correlated with motivated language learning behavior and language learning attitudes.
This is in synch with previous studies, which found similar results.
An interesting finding arose from using nonlinear regression analysis to further analyze the relationship of language anxiety and motivation: namely, that it is highly possible that the link between the two constructs is not linear. This question has not been considered substantially in applied linguistics, and therefore has not been widely tested. The results based on the present sample were found to be significant but not very strong in suggesting that first high levels of motivation are associated with low levels of anxiety, then as anxiety increases motivation falls. As anxiety further increases there seems to be stagnation in motivation levels. Finally, as anxiety reaches even higher levels, motivation falls. This evidence suggests that language anxiety and motivation are not two sides of the same phenomenon as claimed by MacIntyre (2002), but their relationship is more complex and most likely involves the interaction of other variables as well.
In particular to the link between the L2 motivational self system and language anxiety, the ought-to self/other was found to be more tied to the experience of language anxiety than motivated language learning behavior. Since the ought-to self has been theorized to have a prevention focus, the idea that it is closely associated with avoidance goals rather than approach goals links it to language anxiety instead of motivation (Higgins, 1987, 1997; Elliot, 2006).
With respect to the research question concerning language anxiety and language learners’ beliefs, self-efficacy beliefs negatively correlated with language anxiety. This correlation was found to be the strongest among those involving language anxiety and other individual variables. This may also suggest that self-efficacy is more directly related to anxiety in that learners who judge themselves
incapable of completing language learning tasks are likely to feel anxious in a language learning situation. Results also depicted that more anxious learners tend to think of language learning as difficult and they perceive themselves to be less talented than others at learning languages.
The last group of variables that were investigated relative to language anxiety consisted of coping strategies. Here, the assumption that higher levels of language anxiety would appear together with emotion-focused coping strategies was supported.
Similarly, lower levels of language anxiety were found to parallel more use of problem-focused strategies. As the learner does not perceive the conflict (the stressor) to be so threatening as to primarily focus on alleviating the emotion, it is more likely that he/she approaches the conflict as a problem and employs problem-focused strategies of coping.
Finally, from the hypothesized discrepancies, the actual and the ought-to self, differences in teachers’ and learners’ beliefs and teachers’ and learners’ styles, results were not straightforward. According to the findings, self-discrepancies did not prove to be significantly associated with experiences of language anxiety. On the other hand, both differences in self-efficacy beliefs and preference for silent learning and teaching styles were found to correlate significantly with language anxiety in the case of more anxious learners.
In the following part of the data analysis, the variables above were organized into a structural model with the intentions of (a) depicting the direction of the influence of these variables on language anxiety and on one another (see Figure 2) and (b) determining the strength of these influences. The hypotheses of the model posed in Chapter 4 were tested using SEM procedures. The results of the SEM analysis are presented in the next chapter of the dissertation study.
CHAPTER 7:
RESULTS OF STRUCTURAL EQUATION MODELING
Structural equation modeling (SEM) was employed to examine the causal relationships between language anxiety and the elements of the proposed model of the development of language anxiety incorporating the latent and observed variables measured by the instruments described and analyzed above. Literature suggests that before establishing the relationships between latent independent and dependent variables in form of a structural model, the measurement models for each latent variable should be set up, evaluated and modified where necessary (Schumacker &
Lomax, 2004). Measurement models provide information about how much the observed variables measure the hypothesized latent variables and are often referred to as confirmatory factor models that include latent factors and measurement errors (Schumacker & Lomax, 2004). Confirmatory factor analysis models (CFA) aim to
“determine which sets of observed variables share common variance-covariance characteristics that define theoretical constructs or factors (latent variables)”
(Schumacker & Lomax, 2004, p.168). With CFA models the researcher’s goal is to confirm the validity of the theoretical models.
Consequently, in the first part of the chapter, the so-called measurement models are investigated, which are viewed as the constituents of the hypothesized full structural model. The first measurement model includes factors associated with conflicts that promote language anxiety, namely lack of self-confidence, fear of negative evaluation of peers and teachers, fear of participating in speaking tasks in a language class, ought-to/own and actual self differences, ought-to/other and actual self discrepancies and discrepancies between learners’ self-efficacy beliefs and
teachers’ efficacy beliefs about their students’ language learning, and learning and teaching style differences along the dimension of preference for silence. Second, the observed variables indicating language anxiety were tested whether they indeed explain this latent variable in a reliable and valid way. The next group of factors comprised two latent components of coping strategies: problem-focused strategies as measured by the aggregate of problem-focused coping and tension control; and emotion-focused coping strategies as measured by the aggregate variables of self-punishment and letting out emotions.
In the second part of the chapter, the results of testing the full structural model are presented. The full model consisted of the above listed four latent variables as defined by their constituents, three additional observed variables of language learning attitude, motivated language learning behavior, and self-efficacy beliefs and the hypothesized paths that represent the variables’ influence on one another (see also Figure 2).