Capacidad de la memoria de trabajo verbal: Validez y fiabilidad de una tarea de amplitud de lectura / Verbal working memory capacity: Reliability and validity of the reading span task

RESUMEN

La memoria de trabajo se refiere a todos aquellos mecanismos o procesos implicados en el control, regulación y mantenimiento activo de información relevante para la ejecución de tareas cognitivas complejas (Miyake & Shah, 1999); es uno de los principales factores limitantes de las capacidades de alto nivel jerárquico, como el razonamiento o la comprensión lectora. En este trabajo se informa un estudio en el que se analizaron las propiedades psicométricas de una de las tareas más conocidas para la evaluación de la capacidad de la memoria de trabajo, la Tarea de Amplitud de Lectura de Daneman y Carpenter (1980). La tarea consiste en leer un conjunto de oraciones no relacionadas entre sí, mientras se retienen sus palabras finales, y determina el máximo número de palabras finales que un participante puede recordar, que se identifica con su amplitud o span. La prueba se adaptó al español rioplatense partiendo de versiones españolas (Elosúa, Gutiérrez, García Madruga, Luque & Gárate, 1996; Gutiérrez, Jiménez & Castillo, 1996) y se administró en sesión individual a 132 voluntarios, conjuntamente con pruebas estandarizadas de memoria de corto plazo y de trabajo, tales como Amplitud de Dígitos y Amplitud Viso-Espacial de la Wechsler Memory Scale - Revised (WMS - R, Wechsler & Stone, 1987) y Ordenamiento Dígito-Letra de la Wechsler Adult Intelligence Scale III (WAIS III - Wechsler, 1987). La tarea de Amplitud de Lectura mostró una alta fiabilidad. Con respecto a su validez, se obtuvieron correlaciones significativas con tareas de memoria verbal, pero no así con tareas de memoria viso-espacial.

ABSTRACT

In the study of human behavior, working memory has been seen as a central construct (Conway, Kane, Bunting, Hambrick, Wilhelm, & Engle, 2005) since the cognitive revolution took place. Working memory refers to those mechanisms or processes that are involved in the control, regulation, and active maintenance of task-relevant information in the service of complex cognition (Miyake & Shah, 1999), and is considered one of the main limiting factors for complex abilities such as reasoning or reading comprehension. Methodologically, the Reading Span Task (Daneman & Carpenter, 1980) has proven to be both a reliable and valid measure of working memory capacity, and is a widely used measure. The purpose of this study was to develop a local adaptation of the Reading Span Task, based on two previous Spanish versions (Gutiérrez, Jiménez, & Castillo, 1996; Elo súa, Gutiérrez, García Madruga, Luque, & Gárate, 1996), whose psychometric properties were not known. In both cases, the Reading Span capacity was significantly correlated with measures of reading comprehension. However, the psychometric properties of the Reading Span Task were not analyzed. The task requires participants to read a set of unrelated sentences, each of which is between 12 and 14 words long, one sentence at a time, at their own pace. Sentences are arranged in 4 sets of 2, 3, 4, and 5 sentences. There are three different trials at each set-size (number of sentences) level. After having read all the sentences in each set, participants try to recall the final word of each sentence. Working memory capacity is both the maximum number of sentences for which the participant can recall the final word (Span) and the total number of sentence-final words recalled (Conway et al., 2005). The task was administered in one session to 132 undergraduates at the University of Buenos Aires (UBA), along with other standardized short-term and working memory tests, Digit Span and Visuo-Spatial Span from the Wechsler Memory Scale - Revised (Wechsler & Stone, 1987), and Letter- Number Sequencing from the Wechsler Adult Intelligence Scale III (Wechsler, 2003). The Reading Span Task showed high reliability (α = .95). The percentile analysis shows that those participants that obtained a Span of 2 or recalled 6 words fall in the 25th percentile, while those that obtained scores above a Span of 3.5 and recalled more than 24 words fall above the 75th percentile. The correlation analysis shows that the Reading Span Task has positive and significant correlations with verbal memory tasks (measured by Span, Forward Digit Span r = .408; p < .01; Backward Digit Span r = .502; p < .01; Letter-Number Sequencing r = .504; p < .01; measure by Number of sentencefinal words recalled, Forward Digit Span r = .416; p < .01; Backward Digit Span r = .496; p < .01; Letter-Number Sequencing r = .489; p < .01) , but not with visuo-spatial memory tasks (measure by Span, Forward Visuo-Spatial Span r = .169; p = n.s.; Backward Visuo-Spatial Span r = .018; p = n.s.; measure by Number of sentence-final words recalled, Forward Visuo-Spatial Span r = .189; p < .05; Backward Viso- Spatial Span r = .032; p = n.s.). A factor analysis showed the separation of the two factors involved in the tasks a verbal factor whose capacity is affected by the Reading Span Task, Digit Span and Letter - Number Sequencing tasks, and a visuo-spatial factor whose capacity is affected by the visuo-spatial amplitude tasks. Results indicate that the Reading Span Task can be considered an appropriate, valid and reliable measure of the capacity of verbal working memory, useful for doing research on cognitive processes and for neuropsychological assessment.