In this study, the authors examined the hierarchical and multidimensional nature of English self-concept. University students (N = 321) responded to survey items on listening, speaking, reading, and writing self-concepts and a global English self-concept adapted from H. W. Marsh's (1990Marsh's ( , 1992 Academic Self-Description Questionnaire. A confirmatory factor analysis (CFA) found 4 distinct English skill-specific self-concept constructs, demonstrating the multidimensional nature of self-concepts in different skill areas. Hierarchical CFA found that self-concepts of the 4 English skills can be represented by either a higher order English self-concept factor or by a global English Self-concept factor. The correlation between the higher order and global English factors was .97, indicating that they cannot be distinguished as two separate constructs. Within a specific subject domain such as English, academic self-concept can be both hierarchical and multidimensional.In the present study we attempted to examine the multifaceted and hierarchical nature of self-concept within the specific subject domain of English. There have been two major studies of the structure of self-concept (Marsh & Shavelson, 1985;Shavelson, Hubner, & Stanton, 1976). Shavelson et al. proposed a self-concept model that could empirically be scrutinized. According to their model, there is a general self posited at the apex of the self-concept hierarchy, which is divided into academic and nonacademic facets. Under the nonacademic facet are social, physical, and emotional self-concepts in contrast to the academic facet with self-concepts for various academic domains such as English and math. There are self-concepts further down the hierarchy that are even more specific. Since the advent of this model, research on self-concept has shifted to an emphasis on construct validation and its delineation of the hierarchical and multidimensional nature of the self. Marsh and Shavelson (1985) further scrutinized the self-concept model and found that the self-concept structure was, in reality, more complicated than the one originally proposed in the Shavelson et al. (1976) model. Specifically, they found that the multifacets described in the original model were so distinct and diverse by late adolescence that the hierarchy was necessarily weak. Marsh and Hocevar (1985) then demonstrated that students' verbal and mathematical self-concepts are typically so distinct (with a near-zero correlation) that the two constructs could not be incorporated into a general academic self-concept. To ex-

No abstract