ABSTRACT
The assessment process used in identifying impairments is one of forming and testing out hypotheses; the focus of these hypotheses is on testing out the relative intactness of specific components of the model. In assessing the different levels of breakdown using this model, use is usually made of three kinds of evidence. A first source of evidence is the effect of different variables (such as word length, imageability and so on) on performance. This is what Shallice (1988) describes as the ‘critical variable approach’ that ‘seeks to establish the variables that affect the probability that a task will be correctly performed’ (Nickels & Howard, 1995a, p. 1281) by a client. A second source of evidence is the nature of the errors made in different tasks. Where the tasks involve written and spoken word production, errors are made overtly, and can be classified. In comprehension tasks, the nature of possible errors is constrained by the design of the task. For example, in spoken word-to-picture matching, errors in word recognition at a lexical or pre-lexical level that result in misrecognition of a word as another phonologically related word will not be detected when the stimuli use only semantically related distractors. They would, on the other hand, be detected in word-to-picture matching using phonologically related distractors, or in a spoken word definition task.
