A discussion of language testing looks at the relationship between the processes of language learning and language testing, particularly from the point of view of pragmatics theory. It outlines some of the theory of Charles Sanders Pierce and its role in the evolution of linguistic theory, as well as the work of other theorists concerning the nature of knowledge and the role of experience in learning. The paper distinguishes between three perspectives in the testing process (those of the text, author, and audience) and examines the tension between them during language testing. It is argued that the three perspectives must be in appropriate correspondence to each other. Research focusing on the three perspectives as they relate to cloze testing is considered; different forms of tests are viewed as focusing on different perspectives. It is concluded that for language testing research and development to be optimally interpretable, the researchers must take care to control the variables of whichever two perspectives are not the focus in the test in question. A brief bibliography is included. (MSE)