Study 5

findings in Japan (Carreira, 2012; Ando, Fuse, & Kodaira, 2008; Hiromori, 2003), introjected regulation cleanly formed a separate factor with an acceptable internal reliability, though this was clearly the weakest motivating factor with a mean score less than 2. These results indicate the importance of item wordings for effectively conveying theoretical meaning to participants in survey research.

These findings show further problems with the concept of L2-selves theory as explanatory independent of self-determination theory, as they may be safely subsumed within the SDT framework. The items used for ought-to and ideal L2-selves (Papi, 2010; Taguchi, Magid, & Papi, 2009) overlap with the items tested here for introjected and identified regulations, respectively. While this study does not represent a complete empirical investigation of the two competing theories, for all practical purposes, L2-Selves appear to be explainable by organismic integration theory. These findings allow for the final test integrating it with basic needs theory and cognitive evaluation theory in the Japanese elementary foreign language classroom.

variables as second-order factors. As in study 2, this model hypothesized a direct relationship between supportive structure and engagement, mediated by a relationship with students’ basic needs. Based on the high correlations between the latent variables for autonomy, competence, and relatedness, the three were treated as indicators of a second-order latent variable. As a final step to test the relationship between classroom environments, basic needs theory, and organismic integration theory, motivational regulations were treated as outcome variables, regressing on engagement. The theory behind this was students’ active and passive behaviors in class influence their motivations, which can then be measured through the selected survey items.

Figure 6.4. Hypothesized structural model of motivational development.

Four hundred twenty-three fifth-grade students (201 female, 222 male) in western Japan completed engagement surveys following foreign language classes in early February 2013. Motivation surveys were then completed at the end of the school year in March. In reaction to the high correlations between the variables using 4-point Likert-type scales in Studies 2 and 3, a 5-point scale was used with the hope of creating additional room for

variance and alleviating the apparent ceiling effects. As in the previous studies, data was analyzed using MPlus using the same standard fit cutoffs, and Likert data was again treated as ordered categorical and analyzed using robust weighted least squares.

6.6.2 Study 5 Results

Supportive structure was found to correlate strongly with SDT needs, with a weaker though still meaningful direct relationship with classroom engagement. Need satisfaction likewise influenced engagement. In-class engagement equally strongly influenced intrinsic regulation, with weaker effects on identified and introjected regulation, and negatively predicted extrinsic regulation. Gender had significant effects on intrinsic, introjected, and extrinsic regulations, though effects were relatively weak in comparison with other factors.

As in the previous studies, the relationships between variables were quite strong, likely inflated by the same negative skew to the data as seen in the previous models. Model fit was highly acceptable, χ² (503) = 935.448, p < .000, RMSEA = .045 (CI = .041, .05), CFI = .97, TLI = .96. The full model is displayed in Figure 6.5, with the zero-order correlations for latent variables displayed in Table 6.7. As in the previous studies, multicollinearity was an issue, with many students answering 4 and 5 for numerous items. As with other studies of elementary students (Spinath & Steinmayr, 2008), students were likely to give more positive answers. This positive bias and resulting high correlations created a situation where one of the standardized coefficients (2^nd order engagement) was calculated above 1.0, a situation that can occur with distinct but highly correlated predictors (Deegan, 1978; Jöreskog, 1999). At the same time, the results mirror previous work in self-determination theory that does not display the same degree of multicollinearity among predictors (Carreira, Ozaki, & Maeda, 2013; Jang, Kim, & Reeve, 2012).

Figure 6.5. Full process model of classroom engagement and motivation. *p < .05, **p < .01, ***p < .001 Table 6.7. Zero-order correlations for the latent variables with descriptive statistics and internal reliabilities.

1 2 3 4 5 6 7 8 9 10 11 12

1) Supportive structure - .73 .70 .81 .86 .89 .67 .53 .24 -.37 .84 .92

2) Competence - .72 .84 .83 .85 .66 .51 .23 -.37 .87 .88

3) Relatedness - .80 .79 .81 .62 .49 .21 -.35 .83 .84

4) Autonomy - .92 .95 .73 .57 .25 -.41 .97 .98

5) Behavioral engagement - .91 .70 .55 .23 -.40 .95 .94

6) Emotional engagement - .72 .56 .24 -.41 .98 .97

7) Intrinsic regulation - .89 .28 -.61 .75 .75

8) Identified regulation - .19 -.52 .59 .58

9) Introjected regulation - .39 .26 .25

10) Extrinsic regulation - -.42 -.42

11) 2^nd order need satisfaction - 1.00

12) 2^nd order engagement

-Mean 3.72 3.14 3.47 3.05 3.65 3.45 3.54 3.92 1.86 2.21 - -SD .78 .92 .93 .82 .81 .97 .96 1.02 .85 .96 - -Cronbach’s α .71 .78 .78 .58 .75 .84 .78 .84 .78 .75 - -All correlations significant at p < .001

While an alternative model for the data may be constructed with existing intrinsic and extrinsic variables influencing engagement, this model is implausible due to the 5-week interval between the classroom engagement surveys and the motivation surveys. Likewise, while an alternative model showing direct effects from need satisfaction to motivational

outcomes was also tested, the model failed to achieve convergence. Based on these reasons the model presented in Figure 6.5 is believed to represent the most parsimonious fit and interpretation for the hypothesized data.

6.6.3 Study 5 Discussion

As in the previous studies, Study 5 demonstrated that the perceptions of the classroom environment strongly influence students’ engagement. Engagement in turn influences students’ positive feelings of motivation. These results confirm previous work on structure and autonomy-support as crucial to motivating students, and show a pattern of correlation with the motivational regulation factors consistent organismic integration theory. The correlation between engagement and the different motivational orientations decreased in strength as the motivation became less autonomous, negatively predicting heteronomous reasons for language study. These results indicate that teachers’ day-to-day practices promote student attention and enjoyment, which in turn has lasting effects on students’ desire to learn the new language. Regularly providing students with a clear, learning-oriented environment leads to the normalization of in-class engagement, which students then perceive as an intrinsic desire to learn the language. Drawing from previous SDT research, this finding shows further evidence for the validity of self-determination in the Japanese context.

Gender showed weak but measurable influences only on students’ motivation. As other studies have shown (Fryer et al., 2014; Meece, Glienke, & Burg, 2006), male students operate at a motivational disadvantage, showing a stronger relationship with introjected and extrinsic regulations and a negative relationship with intrinsic regulation. Considering the strong and enduring effects of intrinsic motivation on educational outcomes (Reeve, 2012), boys’ general trend towards external motivators is worrisome. However, as gender did not strongly influence in-class need satisfaction or engagement, this influence may indeed be

minimal at the elementary level. More pertinent appears to be how engaged students were in learning activities, directly as a function of teachers’ appropriate educational style.

The flaw within this model comes from the fact that existing motivation over time has not been measured. Without a clear understanding of how students’ existing motivational orientations influence their learning behaviors in class, no concrete conclusion regarding the influence of classroom practice on motivation may be drawn. The research model explored in Chapter 8 aims to answer this question.