The chapter first presents an overview of the study, followed by discussion of the findings on the measures and the proposed hypotheses with regard to prior research.
Then, the chapter discusses implications for pedagogy and theory, followed by conclusions.
Overview of the Study
The aim of the study was to test the veracity of a model hypothesizing research-article (RA) abstract writing ability to be the outcome of a motivated learning process in which identity-based writing motivation interacts with use of self-regulatory writing strategies and a covariate of L2 literacy. To test this model, the study collected data from 185 participants of L2 graduate students majoring in applied linguistics (i.e., master’s and doctoral students) who were recruited on a voluntary basis across 15 universities in Taiwan. The participants underwent a 130-minute procedure involving four measures in sequence, L2 literacy test (55 minutes), RA-abstract writing task (60 minutes), and surveys on identity-based writing
motivation and self-regulatory writing strategy use (15 minutes). L2 literacy test and RA-abstract writing task were given in a counter-balanced order: half of the
participants took the test and then the writing, the other diding the opposite.
In particular, the model was termed as the motivated abstract-writing model (the MAW model), and tested by the 2-step approach of the structural equation modeling.
First, the measures in the study were tested in a measurement model to ensure their psychometric properties. Second, against the backdrop of the measurement model, a structural model of the MAW model was constructed and evaluated. Evaluation on the MAW model was completed by the findings on three major hypotheses, as shown below.
173
Hypothesis 1 (supported): While controlling for L2 literacy and use of self-regulatory writing strategies, L2 graduate students’
identity-based writing motivation has a direct effect on their English research-article (RA) abstract writing ability.
Hypothesis 2(unsupported): While controlling for L2 literacy and identity-based writing motivation, L2 graduate students’ use of self-regulatory writing strategies has a direct effect on their English RA-abstract writing ability.
Hypothesis 3 (unsupported): L2 graduate students’ identity-based writing motivation has an indirect effect on their English RA-abstract writing ability via their use of self-regulatory writing strategies.
These findings are briefed in Figure 5.
Figure 5. The findings on the Motivated Abstract-Writing model.
Solid lines indicate the support for the hypotheses, while the dotted lines indicate the non-support for the hypotheses.
174
Discussion
On the basis of the findings on the MAW model, the following discussion will proceed in an attempt at aligning these findings with the previous studies on the similar issues in question. First, the discussion will address the measures used here in relation to the previous studies, although they were treated as a prerequisite rather than a pivotal concern in the proposed hypotheses. A brief discussion on the measures may provide more background information on how the MAW model is constructed and evaluated. Second, the discussion will address the findings on the MAW model, focusing on the three hypotheses in relation to the previous studies respectively.
The Measures Developed in the Study
As reported in Chapter Four, analyses on reliability, CFA, and construct validity initially substantiate the psychometric quality of the 4 measures designed for the 4 factors. The four measures included GEPT-advanced for L2 literacy, identity-based writing motivation inventory (IWMI) for identity-based writing motivation,
self-regulatory writing strategy inventory (SWSI) for self-regulatory writing strategies, and research-article abstract performance assessment (RAPA) for RA-abstract writing ability. These findings reveal that an acceptable-fitting 4-factor model holds in line with the proposed theoretical construction (i.e., the motivated abstract-writing model, the MAW model). Besides the overall acceptable-fit of the MAW model, these
measures are worthy of note; particularly for the three measures developed for the current study, including IWMI, SWSI, and RAPA.
175
Identity-based Writing Motivation Inventory (IWMI).The preliminary validity
evidence on IWMI suggests the latent variable of identity-based writing motivation can be manifested and measured by five subscales: interest value, utility value, cost, connectedness value, and ability self-concept. Specifically, these five subscales were adopted to operationalize the definition of identity-based writing motivation in the present study, “an identity-development process in which L2 graduate students seek to develop an identity of non-native-English-speaking (NNES) researchers by learning to write RAs acceptable to their disciplinary communities,” as stated in the
Introduction chapter. By these five subscales, IWMI shows a feasibility to
contextualize the latent variable of identity-based writing motivation into learning of RA-abstract writing. In particular, it is feasible to synthesize the identity-based motivation (Oyserman et al., 2002, Oyserman & Destine, 2010) and the
expectancy-value theory in educational psychology (Eccles et al, 1983; Eccles, 2009), with the socio-educational theory (Gardner, 1985, 2007) and the genre theory (Swales, 1990) in L2 learning when describing L2 graduate students’ motivation for RA
writing in Taiwan.
Aside from this overall evidence, it is noteworthy that a few error-covariances were drawn between subscales of IWMI, including the subscales of Interest Value and Cost, the subscales of Cost and Ability Self-concept, and the subscales of Utility Value and Connectedness Value. First, the Cost subscale significantly correlated with the subscales of Interest Value and Ability Self-concept in error terms. As stated in the Results chapter, this correlation may be attributed to how item measures were phrased:
Interest Value (e.g., “I find English research-article writing very interesting), Cost (e.g., I have to give up a lot to do well in writing research-articles in English), and Ability Self-concept (I am good at writing English research-articles). These item measures all involved self-oriented reasons individuals perceived for RA writing in
176
English. Moreover, these self-oriented reasons may be partially related to (a) how actively individuals pursue “personal identity” through a particular task that individuals feel more efficacious in achieving the task, and (b) what subsequent choices individuals will take in a pursuit or denial of this personal identity (Eccles, 2009). However, the findings on an error-covariance between Interest Value and Cost, and that between Cost and Ability Self-concept appear flimsy. Clearly, the finding does not allow an extensive interpretation on how Interest Value, Cost, and Ability Self-concept closely relate to individuals’ pursuit or denial of a personal identity, thus calling for further investigations.
In addition, the Cost subscale was found the least reliable measure as reflected in its low R-squared value and factor loading value. This result however was not
surprising, and actually consistent with Conley’s (2007) statement on the absence of cost in many studies of subjective task value (p. 87). Although Conley did not elaborate on the reasons why these previous studies exclude cost when testing
motivation structure by the expectancy-value theory, it can be speculated that cost did not perform as well as the other subscales regarding psychometric properties.
Theoretically, cost has been identified as a measurable subscale of motivation (Eccles et al., 1983). Empirically, this study found that when addressing L2 graduate students’ motivation for English RA writing, cost appeared elusive in mensurability, as compared to other subscales (i.e., interest value, utility value, attainment value, and ability self-concept) in the expectancy-value theory. Given this finding, it is
understandable that previous studies exclude cost from scale of subjective task (Conley, 2007). If cost is indeed an indispensible element of motivation in theoretical terms, its psychometric properties will require further investigations, such as
improving its mensurbility with better item measures. Accordingly, cost may also become meaningful in practical terms.
177
Second, the Connectedness Value subscale significantly correlated with the Utility Value subscales in error terms. This significant correlation might have been related to how connectedness value was conceptualized here. In the study,
connectedness value is a new construct proposed to reflect context-specific motivation of L2 graduate students. Inspired by the construct of integrativeness (Gardner, 1985, 2007), connectedness value refers to L2 graduate students’ subjective appraisals of reasons and benefits for socializing into a chosen disciplinary community by writing RAs acceptable to the community. Connectedness value also substitutes for
attainment value in the expectancy-value theory. As a context-specific type of attainment value, connectedness value represents a specific value for L2 attainment that encompasses gaining a social identity. According to this definition, connectedness value refers to society-oriented reasons and benefits that individuals appraise for RA writing in English (“Being good at writing English research-articles is important to me because it will allow me to gain a social prestige in my disciplinary community.”)
Likewise, utility value also involves such society-oriented reasons and benefits (e.g., “Being good at writing research-articles is important when I look for a job or pursue further studies.”). It is rather likely that these society-oriented reasons may account for error-covariance in Utility Value and Connectedness Value, as stated in the Results chapter. Moreover, these society-oriented reasons (manifested by utility value and connectedness value) may be partially related to (a) how actively
individuals pursue “collective identity” through a particular task that individuals feel more efficacious in achieving the task, and (b) what subsequent choices individuals will take in a pursuit or denial of this collective identity (Eccles, 2009). However, the finding on the error-covariance between Utility Value and Connectedness Value appears tentative. Clearly, the finding does not allow an extensive interpretation on how closely utility value and connectedness value relate to individuals’ pursuit or
178
denial of a collective identity; further investigations are needed.
Self-regulatory Writing Strategy Inventory (SWSI). The preliminary validity
evidence on SWSI suggests the latent variable of self-regulatory writing strategy can be manifested and measured by three subscales: Cognition Regulation, Motivation Regulation, and Socio-interactive Resources Regulation. Specifically, these three subscales were adopted to operationalize the definition of self-regulatory writing strategy in the present study, “perceived effort that individuals make to effectively regulate cognition, motivation, behavior, and context to better approach L2 writing,”
as stated in the Introduction chapter. By these three subscales, SWSI shows a
feasibility to contextualize the latent variable of self-regulatory strategy into learning of RA-abstract writing in L2. In particular, it is feasible to synthesize the
self-regulated learning theory (Pintrich, 2004; Wolters et al., 2003), L2 writing theory, and RA-writing theory (Petrić & Czarl, 2003; Swales & Feak, 2009) when describing L2 graduate students’ use of self-regulatory strategy for RA-abstract writing in
Taiwan.
Aside from this overall evidence for validity of SWSI, it is noteworthy that an error-covariance was drawn between subscales of Cognition Regulation and Motivation Regulation in SWSI. Specifically, the covariance between the two subscales was negative, indicating that if the participants increased some of
motivation regulation strategies they tended to decrease some of cognition regulation strategies (i.e., efforts that participants make by regulating cognition for RA-abstract writing ). As stated in the Results chapter, a possible explanation for this negative covariance was that once the participants used more motivation regulation strategies they might have less time and energy left for cognition regulation strategies. However, this speculation remains tentative. To date, questions remain regarding (a) what are real causes for the negative covariance and (b) what consequences the covariance will
179
bring to learning achievement. Apparently, these questions call for further research.
Despite those pending questions, the following discussion will proceed by
highlighting Schwinger et al.’s (2012) study that operationalized Wolters et al.’s (2003) self-regulated learning theory and detailed the interrelations between motivation
regulation and effort regulation strategies (i.e., the present study defined cognition regulation as perceived efforts learners make to effectively regulate cognition so as to write better). Particularly for motivation regulation strategies, Schwinger et al.’s (2012) operationalized eight sub-constructs, enhancement of situational interest, enhancement of personal significance, mastery self-talk, performance-approach self-talk, performance-avoidance self-talk, environmental control, self-consequating, and proximal goal setting. Likewise, the present study operationalized motivation regulation strategy by three sub-constructs, relevance enhancement, relative ability self-talk, and self-consequating, adopted from Wolters et al.’s (2003) proposal as well.
Since the present study and Schwinger et al.’s (2012) both followed Wolters et al.’s proposal in operationalizing motivation regulation, a further comparison on the findings of the two studies seems justifiable.
Specifically, Schwinger et al. (2012) argued that students, who only resort to performance self-talk as motivation regulation strategy, are less effective in increasing their task engagement or achievement. In a similar thread of research,
performance-avoidance self-talk has been found not or weakly correlated with engagement and achievement (Lenzner & Dickhause, 2009 cited in Schwinger et al., 2012). Schwinger et al. (2012, p. 8) speculate that performance-avoidance goals seem
“a dubious form of motivation” that often relates to maladaptive outcomes, such as test anxiety (Elliot & McGregor, 1999), self-handicapping (Urdan, 2004), and low academic self-efficacy (Middleton & Midgley, 1997). Based on this conjecture, certain types of self-talk are less effective motivation regulation strategies that do not
180
always promote adaptive task engagement and achievement.
Similarly, the present study found a negative correlation in error terms between cognition regulation and motivation regulation. Notably, this negative correlation occurs in error terms, rather than in factor loadings, indicating the negative correlation is something additional and minor. Yet the negative correlation is apparently different from zero or low correlation between motivation regulation and effort regulation reported by Schwinger et al. (2012). Apparently, what makes the negative correlation here requires further research. In particular, future research should explore what types of motivation regulation strategies are more or less effective in producing adaptive cognitive engagement and achievement in learning.
Research Abstract Performance Assessment (RAPA). As shown in the findings
on the MAW model, RAPA including a timed RA-abstract writing task and two rating scales serves as a valid and meaningful measure for RA-abstract writing ability of L2 graduate students. This finding shows that it is feasible to construct a timed
abstract-writing task and its corresponding rating scales for research purposes.
Specifically, the rating scales were made by translating traditional qualitative analysis on global rhetorical moves and local lexico-grammmatical patterns (Swales & Feak, 2009; Hsieh & Liou, 2008; Pho, 2008) into a more quantitative measure consisting of the two corresponding scales of global move and local pattern.
The two rating scales by themselves may serve as assessment criteria in writing pedagogy. In fact, they can be useful for L2 graduate students and instructors because they embody relative clarity and explicitness of assessment (Jonsson & Svingby, 2007). Usually, a rating scale embodies such clarity and explicitness by elucidating what is important through the scoring criteria and descriptors (Jonsson & Svingby, 2007). Drawing upon scoring criteria and descriptors in the two rating scales, instructors are more likely to conduct “reflective practice” on the effectiveness of
181
their instruction (Beeth et al., 2001; Jonsson & Svingby, 2007). In the same way, students are more likely to conduct self-assessment on their writing ability (Jonsson &
Svingby, 2007). Such self-assessment may be meaningful to L2 graduate students under investigation, since the scoring criteria and descriptors demonstrate relative
“transparency” that may help students learn “why they are doing what they are doing”
(Jonsson & Svingby, 2007, p. 139).
Given its pedagogical and assessment potential, it may be of benefit to implement RAPA similar to the one used in this study as a part of proficiency or diagnostic tests for RA writing ability in post-graduate programs. For example, proficiency tests can be administered throughout the post-graduate programs to assess research-article writing ability in general. A diagnostic test that is developed to reveal students’ strengths and weaknesses in writing ability, to offer detailed feedback that can guide students’ follow-up actions, and to target “specific” instead of general abilities (Knoch, 2010), can be administered at the onset or end of the instruction in post-graduate programs. These applications of RAPA may make possible a
finer-grained assessment on the RA-abstract writing ability of L2 graduate students.
Because the present findings on applicability of the two rating scales remain preliminary, further studies are required to cross-validate their applicability.
Findings on the Testing of Three Hypotheses