The effect of emotional scaffolding on language achievement and willingness to communicate by providing recast

Abstract This research aimed to examine the effect of emotional scaffolding provided by recast on developing elementary EFL high and low anxious learners’ language achievement and their willingness to communicate. Through cluster sampling, the participants (N = 56) whose scores lied ±1 SD in a Key English Test, who were classified as high or low anxious in terms of Foreign Language Classroom Anxiety Scale questionnaire, and who had no significant difference in emotional intelligence in Bar-On Quotient Inventory were selected. Then, the learners were divided into four groups: two experimental and two control groups encompassing learners with high and low levels of anxiety. The difference between experimental and control groups was in teaching emotional intelligence and the teacher’s emotional behavior during providing recasts for the experimental groups. The performances of all groups were also corrected by recasts. Bar-On Quotient Inventory and Willingness to Communicate questionnaire were also administered in the beginning and last sessions to show the development of their emotional intelligence and their tendency to communicate. After fourteen 120-minute sessions, data were analyzed by SPSS. Findings showed that emotional scaffolding had a positively significant effect on students’ learning with high or low levels of anxiety. Meanwhile, learners of both levels of anxiety showed a significantly positive tendency to communicate. Thus, to decrease anxiety especially through receiving recasts, emotional scaffolding should be incorporated in EFL curriculum.


ABOUT THE AUTHOR
Sayyed Mohammad Alavi is a Full Professor in Applied Linguistics at the University of Tehran, Faculty of Foreign Languages and Literature. He received his PhD from Lancaster University in UK in 1997. He published on challenging topics of language testing and assessment, language teaching, and language teacher education in prestigious national and international journals. His current research interests are language assessment, teacher education, and language teaching. Fatemeh Esmaeilifard is a PhD student in TEFL at the University of Tehran, Kish International Campus. She has published some articles in the field of cognitive and metacognitive strategies and their effects on structural development. She is a lecturer at different universities in Iran, Kish Island. Her current interests are in the issues such as English language teaching, emotional strategies, and teacher education.

PUBLIC INTEREST STATEMENT
Language learning processes in a foreign language learning context are believed to be affected by cognitive, affective, and social factors. When a foreign language learning context is viewed as an affective context, some factors such as language learning anxiety, emotional behaviors of learners and teachers, etc. are investigated to explore language learning problems and then suggest a tentative solution to these problems. More specifically, if a classroom is viewed as an emotional setting, learners are expected to exchange their feelings. Therefore, negative and positive emotions may play an integral role in an individual's language achievement. These emotions may also affect learners' willingness to communicate. Having considered the need to investigate the role of emotions in classroom contexts, this study aimed to examine the effect of emotional scaffolding on the language achievement and willingness to communicate through providing recast for learners with high or low levels of anxiety.

Introduction
The polarity of the two main concepts (i.e., cognition and emotion) has been argued for many years. On the one hand, cognition emphasizes on the rational way of thinking, analyzing, logic, but on the other hand, emotion deals with affection and sentimental behaviors. While there is a contrast between thoughts and feelings (Labouvie-Vief, 2015), it can be argued that these two entities interact as well as integrate (Pessoa, 2008).
Cognitive abilities were considered as the only criteria for the individual differences and human success for many years. Therefore, a variety of Intelligence Quotient (IQ) tests were constructed to evaluate intellectual potentials of individuals in specific domain and their intelligence in general ones. However, Salovey and Mayer (1990) based on Gardner's multiple intelligences argued for emotional intelligence. They contended that emotional intelligence can be connected to social intelligence and can affect individual's cognitive abilities. On the other hand, Goleman (1995) stepped forward and claimed that just 20% of individual's intelligence can be indicated by his or her score in an IQ test and 80% is devoted to emotional intelligence. As a result, emotional intelligence plays a vital role in human success (Neubauer & Freudenthaler, 2005).
It seems that research on the concepts such as "affect" and "emotion" has a growing interest during these last two decades (Fredrickson & Cohen, 2008;Linnenbrink, 2007;Pekrun & Schutz, 2007;Zembylas, 2007). Although these notions were posed with the presence of Humanism in the 1970s, no sufficient effort has been done in education (Arnold, 2011). Recently, the term "emotional scaffolding" has drawn the attention of some researchers (Rosiek, 2003;Rosiek & Beghetto, 2009).
It is believed that teachers use a series of strategies to arouse learners' emotional responses to a subject matter by emotional scaffolding (Rosiek, 2003). The most important factor in emotional scaffolding is learners' imaginations. To elicit this factor, as Rosiek and Beghetto (2009) suggested, a teacher can employ metaphors, narrations, emotional examples, and visual representations in order to make a connection between a boring subject matter and learners' emotions. To have a better understanding of such an issue, the role of employing emotional scaffolding in correcting learners' errors during teaching English was investigated in the present study. The learners' errors were also corrected by recast as an implicit corrective feedback selected from Lyster and Ranta (1997) corrective feedback taxonomy.

Theoretical background behind emotion and emotional intelligence
Attention to emotion and affect in educational contexts is not a new phenomenon. It is widely argued that one of the objectives of Humanism is examining the role of socioaffective factors in the process of whole person development. The emphasis on the psychological approach is clearly reflected in Community Language Learning (CLL), Total Physical Response (TPR), and Suggestopedia (Arnold, 1999). The supporters of Humanism believe that anxiety is a barrier in learning and all efforts of a teacher should be to decrease a stressful situation of learning and this can be achieved by giving the affective domain the central role in the learning process (Arnold, 2011).
On the other hand, Gardner (1983) changed the conventional concept of intelligence as a single abstract mathematical intelligence to multiple intelligences and argued that human beings have eight types of intelligence as naturalist, logical-mathematical, musical-rhythmic, existential, bodily-kinesthetic, verbal-linguistic, visual-spatial, and personal intelligence including both interpersonal and intrapersonal intelligences (Gardner, 1983(Gardner, , 2006. Gardner (1999) introduced interpersonal and intrapersonal intelligences under the personal intelligence. However, it is believed that the combination of interpersonal and intrapersonal intelligences can be defined in terms of "emotional intelligence" ( Bar-On et al., 2000;Bay & Lim, 2006;Furnham, 2012;Salovey & Mayer, 1990). In essence, one part of the personal intelligence is devoted to the concept of feelings and emotions what play an important role in emotional intelligence. Therefore, "emotional intelligence is a subset of Gardner's personal intelligences" (Salovey & Mayer, 1990, p. 189).
The early model of emotional intelligence (Salovey & Mayer, 1990) classifies emotional intelligence as a subcategory of social intelligence. It is considered as an ability subcategory by which an individual can accurately distinguish between different types of emotions, control both his or her own emotions and those of the others, and use the earned information for his or activities and thinking. It is worth mentioning that the notion of emotional intelligence dates back to the concept of social intelligence described by Thorndike (1920) who claimed that the secret to have a wisely social relation is to perceive and manage people around us.
However, this ability model proposed by Salovey and Mayer (1990) was later revised and classified into four abilities: a) perceiving emotions: distinguishing accurate from inaccurate emotions and expressing them correctly, being aware of context and culture as two factors in changing emotions, understanding emotions by visual, musical and environmental clues, understanding people's emotions through their verbal and facial expressions and body language b) facilitating thought utilizing emotions: choosing problems which can be solved by using emotions as facilitators of cognition, using mood to produce various cognitive states, giving priority to thought on the basis of the current emotions, and arousing emotions to facilitate judgement and using other people experiences and c) understanding emotions: evaluating emotions in terms of cultures, being aware of the possibility of changing emotions, mixed emotions and their relations, the difference between emotion and mood, the result of the use of emotions and predicting the way may people feel in future in specific conditions, and d) managing emotions: monitoring and managing both his or her own emotions and others' to achieve the objectives, being aware of different emotional strategies to delete, decrease, strengthen emotional reactions, controlling emotional responses to extract the logic behind them, giving attention to desired and undesirable emotions and the way to express them (Mayer et al., 2016). Goleman (1998) developed a performance model of emotional intelligence on the basis of emotional intelligence competencies, i.e., an ability which can be learned and its consequence is flourishing in the workplace. These emotional competencies are underlaid by emotional intelligence. Therefore, to succeed at work, the presence of emotional intelligence is crucial but it is not adequate and "our emotional competence shows how much of that potential we have realized by learning and mastering skills and translating intelligence into on-the-job capabilities" (Goleman, 2001, p. 27). The framework consists of four clusters of emotional intelligence abilities, i.e., (a) self-awareness, (b) selfmanagement, (c) social awareness, and (d) relationship management (Goleman, 2001). Bar-On (2006) also developed competencies model of emotional intelligence and emphasized emotional-social intelligence. In his model, he considered the interrelation between Gardner's theory and Saarni's (1990) description of emotional competence. In fact, Bar-On' s emotional-social intelligence consists of "a number of intrapersonal and interpersonal competencies, skills and facilitators that combine to determine effective human behavior" ( Bar-On, 2006, p. 14). More specifically, non-cognitive model of intelligence that Bar-On suggests encompasses five categories and 15 subcategories related to personality characteristics: 1) self-expression and self-understanding (self-regard, emotional self-awareness, assertiveness, independence, and self-actualization), 2) adaptability (problem-solving, reality-testing, and flexibility) 3) interpersonal skills (empathy, social responsibility, and interpersonal relationship), 4) general mood (optimism and happiness), and 5) stress and emotion management (stress tolerance and impulse control) ( Bar-On, 2006;Bar-On et al., 2000).

Emotion in education
Nowadays, emotions have been considered as factors dealing with leaners' and teachers' achievements in education. It is argued that emotions can develop learners' and teachers' personalities, and affect their health and interest (Pekrun & Schutz, 2007). Zembylas (2007) claimed that before 1990, emotions as an individual phenomenon were the focus on research but at the beginning of 1990s, attention has been paid to social aspects of emotions. Therefore, emotions are built through the interaction between teachers and learners and can be influenced socially, culturally, and politically (Zembylas, 2007). Pekrun (2000) classified six examples of emotions in education: positive academic emotions are joy and hope and negative ones are anxiety, anger, boredom, and hopelessness. There are also other emotions that depend on the context and culture. Emotions such as pride or shame change from one culture to anther (Pekrun & Schutz, 2007).
Emotions must be valued for many reasons. Some of them are assisting in describing both teaching and learning experiences and disclosing the reasonable thinking behind preventing or leaving learning and teaching possibilities (Meyer & Turner, 2007). DeCuir-Gunby and Williams (2007) considered emotions as an internal state related to physical and psychological mechanisms. For instance, emotions are engaged in physiological changes such as "body temperature, heart rate, blood pressure, and oxygen saturation of the blood" (Sutton, 2007, p. 261). However, many elements can identify the type, duration, and intense of emotions, for instance: age, race, class, and gender (Hochschild, 2012). Therefore, emotions do not occur in isolation but in collaboration with other elements. In fact, they are responses to environment and they are as a stimulus for doing affairs (Massey, 2002).
Negative affects impair attention and unsolved problems. For example, if learners lose their face in front of other learners, this negative emotion can hinder their learning (Boekaerts, 2007). Nevertheless, they may experience the same emotion in different contexts. In fact, their current goals can identify that the aroused emotion is either negative or positive. As a result, learners who in advance set themselves reward goals and receive negative affect will be unsuccessful in learning. In contrast, the learners with mastery goals will more likely defeat negative affect and achieve their learning goal (Boekaerts, 2007). Diefenbach et al. (2008) stated that there is a significant relationship between emotions and health. Emotions can even strengthen immune system and can increase life expectancy and longevity in individuals (Fredrickson & Cohen, 2008;Kemeny & Shestyuk, 2008). Meanwhile, the positive emotions direct individuals toward holistic thinking rather than details (Fredrickson & Cohen, 2008). Furthermore, in some studies (Charles et al., 2003;Dewhurst & Parry, 2000), it was shown that negative items arousing negative emotions were recalled in larger numbers in comparison to positive ones.
Teachers were recommended applying coping strategies and regulating their emotions in terms of a situation in which a problem happens (Cubukcu, 2013). Coping strategies help them strike a balance among their feelings (Cubukcu, 2013). The result of teaching becomes effective if they use multiple coping strategies to remove classroom problems (Davis et al., 2008). Davis et al. (2008) posed three coping strategies: task-focused processes, regaining task focus, and emotion-focused processes. Task-focused processes emphasize the strategies used to complete a task successfully e.g., management of time. In emotion-focused processes, the attention is focused on both individuals and the emotion made by the task rather than the task itself. However, when attention returns to a task itself, regaining task focus is applied. Hochschild (2012) also introduced the term "emotional labor" and defined it as monitoring emotions in order to make gesture or facial expressions.
In favor of the fundamental effect of emotion on the process of cognition, Tyng, Amin, Saad, and Malik (2017, p. 1) stated that the impact of emotion can be enormously shown in the process of attention particularly in "modulating the selectivity of attention as well as motivating action and behavior." Then, emotions can play a vital role in learners' attention to a piece of information and retrieve it. They believed that students experience different negative or positive emotions during their learning, taking test, and doing their assignments, deadlines, problem-solving and reasoning such as anxiety, fatigue, disappointment, happiness and so on. These emotions can either enhance or impede learning processes in terms of their duration and intensity. To exemplify, they reported the results of some research indicating the mild amount of stress. Their results revealed that mild stress can facilitate the cognitive and learning process but its excessive amount can stop it (Tyng et al., 2017).
Studies in the field of emotions indicated that there is no sufficient research on learners' and teachers' emotions although attention to this notion has a growing interest during these two last decades (Fredrickson & Cohen, 2008;Linnenbrink, 2007;Pekrun & Schutz, 2007;Zembylas, 2007). Even if there have been investigations in this field, their designs have been limited to the laboratory situations rather than real educational settings (Pekrun et al., 2002;Pekrun & Schutz, 2007). The focus of this research was on the learners' emotions.

Willingness to communicate
The concept of willingness to communicate (WTC) as a situational variable was introduced by MacIntyre et al. (1998). Considering learners in L2 classes indicated that some learners regardless their linguistic knowledge were more eager to communicate in the second language (MacIntyre et al., 1998). They believed that six variables can affect willingness to communicate: communication behavior, intentions, situation, motivation, affective and cognitive context, and interpersonal and intrapersonal context. The relationship between emotional intelligence and the learners' willingness to communicate encouraged the present researcher to consider this ability in her research.

Recast as a corrective feedback
Corrective feedback, on the one hand, deals with Schmidt's noticing hypothesis and on the other hand, is based on Long's interaction hypothesis (Carpenter et al., 2006). According to noticing hypothesis, no learning occurs without some extent of noticing to the form of input (Schmidt, 1990) and corrective feedback can be considered as a tool to trigger noticing and change input to uptake (El Tatawy, 2002). In fact, corrective feedback encourages learners to notice the gap existing between their interlanguage and the target language (Schmidt, 1994).
Corrective feedback can be defined as "information provided to learners about the ill-formedness of their L2 production" (Loewen, 2012, p. 24). Lyster and Ranta (1997) classified six different types of corrective feedback as following: explicit correction, recast, metalinguistic feedback, elicitation, repetition, and clarification requests. Through explicit correlation, a teacher explicitly shows the errors' places and then corrects them. Recast is a repetition of the learners' utterances whose errors are removed while in metalinguistic feedback, a teacher asks some questions in order to draw learners' attention to the rules of target language. In elicitation, a teacher applies some techniques to elicit the correct answer e.g., by making a pause just beside errors and allowing the students to complete the utterance in its correct form. Teachers also correct the errors through the repetition of the learners' utterances with rising intonation. Clarification requests indicate that learners' utterances require to be clearer. A teacher makes use of a clarification request by saying the phrases such as "pardon me". In this study, the focus was on recast as an implicit corrective feedback.
When a teacher reformulates the non-target like utterances of learners, in fact, he/she corrects errors via recasts (Sheen, 2008). Of course, in this case, the meaning and structure of the main utterances do not change (Sheen, 2006). Recasts can be effective in learning and acquisition, if they have these characteristics: 1) focus on the structures suitable for current levels of learners 2) arouse noticing in learners 3) increase learners' knowledge linguistically 4) be accompanied by repetition (Sheen, 2008). Sheen (2006) claimed that the process of error correction by recasts can be done in two ways, that is, in either a full or partial structure. If a teacher repeats the learners' original utterance entirely while its error is removed, he or she uses a full recast. But if this correction is done partially and the teacher focuses on the wrong segment, a partial recast is employed (Sheen, 2006). Many researchers (e.g., Hulstijn, 2005;Mackey & Philp, 1998;Nicholas et al., 2001;Sheen, 2010) contended that recasts are implicit corrective feedback, because there is no explanation of rules in recast (Li, 2013). However, if learners understand correction as an evident process, recasts become explicit (Ellis et al., 2006).
Recasts can be simple or complex. Complex recasts encompass more than one linguistic feature in a communicative course (Mackey & Philp, 1998). According to Sheen (2006), an intensive recast occurs in a particular linguistic item corrected again and again. Sheen (2006, pp. 365-6) also mentioned criteria for the use of a recast as following: 1) recasts are a discourse move that is interactionally identifiable at the level of one turn; 2) recasts can arise in either meaning or form focused interaction; 3) recasts can be more or less implicit/explicit, depending on their form; 4) recasts entail reformulations of one or more errors in a learner's utterance by means of three kinds of repetitions: (a) complete, (b) partial or (c) expanded.
To achieve the aim of this research these null hypotheses were posed: H01. There is no statistically significant difference between the effect of teaching English by emotional scaffolding and traditional way of teaching through providing recasts on developing Iranian elementary EFL high and low anxious learners' language achievement.
H02. There is no statistically significant difference between the effect of teaching English by emotional scaffolding and traditional way of teaching through providing recasts on developing Iranian elementary EFL high and low anxious learners' willingness to communicate.

Method
This research is quasi-experimental. A pre-test and post-test treatment design was employed to evaluate the effect of emotional scaffolding on learners' achievement and their willingness to communicate.

Participants
The present researcher has been one of the English lecturers of the Applied Science and Technology University since 2010. Therefore, this study was more conveniently operationalized in this university for the present researcher. Through cluster sampling, 2 classes were randomly selected. Then, one version of KET was administered to homogenize the learners.
From among 88 students majoring Tourism and Hotel Management, 56 students whose scores in KET (as a means of their English language proficiency's evaluation) were between one standard deviation below and above the mean score, who experienced high or low levels of anxiety in terms of the Foreign Language Classroom Anxiety Scale (FLCAS) questionnaire (as a means of omitting the medium anxious learners and classifying low and high anxious groups in each class) as well as their performances in Bar-On Quotient Inventory (EQ-i) (as a tool for homogenizing the students in terms of their emotional intelligence) were selected as the main sample of this study.
One class incorporated 24 accepted students and another had 32 ones which were randomly classified as control and experimental groups with their own low and high anxious learners. Therefore, this research included four groups: 13 low anxious control group, 11 high anxious control group, 16 low anxious experimental group, and 16 high anxious experimental group. The learners were in elementary level and should pass the English conversation-based courses. The location and the time of holding the classes were the same. However, regardless of the teaching method, the only difference was in the days of holding the classes. The experimental groups' class was on Saturdays while that of the control groups was on Tuesdays in the same university semester. Their age ranges were between 19 and 45 from both genders.

Instruments
The used instruments in this study were as follows: (1) two Key English Test (KET) for level A2 including 56 reading and writing questions (1 hour and 10 minutes), 25 listening questions (30 minutes), and a speaking test (8-10 minutes) in order to homogenize the learners in terms of their language proficiency, to evaluate the learners' language proficiency before the instruction as a pretest and subsequently to compare it with the learners' performances on the second KET as a post test. Both KETs were adopted. The pretest's K-R20 was 0.81 and that of post-test was 0.85.
(2) Bar-On Quotient Inventory (EQ-i) consisting of 90 items in Likert Scales which examines stress management, interpersonal scale, intrapersonal scale, adaptability scale, and general mood scale to identify the learners' emotional intelligence before and after the instruction. The original form of this questionnaire includes 117 items. However, the Persian version of this questionnaire was administered in Iran among the students and was limited to 90 items (Samui, 2003). This 90-item adapted questionnaire was administered among the present sample. Its Cronbach's alpha was 0.80.
(3) the 15-item Willingness to Communicate (WTC) questionnaire adopted from Xie (2011, p. 131) to show the learners' tendency to communicate before and after receiving the treatment. Its Cronbach's alpha was 0.88, and (4) the Foreign Language Classroom Anxiety Scale (FLCAS) questionnaire  validated by Horwitz (1986) and including 33 items in Likert Scales to classify the learners in terms of their anxiety's levels. This questionnaire was adopted. Its Cronbach's alpha was 0.92

Procedure
Study was completed during 1 university semester in Applied science and Technology University, Kish Island branch. Fifty-six elementary students in English majoring Hotel Management and Tourism participated in these conversation-based courses. This study lasted fourteen 120-minute teaching sessions and two 190-minute testing sessions. Each session was held once a week.

Homogenization and pretest sessions
From among 8 conversation-based classes, the head of English department randomly assigned 2 classes to the present researcher who has worked at the university as an English lecturer for 10 years. In the first session, to homogenize the learners of both classes in terms of their language proficiency, KET level A2 was administrated and learners with scores between one standard deviation above and below the mean were selected. The scores of the learners in KET also were considered as a pretest. Therefore, the KET was used both for the homogenization the learners and as a pretest.
To classify the learners of each class into two groups with high and low levels of anxiety, the Foreign Language Classroom Anxiety Scale (FLCAS) questionnaire was administered in each class. According to scoring of FLCAS, the learners whose scores were between 33 to 75 were considered as groups with low levels of anxiety and those with scores above 120 were categorized as high anxious groups. The students who had medium level of anxiety were deleted. Then, Bar-On Quotient Inventory (EQ-i) consisting of 90 items in Likert Scales was administered to identify the learners' strength and weakness in emotional intelligence and their homogeneity in terms of emotional intelligence in general.
After the consideration of the homogeneity among the students of the both classes, the first class was randomly selected as the experimental class and the second one as the control. The high or low anxious learners of each class made the groups of this study. Therefore, in this study, there were 4 groups: (1) the high anxious recast experimental group in the first class, (2) the low anxious recast experimental group in the first class, (3) the high anxious recast control group in the second class, and (4) the low anxious recast control group in the second class. In addition, at the beginning of the second session, all learners were asked to answer the 15-item Willingness to Communicate (WTC) questionnaire.

Training sessions
In the second session, the instructor tried to arouse the learners' emotions about a subject matter in the experimental groups' class. It is believed that "emotional scaffolding engages students' imaginationusing metaphor, visual representations, or narratives of content-in an effort to foster a particular emotional response to academic subject matter" (Rosiek & Beghetto, 2009, p. 176). In order to arouse the learners to speak, the teacher taught the vocabularies of each lesson adapted from the book "Real listening and speaking 1" (Craven, 2008) at the beginning of each session in both classes. That is, both classes received both the same content of the lessons and textbook. After teaching the lesson, the learners of both experimental and control groups were encouraged to give an account of what they learned in the form of a story or a memory. Their errors were also corrected by recasts. However, when they did not utilize the grammatical forms and the speaking strategies taught in terms of their textbook, the teacher asked the questions in such a way to elicit the use of the corresponding rules. This process was repeated from the second to fifteen sessions. The students of the control groups were taught by traditional method and were corrected by recasts without receiving emotional features and emotional scaffolding.
On the other hand, the experimental groups encountered to both general and specific strategies to develop their own emotional intelligence. The instructor also tried to provide recast feedback through emotional scaffolding for the experimental groups with high or low levels of anxiety in all fourteen sessions. In fact, the general strategies (items a, b, c, d, e, h, and i were adapted from Mortiboys, 2005, and items f and g from; Jordan & Le Métais, 1997) related to the teacher's behaviors were listed as following which were practiced in the experimental groups' class: a. To make eye contact with all the learners; b. To memorize the learners' name and addressing them. f. To boost intrapersonal skills in the learners by providing a supportive, safe atmosphere in the class to hearten them to experience the risk of making mistake without being mocked; g. To develop interpersonal skills in the learners by encouraging them to negotiate with the other members of the group to make decisions, solve the problems, and enhance each other's learning (Jordan & Le Métais, 1997); h. To start the session with positive and energetic words, sentences, facial expression, and tone of language such as "I am excited at the prospect of going through this material with you" (Mortiboys, 2005, p. 33) or "I am sure that you will find plenty of relevance and use in today's session and my plan is to make it enjoyable as well as interesting" (p. 33) both in English and Persian languages. These positive words were also used: "fantastic, joy, better, I promise, enjoy, energy, excited, I guarantee, satisfy, delight, definitely, and I am convinced" (Mortiboys, 2005, p. 32); and i. To apply different methods and instruments to teach materials such as whiteboard, PowerPoint, emotional clips, and pen tablets connected to projector.
The specific emotional intelligence strategies were classified and used in this way. In fact, these strategies were intentionally used at the beginning of each session to break the conversation's ice: f. To increase emotional intelligence in learners in pairs or groups, these items were considered on the basis of Mortiboys (2005): • Learners must be in pairs; • One learner speaks and another must listen; • It is possible the listener to speak but he or she must not distract the talker; • The talker should reflect his or her emotions, thinking and; • The listener and talker would change their roles after five minutes; • They say why listening is essential; • They explain what they learned from the members of the group; • They reflect their ideas about the changes they need to have; • They state different ways that make them happier to be both in the class and in the group; • They are encouraged to appreciate the others and use the positive expressions to admire each other such as well done, thank you for . . ., great, excellent, fantastic, and so on; • They must be informed "how their behavior is perceived by, and affects others, and learning to express their feelings about others' behavior in a nonaggressive way" (Jordan & Le Métais, 1997, p. 6); and • Jordan and Le Métais (1997) suggested the DOVE rules of the brainstorming to solve the problems in group to develop interpersonal skills. The learners of this study learned these rules as following: D refers to defer judgement: welcome all opinions and postpone evaluation O refers to off-beat ideas: encourage the creative opinions V refers to vast number of ideas E refers to evaluate the ideas later and enhance the others' ideas g.To give corrective feedback to the learners as following: • To start the sentences with the positive words and sentences; • To soften the negative feedback by using the positive words and expressions such as it'd be good if . . ., can you and could you just . . ., and you did really well but . . ., • To consider feedback as a technique for learning rather than destruction; • To encourage learners to listen and support each other while each individual receives a feedback and; • To distribute time equally for all learners to be corrected.
h. To remove the learners' negative emotions, as a committed teacher, you "have to display your emotional intelligence by dealing with your learners' expectations" (Mortiboys, 2005, p. 49). In fact, when a learner enters a class, he or she brings his or her own expectations from English class, lessons, teacher, and classmate into a new one. It seems that most of the previous experiences are negative due to the learning of a new language, lack of understanding, method of teaching, or the learners' characteristics. Regardless of its reason, a conscientious teacher should change the learners' negative emotions to the positive ones (Mortiboys, 2005). To avoid anxiety and frustration, this process was done: • To hear the learners' expectations and try to lead their negative emotions to positive and productive ones; • To provide a supportive environment for the learners, the present researcher admired their hopes for learning and tried to indicate that their accomplishment would be done by the efforts of both teacher and the learners. It gave them positive feelings; • To respect the learners' opinions; • To assure the learners about the confidentiality of whatever would be posed in the groups; • To ensure all individuals to have opportunity to express their ideas and speak; • To ask the other members of the group to help the learners to overcome his or her fear and negative emotions; and • To provide different solutions for the learners to remove negative emotions as a teacher.

Post-test session
In the sixteenth session, the learners of all groups were administered the KET A2 level to compare the amount of the changes in the control and experimental groups in their language achievement as well as Willingness to Communicate (WTC) questionnaire to indicate the changes in the learners' tendency to communicate. However, Bar-On Quotient Inventory (EQ-i) was only administered for the experimental groups to show the amount of the emotional intelligence's development in the experimental groups after receiving the treatment. This research was based on the quantitative data analysis.

Sample and sampling techniques
To homogenize the participants in terms of their EFL knowledge, one version of KET for level A2 was administered. The learners whose scores lied ± 1 SD were selected. Therefore, eight learners from among eighty-eight ones were deleted. Then, FLCAS questionnaire was administered to classify the learners in two high and low levels of anxiety. As a result, the learners whose scores were between 33 to 75 were considered low anxious while those who received above 120 were the learners with high levels of anxiety. The data showed that twenty-four learners were in medium level of anxiety. So, they were omitted from the learners of this study and fifty-six learners remained. The data are presented in Table 1. The value of skewness (−0.1) in KET indicates that the data is relatively normalized. As McNeese (2016), the skewness between −0.5 to +0.5 shows relatively normalized data.
However, in this research the learners were not assigned to groups randomly due to the practical limitations but the groups were selected as the control and experimental groups randomly. According to Table 2, there is no significant difference (f(3,52) = 1.01, p = 0.39) among groups.
The emotional intelligence of all four groups were evaluated by Bar-On Quotient Inventory (EQi). The value of skewness (0.1) in EQ-i shows that the data are relatively normalized because the skewness is between −0.5 to +0.5. the data are presented in Table 3. The one-way ANOVA on this questionnaire reveals that there is not any statistically significant difference between the control and experimental groups in their emotional intelligence in general (f(3,52) = 1.22, p = 0.3) as showed in Table 4.

The learners' performance on the pretest and post-test
After the homogenization of the groups on the basis of KET in level 2 and EQ-i, the learners' performances, the learners' scores on the same KET were considered as the pretest. Moreover, another version of KET in level 2 was administered and the learners' scores in it was used as the post-test to reveal the differences between groups after instruction. That is, it measures the amount of the control and experimental groups' achievement in the development of English language. To compare the scores of KETs in pre and post administrations, the descriptive statistics were estimated in Table 5. The   comparisons of the groups' Means indicated that all the groups made progress in language learning after the instruction but that of the experimental groups was more than control ones'.
One-way ANCOVA was employed to indicate whether there was a significant difference among groups after the treatment or no. The result of this univariate analysis showed that there was a significant difference between learners' performance in pre and post KETs (f (1, 104) = 3228.86, p < 0.00). The Eta Squared also indicated that the findings could be safely generalized (Eta Squared = 0.97). With regards to four groups, there was a significant difference among groups in the post-test (f (3, 104) = 11.23, p < 0.00) and the findings could be safely generalized due to the large amount of Eta Squared (Eta Squared = 0.24). In addition, the interaction between the learners' scores in pre and post KETs among four groups was also significant (f (3, 104) = 8.36, p < 0.00) and the result can be generalized safely too (Eta Squared = 0.19). This information is shown in Table 6.
The reported results in Table 7 reveal that the performance of experimental groups was significantly different. In fact, the learners with low levels of anxiety who received emotional scaffolding and were corrected by recast showed the best result significantly. After them, the learners in experimental group with high levels of anxiety significantly performed well. It can be concluded that emotional scaffolding had a significantly positive effect on learners with low and high levels of anxiety although this effect was significantly more enormous in the learners with low levels of anxiety. As a result, there is a statistically significant difference between the effect of teaching English by emotional scaffolding and traditional way of teaching by means of providing recasts on developing Iranian elementary EFL high and low anxious learners' language achievement and the first null hypothesis is rejected.
Various speaking strategies were taught to all groups during the semester. Therefore, the speaking test of KET was also considered separately to indicate the learners' actual performance in speaking. Table 8 shows that the experimental groups showed positively significant behavior in speaking in comparison to their control groups. Interestingly, the effect of emotional scaffolding through recast in  the experimental group with high levels of anxiety was considerably more although it is not significantly different from its low anxious counterpart.
Willingness to communicate questionnaire was administered in the second and sixteenth sessions to reveal the changes in the learners of all groups. Table 9 indicates that high anxious experimental learners made progress in their tendency to communicate (Mean = 2.83) after the instruction.
As Table 10 illustrates, there was a significant difference between learners' tendency in pre and post WTCs (f(1, 104) = 166.64, p < 0.00). The value of Eta Squared was 0.62. as a result, the finding could be safely generalized. With regards to four groups, there was a significant difference among groups in the post-test (f(3, 104) = 11.31, p < 0.00) and the finding could be safely generalized (Eta Squared = 0.25). Moreover, the interaction between the learners' willingness in pre and post WTCs was also significant (f (3, 104) = 28.98, p < 0.00) among four groups and its Eta Squared was 0.45. Table 11 shows that the amount of the willingness to communicate in the experimental group with high levels of anxiety was most compared to the other groups and this group is significantly positively different from all the control and experimental groups. The learners with low levels of anxiety also had significantly positive tendency to communicate after receiving treatment in comparison to their own thinking before the instruction and the control group with the low levels of anxiety at the end of the semester. therefore, there are a statistically significant difference between the effect of teaching English by emotional scaffolding and traditional way of teaching by means of providing recasts on developing Iranian elementary EFL high and low anxious learners' willingness to communicate and the second null hypothesis is rejected.  According to Table 12, the experimental groups believed that their emotional intelligence increased significantly after fourteen-session instruction in general. However, that of learners with high levels of anxiety seems significantly more than their low anxious counterpart. Specifically, the analysis of the main scales indicates that both groups significantly improved their emotional intelligence's main scales. This table also reveals that there is no statistically significant difference between the learners with high or low levels of anxiety in interpersonal abilities, stress management, and general mood although their emotional intelligence significantly developed compared to their thoughts before emotional scaffolding. In the case of intrapersonal abilities, the learners with high levels of anxiety made progress significantly rather than those of the low anxious ones. Meanwhile, the adaptability of both groups also increased.  Then, the results of this research can be ascribed to the development of the emotional intelligence and emotional scaffolding.

Discussion
The focus of this study was to investigate the effect of emotional scaffolding through providing recast on the development of Iranian elementary EFL learners' language achievement and willingness to communicate. To achieve the aim of this research, two null hypotheses were proposed. The first null hypothesis was "there is no statistically significant difference between the effect of teaching English by emotional scaffolding and traditional way of teaching through providing recasts on developing Iranian elementary EFL high and low anxious learners' language achievement". The results indicated that emotional scaffolding significantly influenced positively both experimental groups' learning achievement. However, its effect on the learners with low levels of anxiety was significantly more than the high ones. Therefore, the first null hypothesis was rejected. This finding was in line with the results of Sheen (2008), Jang (2011), andJang (2010) in the case of the positive effect of recast on the groups with low levels of anxiety. As Narimanivahedi et al. (2019) concluded, the learners with more emotional intelligence benefited more from recasts rather than other types of corrective feedback. In fact, it may be due to the nature of the recast. In other words, recasts do not interrupt the interaction between learners and teachers so they do not produce extra anxiety.  The second null hypothesis was "there is no statistically significant difference between the effect of teaching English by emotional scaffolding and traditional way of teaching through providing recasts on developing Iranian elementary EFL high and low anxious learners' willingness to communicate". In respect of the learners' willingness to communicate, emotional scaffolding increased the learners' tendency to communicate significantly positively. In fact, the learners with high levels of anxiety and their low counterpart made statistically progress in this field although this amount was more in the learners with high level of anxiety. Thus, the second null hypothesis was also rejected. The result of this analysis regardless of emotional scaffolding is contrary to the findings of Manipuspika (2018) and Baker and MacIntyre (2000). For example, Manipuspika (2018) in her study showed that there was a strong correlation between anxiety and WTC. The higher level of anxiety the learners experienced, the less tendency they had to communicate. She concluded that fear of negative evaluation was one of the most important anxiety of the learners and this factor hindered the high anxious learners to speak. In the present research, emotional scaffolding was incorporated in the process of teaching to decrease the amount of anxiety and fear of negative evaluation especially after receiving the corrective feedback, i.e. recasts.
The findings of this research indicated that emotional scaffolding can decrease the amount of anxiety in learners and increase learners' tendency to speak because both levels of anxiety tended to speak after receiving the treatment compared to control groups. In addition, this effect was more on the high anxious learners. Narimanivahedi et al. (2019) also showed that emotional intelligence can decrease the amount of anxiety enormously in the learners. Then, this research finding was congruent with their findings. They also believed that the awareness of emotional intelligence can positively influence the effectiveness of corrective feedback and corrective feedback preference.
It can be proved that the development of emotional intelligence can change the learners' tendency with high levels of anxiety to more communication. Interestingly, the analysis of learners' speaking skill was compatible with this result. Therefore, teaching of emotional scaffolding can not only boost the communicative willingness of the learners in general and the learners with high levels of anxiety in specific but also their actual performance in speaking. Moreover, the evaluation of the learners' emotional intelligence reveals that they made considerable improvements in EI in general and all five main subscales of EI in specific. Nias (1996) brought three reasons for the importance of emotions in teaching: 1) teachers' feelings are engaged in the process of teaching, their relations to learners, parents, and colleagues; 2) cognition and affection are interrelated and teachers' emotions play an important role in the process of teaching; and 3) cognition and affection themselves originate in and shape social and cultural context. Then, this relationship is reciprocal. In line with this claim, Day and Leitch (2001) emphasized research about the role of emotion in teaching as a professional practice by bringing five reasons: 1) emotional intelligence plays a central role in teaching; 2) both emotion and cognition are necessary in this process; 3) one of the requirements of a successful teaching is emotional health; 4) personality, social environment, and external factors influence emotional and cognitive health; and 5) a need for more research on emotions through profession is felt. However, unfortunately there are no sufficient studies on the field of teachers' and learners' emotions.
Other reasons of attention to nature of emotions in an educational setting are the direct effect of these emotions on learners as humans and the differences and intensity of these emotions on both learners and teachers (Nias, 1996;Schutz et al., 2007). These researchers even found a multidirectional connection between teachers' emotions and the way in which teachers reflect their identity in the classroom. Indeed, a teacher's identity is shaped by the emotional experience that she or he gains during his or her teaching (Nias, 1996). For instance, the experience of a negative emotion can change a teacher's beliefs and accordingly his or her identity. Arnold (2011) stated that if language learning is connected to humanistic and affective features, the result will be more effective. She defined affect as emotions by which a whole person emerges. Then, as she mentioned, affect enhances both between or interpersonal relationship and inside or intrapersonal one by removing threatening atmosphere and increasing self-esteem. Teaching is also considered as "a job which involves interaction among people and inevitably therefore has an emotional dimension" (Nias, 1996, p. 296). Schutz et al. (2007) also emphasized that the transmission of emotions in a classroom context is undeniable. For example, when a teacher shows positive or negative reactions towards learners' achievement or failure, his or her emotions are aroused (Nias, 1996). Jan et al. (2018) showed in their research that there is a negative relationship between emotional intelligence and language anxiety. In fact, when a learner develops his or her emotional intelligence, he or she can not only more manage his or her own emotions and those of the others but also can be more motivated in the development of academic achievement in comparison to the learners with low emotional intelligence (Jan et al., 2018). In the present study, the effect of emotional intelligence was more on high anxious learners rather than the learners with low level of anxiety especially in intrapersonal abilities and adaptability as subscales of emotional intelligence and emotional intelligence in general. Therefore, this result contradicts Jan et al.'s (2018) findings.
In other words, the effect of emotional scaffolding as an umbrella concept and emotional intelligence as emotional scaffolding's subcategory jointly contribute in the development of emotional intelligence of both levels of anxiety. Then, it can be concluded that on the one hand, emotional scaffolding during both the process of leaning and provision of recast and, on the other hand, emotional intelligence training reduced the level of anxiety in the learners especially high anxious ones. This result can stress the importance of emotional intelligence awareness. When a teacher instructs emotional intelligence in the class, the high anxious learners become more aware of its importance in their learning and use it more in this process. However, the detailed comparison of emotional intelligence's subscales showed that both low and high anxious learners were successful in the stress management and this factor was the same in both groups after the instruction. In addition, the subscales (i.e., intrapersonal abilities and adaptability) and the general emotional intelligence that were more in high anxious learners before the instruction developed more after the instruction. Therefore, all subscales as well as general emotional intelligence increased parallel to what they were before emotional intelligence teaching.

Conclusions
This study indicates that a need for attention to the emotional intelligence training has been felt. The findings of this research can encourage teachers to teach different aspects of emotional intelligence in order to help learners to be successful in their learning and to decrease the amount of their foreign language anxiety. Furthermore, receiving corrective feedback is one of the most stressful processes that a foreign language learner in an elementary level can experience.
When teachers provide an implicit corrective feedback such as recast through emotional scaffolding and positive and inclusive environment, students' learning can be increased and the negative aspect of receiving corrective feedback can be decreased. Finally, the most important aim of learning a language is to communicate and specially to speak. The findings of this research showed that emotional scaffolding plays a vital role in facilitating learning and speaking process. The detailed conclusions can be as followings: 1. In this study, both high and low anxious learners developed their language achievement significantly positively. However, the comparison of both levels of anxiety showed that low anxious learners were more successful in their language learning. This may be due to the nature of the recast. In fact, recast is considered as an implicit corrective feedback. When a teacher corrects a learner's error, he or she indirectly provides the corrective feedback (i.e., recast). In other words, to observe the implicitness, the teacher can not directly express his or her feelings during the provision of recast. If so, it should be indirectly. In this situation, both low and high anxious learners can directly receive positive emotion just during teaching not receiving recast itself. Therefore, it may be concluded that emotional scaffolding decreases the amount of the negative emotions such as anxiety in general but its effect on the learners' anxiety during receiving recast was not enough to make the high anxious learners' performance the same as that of the low ones.
2. With regards to the willingness to communication, emotional scaffolding during both teaching and receiving corrective feedback significantly positively influenced both high and low anxious learners the same and there was no significant difference between these two groups. Interestingly, the result of willingness to communicate was similar to the learners' actual speaking. That is, emotional scaffolding can not only increase leaners' tendency to speak significantly positively but also lead to the development of speaking in learners with high or low levels of anxiety.
3. Language achievement includes four main skills (i.e., listening, speaking, reading, and writing). As mentioned, there was a significant difference between high and low anxious learners in the development of their language achievement. The second conclusion also emphasized that speaking skill in both low and high anxious learners developed the same and there was no significant difference between them. Then, another conclusion is that the difference between these two groups was due to the listening, reading, or writing. It is recommended the further research would focus on the other skills to find the exact place of the difference.

Recommendations
It is recommended that teachers incorporate the notion of emotional intelligence while they are teaching different aspects of foreign language. In fact, teachers should inform learners from the positive effect of emotional intelligence on leaning of the abstract content of a foreign language. They should also teach learners how to manage, develop, and apply this intelligent during their study to relieve the stressful situation of the class.
On the other hand, learning without receiving corrective feedback seems ineffective. This claim can stress the essential role of a teacher during providing corrective feedback. According to the findings of this study, it is suggested to EFL teachers to correct learners' errors by using positive words and sentences to eliminate or decrease the negative process of making errors.