633 Studies in Second Language Learning and Teaching Department of English Studies, Faculty of Pedagogy and Fine Arts, Adam Mickiewicz University, Kalisz SSLLT 9 (4). 2019. 633-656 http://dx.doi.org/10.14746/ssllt.2019.9.4.4 http://pressto.amu.edu.pl/index.php/ssllt Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal attributions of grammar success and failure Günter Faber Leibniz University Hannover, Germany https://orcid.org/0000-0001-7099-415X faber@psychologie.uni-hannover.de Abstract Learners’ academic self-concepts and attributions have been widely evi- denced to substantially regulate their educational development. Develop- mentally, they will not only operate in a mutually reinforcing manner. Rather, self-concepts will directly affect learners’ outcome attributions in a particular academic setting. Current research in the English as a foreign language (EFL) context has increasingly analyzed learners’ attributions and self-concepts on a task-specific construct level. Nevertheless, there still exist certain research gaps in the field, particularly concerning learners’ grammar self-concept and attributions. Therefore, the present study aimed at analyzing longitudinal re- lations of prior performance and self-concept with subsequent attributions of grammar success and failure in a sample of preadolescent EFL learners. Find- ings demonstrated that attributional patterns mostly but not entirely de- pended on learners’ grammar self-concept. Poor performing learners holding a low self-concept displayed a maladaptive attribution pattern for explaining both grammar success and failure. Though not with respect to all causal fac- tors, these findings largely confirm the crucial role of task-specific self-concept in longitudinally explaining related control beliefs in the EFL context. Keywords: causal attributions; self-concept; grammar performance; gender; grade level Günter Faber 634 1. Introduction From the perspective of social-cognitive theories, the development of scholastic achievement is considered to be essentially regulated by learners’ individually emerging competence and control beliefs (Schunk & Zimmerman, 2006). Compe- tence beliefs refer to learners’ perceived capabilities to succeed or to fail in a given academic activity or task. They will be closely associated with corresponding control beliefs which represent learners’ perceived likelihood of accomplishing desired aca- demic outcomes by means of own behavioral attempts. In the long term, competen- ce and control beliefs will not only operate in a mutually reinforcing manner. Rather, competence beliefs will directly determine learners’ control beliefs in a certain aca- demic setting which must be seen as an important personal resource to cope with academic requirements in a given educational environment (You, Hong, & Ho, 2011). Academic control beliefs broadly manifest as causal attributions and concern learners’ perceived causes of individually experienced success and failure. They can refer to causes within or outside the individual which can also be seen as more or less consistent over time. Hence, each outcome attribution will reflect a certain cause being internal (or external) and stable (or variable) which furthermore ap- pears more or less controllable (Weiner, 2005). Empirical research substantiated typically occurring patterns in learners’ attributions. Poor performing learners tend to explain failure with a lack of academic ability but own success with external fac- tors whereas moderately or highly performing learners tend to explain success with their ability and failure with a lack of effort or environmental constraints (Faber, 2012a). Thus, poor performers who perpetually attribute failure to internal and sta- ble causes are at risk to develop increasing feelings of uncontrollability and help- lessness (Abramson, Garber, & Seligman, 1980). Likewise, they are prone to display reduced engagement in critical or demanding tasks (Stipek & Mason, 1987). Thus, learners’ preferred attributions should contribute to generating an adaptive or mal- adaptive learning approach (Schunk, 2008). Furthermore, learners’ attributions should be domain- or subject-specific in nature. Accordingly, internal explanations turned out to correlate highest with matched achievement or self-belief scores. With regard to learners’ external explanations, research provided only partial sup- port to their domain- or subject-specificity (Boekaerts, Otten, & Voeten, 2003; Bong, 2004; Marsh, 1984; Vispoel & Austin, 1995). The formation of causal attributions turns out to be dependent upon learners’ processing of individually relevant information, in particular with regard to the perceived consistency, consensus, and distinctiveness of an academic out- come (Kelley & Michela, 1980). For the most part, reference information of this nature should be already represented in learners’ domain- or subject-specific self- concepts (Marsh & O’Mara, 2008). In line with this assumption, the studies having Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 635 concurrently analyzed the relations of self-concept and achievement measures with causal attributions revealed clear evidence for attributions being most profoundly explained by corresponding self-concept rather than by achievement (Bandalos, Yates, & Thorndike-Christ, 1995; Marsh, 1984; Watkins & Gutierrez, 1989). Against this conceptual and empirical background, learners’ causal attrib- utions should claim a crucial role in analyzing and understanding academic de- velopment in the EFL context. Consequently, learners’ explanations of own suc- cess and failure in learning a foreign language should be best explained by their corresponding self-concept. Empirical research on the relation between learn- ers’ control and competence beliefs should warrant most important insights into the cognitive-motivational processing of foreign language learning and reveal important references to implement adequate instructional settings. 2. Attribution research in the EFL context 2.1. Relevant research findings The issue of learners’ competence and control beliefs has gained increasing atten- tion in the EFL context (Gabillon, 2005). Conceptual perspectives on the motiva- tional processing of learning English as a foreign language had directly or indirectly all focused upon learners’ competence and control beliefs (Dörnyei, 2005; Williams & Burden, 1997). However, though being assumed to constitute an important com- ponent of foreign language learners’ motivational orientations (Dörnyei & Ushioda, 2011; Mercer, 2011; Lamb, 2017), their causal attributions had been analyzed less intensely than other constructs. Only in the last decade, empirical EFL research had increasingly considered learners’ causal attributions (Hsieh, 2012). Mostly, EFL attribution research took place at the university level. Relevant studies found differing attribution patterns. In most, but not in all cases, learners assigned success to internal and controllable causes whereas they explained failure in some cases by internal and controllable, in other cases by external and uncon- trollable causes (Demir, 2017; Gobel & Mori, 2007; Mori, 2008; Peacock, 2010; Yaghoubi & Rasouli, 2015; Yilmaz, 2012). However, when considering learners’ com- petence level, analyses produced a more conclusive response pattern. Learners with low proficiency scores reported heightened levels of internal failure attribu- tions whereas learners with high proficiency scores attributed success more strong- ly to internal causes and failure to external factors (Mori, Ming, Nor, Suppiah, & Imm, 2011). Qualitative studies with participants from higher education tracks pro- duced similar results (Gonzalez, 2016; Graham, 2006; Taşkiran & Aydin, 2017). Seemingly fewer studies analyzed EFL learners’ attributions at the school level. Here again, findings revealed differing attribution patterns. High language Günter Faber 636 achievers reported more internal and stable causes than their low achieving class- mates (Sorić & Ančić, 2008), and both high and low achieving learners attributed language outcome most strongly to internal, primarily uncontrollable causes (Erten, 2015a; Erten & Burden, 2014). Likewise, learners’ language achievement was negatively related to external attributions. Therefore, they felt not responsible for a failure outcome (Hsieh & Kang, 2010). Moreover, qualitative research showed learners’ attributions for both success and failure mostly referring to internal causes (Sahinkarakas, 2011; Williams, Burden, Poulet, & Maun, 2004). Some studies also examined the role of L2 self-concepts. In particular, Phothongsunan (2015) showed university EFL learners with high competence beliefs reporting stronger effort attributions than their counterparts with low competence beliefs. Moreover, Mori and colleagues (2011) demonstrated learn- ers with high competence beliefs to trace back language success to internal and failure to external factors. Analyses with school samples demonstrated the for- eign language self-concept to significantly explain learners’ attribution tenden- cies (Erten, 2015a; Erten & Burden, 2014). Most notably, in the study of Hsieh and Kang (2010) learners with lowered competence beliefs tended to attribute test outcomes more strongly to external factors. To some extent, the study of Luo, Hogan, Yeung, Sheng and Aye (2014) lent support to this finding. Furthermore, at all educational tracks EFL research revealed certain gender ef- fects. In particular, some studies reported females to attribute language success more strongly to internal causes than males (Erten, 2015a; Lian, 2012; Peacock, 2010; Tulu, 2013; Yilmaz, 2012). There was also evidence for male learners to explain success more strongly with reference to effort (Mori, 2012). Compared with this, male learn- ers attributed failure primarily to external factors whereas female learners mostly re- ferred to internal factors (Genç, 2016; Mori, 2012; Tulu, 2013; Yilmaz, 2012; Zhori, 2011). Even so, other studies did not find any significant differences between female and male learners (Luo et al., 2014; Yördem, 2016). As the strength and direction of differences appeared to considerably vary across studies, research findings do not al- low for accurately clarifying gender differences in causal attributions. In sum, empirical research on EFL learners’ attributions yielded mixed re- sults. There is evidence for both self-serving and self-blaming attribution pat- terns (Marsh, 1986), each of them producing rather specific consequences in learners’ emotional responses and learning approach (Graham, 1991). Unargu- ably, this inconsistency of findings essentially reflects the diversity of learner samples and educational or cultural contexts relevant studies considered as well as the attribution measures they used. Notwithstanding, in one way or another, all relevant studies substantiated learners’ domain- or subject-specific self-con- cepts to significantly determine their attribution responses. Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 637 2.2. Current research gaps Admittedly, research on EFL learners’ attributions still suffers from certain con- ceptual, methodological, and empirical limitations. First of all, with regard to the developmental importance of early language learning, there is a lack of attribu- tion research in school settings. Furthermore, from the perspective of self-belief research, there is a considerable need for further clarifying the role of learners’ self-concepts. Not least, most studies in the EFL context analyzed learners’ attrib- utions on a subject-specific level but did not test for concurrent performance and self-concept effects. Only few studies had already investigated foreign language attributions with respect to specific language outcomes – for example, in reading, speaking, and listening comprehension (Demir, 2017; Gobel & Mori, 2007; Gra- ham, 2006; Lian, 2012; Yilmaz, 2012). However, following empirical findings on learners’ self-concepts (Arens & Jansen, 2016; Faber, 2012b; Holder, 2005; Lau, Yeung, Jin, & Low, 1999), further research in the field should more intensely ex- plore task-specific attributions,and also scrutinize relevant instruments’ validity, in order to gain more differentiated access to learners’ beliefs to control their out- comes in foreign language vocabulary, listening, reading, writing, or grammar. Most remarkably, there are no appropriate research activities to be found for the issue of EFL grammar learning. This gap must be all the more surprising as the learning of forms is commonly considered to play an important and needful role in the foreign language learning classroom (Nassaj & Fotos, 2004). Corre- spondingly, relevant survey results pointed out the majority of EFL learners per- ceived the mastering of grammar as an important and useful, albeit rather de- manding and not always enjoyable part of their language learning (Jean & Simard, 2011; Kang, 2017; Loewen et al., 2009; Schulz, 1996; Simon & Taverniers, 2011). Moreover, grammar was perceived even as a demotivating factor in EFL instruc- tion (Sakai & Kiuchi, 2009). Therefore, in the course of EFL instruction learners will incrementally develop specific competence and control beliefs concerning their grammar learning in the target language. Accordingly, there is a remarkable back- log to analyze their causal attributions to explain grammar outcomes. 3. Conceptual framework and research questions With regard to these research gaps, the present study aimed at analyzing pre- adolescent EFL learners’ causal explanation of grammar success and failure. In particular, it should challenge findings from previously run analyses of cross-sec- tional data (Faber, 2017b) and examine the longitudinal effects of learners’ per- formance and self-concept on subsequent attributions. Günter Faber 638 The present study was based on a cognitive-motivational modeling of con- struct relations which considered learners’ causal attributions as dependent var- iables. As recent research findings considerably varied across attribution measures each including different numbers, types, and levels of causal catego- ries, their findings could always reflect a certain method bias. In order to evade this problem, the present study narrowed down the causal categories to explain learners’ grammar success and failure and only referred to the basically pro- posed causal factors of ability, effort, task difficulty, and luck (Weiner, 2005). They had been thoroughly substantiated in all relevant studies and evidently represent the causal dimensions of locus of control, variability, and controllabil- ity (Hau & Salili, 1993). Though this limitation restricted the potentially existing causal space of outcome explanations, it was meant to be temporary for the purpose of the present study. Of course, it did not principally rule out an exten- sion of causal factors. Rather, it should warrant an empirical starting point to further refine the framework of relations among grammar performance, self- concept, and attribution variables. Figure 1 Modeling longitudinal relations of prior performance and self-concept with subsequent attribution variables Concerning longitudinal relations among constructs across three meas- urement times, learners’ grammar self-concept at Measurement time 2 was ex- pected to operate as most powerful predictor to explain their attributions at Measurement time 3 (Figure 1) and, thus, it should strongly mediate the rela- tionship between their task-specific performance experiences and their causal beliefs. Following previous findings (Faber, 2017b), high self-concept scores to predict internal success and external failure attributions, and low self-concept scores were expected to predict external success and internal failure attributions. Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 639 Hence, learners’ grammar task performance at Measurement time 1 should not directly affect their causal attributions at Measurement time 3. Even though previously run analyses of cross-sectional data demonstrated the grammar self-concept not being different between female and male learn- ers (Faber, 2017a), the present study precautionally controlled longitudinal re- lations among constructs for potentially operating gender effects (Figure 1). The same applied to learners’ grade level. As the present study took place in EFL fifth and sixth grade classrooms, learners’ performance and self-beliefs were just emerging at grade 5 and developing at grade 6. Therefore, it was worthwhile to compare their performance and self-belief variables across both grade levels in order to test for possible moderator effects. Accordingly, the present study aimed at clarifying the following research questions (RQ): RQ 1: Will attributions of grammar success and failure be substantially affected by learners’ gender? RQ 2: Will attributions of grammar success and failure be substantially affected by learners’ grade level? RQ 3: Will learners’ prior self-concept but not their task performance substantially explain their attributions of grammar success and failure? 4. Method 4.1. Participants The study was conducted with a sample of N = 119 preadolescent learners at second- ary grade level 5 (31 female, 34 male) and 6 (42 female, 12 male) from a German grammar school. In the German tripartite system of strongly selective educational tracks this school type (“Gymnasium”) is the highest track. The female-male ratio sig- nificantly differed between both grade levels (χ2 = 11.359, df = 1, p < .001), as there were more female learners at grade level 6. However, this gender ratio appeared to be representative for the educational track (Blossfeldt et al., 2009). At Measurement time 1, learners’ average age was 11.1 years (SD = 0.4) at grade 5 and 12.1 years (SD = 0.5) at grade 6. At Measurement time 1, fifth-graders had experienced formal EFL instruction for about 6 months, and sixth-graders for about 17 months. Their partici- pation was on a voluntary basis and only with explicit parental consent. 4.2. Procedure Following the longitudinal modeling approach of construct relations (Figure 1), learn- ers’ grammar task performance was assessed in Calendar week 6 (at Measurement Günter Faber 640 time 1). In Calendar week 11 (at Measurement time 2), their grammar self-concept and in Calendar week 19 (at Measurement time 3), their causal attributions of gram- mar success and failure were assessed. All data were gathered in the course of two class periods by two (advanced collegiate) test supervisors who had been instructed in detail. This took place class for class and in the absence of the teaching staff. 4.3. Instruments For assessing learners’ grammar performance, at Measurement time 1 a cloze test was administered (Jensen, 2013). It consisted of ten tasks dealing with the correct use of possessive pronouns (at grade level 5) and the correct use of comparative adjectives (at grade level 6). For both tests, a sum score of correctly solved tasks was used. As the results did not significantly differ between grade levels (inde- pendent samples test: t = 0.018, df = 210, p = .985), they could be considered to truly provide comparable performance measures for further analyses. Learners’ grammar self-concept was measured by means of a newly de- veloped scale that consisted of nine 4-point Likert items (ranging from never to always). Scale items referred to learners’ perceived competencies to under- stand and apply L2 grammar forms or rules, respectively, their perceived mas- tery of grammar tasks in the classroom and their motivational tendency to avoid grammar tasks (Faber, 2017a). The following is a sample item: “During English lesson I manage to catch the grammar rules.” For data from Measurement time 2, a principal component analysis showed all scale items to substantially load on one common factor (with factor loadings ranging from amin = .564 to amax = .802). The scale’s internal consistency (Cronbach’s coefficient alpha) turned out to be sufficient (α = .85). High scale scores indicate learners’ strong beliefs in own grammar competencies. As the scale’s sum scores strongly correlated with the English but not with the mathematics self-concept variable, instrument should partially claim subject-specific validity (Faber, 2017a). At Measurement times 1 and 3, causal attributions were tapped by means of 4-point Likert scale items – four of them concerning a success and four of them concerning a failure outcome in grammar. They all referred to the classical Weiner categories of ability, effort, difficulty, and luck (Weiner, 2005). Learners were asked to rate the importance of each cause to explain the favorable and unfavorable out- come. The following is a sample item: “If I am successful in English grammar, it is because I have practiced hard” (Jensen, 2013). High item scores indicate a certain cause being perceived to strongly affect the grammar outcome. These attribution ratings should absolutely represent most distinguishable information. Accord- ingly, they should correlate only in a moderate manner, at most. Therefore, sepa- rate principal component analyses were run for the success and failure outcome Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 641 with Measurement time 3 data. Their results demonstrated all items revealing suf- ficiently independent information (see Table 1). Finally, learners’ grade level and gender (as dummy variable) were in- cluded in the analyses. Table 1 Principal component analyses (PCA) of attribution items at Measure- ment time 3 F1 F2 F3 F4 Success attributions Ability .990 .031 -.064 -.118 Effort .031 .992 .080 -.096 Difficulty -.064 .080 .994 .041 Luck -.119 -.096 .042 .987 Failure attributions Ability .979 .126 .152 .055 Effort .123 .987 .098 -.018 Difficulty .157 .104 .968 .168 Luck .054 -.018 .159 .986 Note. F = factor 4.4. Data analyses In order to clarify possible grade level and gender effects, a series of univariate analyses of variance (ANOVA) with repeated measures for dependent attribution variables was initially conducted – including grade level and gender as independ- ent factor variables. This particular step was most important for verifying the lon- gitudinal approach of data analyses as well as determining subsequent analyses to run with the total sample or with grade level and gender subsamples, respec- tively. Furthermore, for all performance and self-belief variables overall descrip- tive statistics were calculated. For detecting significant deviations from normal distribution, z-standardized skewness and kurtosis scores were used. To clarify learners’ attribution patterns for grammar success and failure, mean differences of attribution responses were analyzed – namely, whether and how strongly they could be explained by significant effects of learners’ prior grammar self-concept and task performance. Accordingly, for each causal factor a 2Í2 ANOVA with the grammar self-concept and task performance as factor variables was conducted. In view of sample size, both the self-concept and the performance variable were factorized by median split, thus leading to the iden- tification of low and high level subgroups in each case. For all main or interaction effects, partial eta squared (ηp2) as an effect size was calculated. Due to sample size and lack of variance homogeneity, the robust Brown-Forsythe test was used for post hoc comparison (Tomarken & Serlin, 1986). Günter Faber 642 Among all variables, there were missing data to a certain extent. For the grammar self-concept items and the grammar task performance, the amount of incomplete data ranged from 0 to 2.4%; in the attribution variables, they ranged from 0.9 to 2.8%. As these missing values did not produce any systematic pat- tern (Little, 1988), they could still be treated as “missing completely at random” (MCAR test: χ2 = 228.695, df = 224, p = .401). All missing data were estimated by means of the two-step iterative EM algorithm (Enders, 2010). 5. Results 5.1. Measurement time, grade level, and gender effects Initially driven analyses of variance did not substantiate any grade level and gen- der effects. Neither was there a significant difference between attribution re- sponses at Measurement times 1 and 3 (see Table 2). Furthermore, no signifi- cant grade level and gender effects could be found for learners’ grammar task performance (grade level: F = 1.261, df = 1,118, p > .05; gender: F = 3.215, df = 1,118, p > .05) and their grammar self-concept (grade level: F = 2.064, df = 1,118, p > .05; gender: F = 1.660, df = 1,118, p > .05). Consequently, subsequent anal- yses included learner data from the total sample. Table 2 Measurement time, grade level, and gender effects on causal attribu- tions of grammar success and failure Causal factor Wilks λ F df p ηp2 Success ability MT .999 0.137 1,115 .712 .001 MTÍgender .999 0.081 1,115 .777 .001 MTÍgrade level .993 0.765 1,115 .384 .007 Success effort MT .998 0.249 1,115 .619 .002 MTÍgender .995 0.615 1,115 .435 .005 MTÍgender level .999 0.121 1,115 .728 .001 Success difficulty MT .999 0.066 1,115 .797 .001 MTÍgender .999 0.146 1,115 .703 .001 MTÍgrade level .996 0.511 1,115 .476 .004 Success luck MT 1.000 0.019 1,115 .892 .000 MTÍgender .999 0.074 1,115 .786 .001 MTÍgrade level .992 0.911 1,115 .342 .008 Failure ability MT .997 0.313 1,115 .577 .003 MTÍgender 1.000 0.005 1,115 .947 .000 MTÍgrade level 1.000 0.017 1,115 .895 .000 Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 643 Failure effort MT 0.999 .136 1,115 .713 .001 MTÍgender 0.986 1.616 1,115 .206 .014 MTÍgrade level 0.984 1.815 1,115 .181 .016 Failure difficulty MT 0.988 1.428 1,115 .235 .012 MTÍgender 0.985 1.802 1,115 .182 .015 MTÍgrade level 0.986 1.656 1,115 .201 .014 Failure luck MT 0.999 .082 1,115 .775 .001 MTÍgender 0.998 .268 1,115 .606 .002 MTÍgrade level 0.997 .354 1,115 .553 .003 Note. MT = Measurement time 1 vs. 3, ηp2 = partial eta squared 5.2. Descriptive results Descriptive analyses showed learners’ grammar task performance and self-con- cept to be negatively skewed (see Table 3) indicating that most learners achieved higher task scores and held positive competence beliefs. Developmen- tally, this particular result should be most plausible as younger learners still tend to have an optimistic view on own competencies (Helmke, 1999). Luck attributions of success as well as ability and luck attributions of fail- ure showed a significant positive skew. Hence, most learners manifested distinct rejection of item content. They tendentially perveived grammar success not be- ing dependent on luck and grammar failure not being dependent on a lack of own ability or bad luck. Correspondingly, success attributions on luck showed a strong positive kurtosis indicating their distribution to be heavy-tailed on the (left) side of lower response categories. Table 3 Descriptive statistics of grammar performance, self-concept, and attrib- ution variables MT AM SD z skewness z kurtosis Grammar task performance 1 6.57 2.36 -2.55*** -0.95*** Grammar self-concept 2 26.68 4.99 -2.57*** 0.92*** Causal attributions Success ability 3 2.87 1.01 -1.11*** -1.13*** Success effort 3 2.22 1.01 -1.75*** -2.31*** Success difficulty 3 2.72 0.92 -0.57*** -2.04*** Success luck 3 2.22 1.13 9.19*** 22.02*** Failure ability 3 2.03 0.99 2.68*** -1.61*** Failure effort 3 2.34 1.03 0.53*** -2.58*** Failure difficulty 3 2.22 0.88 1.02*** -1.59*** Failure luck 3 2.11 0.97 2.04*** -1.83*** Note. MT = Measurement time, significance: *p ≤ .05, **p ≤ .01, ***p ≤ .001 Günter Faber 644 5.3. Longitudinal relations As ANOVA results showed, only learners’ ability and luck attributions of grammar suc- cess appeared to be significantly explained by their grammar self-concept (see Table 4). Table 4 Mean causal attribution of grammar success and failure depending on prior task performance and self-concept Causal factor MT F df p ηp2 Success ability Grammar task performance 3 2.511 1,118 .116 .021 Grammar self-concept 1 14.580 1,118 .000 .113 PerformanceÍself-concept 2 1.411 1,118 .237 .012 Success effort Grammar task performance 3 0.000 1,118 .999 .000 Grammar self-concept 1 0.467 1,118 .496 .004 PerformanceÍself-concept 2 0.314 1,118 .576 .003 Success difficulty Grammar task performance 3 0.042 1,118 .839 .000 Grammar self-concept 1 0.029 1,118 .864 .000 PerformanceÍself-concept 2 0.012 1,118 .913 .000 Success luck Grammar task performance 3 7.157 1,118 .009 .059 Grammar self-concept 1 8.393 1,118 .005 .068 PerformanceÍself-concept 2 1.959 1,118 .164 .017 Failure ability Grammar task performance 3 10.005 1,118 .002 .080 Grammar self-concept 1 23.903 1,118 .000 .172 PerformanceÍself-concept 2 0.693 1,118 .407 .006 Failure effort Grammar task performance 3 2.733 1,118 .101 .023 Grammar self-concept 1 9.941 1,1187 .002 .080 PerformanceÍself-concept 2 2.157 1,118 .145 .018 Failure difficulty Grammar task performance 3 0.496 1,118 .483 .004 Grammar self-concept 1 2.101 1,118 .150 .018 PerformanceÍself-concept 2 4.067 1,118 .046 .034 Failure luck Grammar task performance 3 1.036 1,118 .311 .009 Grammar self-concept 1 0.053 1,118 .818 .000 PerformanceÍself-concept 2 0.442 1,118 .507 .004 Note. MT = measurement time, ηp2 = partial eta squared Mean ability attributions were strongly dependent on a significant self- concept effect. The effect of the performance factor remained nonsignificant. Across all performance levels, the high self-concept group explained subsequent grammar success more strongly with own ability than the low self-concept group (see Figure 2). In comparison, mean effort and difficulty attributions depended Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 645 Figure 2 Mean success attributions depending on prior grammar performance and self-concept neither on prior performance nor on prior self-concept differences. In turn, mean luck attributions appeared to be affected by a significant self-concept and perfor- mance effect. In the poor performing subgroup, learners holding a low self-concept explained grammar success more strongly with good luck than their classmates with a high self-concept. This difference was statistically significant (F = 8.952, df 1 = 1, df 2 = 61, p < .01). In contrast, in the well-performing subgroup, both learners with a low and a high self-concept rejected good luck as a cause of grammar success (see Figure 3). The apparently existing difference between self-concept groups was non- significant (F = 1.780, df 1 = 1, df 2 = 40, p > .05). Taken altogether, learners with a high self-concept more strongly explained success with internal and thus controlla- ble causes whereas they emphasized an external cause to a lesser extent or rejected it. Learners with a low self-concept less strongly explained a task-specific success out- come with internal causes whereas they emphasized an external cause only in the case of experiencing poor grammar performance. Hence, learners in the high self- Günter Faber 646 concept group apparently felt more responsible for grammar success than their counterparts in the low self-concept group. The low self-concept group appeared to be at risk of discounting potential grammar success by tracing it back to good luck. Figure 3 Mean failure attributions depending on prior grammar performance and self-concept Similarly, learners’ failure attributions were mostly affected by their gram- mar self-concept (see Table 4). In particular, the mean level of ability attributions appeared to be dependent on both a significant effect of the self-concept and the performance factor. Accordingly, across all performance levels it was the low self-concept group explaining an unfavorable outcome most strongly with a lack of own ability (see Figure 3). In comparison, the high self-concept group consid- erably rejected this cause whereby poor performing learners with a high self-con- cept rejected lack of ability as possible failure cause to a lesser extent. With re- spect to learners’ effort attributions, again a significant effect of the self-concept Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 647 variable occurred. Across both performance levels, learners of the high self-con- cept group denied the role of lack of own effort. Remarkably, poor performing learners holding a low self-concept explained a failure outcome with a lack of own ability as well as with a lack of own effort. Those learners who more strongly attributed failure to a lack of effort preferred a lack of ability as another internal cause to a significantly stronger extent (F = 5.155, df 1 = 1, df 2 = 70, p < .05). Furthermore, learners’ difficulty attributions appeared to be significantly depend- ent on an interaction effect. Within the poor performance group, learners with a low self-concept substantially explained failure with task difficulty whereas learn- ers with a high self-concept did not. This difference appeared to be statistically significant (F = 6.307, df 1 = 1, df 2 = 61, p < .05). However, in the well-performing group this difference was statistically nonsignificant (F = 0.160, df 1 = 1, df = 45, p > .05). Concerning learners’ luck attributions for failure, neither the prior perfor- mance nor the self-concept variable explained interindividually existing differ- ences. Across all performance and self-concept levels, bad luck as a possible cause of grammar failure did not play a significant role (see Figure 3). Taken altogether, learners with a low self-concept of grammar competen- cies and poor grammar performance more strongly explained task-specific fail- ure with lack of own ability and effort, to some extent also with the task’s diffi- culty. Thus, they attributed failure to both internal and external causes which they may principally control in the case of effort, at least. In contrast, learners with a high self-concept of grammar competencies apparently could not identify any particular cause to reasonably explain a task-specific failure outcome. In- stead, across all performance levels they perceived internal as well as external causes as less important. Obviously, due to their belief of having appropriate competencies or skills available, they could not really imagine expecting or even experiencing an unfavorable grammar outcome. 6. Discussion and conclusions The main objective of the present study concerned the potentially mediating role of preadolescent EFL learners’ task-specific self-concept to predict their causal attributions of grammar success and failure. Using longitudinal data from two grade levels, learners’ prior grammar self-concept was expected to signifi- cantly predict subsequent attribution differences. Additionally, the impact of gender and grade level on learners’ attributions were analyzed in order to detect possible moderator effects. Günter Faber 648 6.1. Gender and grade level effects The present study did not substantiate learners’ gender to significantly moder- ate their attribution responses. Contrary to relevant research findings neither did females attribute grammar success more strongly to internal causes (Erten, 2015a; Sorić & Ančić, 2008; Yeung, Lau, & Nie, 2011) nor did males explain gram- mar failure more strongly to external causes (Genç, 2016; Mori, 2012; Yilmaz, 2012; Zhori, 2011). Conceivably, gender differences in EFL attributions will yet emerge over time – after learners will have experienced a longer history of suc- cess or failure and, accordingly, will have possibly developed gender-dependent self-beliefs on own strengths and difficulties (Pomerantz, Altermatt, & Saxon, 2002). Moreover, subtle gender effects do not necessarily have to operate at all proficiency levels in the same way (Faber, 2013, 2017b). Further analyses should, therefore, examine differential gender effects in the long term. Likewise, the present study found learners’ attribution responses not to be dependent on grade level. Here again, it should be assumed relevant grade level effects may manifest yet in the longitudinal course of advancing EFL expe- riences (Erten, 2015a; Sorić & Ančić, 2008; Yeung et al., 2011). 6.2. The predictive role of grammar self-concept Concerning the study’s main objective, learners’ attributions were best predicted by prior self-concept differences, in some cases also by additional or interactive effects of prior performance. In particular, learners with high competence beliefs felt responsible for success whereas they could not really explain an unfavorable grammar outcome. Compared with this, learners with low competence beliefs traced back grammar success to external luck and grammar failure to both inter- nal and external causes. Especially the poor performing subgroup with a low self- concept referred to success and failure causes which must be seen as not or not easily controlled. Thus, their attribution responses should potentially induce a sense of personal helplessness (Butkowsky & Willows, 1980; Diener & Dweck, 1978). Altogether, these attribution patterns indicate a self-congruent rather than a self-serving or counterdefensive response type (Marsh, 1986). With all that, neither learners’ prior performance nor self-concept ex- plained their effort and difficulty attributions of grammar success. Likewise, their ability, effort, and difficulty but not their luck attributions of grammar fail- ure were predicted by prior self-concept and, partially, by performance differ- ences. Learners might have processed individually available consistency infor- mation to explain both favorable and unfavorable grammar outcomes and pri- marily focused on the locus of control dimension (Kelley & Michela, 1980). Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 649 Though these findings largely lend support to the explanatory power of task-specific competence beliefs and corresponding competence experiences, not all assumptions of the conceptual modeling perspective (Figure 1) were verified. Contrary to theoretical perspectives (Weiner, 2005), preadolescent EFL learners did not consider causal factors in the most exhaustive way. In particular, they ap- peared to perceive grammar success rather as a result of own ability than of own effort. Obviously, they did not distinguish between both internal causes. Instead, they might have perceived personal causation as a matter of own ability (Nicholls & Miller, 1983). Similarly, in the case of grammar failure poor performing learners with low competence beliefs considered ability and effort in the same way. There- fore, they did not yet realize the causal dimension of variability. Here again, they might have focused on the locus of control dimension. Developmentally, a more differentiated understanding of effort will still emerge in the course of ongoing EFL instruction. Further analyses should clarify this issue. In particular, they should explore the attributional role of feedback practice in the EFL classroom (Schunk, 2008). Possibly, poor performing language learners’ effort attributions merely re- flect their teachers’ commentaries without them being aware of the controllabil- ity perspective. In respect thereof, further research should also analyze the role of learners’ mindsets and, thus, clarify their perceived malleability of foreign lan- guage competencies or abilities (Ryan & Mercer, 2012). 6.3. Strengths and limitations Methodologically, a strength of the present study was the analysis of longitudinal per- formance and self-belief data. Therefore, relations among constructs should be inter- preted in terms of their causal ordering across three measurement times. However, a methodological limitation should have been the analysis of learners’ data from one particular school. Further analyses should necessarily try to replicate the findings in other educational settings. Another methodological limitation was the restricted use of causal factors. Against the background of present findings, further analyses should extend the range of causal factors learners would use to explain grammar success and failure – in essence, as they might tap other internal reasons such as study method or learning strategies as well as external reasons such as teacher grading and instructional support (Gobel & Mori, 2007; Hsieh, 2004). Most of all, the effort cause appears to be highly inferent and, thus, is principally at risk of confounding different information (Sta- bles, Murakami, McIntosh, & Martin, 2014). Accordingly, it should be absolutely un- folded by more proximal operationalizations and differentiated into distinct facets. These effort facets should represent effort experience in a more task-specific manner and address concrete task requirements, learning, and classroom situations learners typically might encounter in the EFL context. Günter Faber 650 Notwithstanding, the present study revealed empirical findings which em- phasize the explanative role of task-specific self-concepts in EFL attribution re- search. In line with relevant research findings (Erten, 2015a; Erten & Burden, 2014; Genç, 2016; Mori et al., 2011), it was not grammar performance per se but learners’ cognitive-motivational processing of grammar success or failure which substantially predicted their attribution responses. Consequently, to more strongly implement an adequate research line, relevant studies should concurrently analyze the relations between self-concept and attribution measures in different areas – that is, not just in grammar but also in reading, spelling, pronunciation, vocabulary, and listening. 6.4. Practical implications Though the present findings must be seen as preliminary in nature, they immedi- ately suggest certain educational implications. In particular, as poor performing learners with low competence beliefs attributed both grammar success and fail- ure mostly to uncontrollable causes, in the case of failure to some extent also to lack of own effort, they are prone to develop expectations of uncontrollability – and, thus, potentially to lessen their engagement, to stagnate with an inadequate learning approach, and to achieve unfavorable performance outcomes (Hsieh, 2012). Therefore, it would be appropriate to implement teaching strategies and learning opportunities in the EFL classroom which should strive for reducing mal- adaptive and strengthening adaptive attributions of grammar success and failure (Erten, 2015b). Accordingly, relevant intervention should be expected to yield most effective results inasmuch as it will apply a combination of direct attribution feedback procedures and task-specific strategy training (Robertson, 2000) whilst teacher effort feedback should heavily refer to acquisition and practice proce- dures priorly introduced and systematically applied in the EFL classroom (Chan & Moore, 2006). In particular, this approach should warrant learners to experience both realistic learning success and control-oriented dealing with failure – espe- cially by supporting and modeling effort attributions. That way, learners should gradually develop adequate competence or efficacy beliefs which, in turn, should facilitate their use of adaptive outcome attributions (Raoofi, Tan, & Chan, 2012). Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 651 References Abramson, L. Y., Garber, J., & Seligman, M. E. P. (1980). Learned helplessness in hu- mans: An attributional analysis. In J. Garber & M. E. P. Seligman (Eds.), Human helplessness: Theory and applications (pp. 3-34). New York: Academic Press. Arens, A. K., & Jansen, M. (2016). Self-concepts in reading, writing, listening, and speaking: A multidimensional and hierarchical structure and its generali- zability across native and foreign languages. Journal of Educational Psy- chology, 108, 646-664. Bandalos, D. L., Yates, K., & Thorndike-Christ, T. (1995). Effects of math self-con- cept, perceived self-efficacy, and attributions for failure and success on test anxiety. Journal of Educational Psychology, 87, 611-623. Blossfeldt, H.-P., Bos, W., Lenzen, D., Hannover, B., Müller-Böling, D., Prenzel, M., & Wößmann, L. (2009). Geschlechterdifferenzen im Bildungssystem – die Bundesländer im Vergleich. München: Aktionsrat Bildung. Boekaerts, M., Otten, R., & Voeten, R. (2003). Examination performance: Are student’s causal attributions school-subject specific? Anxiety, Stress, and Coping, 16, 331-342. Bong, M. (2004). Academic motivation in self-efficacy, task value, achievement goal orientations, and attributional beliefs. Journal of Educational Re- search, 97, 287-297. Butkowsky, I. S., & Willows, D. M. (1980). Cognitive-motivational characteristics of children varying in reading ability: Evidence for learned helplessness in poor readers. Journal of Educational Psychology, 72, 408-422. Chan, K. S., & Moore, P. J. (2006). Development of attributional beliefs and stra- tegic knowledge in years 5-9: A longitudinal analysis. Educational Psycholo- gy, 26, 161-185. Demir, Y. (2017). Turkish EFL learners’ attributions for success and failure in speaking English. International Journal of Contemporary Educational Re- search, 4(2), 39-47. Diener, C. I., & Dweck, C. S. (1978). An analysis of learned helplessness: Contin- uous changes in performance, strategy, and achievement cognitions follo- wing failure. Journal of Personality and Social Psychology, 36, 451-462. Dörnyei, Z. (2005). The psychology of the language learner: Individual differ- ences in second language acquisition. London: Routledge. Dörnyei, Z., & Ushioda, E. (2011). Teaching and researching motivation (2nd ed.). Harlow: Pearson Education. Enders, C. K. (2010). Applied missing data analysis. New York: Guilford Press. Erten, I. H. (2015a). Age related gender differences in causal attributions of Turkish learners of English as a foreign language. ELT Research Journal, 4(2), 129-146. Günter Faber 652 Erten, I. H. (2015b). Attribution retraining in L2 classes: Prospects for exploratory classroom practice. In K. Dikilitaş, R. Smith, & W. Trotman (Eds.), Teacher- researchers in action (pp. 357-367). Faversham: International Association of Teachers of English as a Foreign Language (IATEFL). Erten, I. H., & Burden, R. L. (2014). The relationship between academic self-con- cept, attributions, and L2 achievement. System, 42, 391-401. Faber, G. (2012a). Elementary school children’s spelling-specific self-beliefs: Lon- gitudinal analyses of their relations to academic achievement, school at- titudes, and self-esteem. New York: Nova Science Publishers. Faber, G. (2012b). Measuring self-perceptions of oral narrative competencies and anxiety in the EFL context. Electronic Journal of Research in Educa- tional Psychology, 10, 1343-1382. Faber, G. (2013). Gender and achievement differences in secondary students’ verbal self-concepts: A closer look beyond bivariate comparison. Elec- tronic Journal of Research in Educational Psychology, 11, 665-692. Faber, G. (2017a). A scale for measuring preadolescent EFL learners’ grammar self-concept: Preliminary psychometric and multivariate validation re- sults. In R. V. Nata (Ed.), Progress in education (Vol. 41, pp. 29-81). New York: Nova Science Publishers. Faber, G. (2017b). The mediating role of the foreign language self-concept: Predict- ing preadolescent EFL learners’ causal attributions of grammar success and failure. In J. A. Sørensen & N. S. Schultz (Eds.), Self-perception: Research ad- vances and clinical challenges (pp. 1-82). New York: Nova Science Publishers. Gabillon, Z. (2005). L2 learner’s beliefs: An overview. Journal of Language and Learning, 3, 233-260. Genç, G. (2016). Attributions to success and failure in English language learning: The effects of gender, age and perceived success. European Journal of Ed- ucation Studies, 2(12), 25-43. Gobel, P., & Mori, S. (2007). Success and failure in the EFL classroom: Exploring students’ attributional beliefs in language learning. In L. Roberts, A. Gürel, S. Tatar, & L. Martı (Eds.), EUROSLA yearbook (Vol. 7, pp. 149-169). Am- sterdam: John Benjamins. Gonzalez, A. S. (2016). Attribution theory: Dimensions of causality, stability and controllability according to learners. In C. Gkonou, D. Tatzl, & S. Mercer (Eds.), New directions in language learning psychology (pp. 209-232). Cham: Springer International Publishing. Graham, S. (1991). A review of attribution theory in achievement contexts. Ed- ucational Psychology Review, 3, 5-39. Graham, S. (2006). Listening comprehension: The learners’ perspective. System, 34, 165-182. Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 653 Hau, K. T., & Salili, F. (1993). Measurement of achievement attribution: A review of instigation methods, question contents, and measurement formats. Ed- ucational Psychology Review, 5, 377-422. Helmke, A. (1999). From optimism to realism? Development of children’s academic self-concept from kindergarten to grade 6. In F. E. Weinert & W. Schneider (Eds.), Individual development from 3 to 12: Findings of the Munich longitu- dinal study (pp. 198-221). New York: Cambridge University Press. Holder, M. C. (2005). Fähigkeitsselbstkonzept und Leistungsmotivation im Fremdsprachenunterricht. Bern: Lang. Hsieh, P. H. (2004). How college students explain their grades in a foreign lan- guage course: The interrelationship of attributions, self-efficacy, language learning beliefs, and achievement (Unpublished doctoral dissertation). University of Texas, Austin, USA. Hsieh, P. H. (2012). Attribution: Looking back and ahead at the ‘why’ theory. In S. Mercer, S. Ryan, & M. Williams (Eds.), Psychology for language learning: Insights from research, theory and practice (pp. 90-102). Basingstoke: Pal- grave Macmillan. Hsieh, P. P.-H., & Kang, H.-S. (2010). Attribution and self-efficacy and their inter- relationship in the Korean EFL context. Language Learning, 60, 606-627. Jean, G., & Simard, D. (2011). Grammar teaching and learning in L2: Necessary, but boring? Foreign Language Annals, 44, 467-494. Jensen, L. (2013). Fostering task-specific self-efficacy expectancies in the EFL class- room: An empirical analysis of short-time effects. München: Grin Verlag. Kang, D. (2017). EFL learner and teacher beliefs about grammar learning in Ko- rea. English Teaching, 72(2), 51-69. Kelley, H. H., & Michela, J. L. (1980). Attribution theory and research. Annual Review of Psychology, 31, 457-501. Lamb, M. (2017). The motivational dimension of language teaching. Language Teaching, 50, 301-346. Lau, I. C., Yeung, A. S., Jin, P., & Low, R. (1999). Toward a hierarchical, multidimensi- onal English self-concept. Journal of Educational Psychology, 91, 747-755. Lian, X. (2012). An empirical study on causal attribution regarding listening achievement of Chinese English majors. Contemporary English Teaching and Learning in Non-English-Speaking Countries, 1(1), 23-46. Little, R. J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83, 1198-1202. Loewen, S., Feifei, S. L., Thompson, A., Nakatsukasa, K., Ahn, S., & Chen, X. (2009). Second language learners’ beliefs about grammar instruction and error correction. Modern Language Journal, 93, 91-104. Günter Faber 654 Luo, W., Hogan, D., Yeung, A. S., Sheng, Y. Z., & Aye, K. M. (2014). Attributional beliefs of Singapore students: Relations to self-construal, competence, and achievement goals. Educational Psychology, 34, 154-170. Marsh, H. W. (1984). Relations among dimensions of self-attribution, dimen- sions of self-concept and academic achievements. Journal of Educational Psychology, 76, 1291-1308. Marsh, H. W. (1986). Self-serving effect (bias?) in academic attributions: Its re- lation to academic achievement and self-concept. Journal of Educational Psychology, 78, 190-200. Marsh, H. W., & O’Mara, A. J. (2008). Self-concept is as multidisciplinary as it is multidimensional: A review of theory, measurement, and practice in self- concept research. In H. W. Marsh, R. G. Craven, & D. M. McInerney (Eds.), Self-processes, learning, and enabling human potential: Dynamic new ap- proaches (pp. 87-115). Charlotte: New Age Publishing. Mercer, S. (2011). Towards an understanding of language learner self-concept. Dordrecht: Springer. Mori, S. (2008). Who is to blame? University students’ perception of lack of im- provement in their English ability. JACET Journal, 47, 1-15. Mori, S. (2012). Gender differences in causal attributions in language learning. Kinki University Center for Liberal Arts and Foreign Language Education Journal. Foreign Language Edition, 3, 147-161. Mori, S., Ming, T. S., Nor, N. F. M., Suppiah, V. L., & Imm, O. S. (2011). Attribution tendency and its relationship with actual and perceived proficiency. GEMA Online. Journal of Language Studies, 11, 199-218. Nassaji, H., & Fotos, S. (2004). Current developments in research on the teaching of grammar. Annual Review of Applied Linguistics, 24, 126-145. Nicholls, J. G., & Miller, A. T. (1983). The differentiation of the concepts of diffi- culty and ability. Child Development, 54, 951-959. Peacock, M. (2010). Attribution and learning English as a foreign language. ELT Journal, 64, 184-193. Phothongsunan, S. (2015). Causal attributions of English learning success of Thai stu- dents in an international university. NIDA Development Journal, 55, 101-129. Pomerantz, E. M., Altermatt, E. R., & Saxon, J. L. (2002). Making the grade but feeling distressed: Gender differences in academic performance and in- ternal distress. Journal of Educational Psychology, 94, 396-404. Raoofi, S., Tan, B. H., & Chan, S. H. (2012). Self-efficacy in second/foreign lan- guage learning context. English Language Teaching, 5(11), 60-73. Robertson, J. S. (2000). Is attribution training a worthwhile classroom interven- tion for K–12 students with learning difficulties? Educational Psychology Review, 12, 111-134. Longitudinal effects of task performance and self-concept on preadolescent EFL learners’ causal. . . 655 Ryan, S., & Mercer, S. (2012). Implicit theories: Language learning mindsets. In S. Mer- cer, S. Ryan, & M. Williams (Eds.), Psychology of language learning: Insights from research, theory and practice (pp. 74-89). Basingstoke: Palgrave Macmillan. Sahinkarakas, S. (2011). Young students’ success and failure attributions in lan- guage learning. Social Behavior and Personality, 39, 879-886. Sakai, H., & Kikuchi, K. (2009). An analysis of demotivators in the EFL classroom. System, 37, 57-69. Schulz, R. A. (1996). Focus on form in the foreign language classroom: Students’ and teachers’ views on error correction and the role of grammar. Foreign Language Annals, 29, 343-364. Schunk, D. H. (2008). Attributions as motivators of self-regulated learning. In D. H. Schunk & B. J. Zimmerman (Eds.), Motivation and self-regulated learning: Theory, research, and applications (pp. 245-266). New York: Routledge. Schunk, D. H., & Zimmerman, B. J. (2006). Competence and control beliefs: Distin- guishing the means and ends. In P. A. Alexander & P. H. Winne (Eds.), Hand- book of educational psychology (2nd ed., pp. 349-367). New York: Erlbaum. Simon, E., & Taverniers, M. (2011). Advanced EFL learners’ beliefs about lan- guage learning and teaching: A comparison between grammar, pronunci- ation, and vocabulary. English Studies, 92, 896-922. Sorić, I., & Ančić, J. (2008). Learning strategies and causal attributions in second language learning. Review of Psychology, 15, 17-26. Stables, A., Murakami, K., McIntosh, S., & Martin, S. (2014). Conceptions of ef- fort among students, teachers, and parents within an English secondary school. Research Papers in Education, 29, 626-648. Stipek, D. J., & Mason, T. C. (1987). Attributions, emotions, and behavior in the elementary school classroom. Journal of Classroom Instruction, 22(2), 1-5. Taşkiran, A., & Aydin, B. (2017). EFL learners’ causal attributions and dimension- ality styles for perceived success and failure. Anadolu University Journal of Social Sciences, 17(2), 11-26. Tomarken, A. J., & Serlin, R. C. (1986). Comparison of ANOVA alternatives under variance heterogeneity and specific noncentrality structures. Psychologi- cal Bulletin, 99, 90-99. Tulu, G. (2013). Boys’ and girls’ attribution of performance in learning English as a foreign language: The case of Adama high schools in Ethiopia. Educa- tional Research and Reviews, 8, 2197-2211. Vispoel, W. P., & Austin, J. R. (1995). Success and failure in junior high school: A critical incident approach to understanding students’ attributional beliefs. American Educational Research Journal, 32, 377-412. Watkins, D., & Gutierrez, M. (1989). Between- and within-construct aspects of academic self-attributions: An investigation of the construct validity of an Günter Faber 656 Australian instrument for Filipino subjects. Australian Journal of Psychol- ogy, 41, 291-301. Weiner, B. (2005). Motivation from an attribution perspective and the social psy- chology of perceived competence. In A. J. Elliott & C. S. Dweck (Eds.), Hand- book of competence and motivation (pp. 73-84). New York: Guilford Press. Williams, M., & Burden, R. L. (1997). Psychology for language teachers: A social constructivist approach. Cambridge: Cambridge University Press. Williams, M., Burden, R., Poulet, G., & Maun, I. (2004). Learners’ perceptions of their successes and failures in foreign language learning. Language Learn- ing Journal, 30(1), 19-29. Yaghoubi, A., & Rasouli, Z. (2015). EFL learners’ attributions to English language learning. International Journal of English Language and Literature Studies, 4(3), 130-138. Yeung, A. S., Lau, S., & Nie, Y. (2011). Primary and secondary students’ motiva- tion in learning English: Grade and gender differences. Contemporary Ed- ucational Psychology, 36, 246-256. Yilmaz, C. (2012). An investigation into Turkish EFL students’ attributions in reading comprehension. Journal of Language Teaching and Research, 3, 823-828. Yördem, A. (2016). An investigation into the prep school students’ attributions for success and failure in an EFL context: The case of a state university in Turkey (Unpublished doctoral dissertation). Çanakkale Onsekiz Mart Uni- versity, Çanakkale, Turkey. You, S., Hong, S., & Ho, H.-Z. (2011). Longitudinal effects of perceived control on academic achievement. Journal of Educational Research, 104, 253-266. Zohri, A. (2011). Causal attributions for failure and the effect of gender among Mor- roccan EFL university learners. English Language Teaching, 4, 130-137.