Discrete point testing is related to structuralist linguistics. Now considered out of date, in these tests, every test item is designed to test a discrete point of language. This is based on the theory that language abilities are divided into discrete categories not connected to each other. It was believed that these skills could be learned by repeated practice of these individual pieces, such as learning patterns and memorizing vocabulary as taught in the Audio-Lingual method. (Bachman and Cohen 4). In discrete point testing it is important to make sure that each test item tests only one skill. If more than one ability were tested, then the test item was said to be confounded. One very limited use of discrete point testing is as a preliminary step before an integrative test. For example, I often give a discrete point test on a specific grammar point a week or so before an integrative in order to motivate students to learn a specific structure before I ask them to apply that structure on a more communicative exam.
             Integrative testing is related to the homogeneous competence model and transformational or generative linguistics. This theory claims that linguistic competence is one competence, but is distinct from non-linguistic skills, such as pragmatics (Bachman and Cohen 5). By not separating the language skills, one seeks to test all language skills together. The homogeneous competence model developed around the same time as the heterogeneous competence model, which states that pragmatics are tied to language use. This led to the belief that communicative performance, rather than competence, should be tested, rather than isolated skills. An example of an integrative test is an essay-writing exam where students are tested on syntax, lexicon, semantics, and even pragmatics.
             Stating that a test has construct validity means that the test scores accurately reflect the abilities of the test takers. Different theories of...