The Effect of Sentence-Making Practice on Adult EFL Learners’ Writing Anxiety A Comparative Study

Writing anxiety is one of the most essential factors influencing language learning. The current study is to explore the effect of sentence-making practice on reducing writing anxiety of two classes of adult EFL learners, one in low-intermediate level (LI learners), the other in high-intermediate level (HI learners). Two classes received two-week sentence-making practice, before completing the questionnaire designed specifically for the present study and attending the semi-structured interview. The descriptive statistics and independent t-test demonstrate that both mean scores of two classes are below 3 points, showing that they are in low or moderate anxiety levels. Moreover, the mean scores of HI class are lower than LI class in all four dimensions (classroom experience; while-writing experience; feedback experience; further effects), despite no significant difference, except for in the third dimension (feedback experience). Both statistic results and qualitative analysis with learners’ attitudes towards sentence-making practice illustrate that the learners in two classes have less writing anxiety after sentence-making practice, and this teaching method has also brought beneficial effects on learners’ writing.

sentence writing is less difficult than writing a paragraph or composition, which means that sentence-writing practice could put less pressure on learners, create an easy atmosphere for them to write. Second, sentence-making practice won't take much time than previous traditional writing, during which the teacher and learners have time to discuss and provide timely feedback. Third, the sentence-making practice could let learners experience and apply language points just learned in class.
Hence, the objective of the present study is to investigate and compare the effect of sentence-making practice on writing anxiety, between the group of low-intermediate adult EFL learners and the group of high-intermediate ones, and provide implications for later writing teaching and learning.

Writing Anxiety
The research of anxiety dates back to the 1970s, which is one of the most essential factors influencing language learning. Early research just focused on native language learning. The concept of "foreign language anxiety" was first proposed by E. Horwitz, M. Horwitz, and Cope (1986: 128), distinguishable from generally psychological one, and the former is a unique and complex self-awareness, beliefs, emotions and behaviors that are related to the foreign language learning in the classroom due to the uniqueness of the foreign language learning process (Horwitz, 1986;Guo & Qin, 2010;Li & Li, 2016). Over the past several decades, the research centering on foreign language anxiety is fruitful, which is classified into six categories, i.e., ontology research, correlation research, influencing factors research, comprehensive research, applied research, experimental teaching research, covering different aspects of a foreign language, like, listening, oral, reading, writing and interpreting anxiety (Li &Li, 2016).
Previous studies of L2 writing anxiety have been in growing trend these years, with a total of 1067 formally published papers from 1997 to 2017 (He, 2018). They are mainly related to the following three areas. The first is the relationship between writing anxiety and writing performance. Guo and Qin (2010) found that the overall foreign language writing anxiety and the four anxiety factors of the subjects (i.e., classroom teaching anxiety, conceiving anxiety, avoidance behavior, and lack of confidence) were significantly negatively correlated with the writing scores. It further proved that foreign language writing anxiety is a kind of debilitating anxiety, which hindered the students' writing quality and performance to some extent. The second area of research is teachers' intervention aiming to reduce learners' anxiety during the writing process. Teachers' intervention could be achieved by adopting different teaching techniques, applying alternative writing practices, using various feedbacks, and creating an encouraging writing atmosphere. For instance, Zhang (2005) introduced the feasibility of the Process Approach with the view of removing psychological blockages in English writing. Zhang constructed this approach of writing teaching from four aspects: pre-writing preparation, free draft writing, multi-interactive feedback, and cooperative revising. Compared to traditional product-oriented writing teaching, Zhang considered the Process Approach as an effective way to dispel anxiety and motivate learners' interest in writing. Wu and Gu (2011) explored collaborative writing to effectively reduce learners' anxiety. Similarly, collaborative writing followed the prewriting-while writing-post writing framework. During each stage, activities like Learning Together, cooperative Integrated, Reading and Composition, Think-Pair-Share, Teams-Assisted-Individualization, were integrated into writing, to strengthen learners' cooperation. The statistical results indicated the learners' anxiety level significantly decreased after the experiment. The finding is congruent with Guo (2011). She conducted an empirical study on the effect of the Length Approach on Chinese non-English majors' English writing anxiety. The key point of the length approach is to push learners to write as much as possible, for instance, 500 words. This feature brings positive and error-tolerant writing rubrics. The results showed the reduction of writing anxiety as well as the improvement of writing performance. Also, apart from specific classroom writing teaching approaches, previous studies have resorted to the internet to explore online writing and e-mail to reduce learners' anxiety. The third area is learners' internal factors which influence their anxiety. Previous studies explored the use of self-regulation strategies and the relationship between writing anxiety and self-efficacy. However, despite many previous studies on writing anxiety, these studies mainly focused on undergraduates, and most of them are similar compared with methods or research purposes and contents (He, 2018;Li & Li, 2016). There is little research about the effect on learners with different English levels. Besides, despite the effectiveness of the above research in reducing writing anxiety, there are some weaknesses of those teaching techniques. Guo (2011) pointed out that given the features of the Length Approach, it is more appropriate to high-level English proficiency learners in case of causing avoidance behaviors. Previous writing teaching approaches were constructed in the whole lesson and involved the whole process of writing from preparation to a final product, which occupied much time. However, adult EFL learners have fewer class hours, and even have no specific writing courses for them to practice, compared to undergraduates. Therefore, the current study is to investigate the particular teaching practice to explore whether it has a distinctive effect on adult learners' anxiety, with different levels of English proficiency.

Scales for Measuring Writing Anxiety
Concerning the anxiety scale, Horwitz camp up with a measurement scale as "Foreign Language Classroom Anxiety Scale (FLCAS)", which has been widely used by researchers to quantify learners' anxiety level to date. However, it has been criticized for to a large extent associated with oral language learning and much more related to language learning self-confidence rather than anxiety (Guo & Qin, 2010). Besides, the extremely high reliability has been brought into question suggesting flaws in the scale. Khalaf (2017) developed a self-report scale of English learning anxiety, constructed anxiety in listening, speaking, reading, and writing. However, the aforementioned scales are not sufficient to deeply explore language-skill specific anxiety. Against such background, as for writing anxiety, several measurements have been put forward, like, the Daly-Miller Writing Apprehension Test (WAT; Daly & Miller, 1975), while it was originally developed for first language learners, as well as more concerned with writing self-esteem and writing apprehension. That is why Cheng (2004) constructed the Second Language Writing Anxiety Inventory (SLWAI), referring to ten anxiety scales. SLWAI (Cheng, 2004) consists of three subscales: Somatic Anxiety, Cognitive Anxiety, and Avoidance Behavior, with a total of 22 items, which has a good reliability and validity.
However, the scale items and components of SLWAL cannot cover the second language writing anxiety traits and inducing factors. Furthermore, because of the different language environment, educational background and learner's English levels, SLWAL is not much appropriate to be applied to the English teaching and learning context in mainland China. Bai (2017) constructed a survey scale of the inducing factors for classroom English writing anxiety, which consists of two categories, self-inducing factors (with expression competence, language competence and discourse competence as subcategories) and inducing factors from writing process (with evaluation feedback, written text characteristics and output modes as subcategories). Therefore, the scale developed by the current study referred to both Cheng's SLWAL (2004) and Bai's scale (2017), to investigate the anxiety level between two groups of learners after a period of the particular teaching practice.

Sentence-Making Practice
As for reducing learners' anxiety, Horwitz et al (1986) have suggested two options for teachers, one is to help students learn to cope with the existing anxiety-provoking situation, the other to make the learning context less stressful. Concerning Guo & Qin (2010), there are three basic strategies when it comes to anxiety, which are positive cognitive strategies, positive action strategies, and avoidance behavior. However, it was proved that most learners tended to resort to negative strategies, attempting to avoid English writing activities. Therefore, against such backdrop, teachers have the responsibility to arouse learners' enthusiasm towards writing by applying interesting teaching techniques or varying writing tasks.
Sentence-making is a basic writing skill, and learners make sentences by using new words, phrases, or sentence structures (Liu & Zhang, 2015). Sentence-making practice is free from time and rules, for it can be applied to any procedures during the class, and no fixed rules for learners to obey, no fixed word number, verb tense, or topics, comparing to other teaching techniques. That means teachers could ask learners to make their sentences at the beginning of the class as a way to review what they have learned in the last lesson, or during the class, to attract learners' attention by asking them to write something, or at the end of class, to practice newly learned words and promote their writing output. Hence, sentence-making practice is a much flexible teaching technique, which could create a less stressful atmosphere where learners may not have much anxiety. Besides, concerning the nature of the sentence-making practice, it is the basic writing skill. Adults EFL learners, with their particular characteristics, in this writing practice, could solid their grammar knowledge; learn how to use words that they previously just know their Chinese meaning. Also, since the sentence-making practice is flexible and does not need much time, teachers could increase the frequency of application of this writing practice, by continuously letting learners write several sentences during the class. Meanwhile, with the high frequency of engaging to write, learners may not have more anxiety than before. After making sentences, learners have time to do self-proofreading, peer proof-reading, and group discussion to exchange their sentences, their thoughts, and feelings during writing, then teachers would also give feedback to each of them. This kind of discussion could ease learners' writing anxiety, and let them enjoy English writing. The efforts they take during making sentences by having to include one or two words or phrases could also train learners' imagination.
Sentence-making practice has not been explored in previous studies, let alone the application of this practice to elt.ccsenet.org English Language Teaching Vol. 13, No. 6; reduce learners' writing anxiety. In the two classes, one is high-intermediate English level, and the other is low-intermediate. Learners in both classes have anxiety towards writing, for different reasons. Since all of them are not good at grammar, lack sufficient vocabulary, hardly have time to write after class, and also have little chance to write in previous English learning. Therefore, the present study is comparative research that is going to examine the effect of sentence-making practice on the adult EFL learners' writing anxiety, concerning these two classes. The following are the three research questions: 1. Whether sentence-making practice helps to reduce writing anxiety concerning two classes of adult EFL learners with different English level?
2. If so, is there any difference in each dimension between these two classes?
3. What are the attitudes of participants towards sentence-making practice?

Participants
The current study involved two experimental classes with different English proficiency based on their Test for English Majors-Band 4 (TEM-4) scores. TEM-4 is a large-scale standardized national English proficiency test for university undergraduate English majors in China, which is officially administered by the National Advisory Committee for Foreign Language Teaching (NACFLT). TEM-4 is a reasonably reliable and valid test (Jin & Fan, 2011). Eleven high-intermediate learners are in class one (HI class), with their TEM-4 mean score at 63 points. Seven low-intermediate learners are in class two (LI class), with TEM-4 mean score at 55 points. HI learners have already completed their junior college degree, with the age ranging from 28 to 35. LI learners have just obtained their senior high school degree, aging from 20 to 25. The two classes are now engaging in the University Continuing Adult Education, majoring in English. They have courses as following: English listening and speaking course, comprehensive English course, reading skill course. The researcher, also the teacher, is responsible for these two classes' comprehensive English course.

The Procedure of the Research
In the comprehensive English course, HI class and LI class received two weeks of sentence-making practice (ten lessons in total). During these two weeks of application of this teaching practice, the teacher still followed her original syllabus to teach reading passages, but during which adding the sentence-making practice for learners. Four basic principles were abided by in the implementation.
First, the practice was flexibly used in any period of a lesson. During the beginning of a lesson, the teacher guided students to have a review of what they have learned during the last lesson, like some important words, phrases, and sentence structures. Then the teacher encouraged learners to write two or three sentences which must consist of those reviewed words. During teaching reading passage, the teacher also stopped sometimes to let learners write some sentences covering newly learned words to familiarize those words, as well as arouse their interest and reduce their fear in English writing. At the end of the lesson, the learners were also asked to write some sentences by using words that the teacher appointed.
Second, each time the assigned sentence-making task was under the teacher's judgment of learners 'needs. When the learners readily engaged in the sentence-making writing, the teacher then added more challengeable requirements for the task, like, increase the number of sentences demanded to make, rewrite sentences by using different grammatical structure or by replacing word forms, even lengthen the original sentences by inserting modifiers or clauses. However, when students were encountering difficulties during writing, the teacher suspended the sentence-making practice by providing explicit instruction to deal with these difficulties. If the learners still got stuck in sentence writing, the teacher then presented some sentence examples for learners to imitate. Thus, the whole process was preceded gradually and step by step and the teacher adjusted the requirements timely in line with the learners' responses.
Third, learner-centered activities were involved to ensure learners' active participation in the sentence-making teaching practice. Each time after learners wrote their sentences, the teacher motivated them to first proofread their sentences, to find some errors, whatever on grammatical or semantically or morphological level. Then the teacher encouraged them to exchange their writing with their partner to do peer-proofreading, and have a group discussion. During this process, the teacher would move around the class to provide some timely guidance and feedback, after which learners were encouraged to present and read aloud their sentences to the class.
Fourth, sentence-making teaching practice was employed by the teacher with the hypothesis of endowing them a relaxing environment and providing encouraging chances to write at any time, to make learners not anxious with writing Thus, the number of assigned sentence-making tasks in each lesson had to meet certain criteria to ensure the output of writing. The teacher has to implement the sentence-making teaching practice at least three times and each student has to write at least ten sentences each lesson.
After two weeks of sentence-making practice, the teacher gave out the questionnaire developed in the current study and conducted a semi-structural interview to explore the effect of sentence-making teaching practice on these adult EFL learners' writing anxiety. The procedure of the research is shown in figure 1.

Scale Development
The current study developed a new questionnaire for measuring the learners' anxiety level after receiving the two weeks' practice in two classes. At the outset, the researcher created the preliminary item pool for the scale by referring to Guo & Qin (2010)'s Chinese version of Cheng's SLWAL (2007) and Bai's scale (2017). Since the desired scale is of great different functions compared to previous ones, the scale in the current study is particularly used for examining writing anxiety after one specific teaching practice. To this end, interview and discussion were held on writing anxiety and sentence-making practice with experienced English teachers, educational experts as well as students to add more understanding during the development of the scale. The first version of the scale was presented to the professionals to determine the appropriateness of the created items. Improper and ambiguous items and items with contradictory or repetitive meanings were removed and revised. The results yielded the initial version of the questionnaire, consisting of 21 items in five dimensions: feelings for involving sentence-making practice (4 items), while-writing anxiety (4 items), and writing consciousness development (4 items), linguistic competence development (4 items), feedback anxiety (5 items). This scale adopts five-point Likert scale (1 = strongly agree; 2 = agree; 3 = no strong feelings either way; 4 = disagree; 5 = strongly disagree). The scale was then operated in a pilot study for the sake of exploratory factor analysis to examine the construct validity of the scale.

Exploratory Factor Analysis
This scale was designed particularly for the current study, named the change of L2 writing anxiety after the sentence-making training scale. A pilot study on the initial version of the questionnaire was administered on another class of adult EFL learners. The exploratory factor analysis was conducted utilizing SPSS. In the first round, the Principal Axis Factoring method of extraction was operated to examine the factor structure of the preliminary 21-item L2 writing anxiety questionnaire. We used a factor loading of .50 as the criterion. We found elt.ccsenet.org Vol. 13, No. 6; that items 3, 4, 11 loaded on the first factor (feelings for involving sentence-making practice), items 5, 6, 7 on the second factor (while-writing anxiety), and items 17, 19, 20 on the fourth factor (feedback anxiety). The above three groups of items were thus chosen to form the basis of the subscales of this scale. Besides, we have items 10, 13, 14 on the same factor, and the researcher named it as "further effects" to see whether this practice could render learners gain some benefits from their writing. As such, nine items were deleted and the remaining 12 items were then classified under four factors, renamed as "classroom experience"; "while-writing experience"; "feedback experience"; "further effects". Results of exploratory factor analysis revealed that the scale is satisfactorily valid. Meanwhile, the internal consistency of the 12 items was calculated by Cronbach's coefficient, with a reliability of .832, which suggested that the scale was reasonably reliable. The final version of the questionnaire (including 12 items) is presented in the Appendix.
After the first day of the experiment, the researcher administered the self-developed questionnaire, the result of which revealed the high-level writing anxiety in both classes. That confirmed the homogeneity of two experimental classes and the necessity of implementation of sentence-making teaching practice. After two weeks of the experiment, the questionnaire was conducted to collect quantitative data for comparing the anxiety level of two classes after receiving this specific teaching practice. To avoid practice effect, the sequence order of the questionnaire was adjusted.

Semi-Structured Interview
A semi-structured interview was conducted at the end of the experiment in two classes to explore as well as complementing the understanding of the individualized experience and attitudes towards the sentence-making teaching practice. The interview was unfolded based on the following two introductory questions: 1) after two weeks of sentence-making practice, what are your feelings towards English writing? 2) Do you prefer the current classroom sentence-making practice or not? The whole process was audio-recorded and transcribed, which was then coded through the QDA Miner Lite program under thematic analysis.

Results
At the beginning of this questionnaire, we have our participants do a self-assessment on their whole English level and writing level, with a five-point Likert scale (1 = high; 2 = relatively high; 3 = intermediate; 4 = relatively low; 5 = low). We found that 64% of learners from class one (high-intermediate level) thought their English level was intermediate, and 55% thought their writing level was relatively low. Whereas, as for class two learners (low-intermediate level), 71% believed their English level was relatively low, and 57% claimed that their writing level was low.

Research Question1: Whether Sentence-Making Practice Helps to Reduce Writing Anxiety Concerning Two Classes of Adult EFL Learners with Different English Level?
First, we resorted to SPSS to have a descriptive analysis of their anxiety scores and found that the mean scores of these two groups of learners were 2.52 (low-intermediate) and 2.17 (high-intermediate), which were both lower than 3 points. It means that after sentence-making training, both of the two groups were less anxious toward writing, and their anxiety level was nearly the same. Whereas, we wanted to know whether there was a significant difference between these two groups of learners after experiencing two-week sentence-making practice.

Differences in Classroom Experience
Classroom experience factor is to explore learners' real actions towards the sentence-making practice, including classroom atmosphere, continuing applying this practice, finding more writing chances after class. Table 2 illustrates the descriptive results and independent t-test results of two groups concerning classroom experience. The mean score of LI learners (2.24) was higher than HI learners (2.12), while there was no statistically significant difference between the two groups (t=0.25, df=16, p>0.05). Similarly, there was also no significant difference in each item in this factor. What is needed to be pointed out is that their mean scores were all lower than 3, which means their anxiety level was diminished after this training. Even though they were at different language proficiency levels and English writing levels, both of the two groups indeed were not anxious about writing. Interestingly, scores of LI are lower than HL learners in items 1 and 3. In other words, LI learners more wanted teachers to continue to use sentence-making practice, and they tended to practice their English writing even after class, whereas, HI learners were more into the classroom atmosphere during sentence-writing practice. While-writing experience is to pursue learners' internal feelings during writing, like, the mind goes blank, having a clear mind, and not afraid when knowing their writing would be remarked by the teacher. Table 3 provides the descriptive statistics and independent t-test between two groups concerning while-writing experience and each item in this factor. LI learners' mean scores in this factor and each item were all higher than HI learners. However, this difference was not statically significant when we resorted to an independent t-test (p>0.05). This difference does tell us that HL learners tended to be less anxious during writing, their mind was not going blank, clearer and they were not nervous during writing when knowing their writing would be given remarks by the teacher.

Differences in Feedback Experience
Feedback experience factor aims to explore learners' reactions to the ways their writings are treated. This factor includes three specific items, which are concerned with peer feedback, teacher feedback, as well as writings are shown in the class. Table 4 contains the descriptive statistics and independent t-test of scores between two groups in this dimension as well as each item under it. From table 4 we discovered that LI learners had higher mean scores than HI learners on the whole concerning this factor, and also higher in each item than the HI group. There was a statistically significant difference between the two groups on the whole in this factor (t=3.156, df=16, p<0.05). It reveals that HI learners were less likely nervous and less afraid that their writing being assessed by peers and the teacher, or being presented in the class. As for each item in this dimension, scores on item 1 had a significant difference between two groups (t=2.728, df=16, p<0.05). Most of LI learners were still feeling a bit of fear when their essays were presented to be discussed in the class. Besides, from table 4, we can see that scores on item 3 were the highest in both groups, which demonstrates that, whatever language proficiency levels they are in, the learners were still a little bit more nervous when their writings were examined by the teacher than by peers.

Differences in Further Effects
Sentence-making practice not only lessens learners' anxiety towards writing, as well as may bring long-term effects for learners. Therefore, our questionnaire embraces this factor (further effects) to show how this sentence-making practice benefits their writing, including writing interest and writing knowledge, like grammar and vocabulary. Table 5 shows the descriptive statistics and independent t-test of further effects factor and each item in this factor. We found that the means scores of the two groups were nearly the same, with no statistically significant difference (p>0.05). As for each item in this factor, the scores of HI learners were a bit higher than LI learners, except the item1, while their differences also were not statically significant (p>0.05). Since we oppositely used the Likert scale (1 = strongly agree; 2 = agree; 3 = no strong feelings either way; 4 = disagree; 5 = strongly disagree), the fewer scores they had, the more effects they accepted. Thus, after sentence-making practice, HI learners had a higher interest in English writing, while, LI learners nearly strongly agreed that sentence-making practice improves their use of words and grammar in writing. The mean scores of the two groups were both around 2, indicating learners agreed that the sentence-making practice could bring benefits to them. In a nutshell, the writing anxiety level of HI learners was relatively lower than low-intermediate learners on the whole, even though there was no statistically significant difference between the two groups, except the third factor (feedback experience). It seems that HI learners were significantly less anxious about writing than the other group, not afraid to have their writings examined by peers. Nevertheless, overall, the anxiety level of two groups was indeed moderate after the two-week sentence-making practice, with their mean scores lower than 3 points. Therefore, we can infer that the sentence-making practice is an effective teaching technique to reduce learners' anxiety level, as well as bring other beneficial effects, like arousing writing interest, improve the use of words and grammar in writing. The inferential results we gain from the above descriptive statistics and t-test results could be directly proved by the learners' answers in the semi-structured interview. Concerning positive attitudes, we have seven subcategories under this code according to learners' answers, i.e., sentence-making practice helps to promote writing skills, writing interest, use of words and phrases in sentences, get the hang of grammar and sentence structure, learning output, and cultivate thinking patterns, as well as classroom atmosphere. Figure 2 is the distribution of positive codes in low-intermediate class (LI class), and figure 3 is the high-intermediate class (HI class), demonstrating that nearly 22% of LI learners believed that sentence-making practice could enhance their use of words and phrases in writing. 14% of LI learners considered that the method promotes their writing skills, as well as the use of grammar and sentence structure. Similarly, 28% of HI learners considered that their writing skills and use of words and phrases have been improved. Besides, 7% of LI learners and 14% of HI learners claimed it helped to cultivate thinking patterns in writing.
What is unique is that 7% of LI learners have been more interested in writing after receiving sentence-making practice. 7% of HI learners hold that this method helps them do more output after learning new knowledge, and they like the classroom atmosphere when doing sentence-making practice. Compared to positive ones, there are a few negative answers. We determined three sub-codes under this dimension: monotonous form, incomprehensive, and improper for poor learners. From figures 4 and 5, we can see that 3 LI learners and 1 HI learner stated that sentence-making practice is not comprehensive. For example, "it cannot involve more word usages, I just practice one certain form of a word in writing a sentence" 1 LI learner and 2 HI learners showed that sentence-writing practice is not diverse, only concerning with sentences. For instance, one learner said, "We just write sentences and check our sentences, and there are no more other activities during this method." Besides, 1 LI learner complained that this method was not proper for poor learners. In a word, most of the learners expressed their positive attitude towards sentence-writing practice, like promoting wiring skills, and enabling them to use more precise grammatical structures. Nevertheless, some improvements need to be made in later teaching. Embracing more activities into sentence-making practice, instead of just requiring them to write and discuss. Furthermore, various kinds of language knowledge have to be combined for learners to practice, and we should not just concentrate on one single sentence.

Discussion
The paper was to explore the effect of the sentence-making practice on reducing writing anxiety. The researcher self-developed the questionnaire for the current study to measure the anxiety level of learners with different English proficiency levels, combined with the semi-structured interview. The results showed both high-intermediate learners and low-intermediate learners have less writing anxiety after receiving two-week sentence-making practice with both mean scores lower than 3 points in all four dimensions, confirming the effectiveness of sentence-making practice in both groups. The findings were consistent with previous studies (Guo, 2011, Wu & Gu, 2011. During this teaching technique, learners took advantage of all chances to do sentence-making practice in class and enjoyed the classroom atmosphere ("Classroom Experience"). They enjoyed the writing process with a clear mind and have the urge to write ("While-writing Experience"). They welcomed feedback from either the teacher or peers ("Feedback Experience"). They believed this technique promote their writing skills in the long run ("Further Effect").
In the light of comparing the anxiety factors of learners with different proficiency levels, independent sample t-test showed that even though there was no significant difference, HI learners performed less anxiety level in the whole and in the first two and the last dimensions than LI learners after the experiment. It can be inferred that the sentence-making practice yields a better effect in reducing writing anxiety for learners with a bit higher proficiency. That was verified by qualitative results, where a larger percentage of HI learners expressed positive attitudes and satisfaction with sentence-making practice. Upon the third dimension "Feedback Experience", there was a significant difference. LI learners performed relatively higher writing anxiety compared to HI learners, revealing that LI learners were still afraid to present their writing in front of the class or to be viewed by the teacher after the experiment. This may be attributed to the fact that they were only treated for two weeks. Nevertheless, it gives us implications for adjustment of the sentence-making practice in further teaching, where learners could be more involved in group discussion. The group includes low proficiency learners as well as high proficiency ones, of whom high proficiency ones could be the representative of the group to present their writing in class since they are less anxious about this way of feedback.
Compared to previous research, the sentence-making practice is appropriate to reduce writing anxiety for learners with different proficiency levels. That, in turn, proves that it is a rather basic and practical teaching technique, which can be flexibly employed by language teachers. In traditional writing teaching, learners are affronted with tough writing tasks, like write a whole composition on a given topic. It is difficult for learners, especially when they are not often engaged in writing during language learning. The sentence-making practice starts from simple writing tasks, creating a relaxing environment for learners to write, and the teacher could adjust the amount and frequency of tasks promptly based on learners' responses. It effectively reduced learners' avoidance behavior, since this anxiety factor was a headache and hard to reduce in previous studies (Guo, 2011;Wu & Gu, 2011). Also, the integration of learner-centered activities further promotes their self-confidence and participation. This writing teaching practice enjoys cyclic and repetitive features, making learners familiar with writing and reducing their writing anxiety. It is recommended to use in foreign language writing teaching.

Conclusion
Affective factors are always the most concerning issues in foreign language learning and teaching. Writing anxiety is the critical and decisive notion in writing (Khalaf, 2017), which leads to a negative experience, causes fear in the writing process, and impedes the development of writing ability. Based on the high writing anxiety level of the adult EFL learners in these two classes, the researcher resorted to the sentence-making teaching practice with the hypothesis of reducing writing anxiety. Despite the encouraging results demonstrated from the current study, there are indeed some limitations. First, the study only involved a small sample of participants in a short time, which means the generalizability of the results to other populations has to be made with caution. Second, the current study focused on the effect on writing anxiety and not further explored the relationship with the teaching technique and writing performance. Third, concerning the negative points participants expressed, adjustment of the sentence-making teaching practice is needed. We hope further research could be carried out on more participants and lasting more time to explore its effects on writing anxiety and writing ability. Besides, teachers could add more activities and requirements during the sentence-writing practice; for instance, competition on the number of sentences each group could make on given words or certain grammatical structures, to make it interesting as well as challengeable for learners' future development.