Incongruence Between Learning Style and Written Corrective Feedback Type: Mediating Effect of Implicit Theory of Learning Style

Mediating


Incongruence Between Learning Style and Written Corrective Feedback Type: Mediating Effect of Implicit Theory of Learning Style
Despite a large number of studies on written corrective (WCF), to date, its potential for second language (L2) development is still a controversial issue (e.g., Fazio, 2001;Kepner, 1991;Polio, et al., 1998;Truscott, 1996Truscott, , 1999Truscott, , 2004Truscott, , 2007, among others), among others).Also, the same controversy exists in the right degree of feedback explicitness; the amount that provides students with enough information to form a new hypothesis and does not pose an unnecessary burden on teachers' shoulders.There are studies that report the higher efficacy of more explicit types of corrective feedback (Chandler, 2003;Ellis, et al., 2006;Suzuki, et al., 2019;Van Beuningenet al., 2008, 2012); some other studies, on the other hand, claim the same for implicit types (Ferris & Hedgcock, 2005;Lalande, 1982), and even few others declare no difference (Bitchener & Knoch, 2008, 2009, 2010;Frantzen, 1995;Robb et al., 1986).
To address these issues, recently, some scholars have resorted to individual differences (ID) to explain the contradicting results (Guénette, 2007;Kormos, 2012;Rahimi, 2015;Sheen, 2007;Shintani & Ellis, 2015;Stefanou & Revesz, 2015) arguing that feedback efficacy is highly associated with learners' individual differences, including motivation and attitude, depth of processing, L2 proficiency, and learning styles.One of these factors that seem to directly contribute to learners' benefiting from teacher feedback or language learning, for that matter, is learning style.However, it is only recently that scholars have started to show interest in the relationship between students' learning styles and how well they benefit from WCF (see Rahimi, 2015;Samie, 2021), recommending that teachers should take their students' learning style into account in deciding what feedback type to use when addressing their written errors.
More recently, however, this solution has been argued as not being ecologically valid due to the difficulty of teachers adapting their instruction to students' different styles in natural classroom settings as well as the malleable nature of learning styles and their capacity to be stretched (Zhang, 2013); hence, students can adapt their learning styles and go beyond their comfort zone even if the instruction they receive does not match their styles (Gregersen & MacIntyre, 2014;Young, 2010).
Nonetheless, the proposition of style stretching considers the students to be equally willing to face the challenge of stretching their learning style while the research in the field of psychology shows that it is not as straightforward as this.As Hong et al.(1995) point out, learners are not equally able to stretch their individual styles because they have their own theories about human traits.In other words, students' theories (entity or incremental) about specific traits (say, intelligence, learning style, etc.) determine whether or not and to what extent these traits are fixed or malleable.For instance, students with an entity theory believe that a trait is fixed and uncontrollable while those with an incremental theory consider a trait malleable, increasable, and controllable (Dweck & Leggett, 1988).Hong et al. (1995) call these theories, which learners may not always articulate, implicit theories.
No study, however, has ever focused on the extent to which L2 writers can stretch their learning style in response to the feedback that is against it; there is also no study examining the extent to which their implicit theories of learning style can contribute to style stretching (if any).The only existing studies in the area of L2 writing (Waller, 2015;Waller and Papi, 2017) have focused on the implicit theory of intelligence and have found that it is a strong predictor of student motivation, and students' motivation, in turn, is strongly associated with their feedbackseeking orientation.No study, however, has investigated the extent to which implicit theory of learning style can mediate L2 writers' uptake of WCF and whether or not and to what extent they can respond to the WCF that mismatches their learning style.Hence, the present study probes the extent to which the students with deductive vs inductive learning styles benefit from WCF that is against their learning style (i.e., implicit vs. explicit WCF) and if and to what extent their implicit theory of learning style mediates their style stretching behaviour.

WCF
WCF has been considered an indispensable aspect of writing pedagogy due to its promising and consolidating role in the development of L2 writing skills (Hyland & Hyland, 2006).It is particularly very important in the process approach to writing that entails multiple drafting and revision of each draft.
Research on WCF started with studies that investigated its efficacy in general (e.g., Fazio, 2001;Ferris, 1999;Sheppard, 1992;Truscott, 1996).Later, studies on WCF aimed at identifying the most effective strategies for providing WCF that sparked off a dichotomy of various approaches to WCF-namely, direct vs. indirect and focused vs. comprehensive.

The Contribution of Individual Differences to Feedback Efficacy
Ellis's (2010) componential framework for investigating WCF highlights the contribution of contextual as well as individual difference factors to WCF efficacy.Based on this framework, learning outcomes are the product of the interplay between contextual and individual factors as well as teacher WCF.Several studies have examined the role of contextual and individual factors in feedback efficacy, many of which have focused on the role of affective factors (Cumming, 1995;Ferris, 1995;Ferris & Roberts, 2001;Gram, 2005;Goldstein, 2006;Hedgcock & Lefkowitz, 1994;Hyland, 1998Hyland, , 2011;;Lee, 2004;Leki, 1991;Storch & Wigglesworth, 2010;Swain, 2006;Swain & Lapkin, 2003;Zacharias, 2007).
There are a few studies that have focused on the contribution of cognitive processes to WCF efficacy (Adrada-Rafael & Filgueras-Gomez, 2019;Kim & Bowles, 2019;Qui & Lapkin, 2001;Sachs & Polio, 2007;Stefanou & Reves, 2015).For instance, Kim and Bowles (2019) examined the cognitive processes employed in different types of feedback and investigated whether there is a relationship between the type of feedback, error type, and depth of processing.
The findings showed that the students processed global errors more deeply than local ones in response to reformulation while the depth of the process was the opposite while they were engaged with direct feedback.Also, Sheen (2007) and Stefanou and Reves (2015) showed the contribution of learners' analytic ability to their uptake of WCF.As mentioned earlier, recently, there has been an increasing interest in studying the role of learning style in the uptake of WCF.

Learning Styles
Learning styles, according to Dornyei and Ryan (2015) is a confusing construct and literature suggests various definitions and classifications for this construct (see, for example, Coffield, et al., 2004;Peterson, et al., 2014).Despite these controversies, Dornyei and Ryan (2015) argue that "the current state of confusion is merely due to our insufficient knowledge rather than the scientific inadequacy of the concept" (p.107).In fact, they believe that, at an intuitive level, there is general consensus on the definition of learning styles construct; that is, "different learners can approach the same learning task in quite different ways and…this variation is not indefinite but is characterized by systematic patterns."(p.108).They call these systematic patterns learning styles.The difference mainly lies in the labels and style dimensions.
In the present study, we have adopted the cognitive styles construct offered by Riding (2002).Dornyei and Ryan (2015) find Riding's system of cognitive styles a "powerful and parsimonious system of cognitive styles that …remedied the shortcomings of past styles research while maintaining the attractive features of the concept."(p.115).Riding's construct of cognitive styles has two dimensions: wholist-analytic and verbal imagery.Wholist-analytic dimension determines if the individual sees the concepts as a whole or in parts while the verbalimagery dimension identifies if the individual tests to think in terms of pictures or words.
With regard to learning style classification, since, in the present study, we focus on language learning styles, we have adopted The Ehrman and Leaver classification of learning styles (Ehrman & Leaver, 2003), which is in line with Ridings' construct because, "similar to Riding's theory in that it reorganizes several established style dimensions under a new, comprehensive, and parsimonious superordinate construct" (Dornyei & Ryan, 2015, p. 130).Based on this classification, learning styles can be categorized into 10 sub-dimensions, namely, field dependent-independent, random-sequential, global-particular, inductive-deductive, synthetic-analytic, analog-digital, concrete-abstract, levelling-sharpening, impulsive-reflective.
To the best of our knowledge, only one study has examined if and to what extent learning styles contribute to feedback uptake.Rahimi (2015), using Group Embedded Figure Test for learning style, showed how students with field independence retain teachers' WCF more than those with a dependent style.
Among learning style constructs, the deductive/inductive learning style dichotomy (Ehrman & Leaver, 2003) has been chosen for this study.This construct is of great interest to WCF studies as it coincides with the explicit/implicit dichotomy of CF already explained.As DeKeyser (1994) argues that whereas inductive/deductive and explicit/implicit are "independent in principle, they tend to coincide in practice" (p.188).Based on his proposition, one can expect that in a writing classroom, deductive learners would benefit more from a more explicit type of WCF while inductive learners would be content with a more implicit type.Yates (2000) and Kozhevnikov, et al. (2014) argue that it is an unrealistic goal to expect teachers to precisely measure their students' cognitive styles and then match their teaching to individual students' styles.Rather, they underline the idea of helping students to "self-regulate their learning and flexibly switch between styles, according to situational requirements" (Kozhevnikov, et al., 2014, p. 12).Felder and Henriques (1995), too, point out that the students should be regularly involved in the activities that require the use of their less preferred learning styles; therefore, they need to go beyond their comfort zone and "stretch" their style.

Malleability of Learning Style
To help students stretch their learning styles, Kolb (2015) proposes an intentional mismatch between the learning environment and individuals' style, which holds the potential of flourishing individuals' weaknesses in certain learning styles and therefore lead to the expansion of their style repertoire, enabling them to function in a variety of learning environments.
The suggestion of going beyond the comfort zone, also known as style stretching, entails that this construct is deemed as malleable rather than a fixed trait.One of the models that directly addresses the malleability of learning style is Curry's Onion model (1983).This model categorizes learning style measures into three strata, namely instructional preference, information processing style, and personality style.
Instructional Preference is the outer and least stable layer that refers to the individuals' preference of learning environment, learner expectations, teacher expectations, and other external features.The middle layer, information processing style, assesses an individual's intellectual approach to absorbing information, and as it is not directly influenced by environmental factors, it shows more stability compared to the outer layer despite being modifiable-if aimed by educators.This layer is of great importance to educators as it is stable enough to assess and malleable enough to mould.Learning style falls in this layer, which makes this construct even more suitable for this study.The most inner layer, cognitive personality style, measures an individual's approach to adapting and assimilating information which shows the most stability and the least modifiability.
As mentioned earlier, the present study focuses on the deductive/inductive learning style and aims at examining if and to what extent deductive or inductive learners can stretch their style and benefit from mismatching WCF.However, the question is whether or not we can expect students to stretch their style merely by exposing them to mismatched learning situations.In other words, the question is if the students are equally willing to face this challenge.

Implicit Theories of an Individual's Traits
Although, as mentioned above, it has been recommended that teachers help students stretch their learning styles by exposing them to tasks that mismatch their styles (Kolb, 2015;Kozhevnikov, et al., 2014;Yates, 2000), the idea the learners hold about their traits accounts for how they react when faced with unfavourable situations.These ideas are referred to as implicit theories (Dweck & Leggett, 1988).One of the models that attempt to depict these theories is proposed by Dweck (2000).Dweck claims that this "model can shed light on other important personal and interpersonal phenomena, such as, why contrary to popular opinion, confidence, self-esteem, and past success are not the keys to adaptive functioning" (p.xiii).In fact, individuals might hold two different "theories" about their traits: a fixed, entity theory and a malleable, incremental one (Bandura & Dweck, 1985).These theories, which are "people's basic assumptions about themselves and their world" (Dweck, 1996, p. 69), define individuals' reactions, feelings, and thoughts.For instance, learners with an entity theory of intelligence tend to show a helpless reaction, question their intelligence, and show lower persistence while the holders of incremental theory of intelligence show mastery-oriented response, working harder, and remaining focused in the face of setbacks (Dweck, 2000).Dweck and Leggett (1988) believe that, in addition to learners' different behaviour patterns, individuals' implicit theories account for the goals they set for themselves.Dweck and Leggett introduce two types of goals: (1) performance goals, and (2) learning goals.While individuals with a performance goal show maladaptive behaviour and seek validation of their abilities and "looking smart", those with a learning goal show adaptive behaviour and try to learn something new.Studies conducted by Leggett (1985), and Dweck and Leggett (1988) showed a significant relationship between individuals' implicit theory and their goals; entity theorists tend to hold performance goals while incremental theorists set learning goals.
One of the traits which has been largely investigated, using Dweck's (2000) framework, is intelligence.Many studies show the importance of the implicit theory of intelligence on learners' ability to adapt themselves to challenging situations.De Castella and Byrne (2015), for instance, found entity belief about intelligence to be predictive of lower endorsement of achievement goals, academic self-handicapping, and poorer self-reported academic grades.Elliott and Dweck (1988) also reported that individuals with a learning goal, which is one of the characteristics of individuals with an incremental theory of intelligence, exhibited more challenge-seeking and mastery-oriented responses to failure, and this was regardless of the perceived ability of the participants.Blackwell et al. (2007), conducted two studies on the role of implicit theories of intelligence in seventh graders' mathematics achievement.In the first study, they observed higher achievement in students with the incremental theory of intelligence.In the second study, through intervention, they could promote positive change in students with an entity theory of intelligence.Also, Farrell and Dweck (1985) observed the higher performance of students with learning goals compared to the participants with performance goals when they were presented with new and challenging material.The same pattern is also noticed in other studies (e.g., Dweck & Sorich, 1999;Grant & Dweck, 2003;Henderson & Dweck, 1990;Hong, et al., 1999;Mueller & Dweck, 1998).
The first move to use this theory in the field of second language writing was made by Waller and Papi (2017), who investigated the relationship between language learners' implicit theories of writing intelligence, motivation, and feedback orientation.In this study, they reported a positive relationship between feedback-seeking orientation, motivation, and incremental theory of writing intelligence.Students with an incremental theory of writing intelligence showed more feedback-seeking reactions while students with the incremental theory of writing intelligence showed more feedback avoidance orientation.Also, incremental students manifested higher motivation compared to their entity peers.
Although implicit theories have been frequently used in the field of psychology to explain failure in the face of challenges, in L2 learning, there has been little attempt to exploit the potentials of this theory to explain the conflicting results in employing different approaches to teaching.Considering the potential of implicit theories in depicting "the patterns of vulnerability and hardiness that students display as they confront difficulty" (Dweck, 2000, p.4), more application of this theory seems to be urgent.

Gap in the Literature
Given that learners find it challenging to engage with materials that are presented in a way that pushes them outside their comfort zone (i.e.their learning style), looking at how they handle this challenge through the lens of implicit theories seems to address many unresolved issues in WCF, including whether or not students with different learning styles (deductive/inductive, here) would be equally able to stretch their style of learning to benefit from teacher's WCF.Thus far, however, to the best of our knowledge, there has been only one attempt to investigate the possibility of mediating effect of implicit theory in L2 learning and, more particularly, in written feedback (Waller & Papi, 2017); therefore, to fill the gap in the literature and contribute to WCF domain, and respond to Waller and Papi's call for further investigation, this study was conducted to inspect the relationship between students' implicit theory of learning style and their ability to benefit from WCF mismatching their learning style.

Objectives of the Study
The objective of this study is to investigate the mediating effect of implicit theory on students' uptake of WCF that mismatches their learning style.Moreover, this study aims to probe the difference between two types of learning styles, i.e., deductive, and inductive styles, in coping with mismatched WCF.

Research Questions
To achieve these objectives two research questions have been posed: 1. Is there any difference between inductive learners and deductive learners in terms of their ability to benefit from mismatched WCF? 2. Is there any relationship between learners' implicit theory of learning style and their ability to adapt themselves to the WCF that does not match their learning style?

Significance of the Study
There are some important areas where this study makes contributions.First of all, this study is an answer to the call for more investigation into the moderating effect of individual factors on learners' uptake of WCF (Ellis, 2008); thereby, the findings provide a new perspective into why feedback does not always lead to L2 writing by taking into account not only students' learning styles but also their implicit theory, which is believed to affect learners' goal and predict their behaviour in face of setbacks (see the literature review section).Secondly, in this study, the interaction between different moderating factors (type of WCF and learning style) on L2 learners' written accuracy will be investigated.Thirdly, by bringing to light the interplay between students' implicit theory of learning style and their success in adapting themselves to the provided WCF, the study proposes a more pedagogically sound plan to exploit individual differences in the writing classroom.

Participants and Setting
A convenient sample of 40 (female=25, male=15) English majors at an Iranian university, with an average age of 22, participated in this study.The students were in their fourth year of study, participating in two intact essay writing classes.All the participants had passed two grammar and two writing courses before taking their essay writing course.One of the writing courses focused on paragraph writing and the other on the basics of essay writing where they learned to write descriptive and expository essays.The essay writing course had two important aims: it focused both on improving learners' knowledge of argumentative writing as well as on improving their written accuracy.Both classes were held by the second researcher.
Based on their performance on a learning style questionnaire (see the instrument section, below), the participants were divided into inductive (n=22) and deductive (n=18) groups.Also, the results of their responses to the implicit theory of learning style (see the instrument section) revealed that eight students from the inductive group and nine from the deductive group had an entity theory while 14 students from the inductive and nine from the deductive group had an incremental theory of learning style.In addition, the results of a writing pre-test showed no significant difference between the written accuracy of these two groups (See Table 3

, below).
There was no difference between the treatments the students in the two classes received.In fact, it was the participants' deductivity/inductivity style, rather than their random participation in the two classes was the criterion for grouping them.As the students wrote their essays individually and received their own unique feedback, the researcher could tailor each individual's feedback based on their learning style so that it would be at odds with their preferred type of feedback.

Materials and Instruments
Two instruments were employed in the present study.First, in order to determine the participants' learning style (i.e., inductive or deductive), Cohen, et al.'s (2001) questionnaire was utilized.When administered to our pilot group-49 bachelor students who were also in their fourth year of studying English literature at the same university, the test showed an acceptable level of item consistency, with Cronbach's alpha of .78.This survey consists of 110 items measuring 11 constructs.In the present study, the six items measuring deductivity/inductivity, based on how students deal with grammatical rules, were administered to the participants.Three items of this questionnaire are designed to measure deductivity and three items inductivity.For example, agreement with the statement "I like to start with rules and theories rather than specific examples" indicates deductively whereas the item "I figure out rules based on the way I see language forms behaving over time."distinguishes inductive students from deductive ones.Students were instructed to rate each statement on a 5-point Likert scale from 1 (never) to 5 (always).
The second instrument was an adapted version of Dweck's (2000) implicit theory of intelligence questionnaire administered to identify the students' implicit theory of learning style.The items of the questionnaire were altered to measure learners' implicit theory of learning style (See Table 1).The questionnaire had six items on a 6-point Likert scale from 1(strongly agree) to 6 (strongly disagree).The questionnaire also showed a high level of consistency when administered to the same pilot group mentioned above, with Cronbach's alpha reaching 0.83.Besides, to assess the reliability of this new questionnaire, students' answers to the adapted version were correlated against their answers to the original version.Based on Dornyei (2007) "if two tests correlate with each other in the order of 0.6, we can say that they measure more or less the same thing."(p.223).The correlation between the theory of style and the theory of intelligence was .61which is higher than the needed correlation index to qualify them as the same construct.The materials employed for the present study were two argumentative essays out of a total of 6 essays the participants wrote during an academic semester.The first essay written by the students in week 5 was utilized as the pre-test and the last one written in week 15 as the posttest.The topics of the essays and the texts mainly focused on educational and social issues that were familiar to the students.

Feedback Type
In this study, participants received comprehensive WCF.As for feedback type, since the aim of the study was to compare deductive and inductive learning styles in benefiting from mismatching feedback types, as a more explicit feedback type, we used brief grammatical description (BGD), which involves grammatical explanation, for inductive learners; for deductive learners, we used coded feedback (codes) as a more implicit type of feedback.To provide BGD for the inductive group's error, we left comments on students' errors providing a short grammatical description for each highlighted error in the bubble comment (See Figure 1).For the deductive group, on the other hand, we highlighted the errors and provided codes in the bubble comments in the margin (see Figure 2).We used the coding scheme suggested by Ferris and Roberts's ( 2001) categories (see Table 2) and no further explanation was provided to them other than a chart that clarified the meaning of the codes.The reason why we used indirect (implicit) feedback was first because indirect feedback turns out to be more effective when longterm accuracy improvement is concerned; this type of feedback leads to "increased student engagement and attention to forms and problems."(Ferris, 2003, p. 52).Besides, we needed to provide two types of feedback with different degrees of explicitness for deductive and inductive students; only indirect feedback gave us this possibility.Students wrote their essays in Google Docs and shared them with the instructor.

Data Collection Procedure
The course lasted 16 weeks.At the beginning of the semester, participants completed the two questionnaires introduced above.The first three weeks of the course were devoted to introducing the conventions of text-based argumentative writing through deconstructing samples of argumentative essays and the teacher's introduction of the moves and components of the argumentative essay.In week four, to practice the argumentative genre, students wrote a group essay, on which they only received content feedback and no grammatical error was marked.
Starting from week five, every two weeks were spent on in-class writing of an essay and revising it.That is, in the first week, they wrote an essay in response to a controversial issue discussed in an article, either showing agreement or disagreement on the topic using examples, logic, or other sources.They had two hours to finish the assignment and submit it through Google Docs.Then, in two days, they received WCF on their essays; this feedback, as mentioned earlier, mismatched their learning style; that is, deductive students received coded feedback while inductive ones received BGE.They were then given two hours to revise their linguistic errors and submit the essay again.Since we focused on their accuracy in the subsequent essays and not during the revisions, we gave them the freedom to dictionaries while revising their essays.Then, in the following class session, they received content feedback and revised their writing in class.The same procedure was followed throughout the term until week 16.In total, they wrote six essays and revised them both language and content-wise.

Data Analysis
In order to measure the learners' implicit theory of learning style, a numerical value was allocated to the students' answers (strongly agree=1, agree=2, mostly agree=3, mostly disagree=4, disagree=5, strongly disagree=6).After reversing the items measuring the incremental theory of learning style, the scores on the questionnaire were averaged in order to obtain the students' implicit theory of learning style, with participants holding an average equal to or below 3 being classified as entity, and equal to or above 4 as incremental.Students with an average between 3 and 4 were categorized as neutral and, consequently, their data were removed from the data analysis.
To measure the learners' learning styles, a numerical value was allocated to the students' answers (never=1, rarely=2, sometimes=3, often=4, always=5).After reversing the items measuring deductive learning, the total scores were averaged.The participants with an average above 3 were categorized as inductive, while an average below 3 was associated with being deductive.Moreover, students with an average of 3 were categorized as neutral and, consequently, their data were removed from the data analysis.
Following Zhang (1987) and Chandler (2003), to measure linguistic accuracy, the number of errors per 100 words was calculated.In this procedure, to obtain an error ratio, the total number of errors in each essay was enumerated, divided by the total number of words in that essay, and multiplied by 100.
To answer the first research question, that is, if there is any difference between inductive learners and deductive learners in terms of their ability to benefit from mismatched WCF, we ran a mixed between-within subjects ANOVA.To answer the second question, that is, the effect of the learners' implicit theory of learning style on their ability to adapt themselves to the mismatching WCF, we ran another mixed between-within subjects ANOVA.

Relationship Between Learning Style and Adapting to Mismatching Feedback
Table 3 presents the error means and SDs for the inductive and deductive groups at T1 and T2.As Table 3 reveals, the inductive group has an error ratio mean of 24.91 at T1.This mean has reduced to 15.14 at T2.The deductive group started the semester with an almost similar error mean to the inductive group (25.18), but, unlike the inductive group, their error mean escalated to 26.46 at T2.In order to ensure that the two groups were at the same level concerning their written accuracy at T1, an independent T-test was run to compare their error means on the pretest.The results showed no significant difference between the two groups ( t (40) =.02, p >.05).
To see if the mean differences within and between groups were significant across the two times, a mixed between-within subjects ANOVA was run; this helped us to examine the impact of learning style (i.e., deductive vs. inductive) on participants' ability to benefit from mismatching feedback.Table 4 illustrates the results.F=3.95, p<.05).In order to see if the observed differences between the means of each group across the two times are significant, two paired sample t-tests were run.The results, as expected, show no significant difference between the error means at T1 and T2 for the deductive group (t (17)=.50,p>.05); however, the differences between the means of the two times for the inductive group turned out to be significant (t (21)=3.58,p<.001).
Also, a follow-up independent t-test was run to see if the difference between the posttests of the two groups is significant.The results showed that the posttest error mean of the inductive group (15.14) is significantly lower than that of the deductive group (26.46) (t (38)=3.58,p<.001); hence, the inductive learners could make better use of the mismatched feedback and significantly improved their written accuracy over the treatment.

Relationship Between Implicit Theory of Learning Style and WCF Uptake
Table 5 indicates the error means and SDs for the entity and incremental students at the two times.As Table 5 illustrates, the entity students' error mean ratio at T1 is 25.56.Their T2 error mean has slightly decreased to 24.64.As for the incremental group, the error ratio mean at T1 is 27.69.They have largely reduced their mean of error ratio to 14.72 at T2.In order to ensure that the two groups were at the same level of written accuracy at T1, an independent t-test was run to compare their error means on the pre-test.The results showed no significant difference between the students with an entity theory of style and incremental theory of style (t (38) =.24, p >.05) in terms of error ratio at the beginning of the experiment.
To see if the mean differences within and between groups were significant at the two times, a mixed between-within subjects ANOVA was run.Table 6 illustrates the results.The results show a significant effect for time (F= 4.32, p = .04)and for the theory of learning style (F=5.86,p<.05, η 2 =.20).In order to see if the observed differences between the means of each group across the two times were significant, two paired sample t-tests were run.The results showed no significant difference between the error means at T1 and T2 for the entity group (t (16)=.78,p>.05); however, the differences between the means of the two times for the incremental group turned out to be significant (t (22)=3.94,p<.001).Also, a follow-up independent t-test was run to see if the difference between the posttests of the two groups is significant.The results showed that the posttest error mean of the incremental group (14.72) is significantly lower than that of the entity group (27.69) (t (38)=4.29,p<.001); hence, incremental learners could make better use of the mismatched feedback and significantly improved their written accuracy over time.

Discussion
The results of the study showed inductive students were at a better level of privilege in benefiting from mismatched CF.This finding is at odds with the previous studies which reported a mismatch between learning style and teaching style to be an obstacle to the way of learning (e.g., Charkins et al., 1985;Felder & Henriques, 1995;Felder & Silverman, 1988;Kolb, 2015;Oxford et al., 1991;Wallace & Oxford, 1992).However, the results are in line with those of Liu and Graf (2009), which showed the effective role of learning style type when the medium of instruction favoured a certain group.In their study, reflective learners could adapt themselves better to the mismatched instruction compared to active learners.
The reason why, in the present study, inductive learners were more successful in benefiting from mismatched WCF may be that they were provided with more information than needed to form a hypothesis while deductive learners received less than they required.
The results of the study also showed that there was a significant difference between entity and incremental learners in terms of written accuracy improvement, with the incremental students being more successful in benefiting from mismatched CF.Although the participants were at the same level of written accuracy at the start of the semester, they were not equally successful in rendering this adverse situation an opportunity for learning with the help of other strategies.As Dweck (2000) asserts, "Effort, difficulty, setbacks, or higher-performing peers" (p.3) leads entity students to question their ability and, therefore, perform poorer than their peers.Moreover, Bandura and Dweck (1985) argue that learners with an incremental theory will adopt learning goals more than those with an entity theory.Dweck and Legget (1988) also suggest that those with an incremental theory seek challenges and their theory "is more associated with adaptive motivational patterns" (p.263).
Drawing on the implicit theory model, one may attribute the entity students' lower performance compared to the incremental students to the former showing a helpless reaction to the challenge that receiving mismatched feedback posed on them, questioning their own ability to go beyond their comfort zone.On the other hand, incremental students have shown masteryoriented responses to this setback, trying to overcome it by using different strategies and stretching their learning styles.As Dweck's (2000) studies showed, incremental students, deem their intelligence as growing and, therefore, view received feedback and challenges as an opportunity for learning, while to entity students, who deem their intelligence as a fixed trait, this is a threat to their ego and self-esteem.Given that there was a high correlation between our instrument measuring the implicit theory of learning style and Dweck's implicit theory of intelligence questionnaire, the same might hold true for learning style; hence, it is no surprise that mismatched feedback has been less beneficial to the entity group than the incremental group.
This result is in line with the previously conducted studies on the implicit theory of intelligence.Henderson and Dweck (1990), Blackwell et al. (2007), and Stipek and Gralinski (1996) also found that holding an entity theory of intelligence had an adverse influence on learners, leading them to obtain lower academic achievements compared to the incremental learners.Waller and Papi (2017), also using Dweck's (2000) model, reported the incremental students to have higher motivation and lower feedback avoiding orientation compared to their entity counterparts.In sum, given that entity and incremental students' performances were different, it appears that simply exposing students to mismatched feedback does not lead to style stretching.

Conclusion
Even though style matching seems an ideal approach to recognition of learners' differences, by stretching style, we can help "learners become more empowered in an array of language learning contexts and expand their comfort zones to embrace a larger style repertoire" (Gregersen & MacIntyre, 2014, p.185).As mentioned earlier, contrary to the speculation of Hayes and Allinson (1997), who contend that "exposing learners to learning activities that are mismatched with their preferred learning style will help them develop the learning competencies necessary to cope with situations involving a range of different learning requirements" (as cited in Gilakjani, 2012, p.54), in this study, exposing learners to mismatched WCF did not lead to style stretching (evident by lack of progress of deductive groups who were at the most disadvantage).The result of this study, however, seems to introduce an effective factor in learners' success in style stretching.The progress that learners with an incremental theory of learning style showed indicates that implicit theory is a critical factor in the extent to which L2 writers would benefit from WCF that mismatches their preferred learning style.
In sum, the results of the study revealed that exposure to unfavored learning situations does not lead to style stretching for everyone.It appeared that one's theory of learning style is a determining factor whether and to what extent learners can stretch their styles and benefit from mismatched WCF.

Pedagogical Implications
This study was conducted in search of a more pedagogically sound approach to exploiting and honouring students' learning styles.In this quest, the results showed the decisive influence of the implicit theory of learning style on learners' ability to stretch their style.This result is of interest to teachers who find tuning their WCF to students' preferred learning style unrealistic and challenging.In other words, instead of fine-tuning feedback to students' learning styles, teachers are recommended to promote incremental views in their students by raising awareness of the malleable nature of learning style.
Therefore, it could be expected that writing teachers exploit this effect by asking students to write an essay arguing in favour of learning style malleability to promote this belief in learners' minds and leading them to change their implicit theories to be more in line with scientifically proven theories.

Limitations and Suggestions for Further Research
The present study suffers from a few limitations.First of all, in this study, a convenience sampling method was used, which decreases the generalizability power of the findings of the study.Secondly, the small sample size, especially entity students, might have affected the findings of the study.Hence, replication of this study in other contexts and with a larger sample size may yield different results.Thirdly, in this study, only one of the many constructs of learning styles (deductive vs. inductive) was investigated.Further research is required to see if students with other types of learning styles and an incremental theory would also benefit from mismatched WCF more than those with an entity theory.
In addition, bearing in mind that previous researchers have shown that students' implicit theory can be manipulated (Hong et al., 1999), future research can focus on this aspect and its result on students' uptake of mismatched feedback.In addition, interviews might also give more depth into understanding the barriers learners face when receiving mismatched feedback, and even it might provide clues into how their implicit theories were moulded.
Correspondence should be addressed to Mohammad Rahimi Email: rahimim@shirazu.ac.ir Figure 1 Sample of BGD WCF

Table 1
Dweck's (2000)original measure of implicit theories of general intelligence and adapted implicit theories of learning style

Table 2
Error categories and codes (Adopted fromFerris & Roberts, 2001, p.169) Errors in sentence /clause boundaries (run-ons, fragments (frag), comma splices, word order (wo), omitted words or phrases, unnecessary words or phrases, and other unidiomatic sentence construction.Spelling (SPL)Errors in spelling (other than those already classified as word choice) Other Errors that do not fit into previous categories (may include capitalization (cap), punctuation (p), lower case, and so on)

Table 3
Error ratio of the inductive and deductive groups across the two time periods

Table 4
Repeated measures for deductive & inductive groups

Table 4
reveals there is a significant effect for both Time (F=5.00, p<.05) and Learning style (

Table 5
Error ratio of the deductive & inductive groups across two time periods

Table 6
Repeated measures for incremental & entity groups