The purpose of this study was to examine the causal relationships among the latent variables of self-identity, self-determination, self-confidence, attitude, and English proficiency. The expectation was that it would check (a) whether EFL learners’ affective variables such as attitude and cognitive variables (e.g.,
self-confidence) may have an impact on both intrinsic and extrinsic self-determination;
(b) whether EFL learners’ intrinsic/extrinsic self-determination plays a significant role by having a direct effect on English proficiency, and self-identity, and an indirect effect on self-identity as mediated by English proficiency; (c) whether EFL learners’
English proficiency might impact their self-identity (i.e., the effects of linguistic outcomes on nonlinguistic outcomes); and (d) whether the attitudes of EFL learners have a direct or indirect influence on their self-identity.
The rationale for constructing such a structural model was based on the importance of the social psychological perspectives of self-identity because it is assumed that language learner self-identity is closely related to successful acquisition of foreign/second language(s) (Anwaruddin, 2012) and that self-identity is interrelated with other latent variables in the social-psychological field. It is thus worth addressing the nature of the social psychological variables of attitudes, self-confidence and self-determination as antecedents that have an influence on language learner self-identity in the EFL context.
The second purpose was to investigate the internal construct of each latent variable of the study, which includes self-identity, self-determination, self-confidence, and attitudes, along with the examination of the reliability and validity of each latent variable.
Finally, the third purpose was to explore the variance or difference of EFL learners’ self-identity, self-determination, self-confidence, attitudes and English proficiency based on their gender and English proficiency level.
This chapter will summarize the major findings of the study and interpret these findings in the context of SLA, applied linguistics and educational psychology research.
Summary and Discussion of the Major Findings
The final model in this study provided a good fit to the data, supporting the hypothetical interrelationships among English learning, self-identity English learning, English learning self-determination, English learning self-confidence, English
learning attitude, and English proficiency in the EFL classroom context. The full structural equation model of the attitude-driven and self-determination-mediated English learning self-identity integrated some of the most important social
psychological elements of L2 acquisition research into one large-scale causal study.
The first part of the discussion will center on the systematic causal relationship of the affective and cognitive variables (i.e., attitude and self-confidence) towards social psychological variables (i.e., intrinsic and extrinsic self-determination) and English proficiency, before leading to the variables of self-identity.
The Direct Effects among the Latent Variables of the Full model
The first major concern of the present research focused on the direct effects among the latent variables. As shown in Figure 28, the direct causal relationships can be found in the paths of attitude-confidence, attitude-motivation, attitude-performance, attitude-identity, confidence-motivation, confidence-performance,
motivation-performance, and motivation-identity.
English Learning Attitude
The final model suggests that the latent variable of English learning attitude has
both significant direct and indirect effects on the other variables, and the first major concern of the current study was to discuss the direct relationships among the factors.
In the final model (Figure 28), the English learning attitude appears to be the original engine, or, an initial contributing factor to drive the model. Under the affective domain, it is clear that attitude has significant and direct effects on all the
self-determination-related regulation/motivation variables. Specifically, both intrinsic motivation and extrinsic motivation can be predicted by attitude β = .45 (p < .05; R2 = 0.21; Cohen f2 = 0.27) and β = .42 (p < .05; R2 = 0.18; Cohen f2 = 0.22), respectively.
These two causal links, with both medium effect sizes, demonstrate moderate
strengths of the direct causal links, showing that attitude has the predictive power over intrinsic and extrinsic motivation; in other words, the enhanced intrinsic and extrinsic motivation can be meaningfully explained by positive attitude.
Furthermore, A-motivation can also be negatively predicted by attitude with a regression coefficient of -.85 (p < .05; R2 = 0.73; Cohen f2 = 2.70). The strongest causal impact in absolute value was observed on the link from attitude to
A-motivation. The R2 value of the causal link from attitude to A-motivation was approximately 3.5 times larger than that from attitude to intrinsic motivation, and approximately 4 times larger than that from attitude to extrinsic motivation. Such results suggest that language learning attitude plays a critical role in determining EFL learners’ motivational levels in learning a language, and such effects are particularly pronounced in the circumstances in which A-motivation is likely to occur. Importantly, this predictive power is present in a pure sense on the grounds that self-confidence provides little explanatory power over A-motivation (β = -.07; p > .05; R2 = 0.005;
Cohen f2 = 0.005).
These results are consistent with most academic research on the issue between attitude and motivation. For example, by applying the analysis of structural equation
modeling, Gardner et al. (1997) demonstrated a causal relationship from attitude towards language learning and motivation, indicating that high levels of motivation are derived from a favorable attitude toward the learning situation. Dörnyei (2003) also claims that enhancing the learners’ L2 related values and attitudes can generate the initial motivation, and it can be further realized that attitude is a contributing factor to motivation. More recently, Gardner (2006) has proposed that the motivation factor is characterized by a high level of motivational intensity to learn the target language, and favorable attitudes toward the target culture. The current study not only confirms the significance of attitude in influencing motivation, but also extends the previous research findings (e.g., Gardner et al., 1997). With respect to the connection between attitude and self-confidence, based on Clément’s theory of language learning motivation (e.g., Clément, 1980; Clément, Dörnyei & Noels, 1994), Csizér and
Kormos (2008) propose a link between self-confidence and language related attitudes.
Although Gardner’s model supports the indirect effects of attitude on self-confidence via motivation, it does not test the possibility of the direct effects from attitude to self-confidence. To shed light on this unknown, this as yet unconfirmed direct causal relationship was further tested in this study with the results corroborating the causal link. Hence, it seems that attitude not only predicts affective variables such as motivation, but also cognitive variables such as self-confidence.
Moreover, as mentioned above, attitudes toward English learning have direct and significant influences on confidence among English learners. According to the
regression coefficient, the value (from attitude to self-confidence) is .58 (R2 = 0.34; f2
= 0.52), indicating the English learning attitude serves as a strong predictor of English learning self-confidence. According to the past research, Clément and Kruidenier (1985) found that the effects of attitude on self-confidence were indirect and mediated by language contact. Another study by Yoshima et al. (2004) yielded similar results,
revealing the indirect effects of attitudinal factors on communicative confidence via motivation. However, in contrast with the results of those two studies, the current study discovered that attitude had a strong and direct impact on self-confidence, a finding with important implications regarding the strong predictive power of attitude--specifically, that attitude shapes self-confidence during the process of language acquisition.
Furthermore, mediated by self-determined-related regulation variables as well as self-confidence, attitude can still exert significant and direct effects upon English proficiency (β = .29; p < .05; R2 = 0.09; Cohen f2 = 0.10). Although the effect appears small (i.e., the slight magnitude of the direct causal link), it is nonetheless clear that English proficiency can be ultimately and meaningfully explained to some extent by attitude, as it is conditioned and mediated by both affective and cognitive factors. This finding again corroborates the pivotal role of attitude in orienting EFL learners’
motivation, self-confidence, and linguistic outcome. Whereas the causal link between attitude and English proficiency was not tested in Gardner’s model, the current study confirms that this causal link can be upheld with practical significance (i.e., small effect size). Likewise, this significant path coefficient sheds new light on the causal chain of attitude-motivation-proficiency (cf. Gardner et al., 1997). The findings of the current study mesh with Noels et al.’s (2000) idea, who proposed that affective variables, such as attitudes, orientations, anxiety and motivation, are important in predicting L2 achievement. From a pedagogical standpoint, the concept that attitude can directly lead to L2 achievement may bear the important implication that, by fostering EFL learners’ positive attitude towards the target language, their achievement can be predicted.
In addition to the latent variables of self-confidence and self-determination, attitudes toward English learning also have a significant direct impact on English
learning self-identities. With direct predictive power (β = .49; p < .05; R2 = 0.24;
Cohen f2 = 0.32), the final model reveals that English learning attitude can be
considered a positive contributor to influence English learning adaptive self-identity, which includes additive self-identity and productive self-identity. The medium effect size further implies a moderate strength of the direct causal path, in which attitude may have a modest impact on adaptive self-identity; in other words, by controlling for the variances of other factors in the final model, an increasing value of English
learning attitude will result in English learning adaptive self-identity being meaningfully increased to a moderate level.
Moreover, such findings make plausible the notion that EFL learners with a more positive attitude are able to either maintain their L1 and L2 self-identities
simultaneously without conflicts, or reinforce the development of self-identities of both languages simultaneously. Similar outcomes were also found in subtractive self-identity, with the regression coefficient β = .37 (p < .05; R2 = 0.14; Cohen f2 = 0.16), revealing the possibility that a positive attitude will lead to the construction of subtractive self-identity, though the predictive effects are less powerful than for adaptive self-identity. The medium effect size (comparatively lower than that of adaptive self-identity) suggests that the strength of the direct causal relationship was only modest. This is because attitude exerts only a modest effect on subtractive self-identity. By controlling for the variances of other factors in the final model, the English learning subtractive self-identity also significantly grew to a modest degree when the value of English learning attitude became stronger, with only a half of the effect regarding the value of adaptive self-identity.
In particular, the data-driven results can clearly indicate that a stronger level of attitude will throw the light on adaptive self-identity (i.e., additive self-identity and productive identity) and subtractive self-identity, and the values of R2 and Cohen f2 of
the causal link from attitude to adaptive self-identity were approximately 2 times larger than those of attitude on subtractive self-identity. Such results imply that, in general, learners with a positive attitude are more likely to develop L1 and L2
self-identities (i.e., additive self-identity and productive self-identity) simultaneously (β = .49; p < .05; R2 = 0.24; Cohen f2 = 0.32), while it is still possible that learners’ L1 self-identity is replaced by L2 self-identity (i.e., subtractive self-identity; β = .37; p
< .05; R2 = 0.14; Cohen f2 = 0.16).
In contrast, different attitudes have significant direct but negative effects on split self-identity. With the regression coefficient β = -.40 (p < .05; R2 = 0.16; Cohen f2 = 0.20), the medium effect size indicates a moderate magnitude of the direct causal link, in which attitude can have a modest impact on split self-identity; this means that by controlling for the variances of other factors in the final model, English learning split self-identity was moderately and negatively influenced by attitude when the value of English learning attitude increased. To be more precise, English learning attitude plays a negative role in predicting English learning split self-identity to a mediocre level. Likewise, attitude has negative direct effects on zero self-identity: the
regression coefficient β = -.32 (p < .05; R2 = 0.11; Cohen f2 = 0.12) and the small effect size (comparatively lower than that of split self-identity) imply a mild strength of the direct causal link, in which, with the variances of other factors under control in the final model, a slight level of zero self-identity will be meaningfully amplified when the value of English learning attitude diminishes to a certain level.
With regard to these two negative paths from attitudes towards split self-identity and zero self-identity, the values of R2 and Cohen f2 of the model show that attitude was a stronger predictor for split self-identity than for zero self-identity, and among the four variables of self-identity, the strongest one centered upon adaptive
self-identity, while the weakest one was zero self-identity.
On balance, the positive causal relationship between English learning attitude and English learning self-identity appears to be genuine, and the findings of the studies are consistent with existing research. Taylor (2000), for example, investigated the attitudes of pupils toward language exchange programs, and suggested that pupils under the influence of peer-pressure and other incentives will take an interest and be more willing to take part in language exchange activities to develop their identities as regards the target language and culture. Chun (2001) reported similar findings by analyzing a conversation in which a Korean-American speaker used features of African-American English during the process of language contact, thus revealing that attitude within the community can have a strong influence on the degree to which dialect assimilation occurs, which, in turn, may affect the development of self-identity.
Ellis (2008) further pinpointed the fact that attitudes learners hold towards the learning of a particular L2 reflect the intersection of their views about their own self-identity and those about the target-language culture. The power of attitude in shaping one’s linguistic self-identity is profound and should not be ignored in the EFL context.
English Learning Self-Confidence
The final model reveals that the latent variable of English learning
self-confidence is another important element which directly influences English learning self-determination. The final model suggests that self-confidence is a
significant direct predictor for intrinsic motivation (intrinsic self-determination), with the regression coefficient β = .49 (p < .05; R2 = 0.24; Cohen f2 = 0.32); therefore, based on the medium effect size, such outcomes can be understood as having a direct causal path of modest strength. In this case, self-confidence can shed light on intrinsic motivation, which implies that by controlling for the other factors of the final model, when the value of self-confidence increases, the intrinsic motivation will also be
substantially raised to a moderate level. Self-confidence shows a similar direct impact on extrinsic motivation (extrinsic self-determination), with the regression coefficient β
= .58 (R2 = 0.34; Cohen f2 = 0.52), and the large effect size signifies a strong direct causal relationship and self-confidence may powerfully influence the level of extrinsic motivation. Specifically, by controlling for the variances of the other factors in the final model, with the increase of the value of self-confidence, extrinsic motivation will substantially ascend to a vigorous level because the language learners strongly believe they can get the external rewards.
From a social psychological perspective, self-determination theory encompasses the needs for autonomy, competence and relatedness. The findings of this study reveal that self-confidence has the capacity to predict such needs during the process of second language acquisition. Furthermore, within the ESL/EFL context, learners with greater self-belief will find that confidence will not only give them more enjoyment and satisfaction during the learning process, but also strengthen their external
regulation to assist them in achieving their learning goals. In brief, this study focused on the positive and direct causal links between self-confidence and different types of motivation, including intrinsic, extrinsic and A-motivation.
However, while the findings that suggest self-confidence has a directly negative impact on A-motivation are interesting, the results did not reach a level of significance (R = -.07; p > .05; R2 = 0.005; Cohen f2 = 0.005). Yet, learners with more confidence regarding second language acquisition seem to encounter less A-motivation, which suggests that learners without any motivation towards the language cannot be
significantly affected by the level of self-confidence. Such results are also mentioned by Sakai and Kikuchi (2009), who refer to several studies where researchers have included also internal forces in SLA, such as a learner’s lack of self-confidence and negative attitudes of the learners in addition to external forces. They point out that
Dörnyei (2001) listed reduced self-confidence and negative attitude towards the foreign language as sources of A-motivation, but only for lower-proficiency learners and not generalizable to the overall population.
As for the three causal links, Clément et al.’s (1985) study suggested that self-confidence should influence motivation. By applying regression analysis, Csizér and Kormos (2008) proposed similar findings suggesting that language students’
motivated learning behavior was largely affected by students’ linguistic
self-confidence and their integrative motivation. Furthermore, Gardner (2006) also considered that L2 learners are motivated individuals when they “exhibit positive affections, and demonstrate self-confidence (self-efficacy)”, and further claimed that the frequency and quality of language contact can give rise to individual differences in self-confidence with the language, which would in turn influence the individual’s motivation to learn the language (Gardner, 2010).
This study offers a more detailed picture of this causal link. That is, the value of this study rests on the elaboration of the causal links between self-confidence and different types of motivation including the direct paths of intrinsic, extrinsic and A-motivation. More specifically and importantly, both intrinsic and extrinsic motivation can be meaningfully depended on one’s self-confidence, whereas A-motivation cannot. There are two plausible explanations for this.
These two potential explanations derive from related studies from the fields of applied linguistics, as well as social and educational psychology. First, A-motivation, which is independent of intrinsic and extrinsic motivation, presents the absence of either intrinsic or extrinsic reasons for performing an activity. It refers to the
phenomenon in which the individuals see no correspondence between their action and the consequences of that action. However, self-confidence relates to learners’ belief in themselves and their ability to complete a task. Therefore, if learners find that their
actions, in this case their efforts to learn a second language, have no consequences then it is difficult for them to evaluate the efficacy of those actions. The second possible theory that could support the findings of the current study is that
self-confidence, based on the essence of self-efficacy and personal belief towards the goal, represents little about concept of interests. However, A-motivation, which stems from a lack of interest in doing the action (Deci & Ryan, 2012), is different from self-confidence; there is thus no connection between the two variables.
Another direct prediction regarding self-confidence is connected to English proficiency. This causal link was established by controlling for the effects of both attitude and self-confidence on motivation, and the effects of attitude on proficiency.
This causal link is positive and should not be ignored: the regression coefficient was β
= .23 (R2 = 0.055; Cohen f2 = 0.058) and the small effect size reveals the mild strength of a direct causal path in which self-confidence will have slightly positive effects on proficiency. The insight her is that by controlling for the variances of other factors in the final model, learners with greater self-confidence about English learning will achieve a slight but meaningful increase in their English Proficiency. Hacker, Dunlosky, and Graesser (2009) provide similar ideas to support the findings of this study. Knowledge and beliefs about cognition include constructs such as self-efficacy
= .23 (R2 = 0.055; Cohen f2 = 0.058) and the small effect size reveals the mild strength of a direct causal path in which self-confidence will have slightly positive effects on proficiency. The insight her is that by controlling for the variances of other factors in the final model, learners with greater self-confidence about English learning will achieve a slight but meaningful increase in their English Proficiency. Hacker, Dunlosky, and Graesser (2009) provide similar ideas to support the findings of this study. Knowledge and beliefs about cognition include constructs such as self-efficacy