The Working Memory Token Test (WMTT): Preliminary Findings in Young Adults With and Without Dyslexia Shimon Sapir, Ph.D. and Ravit Cohen Mimran, Ph.D. Department of Communication Disorders Faculty of Social Welfare & Health Studies, University of Haifa, Israel sapir@research.haifa.ac.il Abstract The purpose of the study was to assess the validity of a new version of the Token Test (TT) which is aimed at detecting deficits in auditory verbal working memory (henceforth WMTT). Twenty eight young healthy adults, 14 with and 14 without dyslexia, all paid volunteers, took part in the study. The WMTT was found to be positively and significantly correlated with tests of digit span forward and backward, pseudoword reading performance, but not with real word reading, performance IQ, and the d-2 attention measure. The WMTT also appeared to be a more sensitive method than the original TT to detect differences in auditory verbal working memory between the two subject groups. These main findings, albeit preliminary, indicate that the WMTT is superior to the original Token Test (memory parts) in assessing auditory verbal working memory deficits. 1. Introduction The Token Test (TT) [1] has been used extensively in research and clinical practice to screen individuals with deficits in auditory verbal comprehension and short term auditory verbal memory. According to the model of working memory proposed by Baddeley [cf. 2] verbal working memory depends on two sub-systems: a temporal verbal acoustic storage system, also called the phonological loop (PL), and a working memory capacity (WMC). The PL is assumed to be necessary for the immediate retention of sequences of strings of phonemes (words and nonwords). The WMC is assumed to be responsible for attention control of working memory (also called the central executive) where there is a need to combine simultaneous processing and storage [3]. We have revised and computerized the memory part of the TT, so that it can also help detect deficits in WMC in addition to PL deficits. We call this revision the Working Memory TT, or WMTT, which includes items that test both auditory verbal memory span and auditory working memory. 1.1. The WMTT The original TT tested auditory verbal memory by having the subject point to tokens of 5 different colors, 2 shapes (circle, squares), and 2 sizes (small, large), in response to commands such as "touch the blue circle" (a one dimensional command, or 1-D, since the tokens differ only by color), or "touch the yellow square and the red circle" (a two dimensional command, or 2-D, since the tokens differ by color and/or shape), or "touch the small red circle and the large green square" (a three dimensional command, or 3-D, since the tokens differ by size, and/or color and/or shape). Such commands require that the subject stores the words in memory in the same order that they appear in the command. To include commands that test working memory, we used the words "before" and "after" at the beginning or between two parts of the command, so that the order in of the instructions within a command did not correspond to the order in which these instructions were to be performed by the subjects. For example, the command "before you touch the red circle touch the green square" or the command "touch the red circle after you touch the green square" both require that the subject first perform the second instruction and then the first instruction in the command. We call these types of commands non-serial, and their performance requires manipulation of the words stored in short term memory, as well as some form of inhibition and vigilance to ensure that the appropriate order of performance is achieved in spite of the difference in order of instructions. Thus, the performance of these commands requires use of executive functions, which are most characteristic of working memory [2]. In contrast, when the words "before" and "after" are inserted in the commands such that the execution of these commands is in the same order as the instructions (i.e. serial commands), the task is less likely to involve working memory, and most likely to involve auditory memory span. 1.2. Working memory and reading There is considerable evidence that individuals with reading disabilities (RD) have an impaired working memory ability for verbal information [4, 5]. However, researchers do not agree where the working memory deficits of individuals with RD originate with respect to the model of working memory proposed by Baddeley and his colleagues. In particular, it is not clear whether the memory problems of individuals with RD include mainly deficits in temporary storage of verbal information [6,7] or are realized mainly when there is a need to activate the central executive [5]. The present study addresses this controversy by assessing the verbal working memory model as a whole rather than focusing on particular aspects of it, as has been done by much of the research to date. The primary purpose of this investigation was to assess the validity and sensitivity of the WMTT, by correlating it with the forward and backward digit span tests (FDST and BDST, respectively), and by correlating it with pseudoword reading (PWR) (which has been shown to be strongly correlated with auditory working memory) and real word reading (RWR). The inclusion of the dyslexics in the study is for two main purposes: to form part of a continuity of the level of performance on memory tests, and to assess the extent to which the WMTT can contribute to the differentiation between those individuals with DD and phonologic processing 10.21437/Interspeech.2005-431