The Reliability and Validity of Short-term and Working Memory Pointing Tasks Developed for Clinical Populations with Speech and Language Disorders

Background & Objectives: Short-term and working memory capacity theories have gained considerable attention as underlying cognitive mechanisms, deficiencies in which may account for language processing difficulties in individuals with speech and language disorders. However, there are limitations to the employment of short-term and working memory tasks developed for assessing normal cognitive processing for the evaluation of speech and/or language-impaired clinical populations. The current study developed pointing span measures that do not require verbal output and thus can be used to assess short-term and working memory capacities in special populations with verbal-output deficits. The specific aim of the study was to examine psychometric properties such as test-retest reliability, concurrent and construct validity, and internal item consistency in newly developed pointing span measures. Methods: Digit forward and backward pointing span measures were developed by adapting the most commonly used digit forward/backward recall tasks to pointing measures. An alphabet pointing span task and a subtract-2 pointing span measure were developed as working memory tasks. Fifty-five normal individuals participated in the study and performed all of the tasks. Results: The test-retest reliability results revealed that the range of Pearson's correlation coefficients was 0.43 similar to 0.77. However, test-retest reliabilities increased up to 0.86 when a composite measure was used with a combination of at least two different tasks. Concurrent validity results suggested that the pointing version of the digit span measures shared only 25 similar to 40% of the total variance with the standardized measure of the digit span-recall. As short-term memory measures, digit and word forward pointing span measures were highly correlated (r=0.79). A principal component analysis revealed that the one-factor solution accounted for 66% of the total variance for four different working memory pointing tasks. Discussion & Conclusions: The current results were consistent with those from Waters and Caplan (2003), reporting composite measure increased stabilities and reliabilities for the working memory measures and all of the working memory tasks loaded onto a single factor. Further studies are required to examine the effects of age and output-modality on these tasks.