Willingness to speak English among foreign language learners: A causal model

Abstract The present study is an attempt to investigate the extent to which Iranian EFL learners are willing to speak English in language classrooms. Moreover, the relationship between willingness to speak (WTS) and language learning anxiety, language learning motivation, and self-perceived speaking ability is sought. Furthermore, the study seeks the extent to which WTS contributes to an improvement in the learners’ speaking ability. One hundred and fifty-six EFL learners selected based on purposive sampling took part in the study. A packet of questionnaires were employed to measure WTS, language learning anxiety and language learning motivation. Further, all participants took part in an IELTS speaking test in order to find out if their WTS is related to their speaking performance. Results showed that learners were fairly willing (but not highly willing) to speak English in class. Moreover, it was found that WTS is negatively correlated with language learning anxiety, while positively correlated with language learning motivation and self-perceived speaking ability. Several factors were identified as having an influence on learners’ WTS, including topic of discussion, effect of interlocutor, shyness, self-confidence, teacher, and classroom atmosphere. Finally, learners took part in a speaking test, through which, it was found that those who were more willing to speak got a higher speaking score. Based on the findings, a foreign language classroom WTS model of factors underlying willingness to speak English is proposed. Results can have pedagogical implications for language learners and teachers.


PUBLIC INTEREST STATEMENT
The issue of willingness to communicate has always been a thorny issue among language educators. There might be situations in which individuals may exhibit different degrees of willingness to communicate with others. In language classrooms, more specifically, this issue is more evident as some language learners are not as willing as others to speak up and take part in classroom discussions and activities. The present study is an endeavor to explore the issue of willingness to speak in a foreign language classroom with the hope to come up with factors that can promote or hinder language learners' willingness to speak.

Introduction
There are many factors that influence language learning process. To many language practitioners, motivation is a key element in the process of learning a second/foreign language. However, although necessary, it is not adequate. In fact, in an authentic, real situation, particularly outside the language classrooms, where the language learners are exposed to the language, having a high degree of motivation and even benefiting from an acceptable proficiency level may not necessarily suffice to initiate communication. In such a situation, the language learners have to start or participate in communication and then maintain it effectively (Matsuoka & Evans, 2005).
In communicative classrooms, teachers who favor communicative language teaching like to have motivated students who demonstrate high degree of willingness to communicate in the second/ foreign language. A lack of willingness results in ineffective interaction and language production (Freiermuth & Jarrell, 2006). So, it is true that the notion of Willingness to Communicate (WTC), which is actually the intention and desire to initiate communication, plays a key role in learning a second/foreign language (MacIntyre, Clément, Dörnyei, & Noels, 1998). Some researchers (e.g. MacIntyre, Baker, Clément, & Donovan, 2003;MacIntyre et al., 1998) have argued that a fundamental goal of L2 education should be the encouragement of willingness to communicate in language learning, because WTC is expected to facilitate the language learning process so that higher WTC among students leads to increased opportunity for practice in L2 and authentic language use.
When the goal of language teaching is to encourage learners to communicate effectively, it is really crucial to find out why some learners are reluctant to speak in language classrooms (MacIntyre, 2007). Such reluctance will make learners lose their interest in learning the language. They will thus notice they are not making any progress in their learning. The fact is that in an EFL environment such as Iran, learners are not exposed to English language outside classroom. There are not ample opportunities for the students to practice their language, and their practice is limited to classroom environment. If they do not take the opportunity to speak in the classroom, there is no other chance for them to practice their language. This will make the learners unable to develop their speaking ability.
What the researcher of the current study has witnessed in language classrooms is that when presented with an opportunity to speak, most language learners are not that eager to take part in classroom discussions. This has always been a concern among language educators, since after all, language learners need to practice in order to learn a foreign language successfully. Otherwise, they will not be able to use what they have learned for communication in real situations. Many language learners complain that they are not good at speaking skill. In fact, this could be a result of their unwillingness to speak and inadequate speaking practice in language classrooms. Nevertheless, to many language learners, learning to speak is much more important than learning to read and write (Ya-Ni, 2007). Clearly, the mastery of speaking skill is a priority for many language learners, and learners often measure their degree of success on the basis of their improvement in speaking ability. Using MacIntyre et al.'s (1998) WTC model as a framework, the current study is expected to shed some light on this concern. The present study is, therefore, one of the first attempts to look into willingness to speak English among Iranian EFL learners with the goal to add to the existing literature on this area and fill the gaps. It is worth mentioning that the present study uses the narrow definition of WTC, which is actually oral communication. As such, the term "Willingness to Speak" (WTS) is introduced to replace WTC.
Previous literature has illustrated that learners' perceived linguistic competence and lack of anxiety are related to their WTC (Baker & MacIntyre, 2000;Cetinkaya, 2005;Hashimoto, 2002;MacIntyre & Charos, 1996;Weaver, 2010;Yashima, 2002). It has also been recognized that students' motivation is related to their WTC (Baker & MacIntyre, 2000Cetinkaya, 2005;Peng, 2007;Yashima, 2002;Yu, 2009). With regard to the importance of language learning anxiety and language learning motivation, MacIntyre (2007) contends that these two variables should be taken into consideration in WTC research as they are of paramount importance among the factors affiliated to WTC, and are proved to be most immediately responsible for one's WTC. In terms of gender, research has shown the influence of gender on WTC (Donovan & MacIntyre, 2004). However, there is both lack of research in this regard as well as inconsistency in the findings.
There has been plenty of research conducted on the antecedents on willingness to communicate construct and the interrelationships among the variables that are thought to have an impact on individuals' degree of willingness to communicate. Burroughs, Marie, and McCroskey (2003), too, examined the relationships of communication apprehension and self-perceived communication competence with willingness to communicate among Micronesian and American students who were studying in Micronesia. The participants were 131 undergraduate college students. Results revealed that Micronesian students reported that they had less communication competence than the American peers. Moreover, they were seen to be more apprehensive in communication and hence less willing to communicate than the American counterparts. Results confirm the existence of a positive correlation between self-perceived communication competence and willingness to communicate, and a negative correlation between communication apprehension and willingness to communicate.
In a study conducted in an EFL setting, Cetinkaya (2005) tried to examine Turkish college students' degree of willingness to communicate whenever they had the opportunity to communicate. The results of the study demonstrated that students were to some extent willing to communicate in English, and that they were more willing to communicate in pairs or small groups rather than large groups. Furthermore, a negative correlation was found between students' perception of their language proficiency and their communication apprehension. In other words, the students believed that the higher their language proficiency, the lower their communication anxiety will be. It was found that willingness to communicate is directly affected by perceived communicative competence and indirectly influenced by motivation via self-confidence. Hodis (2009) tried to analyze how willingness to communicate, communication apprehension, and self-perceived communication competence change over time. Employing a longitudinal design for the study and asking 1,187 undergraduate university students to participate in the study, the researcher illustrated that when learners' self-perceived communication competence increased, their willingness to communicate increased as well. Moreover, a negative correlation was proved to exist between communication apprehension and self-perceived communication competence. Finally, it was concluded that throughout the semester, students' willingness to communicate and self-perceived communication competence increased, while their communication apprehension decreased. Yu (2009) examined willingness to communicate among Chinese EFL learners, and how it is related to communication apprehension and self-perceived communicative competence. Using a quantitative research method and administering a battery of questionnaires to a group of 235 Chinese university students, the researcher sought the objectives of the study. Results confirmed that willingness to communicate, communication apprehension, and self-perceived communicative competence are interrelated.
In another recent study, which was conducted in the Iranian EFL setting, Alemi, Daftarifard, and Pashforoosh (2011) found out that learners' anxiety did not affect their WTC and the way they will decide whether or not to take part in communication. Such finding is interesting and is in contrast with previous research (e.g. MacIntyre et al., 2003;Yashima, Zenuk-Nishide, & Shimizu, 2004) in which a correlation was found between willingness to communicate, perceived competence and language learning anxiety.
To date, all the research studies conducted on WTC and its related predictors have emphasized the importance of creating willingness to communicate or speak among language learners and how such willingness is influenced by or related to some factors. However, an area of research that is neglected in the literature is how willingness to speak can be extended to a specific language skill such as speaking. Given the fact that the goal of modern language pedagogy is to train language learners who are ultimately able to use the language effectively for communication, it is not clear in the related literature on WTC whether willingness to speak contributes to a better speaking ability among language learners. It is generally believed that reticence in speaking activities prevents learners from making progress in their speaking ability and hence arriving at the outcomes the language course purports to achieve. In fact, language learning is proved to be more effective when learners are made willing to use the language (Swain, 2005). Swain's research highlighted the fact that learners who are passive and reticent in language classes use the language less frequently and hence make slower progress. When they see they are making slow progress, this will negatively influence their self-confidence in speaking activities.
The problem is worse in Asian countries. According to Tsui (1996), as compared to their Western counterparts, Asian learners are perceived more reticent in language classes. Littlewood (2002), too, argues that Asian students are not highly willing to speak in language classes, and that such attitude is not innate, but is a result of growing up in a cultural and educational environment in which teacher is not seen as a facilitator of the learning process but as an authority. The present study, therefore, furthers one's understanding of WTC by making an attempt to extend willingness to speak to speaking skill. Such being the case, the study seeks answers to the following questions: (1) To what extent are Iranian EFL learners willing to speak English in language classrooms?
(2) What is the relationship between Iranian EFL learners' willingness to speak and their language learning anxiety, language learning motivation, and self-perceived speaking ability, and which variable is the best predictor?
(3) How do males and females differ in terms of their willingness to speak?
(4) What is the relationship between Iranian EFL learners' willingness to speak and their speaking ability?

Method
A total number of 156 Iranian EFL learners (50 males and 106 females) preparing for IELTS took part in the study. A number of instruments were used to collect the required data for the study. Willingness to speak was measured by a 27-item questionnaire that sought students' perception and degree of willingness to speak English in different situations in language classrooms. The questionnaire was adopted form Cao and Philp (2006), McCroskey (1992), and Lisa (2006). Language classroom anxiety was measured by the Foreign Language Classroom Anxiety Scale (FLCAS) developed by Horwitz, Horwitz, and Cope (1986). It consists of 33 items which investigate the degree of anxiety that learners experience in foreign language classrooms. Language learning motivation was measured through a 22-item questionnaire developed by Gardner and Lambert (1987). Moreover, following Weaver (2010), learners were asked to rate their speaking ability on a "high, medium, low" scale. The purpose is to find out students' perception of their own speaking ability, as it accounts for one objective of the study. Table 1 illustrates the reliability index of each questionnaire used in the study. Moreover, the participants of the study took part in a speaking interview test to measure their speaking ability and hence meet one of the objectives of the study. To ensure the reliability of the speaking test and the scores obtained, two trained IELTS examiners were recruited to rate the individuals. In order to ensure the reliability and the extent of the agreement between the two judges, inter-rater agreement was measured. The result of Intra-class Correlation Coefficient (ICC) showed perfect agreement between the two interviewers (ICC = 0.856).

Research question one
The first research question of the study sought the extent to which Iranian EFL learners are willing to speak in foreign language classrooms. Table 2 illustrates the results pertaining willingness to speak questionnaire. As can be seen in Table 2, the respondents' mean on the willingness to speak questionnaire is 3.77, which is indicative of the fact that the respondents were willing (but not highly willing) to speak English in language classrooms. The questionnaire measured the learners' willingness to speak on a scale of 1-5 from definitely not willing to definitely willing. The mean achieved in the descriptive analysis of the data reveals that they were more than moderately willing to speak English in class. Such finding is in line with Cetinkaya (2005) who found that the participants in the study were somewhat willing (but not very wiling) to speak in class. The findings of Jackson's (2008, 2009) research, too, show that more than half the participants were willing to communicate in English classes and had positive attitudes toward speaking with others. Item analysis of the questionnaire provides a better picture of learners' willingness to speak. Table 3 demonstrates the results of item analysis in ascending order. Table 3 starts with the items that show learners' least degree of willingness to speak and moves toward those that display learners' greater willingness. Starting from top of the table, one can see that the mean of item 17 (volunteer to speak individually in class) is the lowest (3.32), which means learners showed least degree of willingness to volunteer to speak individually in class, as compared to other items. On the contrary, the item in which students expressed the highest degree of willingness is item 16 (Speaking about a topic I am interested in), whose mean is 4.45. The following is a thorough description of the main results of the data.

Task type
Several items in the willingness to speak questionnaire address the issue of task type. Three items concerned learners' degree of willingness to speak individually in class, speaking in pairs, and speaking in groups. Descriptive statistics of the data show that learners showed less willingness to speak individually in class; however, when it comes to talking in groups (either small or large) or in pairs, they are more willing. A comparison of talking in groups (both small and large) and talking in pairs shows that the mean of talking in large groups (3.58) is less than those of talking in small groups and in pairs (3.96 and 4.16, respectively). This indicates that students are more willing to speak when they are in pairs or small groups rather than large ones. In line with this, Item 9 asks students to express their degree of willingness in terms of giving a lecture in front of class. Students showed less degree of willingness in giving a lecture in front of class, in comparison with other items and other situations. This indicates that learners are not highly willing to present a lecture in front of class where everybody is watching them closely. Cetinkaya (2005) showed that the individuals in his study preferred to speak in dyads or groups rather than speaking individually in front of a large group of people. Cao and Philp (2006), too, contend that running a pair-or group work in class gives individuals greater opportunity to speak than in the case of speaking individually.
The first item of the willingness to speak questionnaire asks the respondents to express their degree of willingness to volunteer to answer when their teacher asks a question. The mean of this item is 3.54, which is relatively low in comparison with other items. In contrast, when it comes to volunteer to participate in classroom discussion (item 7), students show more willingness to speak (x = 3.7). Similarly, Item 25 asks students to express their willingness to speak when no one else is speaking. The mean of this item is 3.61, which, similar to item 1, is relatively low. With respect to asking a question in class (item 5) and presenting their opinions in class (item 6) and helping other students answer a question (item 8), students display similar degree of willingness. The mean of these three items is 3.73, 3.74, and 3.69, respectively.

Seating location
Items 20 and 21 requested students to express their degree of willingness to speak when they are sitting in front of classroom as compared to sitting at the back of the classroom. The mean of item 20 (speaking when I am sitting in the back of the classroom) is 3.44 while that of item 21 (speaking when I am sitting in front of the classroom) is 3.62. This means that students are more willing to speak when they are sitting in front of the classroom.

Topic of discussion and interlocutor
Two items in the questionnaire (items 19 and 26) asked respondents to express their willingness to speak with their classmates or their teacher when they have different opinions. The mean of item 19 (discussing a topic with my friends when our opinions are different) is 3.72, which is lower than that of item 26 (discussing a topic with my teacher when he or she has a different opinion), which is 3.45. This is indicative of the fact that the respondents are more willing to discuss a topic with their classmates when they think differently. Such finding is in line with previous research. In Nagy and Nikolov (2007), students noted topic of discussion as a major factor that made them unwilling to speak English, either because they did not know anything about the issue, or they did not have any opinion about the topic or did not understand the topic. Moreover, 17 percent of the participants in Woodrow's (2006) study referred to topic of discussion as an influential factor that made them reticent in class. Kang (2005) found out that the participants in the study felt more secure while discussing a familiar topic. He argued that when the learners were discussing a topic about which they had some background knowledge, they felt more secure and comfortable as a result of having some ideas to discuss, as well as having no fear of comprehension problems.

Fear of negative evaluation
Item 22 of the questionnaire requires the respondents to evaluate their degree of willingness to speak when they know their speaking is graded. The mean of this item, according to Table 2 is 3.8. More than 33% of the respondents are unwilling to speak or moderately willing to speak when they are aware that their speaking will be graded. To support such findings, MacIntyre et al. (2001) managed to show that learners' willingness to communicate decreases if they know they are being formally evaluated. The Iranian EFL learners in Jamshidnejad's (2010) study, too, reported that they are over-concerned with making mistakes and being negatively evaluated by others, and that the fear of making mistakes increases the chance of losing one's face.

Fear of correctness of speech
However, in item 27 (speaking when I am sure that my answer is correct), the mean is higher (4.23) which shows learners are more willing to speak when they are confident of the correctness of their response rather than the time they know their speaking is being graded by their teacher. Above 80% of the respondents expressed their willingness to speak when they are sure that their answer is correct. Similar to this is Lisa (2006) who found out that 85 percent of the individuals in the study are more willing to talk only if they are sure their answer is correct.

Effect of topic of discussion
Five items in the questionnaire (items 15, 16, 18, 23, and 24) discuss the issue of the influence of the topic of discussion on the learners' degree of willingness to speak. Item 15 concerns topic familiarity and asks the respondents to express their willingness when they discuss a topic they are familiar with. This item receives a high mean value of 4.28, which indicates that learners are most willing to speak about a topic they have familiarity with. Similarly, item 16, which has the highest mean value (4.45), addresses the issue of topic interest. The learners seem to have the greatest degree of willingness to speak about a topic they are interested in.

Interlocutor effect
Another category the researcher arrived at was the effect of interlocutor on the learners' degree of willingness to speak. Items 11 and 12 address the issue of the sex of the interlocutor. They get the students to show their level of willingness to speak with somebody who is of opposite sex or somebody who is of the same sex as them. The mean of item 12 (speaking to a classmate who is of the opposite sex) is 3.68, while that of item 11 (speaking to a classmate who is of the same sex) is 3.76. This proves that the sex of interlocutor is relatively important to the students and they prefer to talk to somebody who is of their sex rather than somebody who is of the opposite sex.
Items 13 and 14 deal with the age of the interlocutor. These two items ask the respondents to express their degree of willingness to speak with a person who is older or younger than them. Results show that the mean of item 13 (speak to a classmate who is older than me) is 3.8, while the mean of item 14 (speak to a classmate who is younger than me) is 3.68. The mean difference shows that students are somewhat more willing to speak with a classmate who is older than them. In fact, 59% of the individuals expressed their willingness to speak with a classmate who is younger than them, while more than 64% of them preferred to speak with an older classmate. The important role of the interlocutors' characteristics such as age and gender in one's willingness to speak is supported by Jamshidnejad (2010) who argues that the presence of an opposite sex and age in a conversation may motivate the learner to maintain an acceptable level of accuracy probably in order to demonstrate one's higher proficiency level. In other words, the presence of an interlocutor from an opposite gender pushes the speaker to be more careful with the accuracy of the sentences one makes.

Research question two
The second research question of the study comprises several parts, which are dealt with one by one below.
One part of the second research question of the study seeks the relationship between learners' willingness to speak and their language learning anxiety. As stated earlier in Chapter Three, learners' language learning anxiety was examined through Foreign Language Classroom Anxiety Scale (FLCAS). Table 4 demonstrates the descriptive results.
As suggested by Horwitz et al. (1986), to determine a student's anxiety level, it is necessary to add up their responses to all the questions, remembering to first reverse-score the items that need reverse-scoring, then divide the total by 33 (the total number of questions). Students with averages less than 3 should be considered not very anxious, while students with averages between 3 and 3.99 are fairly anxious. Students who average 4 and above are probably highly anxious, and it is necessary to begin to work with them to find a way to reduce their anxiety. As can be seen in Table 5, the majority of the respondents (86.5%) reported they were not very anxious in the foreign language classroom. This is while, 13.5 percent of individuals reported they were slightly anxious. Surprisingly, no single student reported to be anxious while in language classroom.
In order to meet the objective of the study, which was determining the degree of relationship between learners' willingness to speak and their language learning anxiety, Pearson product moment correlation was run. The correlation results are tabulated below. Table 5 depicts that a negative correlation (r = −.37) exists between willingness to speak and language learning anxiety among the participants of the study. The correlation is significant at the level of .01. This means that as learner' foreign language classroom anxiety decreases, their degree of willingness to speak increases, and vice versa. Previous research in the Iranian EFL context also showed that the existence of anxiety in language classrooms is a debilitating factor that can have a negative influence on the learners' performance (e.g. Rezazadeh & Tavakoli, 2009;Riasati, 2011;Rouhani, 2008). It goes without saying that creating a less threatening environment in class puts the learners at ease and when they feel relaxed, they will be more willing to participate in classroom discussions and activities

Willingness to speak and language learning motivation
The second part of the second research question of the study looks into the relationship between willingness to speak and language learning motivation. As stated earlier in Chapter Three, learners' degree of motivation was assessed using a motivation questionnaire. Table 6 demonstrates the descriptive statistics on the learners' motivation degree.
As can be seen in Table 6, a considerable number of respondents (54.5%) reported that they were fairly motivated to learn the language, while 43.6 percent of them were slightly motivated and a negligible number of 3 percent were not motivated to learn the language. In order to determine the relationship between learners' degree of willingness to speak and their language learning motivation, Person product moment correlation was run. The correlation result is presented in Table 7.
As Table 7 demonstrates, there is a positive correlation (r = .22) between willingness to speak and language learning motivation. The correlation is significant at the level of .01. This is indicative of the fact that these two variables are positively correlated, and that learners' willingness to speak increases as their motivation to learn the language rises. Research in the Iranian EFL context (e.g. Chalak & Kassaian, 2010) has also proved that the learners benefit from a fairly high degree of motivation to learn the language, and that the majority of learners are interested in learning language as it allows them to speak and communicate with English people.
The result of Hashimoto's (2002) study is in agreement with that of the present study. Hashimoto found a positive correlation between motivation and willingness to communicate, indicating that the students who benefit from a higher degree of motivation for learning the language are more willing to communicate and thus use the language more frequently in class.
The third part of the second research question of the study aimed at determining the relationship between learners' willingness to speak English in language classroom and their self-perceived speaking ability. As mentioned earlier, the participants of the study rated their speaking ability on a scale of high, medium, and low. As such, to measure the relationship between these two variables, Spearman correlation was used. The results are demonstrated in Table 8.
As can be seen in Table 9, there exists a positive correlation (r = .19) between willingness to speak and self-perceived speaking ability. This correlation value is significant at .05 level. This indicates  that the learners who have a higher perception of their speaking ability or rate their speaking ability higher have a greater willingness to speak in language classrooms.
For further investigation, the degree of difference between learners' willingness to speak in terms of their self-perceived speaking ability was determined. The purpose was to see if learners' degree of willingness to speak English in language classrooms differs in terms of their self-rated speaking ability. As mentioned above, there were three groups of learners in terms of their self-rated speaking ability (Low, Medium, and High). As such, in order to test their differences on willingness to speak, analysis of variance (ANOVA) was used. This choice is supported by Coolidge (2006), who sees ANOVA as a powerful and robust statistical test that examines the differences between three or more groups' means. Table 9 shows the descriptive statistics of learners in terms of their self-rated speaking ability. Table 10 shows the result of ANOVA.
Results of ANOVA, as can be seen in Table 10, show that willingness to speak differs significantly across three groups based on their self-rated speaking ability (F(2, 153) = 3.34, p ≤ .05). This indicates that the three groups who were divided in terms of their self-rated speaking ability differ significantly on their degree of willingness to speak English in language classrooms. However, in order Table 7

. Correlation between WTS and motivation
Note: WTS = Willingness to speak. *Correlation is significant at the 0.01 level (two-tailed).

Table 8. Correlation between WTS and self-perceived speaking ability
Note: WTS = Willingness to speak; SPSA = Self-perceived speaking ability.
*Correlation is significant at the 0.05 level (two-tailed).  to see which of these groups differ from each other, a post hoc test of multiple comparisons was run.

WTS SPSA
The results are shown in Table 11.
As can be seen in Table 11, post hoc analysis using Tukey shows no statistically significant difference exists between learners who rated their speaking ability as high (Mean = 4.06, SD = .51) and those who perceived it as medium (Mean = 3.74, SD = .58). A similar finding can be detected between those learners who perceived their speaking ability as low (Mean = 3.61, SD = .71) and those who rated it as medium (Mean = 3.74, SD = .58). However, a significant difference can be observed between the learners who perceived their speaking ability as high (Mean = 4.06, SD = .51) and those who rated it as low (Mean = 3.61, SD = .71). Results show that those who rated their speaking ability high are apparently more willing to speak English in language classrooms than those learners who perceived their speaking ability as low. Such finding can be explained in terms of the fact that as suggested by previous studies (e.g. Cetinkaya, 2005;Hashimoto, 2002;MacIntyre, Baker, Clément, & Donovan, 2002), an individual's higher perception of his speaking ability will increase one's selfconfidence and reduce anxiety. These will, in turn, enhance the learner's willingness to speak. Kim (2004) found out that a decrease in one's level of anxiety, combined with a high perception of communication ability, makes a person self-confident and hence more willing to communicate.
Another part of the second research question of the study aims at figuring out which of the independent variables, that is learners' language learning anxiety, language learning motivation, and their self-rated speaking ability best predicts the dependent variable of the study, namely learners' degree of willingness to speak English in language classrooms. To this end, Multiple Regression Analysis was conducted on the data with language learning anxiety, language learning motivation, and their self-rated speaking ability as predictor variables and willingness to speak as criterion variable. Results are illustrated in Tables 12 and 13.  As can be seen in the regression analysis Table 13, a significant model is obtained (F [3, 152] = 9.92, p ≤ .001), which implies one or more variables are related to willingness to speak. This model indicates that about 16 percent of the variation in willingness to speak English is accounted for by language learning anxiety, motivation and self-perceived speaking ability. Moreover, the results of the regression analysis show that anxiety is the best predictor of willingness to speak (B = −.31, p ≤ .001).

Research question three
The third research question of the present study aims to find out if males and females are different with regard to their degree of willingness to speak. In other words, the question seeks to figure out if males or females are more willing to speak in classrooms. To this end, independent sample t-test was run to investigate the difference between the two groups. Results are illustrated in Tables 14  and 15.
The willingness to speak mean score of males is 3.73 with standard deviation of .48. For females, on the other hand, the mean score is 3.79 with the standard deviation of .64. Analysis of the independent sample t-test investigated if there was a statistically significant difference in willingness to speak among males and females. Table 15 shows the result.
As can be seen, there was no statistically significant difference between males (M = 3.73, SD = .48) and females (M = 3.79, SD = .64) in terms of willingness to speak (t(154) = −.564, p = .557). Put it simply, males and females do not differ significantly with respect to their degree of willingness to speak English in language classrooms. Donovan and MacIntyre (2004) found no significant difference between junior high school male and female students in terms of willingness to speak. However, MacIntyre et al. (2002) observed a significant difference between boys and girls in terms of willingness to communicate. They concluded that girls show an increase in the degree of willingness to communicate as they move from one grade to another, while the degree of willingness to communicate among boys remains unchanged.

Research question four
The forth research question of the study seeks the relationship between leaners' willingness to speak and their speaking ability.
As stated earlier, students' speaking performance was measured through IELTS speaking tests and rated by two certified IELTS examiners. The two raters marked the participants' speaking performance independently. The two speaking scores obtained were then averaged in order to come up with each participant's speaking score. Tables 16 and 17 demonstrate frequency and descriptive statistics of the IELTS speaking test scores.
As can be seen, the participants' IELTS speaking scores range from 4.00 to 8.50 and the overall mean score of the speaking test is 6.30 with the standard deviation of .80. Based on the mean score obtained here (M = 6.30) and the interpretation of the scores which was discussed earlier, participants of the current study can be categorized as competent readers. Based on the interpretations, these kinds of students have generally effective command of the language despite some inaccuracies and misunderstandings. They can use and understand fairly complex language, particularly in familiar situations.
To seek the relationship between these two variables, Pearson product correlation was run, the result of which is shown in Table 18.
As can be seen, there is a strong positive correlation (r = .64) between willingness to speak and speaking ability. The correlation value is significant at .01 level, indicating that those learners who got a higher score in the willingness to speak questionnaire scored higher on the speaking test. Other research (e.g. McCroskey, 1992;McCroskey & Richmond, 1987) managed to show that the learners who got a higher score on willingness to communicate questionnaire had more active participation in classroom activities than those who received a low score.

Conclusion and implications
Quantitative findings of the study showed that there is a relationship between willingness to speak and language learning anxiety, language learning motivation, and self-perceived speaking ability. It was also shown that more willingness to speak leads to a better speaking ability. Further, findings revealed a number of factors influencing willingness to speak English in language classrooms. With this in mind, a classroom WTS model is created to account for the complexities underlying this phenomenon in language classrooms. The model is presented in Figure 1.
The proposed EFL classroom WTS model represents the factors that are related to or influence willingness to speak English among Iranian EFL learners. As discussed earlier, quantitative findings of the study demonstrated a relationship between willingness to speak and language learning anxiety, language learning motivation, and perceived speaking competence (ability). Findings also revealed a number of factors that influence willingness to speak. The factors that were identified as having an impact on their willingness to speak are divided into two larger categories, i.e. environmental (situational) factors and individual factors, as illustrated in the model above. The environmental factors include factors that exist in the classroom environment and influence the learners' degree of willingness to speak. These include task type, topic, interlocutor, teacher, classroom atmosphere, and seating arrangement. The individual factors, on the other hand, refer to the individuals' personal characteristics and include learners' personality, self-confidence, the degree of opportunity they have in language classes, fear of evaluation, and fear of correctness of their speech. It was shown that these factors influence learners' willingness to speak English in language classrooms. Finally, when a higher degree of willingness to speak is observed among language learners, students tend to participate more actively in classroom activities and discussions (Cao, 2009). A novel element in the study is the attempt that was made to make a link between willingness to speak and speaking performance. Findings showed that learners who benefit from a higher degree of willingness to speak achieved better scores on the speaking test, which can be attributed to the greater degree of practice and more language use they had in language classes. As shown in the literature (e.g. Cao & Philp, 2006;MacIntyre et al., 2001), a higher degree of WTS facilitates language learning process and leads to more practice and language use. However, there are a number of factors arrived at in the present study which do not exist in MacIntyre et al.'s WTC model. Since the current study focused on foreign language classroom setting, the classroom environmental factors identified as influencing willingness to speak are classroom-specific, such as task type, seating location, teacher, classroom atmosphere, and perceived speaking opportunity. Such factors are not identified in MacIntyre et al.'s model which is applicable to general L2 conversation contexts (Cao, 2009). As such, it can be claimed that the current study is distinctive in that it has produced a new WTS model that suits foreign language classroom settings. It is hoped that this study would shed more light into the complex and dynamic nature of WTC. A comparison between and MacIntyre et al. 's (1998) heuristic pyramid WTC model and the empirically grounded WTS model in the present study reveals that unlike MacIntyre et al.'s model that suits all conversational contexts, the model proposed in the study applies to language classroom contexts. For this reason, it is referred to as WTS model, which is actually a new term in this area of research and could thus be regarded as a contribution to the existing body of knowledge.
As willingness to speak is shown to be a predictor of better speaking performance, it seems necessary for language teachers to identify the possible ways through which learners' degree of willingness to communicate can be enhanced. The results of the study suggest that willingness to speak is influenced by several factors. Thus, teachers should not attribute students' reticence to one single factor such as personality or shyness. They need to be aware of the factors that could encourage or discourage communication among learners. Therefore, it is really necessary for language teachers to promote factors that facilitate communication and remove those that hinder communication. They should also be mindful of the interactions between variables while planning learning activities (Cao & Philp, 2006).