Discusses the three main approaches to defining language knowledge and use: (1) the structural approach, (2) the functional approach, and (3) the general proficiency approach. Asserts that each approach has specific consequences for language testing and that no one approach can claim to be the only way of representing that knowledge. (Author/SED)

Presents the theory of general language proficiency and looks at the construct validation of cloze tests and C-tests. Describes the defects of classical cloze procedures. Gives an example of the C-Test and discusses its empirical validity. Concludes that C-tests are authentic tests of the construct of general language proficiency.