An On-line Questionnaire Survey on Students’ Views and Teachers’ Practices in Corrective Feedback in Teaching Chinese to Speakers of Other Languages

IDEAS


Introduction
The research on the correlation between CF and views (beliefs/ attitudes/ preferences) of students and teachers in second language acquisition(SLA) or in foreign language learning and teaching(FLLT) has been conducted for many years by scholars or educators in the world, and a great deal has been achieved (Akiyama, 2017;Cinkara & Galaly, 2018;Couper, 2019;Horwitz, 1985;Xuan & Murray, 2020;Zhang & Rahimi, 2014). Entering the keywords corrective feedback between 1970 to 2020 in Bing Academic, a Microsoft search engine for academics, we obtained 178,000 academic results in English. They covered fields in linguistics, grammar, psychology, pedagogy, pragmatics, cognition, learning disability, computer, etc. and different aspects of learning, such as speaking, writing, listening, and communication, etc. Horwitz's (1985) study suggests that a systematic assessment of student beliefs would increase student learning and satisfaction in the foreign language methods class. Horwitz's (1988) creation of the well-known Beliefs About Language Learning Inventory (BALLI) questionnaire has greatly stimulated scholars interest in L2 learner beliefs . Researches indicate that learners' beliefs in SLA or FLLT might be accountable for learners' behaviors, learning strategies, motivations, and proficiency (Ghabanchi & Meidani, 2012;Zhu & Wangle, 2019).

Eight Issues in CF Study
To study students' and teachers' views and CF types is to a certain extent to find clues to the following eight issues based on the early five fundamental issues (Ellis, 2009;Hendrickson, 1978): 1. Whether to correct; 2. What to correct; 3. Who is to correct; 4. When to correct; 5. How to correct; 6. Which type of CF is the most effective; 7. How often is to correct; and 8. What effects of CF.
Forty-eight thousand two hundred results relevant to the eight issues in CF study between 1970 to 2020 on Bing Academic prove that although the answers to the eight questions are controversial, this does not prevent scholars from researching them. For decades of years, scholars have been revolving around the issue of CF and learners ' and teachers' beliefs, attitudes, preferences, views, etc. to explore the correlation between them with the common goal to improve efficacy of L1 or L2 or FLLT.
Whether to correct? For the first question, is it necessary to correct the errors? When unable to recognize their own errors, students need some professional assistance from their teachers. A survey on college students' attitudes toward error correction indicates that students do want to be corrected, and their eagerness to be corrected is more than what teachers feel they should be (Hendrickson, 1978).
What to correct? For this questions, different answers are provided by different scholars or teachers in different classes. Most researchers suggest that teachers should not correct all the errors but correct those that interfere with the meaning of a message or comprehension of the intended meaning of the speaker or writer. Teachers are expected to focus attention on a few error types rather than try to address all the errors learners made (Ellis, 2009). Errors should be corrected when the goal is learning, but not at all times, and it is useless for acquisition (Amara, 2015). Correction of global errors is more crucial than that of local one because it clarifies the intended message more than the correction of several local errors in the same sentence (Burt, 1975). Phonemes and words are what to be corrected in pronunciation (Couper, 2019).

Who is to correct?
The third question is about whether it is teachers or learners themselves or peers that should correct the errors. Earlier studies reveal that correcting students' errors is mainly teachers' responsibility (Hendrickson, 1978). But later, being more proficient in pedagogy, teachers do not correct learners' errors all the way. They tend to lead learners to find problems themselves so that they can be aware of the importance of autonomous learning and achieve more. Chandler's (2003) findings indicate that students feel that they learn more from self-correction. Besides, teachers encourage peer correction to let learners to help each other and learn from each other. Sato's (2013) study reveals that positive beliefs lead to positive peer interaction and peer CF.
When to correct? Concerning this question, the heated debate over immediate or delayed correction has never ceased. Ölmezer-Öztürk & Öztürk's research (2016) displays that immediate feedback is not felt comfortable, and students are discouraged from speaking in a classroom atmosphere. Amara's (2015) findings show that the timing of correction is determined by the type of errors committed. For pronunciation or grammatical errors, immediate correction is preferable. But for communicative purpose, delayed correction is more preferred. Li et al (2016) find both the immediate and the delayed feedback result in gains in grammaticality judgment test scores, with immediate feedback showing some advantage over delayed feedback.
How to correct? For the fifth question, different correction methods are applied in different situations. Couper (2019) provides diverse correction methods used by participants. Yoshida (2008) finds that teachers choose recasts correction methods because of the time limitation of classes and their awareness of learners' cognitive styles. They also choose elicitation or metalinguistic feedback when they think learners can work out correct forms themselves. However, for most of the learners, they prefer to be able to think about their errors and the correct forms before receiving correct forms by recasts. Cinkara & Galaly's (2018) finding reveals that students show high preference of specific written feedback to facilitate the correction of mistakes. Omidpour & Bavali's (2017) study suggests that teachers with normative styles are likely to use oral CF techniques more frequently than those with higher diffuse-avoidant style score.
Which type of CF is the most effective? Concerning this question, the answers also differ from different angles, different learners and teachers. Li's (2010) findings indicate that there is a medium overall effect for CF, and the effect of implicit feedback is better retained than that of explicit feedback. On the other hand, Maleki & Asl (2015) confirm the efficacy of explicit feedback strategies than that of implicit. Learners who use explanation as an explicit CF strategy achieve higher scores than those who use recast and error code feedback strategies. Ölmezer-Öztürk & Öztürk (2016) argue that different types of CF might result in quite different students' perceptions. For example, recasts and clarification requests are perceived as ambiguous, and meta-linguistic feedback as anxiety-provoking and difficult to comprehend. Chandler's (2003) study proves that direct correction is best for producing accurate revisions in writing, and students prefer it for its being the fastest and easiest. Masantiah et al (2018) find that different CF is supposed to be used for different learners considering their basic knowledge.
How often is to correct? By how often is to correct, we mean the frequency of a teacher's maximum correction in one class that is acceptable to both teachers and students. After examining individual versus team differences, Pyke & Sherlock (2010) find that teams tend to receive a greater amount of CF, whereas individuals require greater motivational feedback (providing motivation to the learner to perform a task or to learn). However, it seems that few researches in the previous studies were on this issue, and it was not singled out for elaboration, but was mingled with whether or which or what to correct.
What effects of CF? Whether CF contributes to L1 or SLA or FLLT is another controversial issue. Much research evidence in L1 or L2 indicates that teachers' feedback focusing principally on correcting the errors is not likely to produce substantive or even measurable improvement in the quality of students' learning (Baleghizadeh & Dadashi, 2011). Results of devoting much valuable time to correcting students' mistakes may not be satisfactory as was expected (Hendrickson, 1978;Truscott, 2007). Interestingly, some studies in writing have proven the effectiveness of CF in promoting acquisition (Ellis, 2009;Kang & Han, 2015;Tanveer et al, 2018). Different from the above studies, this article focused on the major CF issues by making a qualitative and quantitative comparison study between the types of CF students expect the teacher to use and those teachers actually use in the context of TCSOL, aiming at probing into the depth of CF and contributing to researches in TCSOL, L1, L2 or FLLT.

Research Design
In this research, the authors used an online questionnaire survey(Note 1) to investigate congruence and incongruence between students' views and teachers' practices in CF in TCSOL.

Instrument
Wen juan xing, an online questionnaire survey software (SPSS) in China was used separately used to distribute, collect and analyze the data of students and teachers. Then, the proportion and correlation analysis between students' views and teachers' practices in CF in TCSOL was further evaluated by Pearson chi-square test accomplished by cross-tabs.
To ensure the validity of the survey, a 5-point Likert scale was used to measure students' views and teachers' practices in CF in TCSOL. Two types of scale were adopted: for students, it was from strongly disagree to strongly agree. For teachers, it was from not true of me at all to always true of me. The reliability coefficient values of students (0.77) and teachers (0.62) are over 0.6. The number of respondents of students(97) and teachers(23) might be one of the major factors that lead to the different results of Cronbachαcoefficients.

Participants
Among 97 international students who participated in this questionnaire survey, 54.64% were male and 45.36% female.

Materials
The questionnaire survey for students comprised three parts. The first part was about demographic data composed of 4 issues: gender, nationality, length of Chinese study, and the university where he or she was learning. The second part was composed of 24 items. It covered the students' cognition of committing verbal errors in Chinese classes, CF 7-W issues in whether, what, who, when, how, how often, and what effects. The third part included 11 items concerning students' psychological responses to CF in Chinese classes and learning effects of CF. In each of the above items, there are five choices for them to choose from: 1. strongly disagree; 2. disagree; 3. not sure; 4. agree; 5. strongly agree.
For teachers, the questionnaire survey contained three parts, too. The first part was also about demographic data composed of 4 issues: gender, age, the major of the final degree, and length of TCSOL. The second part was the same as the students'. The third part included 4 items concerning teaching effects of CF. In each of the above sentences, there are five choices for them to choose from: 1. It is not true of me at all; 2. It is not true of me; 3. not sure; 4. It is usually true of me; 5. It is always true of me.
In the questionnaire, the items of the same category were not arranged in sequence but scattered through the whole questionnaire. Some items were affirmative, some negative to avoid the subjects' inertial choosing, consequently affecting the quality of the survey. But in this article, the items are arranged alphabetically and taxonomically for the sake of convenient reading.

Results and Analysis
The survey started from November 2018 and ended in January 2019. After 3 months survey, we obtained the valid results from 97 foreign students from all over the world and 23 Chinese teachers. We compared students' views with teachers' practices in CF in Chinese classes, and obtained the following results.

Comparison on the Cognition of Committing Verbal Errors
The first comparison was about students' and teachers' cognition of committing verbal errors (see Table 1a). *T 0.00(0) 0.00(0) 0.00 (0) 43.48 (10) 56.52(13) *S: students; *T: teachers *verbal: spoken and written language 100% of teachers against 87.63% of students thought it natural for foreign students to commit verbal errors in the process of Chinese learning. 10.31% of students were not sure about that, and 2.06% disagreed or strongly disagreed to it. But generally, like teachers, a majority of students agreed to it.
Pearson chi-square test was used for further analysis (see Table 1b). The result reveals that the asymptotic significance (2-sided) is more than 0.05, indicating that students' and teachers' cognition of committing verbal errors is not significantly different, hence this result accepts H 1 .

Comparison on Whether to Correct
Then the authors compared students' view and teachers' practice in whether it is necessary to correct students' verbal errors (see Table 2a).  (13) For the necessity of CF, 86.6% of students agreed or even strongly agreed to whether to correct. Similarly, 95.65% of teachers did usually or always practice it.
Pearson chi-square test was used for further analysis (see Table 2b). .250 Pearson chi-square test reveals that the asymptotic significance(2-sided) is more than 0.05, i.e. this result accepts H 2 .
S T 1.03(1) 0.00 (0) 0.00(0) 0.00 (0) 5.15(5) 0.00 (0) 55.67 (54) 73.91 (17) 38.14 (37) 26.09(6)  g. Teachers' error correction should focus on grammatical errors. S T 5.15(5) 0.00 (0) 12.37 (12) 39.13 (9) 27.84 (27) 43.48 (10) 48.45 (47) 17.39 (4) 6.19 (6) 0.00 (0) As for a, students' view was quite different from teachers' practice. 77.32% of students expected the teachers to correct every mistake they made, but only 17.39% of teachers usually or always practiced it. Similarly, in b and c, 67.01% and 65.98 % students expected teachers to correct their pronunciation and grammar mistakes respectively in class even if they did not interfere with comprehensibility, however only 21.74% and 30.43% of teachers respectively did so. For d, 86.96% of teachers corrected common or habitual errors, but only 46.39% of students agreed or strongly agreed to them, 37.11% of students against 8.70% of teachers were not sure about it. For e, compared with 89.70% of students who agreed or strongly agreed to oral correction, 73.91% of teachers practiced it. 26.09% of teachers against 8.525% of students were not sure about it. But for f, the percentage of students' (93.81%) view and that of teachers' (100%) practice in correcting students' written errors were almost congruent with each other. As to g, 54.64% of the students agreed or strongly agreed to focusing on grammatical errors, but only 17.39% of teachers practiced it. 43.48% of teachers against 27.84% of students were not sure about it.
Pearson chi-square test was used for further analysis (see Table 3b). We can see that all the asymptotic significances(2-sided) are less than 0.05 except f, indicating that students' view and teachers' practice in what to correct are significantly different in most cases, hence the results accept written errors, but reject H 3 in most cases.

Comparison on Who is to Correct
Who is to correct? Is it the teacher or the student himself or herself or the peer that should correct the errors (see Table 4a)?  (7) 2.06(2) 8.70 (2)  d. Teachers are supposed to guide students to find the correct answer by themselves.
S T 0.00(0) 0.00 (0) 2.06(2) 17.39 (4) 10.31 (10) 8.70 (2) 55.67(54) 34.78 (8) 31.96(31) 39.13 (9) The table indicated that 76.28% of students disagreed or strongly disagreed to a, in other words, they preferred teachers' correction. This percentage formed a sharp contrast with that of teachers' (17.39%). 11.34% of students against 47.83% of teachers were not sure about it. 77.32% of students agreed or strongly agreed to b, but only 47.83% of teachers believed so, and 30.44% of teachers did not think it true for them. Peer correction in c was agreed or strongly agreed to by 27.83% of students, and practiced by 39.13% of teachers. 34.02% of students against 43.48 of teachers were not sure about it. In d, 87.63% of students approved of self-correction, and 73.91% of teachers practiced it.
Pearson chi-square test was used for further analysis (see Table 4b). It is found that all the asymptotic significances (2-sided) are less than 0.05 except c. indicating that students' view and teachers' practice in teachers' correction and self-correction are significantly different, hence the results reject teachers' correction and self-correction, but accept that of peer correction, partially accept H 4.

Comparison on When to Correct
In this research, delayed CF was not examined because of the limited length of the paper. The comparison only involved immediate CF in pronunciation, vocabulary, grammar, language use, reading aloud, and answering questions (see Table 5a). were not sure whether they should correct students' mistakes immediately respectively in vocabulary, grammar, reading aloud even by interrupting them. None of the teachers corrected students' mistakes immediately even by interrupting them when they were answering questions in g, for 78.26% of teachers it was not true at all or not true, 21.74% were not sure about it; but a great majority of students (82.47%) agreed or strongly agreed to it, only 8.24% disagreed or strongly disagreed to it.
Pearson chi-square test was used for further analysis (see Table 5b). It is found that all the asymptotic significances(2-sided) are less than 0.05, indicating that students' views on immediate correction are significantly different from the teachers' practices, thus the results reject H 5 .

Comparison on How to Correct
The sixth comparison was CF type preferred by students and practiced by teachers (see Table 6a).  (0) 14.43 (14) 8.70 (2) 29.90(29) 52.17 (12) 32.99(32) 30.43 (7) 17.53 (17) 8.70 (2) It was found in a that direct CF was agreed or strongly agreed to by a great majority of students (81.44%), but only 26.09% of teachers practiced it. 56.52% teachers did not practice it. 57.73% of students accepted direct correction in b, but only 21.74% of teachers practiced it, and 39.13% did not practice it. In c, in order not to hurt students' self-esteem, indirect CF was agreed or strongly agreed to by 50.52% of students, but practiced by 39.13% of teachers. 29.90% of students against 52.17% of teachers were not sure about it.
Pearson chi-square test was used for further analysis (see Table 6b). It is found that the asymptotic significances (2-sided) in a, b are less than 0.05, indicating that students' view and teachers' practice in direct correction are significantly different, hence the results reject the direct correction. But the asymptotic significance (2-sided) of c is more than 0.05, indicating it accepts the indirect correction. Therefore, the results partially accept H 6 .

Comparison on How Often is to Correct in One Class
The seventh comparison was about students' view and teachers' practice in the maximum correction frequency in one class. We provided the options 1-2 times, 3 times, 4 times, 5-10 times, limitless, and not sure for them to choose from (see Table 7a).  (5) 13.40 (13) 52.17 (12) 3.09 (3) 0.00 (0) 31.96 (31) 0.00 (0) 24.74 (24) 8.70 (2) 13.40 (13) 17.39 (4) This table revealed that 73.91% of teachers' practice in the maximum correction was no more than three, but only 26.80% of students agreed to it. 31.96% of students accepted 5-10 times correction, 24.74% of them even maintained there should be no limit. The proportion of being not sure was similar to each other.
Pearson chi-square test was used for further analysis (see Table 7b). .003 It is found that the asymptotic significance(2-sided) is less than 0.05, indicating that students' view and teachers' practice in the maximum correction frequency in one class are significantly different, hence the result rejects H 7 .

Comparison on CF efficacy
As for this part, it is difficult for the researchers to give the same options for the subjects to choose from because it concerns positive or negative effects of CF on students' psychology and students' learning effects, which quite differ from the actual teaching efficacy confirmed by teachers. Therefore, the data were calculated separately.

CF Psychological and Learning Effects on Students
This part concerned students' psychological responses to CF and learning effects (see Table 8).  (37) We assumed that teachers' keeping correcting would make students unwilling to speak or afraid to speak. However, the results in a and b showed that 56.7% and 70.11% of students disagreed or strongly disagreed to them respectively. We assumed that teachers' correction would make students feel humiliated, nervous and anxious, shy, and even reluctant to answer the questions for a period of time. On the contrary, 76.29%, 70.10%, 64.95% and 70.10% in c, d, e, and f disagreed or strongly disagreed to it respectively. What's more, 56.70% felt lucky when their teachers corrected their errors in class, and 64.95% felt grateful to their teacher for correction, as were revealed in g and h. From i, j and k, it was learned that students learnt much whether from teachers' correction of other students' mistakes(75.25%) or from teachers' correction of their own mistakes(82.47%). Teachers' correction is beneficial to students' Chinese learning(92.78%). In other words, these results showed that CF mainly produces positive psychological effects and better learning on students.

CF Teaching Effects by Teachers
As for CF efficacy conformed by teachers in Chinese classes, we focused on the actual effect of CF in the teaching process. Will students commit the same mistakes after correction (see Table 9)?  (11 21.74(5) 17.39(4) 0.00 (0)  b. I try in vain to correct students' mistakes in class because they will be repeated 4.35(1) 39.13(9) 43.48 (10)  As a indicated that 60.87% did not think that CF would affect students enthusiasm in answering the questions although 21.74% were not sure of it, and 17.39% approved it. b displayed that 13.04% of teachers thought that CF was useless, and students would repeat the mistakes. But 43.48% were not sure about it, and less than half of the teachers (43.48%) didn't think it true to them. c indicated that 39.13% believed that the mistakes were not repeated in class, but were repeated after class. However, 47.83% were not sure about it. 13.04% didn't think it true at all for them. d revealed that only 17.39% of teachers believed that students would not repeat the corrected mistake, 52.17% were not sure of it, and 30.44% did not approve it. The results suggested that it was still not certain that CF was very pedagogically effective, and students would not repeat the mistakes after correction. Therefore, the results in 4.8.1and 4.8.2 partially accept H 8.

Discussion
Results revealed that both students and teachers conformed to each other in cognition of committing verbal errors in the process of Chinese learning. They were congruent with each other in whether to correct, which confirmed the results in the early study (Hendrickson, 1978). However, students' views and teachers' practices in what to correct were quite different in many aspects. Most students (77.32%) hoped that teachers could correct every mistake whether they interfered with comprehensibility or not, but only 17.39% of teachers practiced it. A great majority of teachers (86.98%) just focused on common or habitual errors. More students (54.64%) than teachers (17.39%) paid attention to correcting grammatical errors. Of course, they coincided with each other in written errors. These results considerably matched those controversial researches in early times (Ellis, 2009). As for who is to correct, most students preferred teachers' correction, but fewer teachers practiced it. More students than teachers approved self-correction. Although they consisted with each other in peer correction, the percentages of view (27.83%) and practice (39.13%) were not high. These results agreed to some extent with those in the previous studies (Chandler, 2003;Ellis, 2009;Hendrickson, 1978;Sato, 2013). With regard to when to correct, students' view on immediate correction were significantly different the teachers' practice. Most students preferred immediate correction. As far as how to correct is concerned, it is a very complicated issue, for it could be understood in a broad or narrow sense. In the broad sense, it refers to direct or indirect CF. But in the narrow sense, it involves specific CF methods and teachers' attitudes. In this study, broadly speaking, students and teachers consisted with each other in indirect correction, but not in direct one. Practically, how to correct is contingent on the type of courses, the level of learners, the beliefs of learners and teachers. This result agreed with the earlier researches, too. In terms of the maximum correction frequency in one class, students' view and teachers' practice were also measurably different. 73.91% of teachers' practice in the maximum correction was no more than three, but 56.65% of students accepted 5-10 times correction or limitless correction. As for the CF efficacy, from both students' and teachers' perspectives, CF mainly generated positive psychological effects and better learning effects on students, but it did not prove distinctly that CF was very effective, and students would not repeat the mistakes after correction.

Pedagogic Implications
Based on the results discussed above, the present study on CF might have implications for TCSOL, L1, L2 or FLLT. To satisfy students' CF need and minimize their repetition of mistakes, it is suggested that teachers correct as many mistakes as possible. Next, compared with peer correction, most students preferred teachers' correction because they were considered authoritative and their corrections tended to be reliable. Thus teachers are suggested to spend more time in CF. Besides, since self-correction under teachers' guidance was welcomed by the majority of students, it should be practiced more in and after class. Furthermore, most students reported a preference for immediate correction of pronunciation, vocabulary, grammar, language use, reading aloud and answering questions without caring about being interrupted, such results may eliminate teachers' uncertainty about when to correct mistakes or reduce their delay in immediate correction. Finally, as to frequency of CF, teachers can perform 5-10 times correction or even more in one class period without worrying that it would result in negative effects of CF.

Limitations and Prospective Researches
This article compared students' views with teachers' practices in CF in the context of TCSOL. As CF is an indispensable part in the process of teaching, our findings may be of some help to TCSOL, L1, L2 and FLLT.
But there are still some limitations in the research. Firstly, the research methodology was simple. It was only based on the network survey, lacking of class observations or students' performances. As a result, the most effective type of CF was not examined. Secondly, it just involved part of the contents in CF research for the limited length of the survey, and it did not cover all the detailed or specific issues. For instance, delayed correction was not involved. Thirdly, students at different levels and teachers teaching different courses were mixed together to compare, they were not divided into groups according to levels and courses. This might influence the results of the survey. Finally, cultural factors were not covered to go into the deeper side of the research.
Prospective researches may require diverse research methods and more detailed design to explore the relationships between students' views with teachers' practices in CF in the context of TCSOL. They can be interdisciplinary, taking more psychological and cultural factors into account, instead of a pedagogical longitude study.

Conclusion
To learn or teach a foreign language or L2 well is a hard journey. SLA study reveals that result of L2 acquisition is an implicit system, which concerns the "mental representation". Instruction can only affect explicit knowledge, but acquisition involves the development of implicit knowledge, and few scholars, if any, believe that explicit knowledge turns into implicit knowledge (VanPatten et al., 2020: 237, 288). Hence to improve the efficacy of CF and the quality of language learning and teaching, there is still a long way to go.