The measurement of stability in aphasia recovery: implications for language modelling

Abstract

Background: Performance stability is an implicit assumption within theoretical explanations of aphasia. The assumption being that when completing language processing tasks, performance will be stable from moment to moment and day to day. Theoretically, aphasia is most commonly viewed within a modular framework. that is, language processing is carried out by specific, specialised language processing modules. Aphasia is thought to result when one of these modules is dammed leading to a unique pattern of performance results. Implicit to this view of aphasia is stability, once damaged, the module will no longer be accessed and any process using the module will be impaired. This theory of aphasia is widely held within both research and clinical communities and underlies many of our approaches to the assessment and treatment of aphasia. However more recently researchers have been expressing doubts about the validity of assuming stability in aphasia performance. Instead variability in performance is being reported and alternative explanations of aphasia are being provided. One of these considers aphasia to result from a reduction in or the inefficient allocation of cognitive resources. Aims: This research explored variability in aphasic performance, with the aim to examine variability over a range of tasks and time periods. Methods and Procedures: Eight aphasic and ten non-brain damaged individuals participated in eight sessions. Within these sessions they completed a spontaneous language task; which contained four different narrative genres. a lexical decision task and a simple reaction task.\Performance on these tasks was examined for three different levels of variability; inter session variability (across session means for time measures), intra session variability (across items for time measures) and inter session variability (item to item accuracy for lexical decision). The three different levels of variability examined performance on the same task across days and within the same task on the same day. To determine whether the change in the performance of aphasic individuals was in the same range and followed the same pattern of change and variation demonstrated by the non-brain damaged participants, the pooled results of the non-bruin damaged individuals' performance developed a 'normal' range of performance. Using the group's data the results of each of the aphasic individual was then converted to a z- score. Outcomes and results: The results demonstrate that for all aphasic individuals, across the three tasks and three time periods examined, variability is a regular, if not universal feature of aphasia. Conclusions: Stability in aphasic performance cannot be assumed. Instead research and clinical approaches must establish stability or consider the impact of variability before conclusions about performance can be drawn. The presence of variability also calls into questions the traditionally held view that aphasia results from the selective impairment of specialised language processing modules. Instead an alternative mechanism for impairment must be considered. The resource allocation view of aphasia was explored and found to explain the performance of aphasic individuals within this study.