MT2 = 2.75). This finding reveals that learners' fear of negative evaluations from other people, such as their teachers and peers, was the variable most positively affected by the utilization of anxiety-reducing strategies.
Group differences in the anxiety-reducing practices of teachers in response to treatment, independent of preexisting group differences If we compare the statistics found in Table 6 with those in Table 10, we observe that all of the variables that did not significantly differ between the experimental and control groups at T1 became significantly different between the two groups at T2 (see Table 10). This result indicates that there was a statistically significant effect of the condition factor on all teachers' variables at T2 despite that it had no significant effect on any of them at T1. The adjusted mean scores (marginal means) of the significantly changed variables suggest larger changes in favor of the experimental group rather than the control group at T2 (e.g., promoting self-confidence: marginal mean (MM) = 4.29, SD = 0.30 for the experimental group; MM = 2.69, SD = 0.80 for the control group). The sizes of the condition effect (g 2 p ) were quite large for all of the affected variables at T2 and were considerably larger than the g 2 p at T1. The significant F values, which were larger in all of the variables at T2 compared with T1, confirmed that the effect of the condition was significant and sufficiently large for these changes not to be accidental. All of these findings suggest that the utilization of anxiety-reducing strategies in EFL instruction resulted in positive changes in most of the teaching behaviors of the teachers in the experimental group compared with those in the control group; this difference between the two groups was likely attributable to the lack of intervention in the latter group. The condition factor had the largest effect at T2 on demonstrating proper teacher behavior (F (1,9) = 75, p < .001, g 2 p = .89), indicating that teacher behavior was the variable most positively affected by the utilization of anxiety-reducing strategies. This finding confirms that the intervention that occurred in the experimental classes by utilizing anxietyreducing strategies had positive effects not only on learner anxiety, which we had anticipated, but also on teacher's behaviors.
Group differences in learner anxiety attributable to treatment, independent of preexisting group differences A condition between-subject ANCOVA was performed to determine the effect of the condition factor on learner anxiety at the end of treatment after specifying the T1 scores as covariates to control for their likely effects on T2 scores.
No preexisting differences were identified between the learners in the study groups in terms of their anxiety at T1 (beginning of treatment), except for self-confidence, which suggested that learners in the two groups had similar FL anxiety levels; nevertheless, this factor had a significant effect on learner's anxiety at T2 (end of treatment). The effect of treatment on learners' overall anxiety was large (F(1, 465) = 210, p < .001, g 2 p = 0.31; Table 11). The adjusted means suggests a large positive change in learner's anxiety in favor of the experimental group in this variable (MM = 2.83, SD = 0.44 in the experimental group; MM = 3.28, SD = 0.45 in the control group). The figures in Table 11 indicate that the component of learner anxiety most affected by experimental treatment was fear of negative evaluation (F(1, 465) = 134, p < .001, g 2 p = 0.22). It is remarkable to  find that fear of negative evaluation, which has been regarded as the main source for anxiety for Saudi EFL learners according to the findings of the first stage of the current study, has been the variable that was most positively affected by the implementation of anxiety-reducing strategies. This confirms the success of the strategies utilized to target this source of anxiety. Significant differences were also identified between learners in the two groups in terms of their self-confidence at the end of the treatment (F(1, 465) = 20, p < .001, g 2 p = 0.04), with marginal means in favor of the learners in the experimental group (MM = 3.28, SD = 0.54 in the experimental group; MM = 3.07, SD = 0.55 in the control group), despite the higher self-confidence of the learners in the control group before the intervention began (see Table 7).
These analyses were able to capture the net effects of experimental treatment by identifying the key posttreatment differences between the experimental and control participants at T2 that remained after reducing for (statistically removing) the effect of any group differences at T1; consequently, the analyses confirm that the experimental intervention positively affected EFL learner's anxiety in the experimental group. Dörnyei (2007) assumed that any significant differences between the study groups in the posttest scores after ANCOVA should be related to the events that occurred after the pretest, because this analysis controls for (statistically removes) the effects of the preexisting differences between the two groups. Building on this assumption, the findings presented here clearly indicate that the anxiety-reducing strategies utilized in the experimental group were a cause of the significant positive decline in FL anxiety and of the enhanced self-confidence of learners in the experimental group at the end of the eight-week English course. This finding is reflective of the causal relationship between the anxiety-reducing practices of teachers and the FL anxiety of learners (that is, learners feel less anxious with teachers who are better able to reduce their anxiety).
The design of this study, which involved the utilization of an experimental intervention with prepost evaluations, careful sampling procedures, data collection procedures, and laborious statistical analyses, aimed to obtain rigorous and unequivocal results. All of these procedures gave us the confidence that the anxiety-reducing practices of the experimental EFL teachers were the sole cause of the positive changes (decreased levels of FL anxiety and increased self-confidence) among learners in the treatment classes.
This study suggests a number of implications and recommendations for EFL instructors, EFL curriculum designers, and academic institutions in Saudi Arabia as well as for future researchers in the field. EFL teachers should behave as models and should avoid antagonistic behavior toward students, such as aggressively criticizing, overcorrecting errors or blaming students when they make mistakes, engaging students in competition, and offering public comparisons of the performance or grades of different students. Teachers should promote cooperative learning in which the students work together instead of competitively, show faith in students' abilities to succeed and make use of encouraging expressions for this purpose, provide students with positive feedback and appraisals of their performance, provide students with more control over their learning, and involve students in decision-making. Teachers should also tackle learners' beliefs and misconceptions that can evoke their feelings of anxiety.
This study recommends that EFL curriculum designers in Saudi Arabia should include anxiety-reducing strategies within the content of the prescribed curriculum provided to students. The content of the prescribed book should allow students to contribute to their learning and should allow their ideas to be sought and used. The curriculum should include tasks that encourage cooperative learning, in which students work collaboratively and exchange personal experiences and ideas. The designers should focus on the quality rather than the quantity of the curriculum content, and they should attempt to reduce the density of the EFL curriculum content to diminish teachers' fears that they will not be able to cover all of the content in time. This reduction will provide teachers with sufficient opportunities to make use of anxiety-reducing strategies in their classes.
Schools must remove the obstacles that evoke learners' anxiety by decreasing the strict institutional norms that they usually impose. Instead, schools can grant both teachers and students the freedom to run classes using their own methods. Schools must also find practical solutions for the problem of overcrowded EFL classes, which is a real barrier to the proper utilization of anxiety-reducing strategies in these classes.
Some limitations to the study presented in this article should be acknowledged. One limitation of this study concerns the lack of female participants in the second stage. Although a study of this type with female participants in the sample is likely to have similar or even better results, future research endeavors should include a female population of language learners to establish this conclusion empirically. Another limitation of this study is that it did not investigate the effectiveness of anxiety-reducing strategies on learners' EFL achievement. Thus, future research efforts should examine how the utilization of such strategies would affect actual learner achievement. This could be taken forward by identifying practical means by which to utilize anxiety-reducing strategies in classroom settings and to evaluate their effects on learners' anxiety as a first step, and on their actual achievement as a subsequent step. If it turns out that the implementation of anxietyreducing strategies in the language classroom has resulted not only in low anxiety in learners, but also leads to better achievement, then the implications of such a study would be very far-reaching indeed. A third limitation is that this study does not examine the interaction domain in the classroom and its influence on learner anxiety. This area may be influenced by the educational culture, and it may be necessary for future research endeavors to subject it to further investigation.

Conclusions
The experimental intervention applied in this study resulted in a significant decline in learner's anxiety and an increase in self-confidence over time only among the experimental learners (as opposed to the control learners). This result remained significant even after we controlled for pretreatment group differences. The major contribution of this study is that it is the first empirical investigation to experimentally establish a positive causal relationship between anxiety-reducing strategies and student's FL anxiety. Hence, this study reinforces the fundamental role of teachers in reducing the learners' anxiety. The study is expected to have far-reaching implications not only for FL language teaching and learning practices in Saudi Arabia but also for FL/SL theory more generally. Based on the findings of the present study, this article presents implications for EFL practitioners in Saudi Arabia and also sheds some light on possible future research endeavors that could be of interest for researchers in the area of FL anxiety.

Notes on contributor
Fakieh Alrabai is an assistant professor of applied linguistics at the Department of English, King Khalid University, Muhayel Aseer campus. Dr Alrabai is the author of many books and journal articles in the areas of foreign language motivation and language anxiety.