A Comparison of Constructs
3. The present study 1 Aims
The aim of this study was to compare and contrast Mihaljević Djigunović's and Dörnyei's conceptualisations of L2 motivation with the purpose to explore which of these theoretical concepts might be better applicable to empirical research of motivation for learning English and German as L2 in a Croatian socio-educational context. In order to fulfil this aim, Mihaljević Djigunović's and Dörnyei's questionnaires designed to measure L2 motivation were investigated with respect to two subtypes of validity – convergent validity (as a subtype of construct validity) and incremental validity. More precisely, the present study seeks to answer the following questions:
a)! Is there a correspondence between Mihaljević Djigunović's and Dörnyei's concepts of motivation; that is, to what degree are the two measures of constructs related?
b)! Does Dörnyei’s measure of motivation contribute more information about learners’ motivation, beyond what might be obtained by Mihaljević Djigunović’s measure of motivation?
3.2 Participants
The sample included learners of two foreign languages in Croatia: German and English. As previous studies have shown, learners of English, in contrast to learners of German, are more motivated for language learning which can be attributed to differences in various dimensions of a school and out-of-school learning context in Croatia (Bagarić, 2007; Karlak, 2013). Unlike German learners, learners of English have considerable contact with English and the culture of various English-speaking communities through the media, so that the context of learning English in Croatia may be described as having many features of a L2 learning context in terms of the amount and quality of input readily available outside the classroom (Pavičić Takač & Bagarić, 2011). Such out-of-school learning context enables them to acquire English also unconsciously (Bagarić, 2001) through authentic, unsimulated language use, which is
generally not the case with German. This may influence learners’ personal motivational dimensions for learning the target language over the years (Mihaljević Djigunović & Bagarić, 2008; Karlak, 2013). In order to avoid the target-language based bias, learners of both English and German were included in the study.
A total of 468 learners from four Croatian secondary schools participated in this study: 236 learners of English and 232 learners of German. All participants were Croatian, aged between 15 and 19.
Learners’ final course grade is taken as an indicator of FL proficiency. In Croatian educational system, grades range from 1 to 5, with 5 being the highest, 2 the lowest passing grade, and 1 the failing grade. Participants’ average grade in the foreign language is 3.62 (SD=1.083). More specifically, English learners’
average grade is 3.63 (SD=1.084) and German learners’ average grade is 3.61 (SD=1.084). This difference was not found to be significant (t(431)=.207; p>.05).
3.3 Instruments and procedures
The instrument used in the present study to investigate motivational dimensions based on Mihaljević Djigunović’s (1996, 1997, 1998) motivational framework (see Section 2.1) was the Types and Intensity of Motivation for learning EFL Questionnaire (Mihaljević Djigunović, 1998). Its 38 items are distributed over five scales (with the scale name abbreviation given in parentheses):
1.! Pragmatic-communicative (MDprag-com) – 12 items, reflecting learners’ views of knowledge of English as instrumentally and professionally valuable in the context of international communication. Some items in this scale also include elements of integration, but, as Mihaljević Djigunović (1997) contended, the scale referred to a general type of integration into the international community rather than integration into another linguistic-cultural group.
2.! Affective (MDaffect) – 6 items, referring to the learners’ wishes to learn English because they find the language aesthetically or emotionally appealing (e.g. they like the sound of the language).
3.! Integrative (MDinteg) – 4 items, looking into learners’ wishes to be integrated into an English-speaking linguistic and cultural community.
4.! Teaching setting (MDteach) – 9 items, assessing the potentially negative influence of the following factors: teacher’s behaviour, teaching materials, teaching and assessment methods.
5.! Learning difficulties (MDlearn) – 7 items, measuring the negative influence of bad grades, learners’ pre-existing experience and lack of background knowledge, beliefs about themselves, their wishes and their parents’ wishes.
The statements were followed by a five-point Likert scale: (1) strongly disagree, (2) slightly disagree, (3) neither agree nor disagree, (4) slightly agree, (5)
strongly agree. The questionnaire had been validated in participants' L1 in their learning context.
The motivational dimensions rooted in Dörnyei’s L2 Motivational Self System were measured by a questionnaire based on the Taguchi et al.’s (2009) version of Dörnyei et al.’s (2006) instrument as well as the version validated in the Croatian context (Martinović, 2014). The final questionnaire used in this study, which we will refer to as Dörnyei’s questionnaire, contained 92 items addressing motivational dimensions that were relevant for the profile of participants who participated in the study. This questionnaire was a compilation of different versions of Dörnyei et al.’s instrument, but the original item wording was kept to ensure that the same construct was measured. The items were spread over 16 scales:
1.! Ideal L2 Self (Didself) – 9 items, referring to learners’ ideal pictures of themselves learning L2.
2.! Linguistic self-confidence (Dling) – 3 items, assessing learners’
confidence in their linguistic abilities.
3.! Instrumentality-promotion (Dpromo) – 11 items, measuring the extent to which learners believe learning L2 is necessary for them to meet some personal goals (e.g. getting a good job, making more money) and to become successful in future.
4.! Travel orientation (Dtravel) – 2 items, measuring the extent to which learners wish to learn L2 in order to travel.
5.! Cultural interest (Dculture) – 4 items, exploring learners’ interest in the cultural products of L2 culture (e.g. books, magazines, music).
6.! Interest in the L2 (Dinterest) – 3 items, investigating the affective dimension that learners possess about the L2 (e.g. they like the rhythm of L2).
7.! Attitudes toward L2 Community (Dcommun) – 3 items, assessing learners’ attitudes toward the country and its citizens whose language they are learning.
8.! Integrativeness (Dinteg) – 2 items, examining learners’ attitudes toward native speakers of L2 and their culture.
9.! Ethnocentrism (Dethno) – 2 items, investigating learners’ attitudes toward L2 culture, its values and customs.
10.!Attitudes towards learning English (Dteach) – 5 items, exploring motives related to the learning situation (e.g. atmosphere of L2 classes).
11.!Criterion measure (Dcriter) – 9 items, assessing learners’ intended efforts to learn English.
12.!Ought-to L2 Self (Douself) – 11 items, exploring learners’ beliefs about how necessary learning L2 is for them (as an obligation, duty or responsibility) in order to avoid some negative outcomes.
13.!Instrumentality-prevention (Dprevent) – 6 items, referring to regulation of duties and obligations such as studying L2 in order to get good marks, pass the course or to graduate.
14.!Parental encouragement/Family influence (Dparent) – 8 items, looking into the active or passive role of parents/family members in the process of learning L2.
15.!L2 Anxiety (Danxiety) – 10 items, examining the role of various sources of L2 anxiety (e.g. speaking L2 in class/public, speaking to a native speaker, making mistakes).
16.!Fear of assimilation (Dfear) – 4 items, exploring learners’ beliefs about internalization and its negative influences on them, their values and their native language.
The majority of these scales contained statement-type items followed by a six-point Likert-type scale (from 1- strongly disagree to 6 – strongly agree). The 15 items referring to L2 anxiety and cultural interest were question-type items based on 6-point Likert-type scale ranging from 1 (not at all) to 6 (very much).
Participants were additionally asked to complete several questions pertaining to some demographic characteristics (e.g. age, gender, previous language learning experience, exposure to the target language outside the classroom). Both questionnaires (omitted due to space limitations) are available from the authors on request.
The questionnaires were distributed in Croatian, i.e. participants’ L1, and were administered on two separate occasions during regular FL classes.
Participants were informed about the purpose of the study and were told that their participation was anonymous and voluntary. It took them on average 15 minutes to complete Mihaljević Djigunović’s questionnaire, and about 25 to complete Dörnyei’s questionnaire.
Cronbach’s alpha for internal reliability of the questionnaires was .965 for Dörnyei’s questionnaire, and .826 for Mihaljević Djigunović’s questionnaire.
The following table presents a summary of the number of items and the Cronbach’s alpha coefficients for each scale in the final version of the instrument. As can be seen, all coefficient values indicate that both instruments worked well in terms of their internal consistency. All but one of the scales in both questionnaires reached the .70 threshold, which points to focused scales and acceptable homogeneity. The fact that integrativeness failed to reach the .70 threshold was expected because a similar tendency was noted in previous studies (Dörnyei & Taguchi, 2010, pp. 122-124). To answer the research questions, correlational analyses and multiple regressions were conducted. The statistical analysis was performed using SPSS (Version 20).
Table 1. Cronbach’s alpha for internal reliability of Dörnyei’s and Mihaljević Djigunović’s questionnaire
Dörnyei’s questionnaire
Mihaljević Djigunović’s questionnaire Name of scale Items α Name of scale Items α
Didself 9 .962 MDprag-com 12 .925
Dling 3 .885 MDaffect 6 .929
Dpromo 11 .920 MDinteg 4 .751
Dtravel 2 .869 MDteach 9 .781
Dculture 4 .858 MDlearn 7 .752
Dinterest 3 .754
Dcommun 3 .875
Dinteg 2 .635
Dethno 2 .696
Dteach 5 .879
Dcriter 9 .861
Douself 11 .868
Dprevent 6 .694
Dparent 8 .865
Danxiety 10 .920
Dfear 4 .814
TOTAL 92 .965
38 .826 3.4 Results
In order to investigate the convergent validity of the instruments, the relationship between questionnaire scales was investigated using Pearson product-moment correlation coefficient. The correlations are presented in Table 2. As can be inferred from Table 2, there are very few statistically insignificant correlations. A content analysis of the scales within the two instruments revealed potential similarities in the underlying construct. Indeed, these were confirmed by large correlations (>.5) between MDprag-com and Dpromo (.770), Dtravel (.676) and Dculture (.768) all of which reflect valuing learning a L2 in order to attain a personal goal or to become part of international community. Similarly, large correlations were found between MDaffect and Dinterest (.687), which was expected because these scales refer to learners’
affective perception of the L2. There was a medium positive correlation between MDinteg and Dinteg (.451) (both mirroring learners’ wish to integrate into the L2 community).
Table 2. Correlations among the scales of Dörnyei’s and Mihaljević Djigunović’s questionnaire
MDprag-com
MDaffect MDinteg MDteach MDlearn Didself .758** .715** .515** -.164** -.615**
As for the scales assessing potential negative influences, MDteach was statistically significantly correlated with Dteach .371) as well as Dcriter (-.289). The three subscales all measure learners’ attitudes towards the learning situation, i.e. the teacher, teaching materials or classroom atmosphere. The subscales exploring learning difficulties learners may experience due to anxiety, lack of knowledge or negative influence of the L2 on their L1 or their own values were also statistically significantly correlated: the correlation between MDlearn and Danxiety was .446 and between MDlearn and Dfear was .134.
There are also several low correlations in the matrix which indicate relatively little relationship between the scales. This suggests that these scales may contain distinct motivational dimensions (e.g. Dprevent, Dparent, Douself and Dfear). Interestingly, the Didself scale is highly correlated with all those scales in Mihaljević Djigunović’s questionnaire that contain an aspect of personalisation. On the other hand, it shows low correlations with the scale referring to teacher, teaching materials and methods. A similar pattern can be noticed with the Dling scale. Both Didself and Dling scales contain items associated with learners’ perception of themselves as L2 learners which is obviously a motivational dimension inherent to both L2 motivation measures, the only difference being the extent to which it is explicitly stated in individual items: whereas this dimension is articulated explicitly in Dörnyei’s questionnaire, it is implicitly contained in the items related to pragmatic-communicative and integrative motives in Mihaljević Djigunović’s questionnaire.
To address the second research question, hierarchical multiple regression analyses were performed with course mark as the dependent variable and Dörnyei’s and Mihaljević Djigunović’s scales as predictor variables. Table 3
shows the results of the analysis when Mihaljević Djigunović’s scales were entered in the first block, and Dörnyei’s scales in the second block.
Table 3. Model summary of hierarchical multiple regression 1
Model R R2 Change F Change Sig F Change
1 .515 .265 28.044 .000
2 .607 .103 3.820 .000
The results in Table 3 show that the variables in Block 1 (MDprag-com, MDaffect, MDinteg, MDteach, MDlearn) explained 51.5% of the variance.
After Block 2 variables were entered (Didself, Dling, Dpromo, Dtravel, Dculture, Dinterest, Dcommun, Dinteg, Dethno, Dteach, Dcriter, Douself, Dprevent, Dparent, Danxiety, Dfear), the model as a whole explained 60.7% of the variance. R2 change value was .103 indicating that the variables in Block 2 (i.e. Dörnyei’s scales) explained an additional 10.3% of the variance in the course mark. This is a statistically significant contribution (as indicated by the Sig F Change value).
Table 4 presents the results of the analysis in which Dörnyei’s scales were entered in the first block followed by Mihaljević Djigunović’s scales in the second block. Mihaljević Djigunović’s scales also made a statistically significant contribution to the variance (6.8%).
Table 4. Model summary of hierarchical multiple regression 2
Model R R2 Change F Change Sig F Change
1 .548 .301 10.160 .000
2 .607 .068 7.999 .000