Multidimensionality and Hierarchical Structure of the English Vocabulary Learning Attitude Scale

Article excerpt

The last three decades have seen researchers recognize and show growing concern about the importance of attitude when measuring the success of foreign language learning (Bartley, 1970; Briem, 1974; Chiachiere, 1993; Clement, Dornyei, & Noels, 1994; Corbin & Chiachiere, 1995; Gardner, 1985; Masgoret & Gardner, 2003). In numerous studies it has been shown that attitude toward foreign language learning is positively associated with foreign language proficiency and achievement (Coakley, 1985; Martinez, Aricak, & Jewell, 2008; Raymond & Roberts, 1983; Shaver, 1990). Raymond and Roberts (1983) found a moderate but positive relationship (r = .25) between foreign language attitude and college foreign language grades. Likewise, stronger associations were found for specific language skills such as reading (r = .43) and listening (r = .46; Raymond & Roberts). In a regression study, Martinez et al. (2008) also demonstrated that reading attitude significantly predicted reading achievement (3 = 0.22). Indeed, the aforementioned researchers have suggested that the role of attitude should not be neglected in the process of foreign language learning. Bartley (1970), in discussions of the importance of attitude, argued that "attitude toward learning is probably the most important factor in academic success" (p. 383).

Hence, a number of attempts have been made to develop reliable and valid instruments to measure learners' attitudes toward foreign language learning (Briem, 1974; Corbin & Chiachiere, 1995; Raymond & Roberts, 1983; Shaver, 1990). These early pioneering researchers, arguably, have contributed valuable insights into the way in which a language learning attitude scale should be established (e.g., Corbin & Chiachiere, 1995; Raymond & Roberts, 1983). For instance, Corbin and Chiachiere (1995) provided evidence for factorial validity of scores on Chiachiere's (1993) foreign language learning scale through exploratory factor analysis technique. Likewise, Raymond and Roberts demonstrated both discriminant and predictive validity of the intended scale through correlation and regression analyses. These early studies are, however, exploratory in nature and lack a priori theory to support the construct validity of the related attitude scales.

Most contemporary social psychologists tend to agree that attitude can be defined as one's overall evaluation of a stimulus object, albeit with some heterogeneity on the formal definition (Ajzen, 2005; Eagly & Chaiken, 1993). Attitude as reflected by its evaluative nature has often been theorized as either a multidimensional construct, consisting of cognition, affect, and conation (Ajzen, 2005; Bagozzi, 1978; Breckler, 1984; Eagly & Chaiken, 1993; Kothandapani, 1971; Ostrom, 1969; cf. Bagozzi & Burnkrant, 1979, 1985) or simply a onedimensional construct (Dillon & Kumar, 1985). A unidimensional perspective of attitude posits that there is no need to differentiate between any subtraits on the basis of evaluation of a stimulus object. By contrast, a multidimensional perspective of attitude suggests that a tripartite distinction can "help researchers evolve an understanding of the conditions under which attitudes truly have varying numbers of components" (Eagly & Chaiken, 1993, p. 14). According to Ajzen (2005), the cognitive component of attitude reflects beliefs of or thoughts about the attitude object; the affective component of attitude refers to feelings or emotions associated with a stimulus object; conation consists of behavioral inclinations, intentions, or commitments. On the prediction of the occurrence of behavior, researchers have found that the three components may have different effects. For instance, Bagozzi and Burnkrant (1979) found that the predictive power of the affective component ([beta] = .65) on religious behavior was approximately three times as forceful as the cognitive component ([beta] = .23). In other words, the occurrence of the religious behavior can be explained by the cognitive component for only 5% ([[beta]. …