A Critical Review of Empirical Research on the Effect ( s ) of Written Feedback on Writing

This paper aims to review empirical studies on the relation and effect of teachers’ feedback on second language (L2) writing development. The purposes of this review are two folds; (a) to present and compare the findings of these studies, and (b) to make certain recommendations for future research based on the limitations and gaps in literature. Thus, this review does not merely provide a summary of previous studies, but it also includes a critical analysis of these studies. The approach employed is based on comparing the findings of these studies. We also question both the internal and external validity of these studies through examining their research methodology. Keywords— Feedback, L2 writing skill, critical review.

Observation (4 teachers were observed in 2 lessons) Quasi experiment (46 subjects) Interviews (7 faculty members) Questionnaires (both to teachers and students) Main conclusions The effect of feedback on students' writing accuracy and comp lexity is not significant No significant effect of the type of feedback on students' writing Acknowledge d limitations Duration of the treatment might not be enough ; 10 weeks Not enoughsample size Koen, Bit zer and Beets, 2012 Methodology: population and research instruments perspectives and practices of feedback in a Life Skills classroom are explored in order to reframe the purpose of lecturer feedback in promoting the quality of students' learning How do students in a Life Skills classroom experience feedback?
In attempting to answer this question, we explore how feedback, a key issue in assessment, can be used to inspire students to learn.A basic interpretative qualitative approach employing focus groups and semistructured interviews created the opportunity to exp lore how finalyear students in a Life Skills classroom experience four feedback issues.

Main conclusions
The findings suggest that students need to learn how to convert feedback into enhanced knowledge in order to understand precisely what aspects they need to improve it is not feedback in itself that will improve learning, but  The present study is a qualitative research descriptive statistics: the frequency of use of teachers' actual practices regarding the focus, the amount, and the explicitness of WC the use of a pilot study as it aims at exp loring and describing the beliefs of writing teachers about the use of WCF and their actual practices.

Main
Teachers' beliefs were greatly congruent with their practices conclusions regarding the amount and focus of WCF.Teachers' beliefs were incongruent concerning the exp licitness of WCF, the use of positive feedback, and the source of WCF.Several contextual factors related to the university overall context (e.g.t ime allocated to cover the syllabus), teachers (e.g.teaching experience), and students (e.g.proficiency levels) were found to affect teachers' practices As for the relationship between students' preferences and teachers' practices, complete congruence was found in terms of teachers being the key providers of feedback.students' preferences were incongruent with teachers' practices regarding the explicitness of WCF, the focus of WCF, and the provision of positive feedback Acknowledge d limitations

Teacher participants knew what this research study was investigating. one might argue that their responses might have been influenced
The topics of the students" essays varied fro m one teacher to another, which may have influenced teachers' feedback practices and it has thus biased the overall findings Students who participated in the present study were not at the same level (i.e. level three and level five students).This may have affected teachers' practices concerning the amount, focus, and exp licitness of WCF Kaewkasi, 2013, EFL

Methodology: population and research instruments
To investigate and compare the effects of direst corrective feedback with written meta linguistic explanation and direct corrective feedback with oral meta linguistic explanation 83 third year students Research instruments: proficiency test, pre and post tests: an identical pre and post test was administered before and after receiving the treatment to investigate the students' accurate improvement Focus on edpast and fragment.

Main conclusions
The The results show that there is not enough evidences in favor of erro r correction based on the subjects' action in the class within a limited time The results also portray that there is no direct relation between the number of the errors and receiving feedback in the form of erro r correction in experimental group Acknowledge d limitations Abdollahifa, 2014, EFL

Methodology: population and research instruments
This study hence seeks to discover the possible effects of teachers' interactional feedback on the learners' performance and motivation A group of 20 intermediate learners was randomly div ided into experimental and control group.The control group received only corrective feedback on form and organizational structure, whereas the experimental group also received additional co mments fro m the teacher about the ideas involved in the writing

Main conclusions
The study was successful to indicate the positive effect of interactional feedback both on the performance and motivation of the learners with a better teacher-learner rapport witnessed.

Acknowledge d limitations
The low nu mber of participants in the study which decreases the generalizabiliy of the study.The validity of the study could have been increased if type of the CF was decided in terms of directness This table provides a summary of the main recent studies that have investigated the issue of feedback (especially teachers' written feedback) in relat ion to writing.Based on an in-depth review of these studies, the following points should be considered: developing students' writings are not consistent.In relation to the objectives of the current study, we believe it would be useful to distinguish between these studies based on their main findings.In other words, we would like to present the studies that are for and the studies that are against the provision of feedback.Thus, the studies which have found a positive feedback on writing are p resented first followed by the studies which have foun d no connection between feedback and the development of students' writ ings.The choice of such order is driven by organization matters only.

III. DEB ATE ABOUT THE EFFECTIVEN ESS OF FEEDB ACK ON WRITING
There has been a long debate about the implementation of corrective feedback in L2 classes.Researchers have advanced both theoretical and practical argu ments to support their views either for or against using corrective feedback.In this study, we aspire to contribute to the development of second language learning through investigating the possible relation between corrective feedback and the development of students' writ ing.This article reviews the literature on the issue in order to shed some light on this debate through presenting the different arguments in favor of o r against the usefulness of corrective feedback.Researchers have focused on the effect of feedback on the development of students' accuracy.This can show that the focus is still on the final product of students rather than on the writing process.This implies that the focus on writing process may be more help ful.Teachers and researchers should provide feedback on the content, steps, first drafts, and early stages in the writing process.Liu (2008) notes that although both direct and indirect feedback help students self edit their drafts, direct feedback does not help students to improve in different papers.Besides, the findings of Ma (2006) suggests the effectiveness of feedback as it has been found that students in both feedback conditions experienced change in their writing accuracy as a consequence of receiving feedback and taking writing courses.Both form and content related feedback positively affect L2 students writing accuracy.Bitchener, Young and Cameron (2005) support the findings for the effect iveness of feedback in L2 writ ing.It has been found that the combination of full direct feedback and teachers' written conferencing has been proven to be effective in helping students improve their writing accuracy especially their use of simp le past and articles (treatable errors as opposed to untreatable errors such as the use of prepositions).The findings of Tootkaboni and Ashouri (2014) also support the use of feedback although they have found no significant difference between direct and indirect feedback.Maleki and Eslami (2013) have also noticed a strong link between providing feedback to language learners and their writing accuracy.In the study of Maleki and Eslami (2013), the long term effects of direct and indirect feedback have been examined.It has been concluded that indirect written corrective feedback has more positive effects on L2 students' writing abilities.In relation to the long term effects of feedback, Van Binglan and Jia (2011) conclude that correction and exp licit explanation can be more helpful for long term progress in writing accuracy.Moreover, based on the study of Farrokh i and Sattarpour (2012), it has been noticed that both experimental groups outperformed the control group in terms of accurate use of definite and indefinite articles.Kaewkasi (2013) also notes that written corrective feedback helps students to use ed past.Implicit and explicit feedback can also help students to decrease the number of errors after the treatment (Falhasiri and Tavako li, 2011).The results of Abdollahifa (2014) have succeeded to show that there is positive impact of interactional feedback on the motivation as well as performance with a better teachers' students' rapport.Finally, Chandler (2003) indicates the positive effects of some forms of feedback including direct correction and underlining of error.A ll these studies demonstrate that feedback has a pos itive effect on the development of students writ ing accuracy.However, some researchers, as will be discussed in the next sub section, are not fully convinced of the role of feedback in enhancing students writing skill.

EVIDENCE AGAINS T THE EFFECTIVEN ESS OF FEEDB ACK ON WRITING Many researchers
are not yet convinced of the role that feedback can play in helping students improve their writing abilities.They argue that there is not enough research evidence for the effect of feedback on writing.They justify their point of view by the lack of consistent results regarding the effect of feedback on writ ing.Mahwah (2012) has found that the effect of feedback on students' writing accuracy and complexity is not significant.Another conclusion of this study is that no significant effect of the type of feedback on students writing develop ment is noticed.Truscott (1996Truscott ( , 1999Truscott ( , 2004Truscott ( , 2007Truscott ( , 2009) also argues against the use of grammar correct ion in L2 writing and that grammar correct ion has no place in L2 writing classes and therefore should be abandoned.Ghabanchi (2011) also supports the view against grammar correction in writing classes.The results of this study demonstrate that there is not enough evidence in favor of g rammar correction.Researchers against feedback provision maintain that there is no direct relation between the decrease in the number of errors in students writing and receiving feedback.Other studies, such as Koen et, al (2012) and Evans, et, al, (2010) advocate the view that it is not feedback in itself that will imp rove students writing, but the way students understand what to do with such feedback.It is, thus, reco mmended that students be trained how to convert the feedback they receive into enhanced knowledge in order to know precisely what aspects of feedback can be more useful.

IV. LIMITATIONS AND GAPS IN LITERATURE
This section provides a critical rev iew especially of the emp irical studies that have investigated the effect of feedback on writing.Before d iscussing the limitations of these studies, it has to be noted that some researchers have acknowledged the limitat ions of their studies while others haven't.Based on a deep analysis of these studies, it can be noticed that most of the researchers have investigated the effect of feedback or some specific form(s) of feedback on students' writing accuracy.Such studies include Ma (2006), Bitchener, et. al (2005) while some other studies focus on some specific errors.Another remark is that although the reviewed studies have tried to co me up with mo re reliab le data to support their conclusions, they still have different limitat ions which can be a threat to their internal valid ity.These limitations are mainly related to focus of these studies, sample size, data collection methods, the tests, treatment time, and other issues related to methodology.
Most emp irical studies rely on co mparing the performance of the participants in the pre and post tests.They make conclusions about the effectiveness of feedback with reference to the number of errors students make in the pretest and posttest.Researchers conclude that there is a positive effect of feedback if they simply find that the number of students' errors decreases in the posttest.However, Truscott (1996Truscott ( , 1999Truscott ( , 2004Truscott ( , 2007Truscott ( , 2009) ) points out that a reduction in the number of errors cannot be considered as a proof of learn ing (learning how to write in this case).Another limitation of some of these studies is related to the approach adopted to test the effectiveness of feedback on students' writ ing abilities.In other words, the participants were required to rewrite (in the posttest) the same drafts or to write on the same topics.In other cases the pretest and the posttest were the same.This can be a real threat to the internal validity of these studies as familiarizat ion is one of the threats to research validity (Campbell and Stanley, 1963).This imp lies that there might be an imp rovement in students 'writings not because of the feedback provided but simp ly because the students have become familiar with the topic.Such studies include Kaewkasi (2013), in wh ich the pre and post tests were identical.It is widely recognized that research validity and reliab ility are main ly related to its methodology.In other words, the more one varies the sources and methods to collect data, the more they can test the research hypotheses with confidence.Adopting a mixed method approach provides the research with both qualitative and quantitative data.Mubarak (2013), Liu (2008), and Ghabanchi (2011) have relied on both qualitative and quantitative data.However, some studies either rely on qualitative o r quantitative data: Van Binglan and Jia (2010) and Maleki and Eslami (2013) used quantitative data only while Koen et, al (2012), Evan et.al, and Alkhatib (2015) based their conclusions on qualitative data.Sample size and sampling procedure are also key elements in research.The conclusions of a given study are mo re likely to be valid if the samp le size is representative.It is representative when the number of participants is at least 30 or more and they are randomly selected (Hatch and Lazaratton, 1991).In most of the reviewed empirical studies the participants were not randomly selected.As for the sample size, many studies have relied on limited number of participants including the study of Mubarak (2013) Most of the reviewed studies have dealt with the effect of feedback on the develop ment of learners' accuracy with more focus on a specific aspect of writing such as the use of simp le past or the use of articles.Truscott (1996Truscott ( , 1999Truscott ( , 2004Truscott ( , 2007Truscott ( , 2009) ) argues against some studies that advocate the use of corrective feedback.He concludes that the fact that students manage to eliminate so me errors in subsequent drafts is not a proof of learning.However, writing need to be conceptualized as a set of various skills.
Writing is considered as a means for learners to express and share their ideas.It is a means of communication.That is why; more emphasis has to be on the communicative aspects of writing.Besides, the perception of feedback has to be reconsidered.At the level of methodology, the experiments in future studiesshould be based on comparing the performance of the control and the experimental g roups in both the pre and post tests, whereas in many studies there are no control groups.Future studies have also tried to overco me the limitat ion related to the treatment duration.Many researchers have to acknowledge and take into account the possible effect of the limited period of the experiment on the effectiveness of the treatment.We believe that this that the treatment would have greater effect if the period were longer.
Many researchers have recommended testing the effects of feedback in a new p iece of writing or in delayed post tests.
Based on this, researchersare invited to investigate the effectiveness of feedback through requiring students to write on a new topic (the pre-test topic should be different fro m the post-test one and different fro m a ll other topics covered during the treatment period).Population size is also another difference.So me of the emp irical studies have relied on a limited nu mber of part icipants (sometimes 8 o r 10 students).However, mo re than 110 students participated in the experiment.The participants were randomly assigned to experimental and control groups.
Research on the relat ion or effect of feedback on writing has generally depended either on qualitative or quantitative data.It is rarely when researchers use a mixed method approach.Co mbining these methods helps to be more confident in the findings and in the process of interpretation.
The participants' language proficiency is a pro minent variable in applied linguistics studies.However, mostly all the reviewed empirical studies have disregarded the influence of the learners language proficiency on the differences that might be noticed in the performance of students in the tests.Language proficiency is a determinant variable that need to be controlled for.

VI. CONCLUSION
This review has been about the value and efficacy of written feedback on the development of students writing skill.This review provided a ground for future research on feedback and writing.Th is article presents a summary of previous studies in order to draw attention to the debate regarding the effectiveness of feedback.The art icle also makes some recommendations for future research based on the limitations and gaps discussed in the critical review.
fu ll, exp licit written feedback and one-to-one conference feedback enabled them to use the past simple tense and the definite article with significantly greater accuracy in new p ieces of writing than was the case with their use of prepositions direct oral feedback in co mbination with direct written feedback had a greater effect than direct written feedback alone the combined feedback option facilitated improvement in the more ''treatable'', rule -governed features (the past simple tense and the definite article) than in the less ''treatable'' feature (prepositions) results of the present study indicate that WCF either WM E o r OM E are effective measures in helping students effectively learn both reactive and proactive FoF can help improve the writing skill of the students.Students who received the proactive FoF significantly outperformed the students in the other group which can underline the effectiveness of this type of FoF.The students in reactive FoF class could not present the same level of skill four months after the intervention finished.That is, proactive type of FoF is significantly mo re effective in imp roving writ ing skill among the students.
Although the issue of writing and feedback has attracted researchers from different disciplines, most empirical studies have not yet succeeded to come up with consistent results that clearly highlight or co mp letely reject the effect of feedback on the development of learners' writ ing abilities.It has also to be noted that based on the number of emp irical studies that have examined the effect of feedback on writ ing, most research in the L2 context is recent.This imp lies that this research area has not been yet adequately investigated. Serious attempts have been made to better understand the possible link between feedback and writing based on the number of published articles and papers starting fro m the beginning of this century.However, many researchers have investigated the effect of different forms of feedback on the development of learners' writing skills although there is not yet a complete agreement on the relation between feedback and writing. Despite the fact that researchers have investigated writing in relation to feedback fro m different perspectives, no study is free fro m some limitations as shown in the table.Based on these shortcomings, a lot of reco mmendations have been proposed for future research.
Maleki and Eslami (2013)riate research design is another key issue in research.Most of the reviewed emp irical studies are quasi-experimental.It is said that the pretestpost test design is more appropriate.Ho wever, in some studies likeMa (2006)no control group is used.Another limitation is related to the time of the treat ment and the number of papers co mpared.As for the treat ment time,Mubarak (2013),Liu (2008), andMa (2006)acknowledge that the duration of the treatment is not sufficient.Concerning the number of papers considered,Liu (2008)compares only two drafts and only four papers are evaluated inMa (2006).Another limitation is related to language proficiency of the participants as a confounding variable.Researchers are required to control for this variable regard ing its great effects on differences in performance of students.The participants should constitute homogenous group.Yet, nearly all the reviewed studies in this paper have neglected this variable exceptMaleki and Eslami (2013)andKaewkasi (2014).