The influence of working memory on reading growth in subgroups of children with reading disabilities

Abstract

This 3-year longitudinal study determined whether (a) subgroups of children with reading disabilities (RD) (children with RD only, children with both reading and arithmetic deficits, and low verbal IQ readers) and skilled readers varied in working memory (WM) and short-term memory (STM) growth and (b) whether growth in an executive system and/or a phonological storage system mediated growth in reading performance. A battery of memory and reading measures was administered to 84 children (11–17 years of age) across three testing waves spaced 1 year apart. The results showed that skilled readers yielded higher WM growth estimates than did the RD groups. No significant differentiation among subgroups of children with RD on growth measures emerged. Hierarchical linear modeling showed that WM (controlled attention), rather than STM (phonological loop), was related to growth in reading comprehension and reading fluency. The results support the notion that deficient growth in the executive component of WM underlies RD.

Introduction

Cognitive impairments are important correlates of functional outcome in children with reading disabilities (RD). Of these impairments, working memory (WM) has been the focus of extensive research efforts because it plays a central role in several domains of cognition, including language comprehension, fluid intelligence, writing, and arithmetic (Gathercole, Alloway, Willis, & Adams, 2006; for a review, see also ; Swanson & Siegel, 2001). Furthermore, WM impairments have been related to specific aspects of RD such as problems in reading comprehension (Swanson, 1999). Thus, there is evidence that WM impairments may play a critical role in mediating some of the academic problems in children with RD (e.g., de Jong, 1998, Gathercole et al., 2006, Swanson and Berninger, 1995, Willcutt et al., 2005). Although WM is integrally related to a number of academic behaviors, relatively few studies have been undertaken to systematically explore whether growth in WM underlies growth in reading performance.

One possible framework to capture growth in WM and its influence on reading is Baddeley, 1986, Baddeley, 1996, Baddeley, 2000 multicomponent model. Baddeley and Logie (1999) described WM as a limited capacity central executive system that interacts with a set of two passive store systems used for temporary storage of different classes of information: the speech-based phonological loop and the visual sketchpad. The phonological loop is responsible for the temporary storage of verbal information. Within the phonological store, items are held for a limited duration and are maintained via the process of articulation. The visual sketchpad is responsible for the storage of visual–spatial information over brief periods and plays a key role in the generation and manipulation of mental images. Both storage systems are in direct contact with the central executive system. The central executive system is considered to be primarily responsible for coordinating activity within the cognitive system, but it also devotes some of its resources to increase the amount of information that can be held in the two subsystems (Baddeley & Logie, 1999). A recent formulation of the model (Baddeley, 2000) also includes a temporary multimodal storage component called the episodic buffer.

In one of the few developmental studies testing Baddeley’s WM model in children with RD, Swanson (2003) compared participants with RD with skilled readers (N = 226) across four age groups (7, 10, 13, and 20 years) WM tasks. As expected, participants with RD were inferior to skilled readers at all age levels, but these differences increased with age. The age-related differences between ability groups were maintained under cuing conditions and when reading and arithmetic skills were partialed from the analysis. Skilled readers showed age-related increases in WM, whereas the trajectory of growth for children with RD showed minimal age-related changes in span level across ages 7–20 years.

These findings raise the question as to whether constraints in WM growth constrain growth in reading. Although the literature is clear that impairments in WM in children with RD are related to reading performance, whether growth in WM underlies poor growth in reading skills has not been tested empirically. Gathercole, Tiffany, Briscoe, and Thorn (2005) conducted one of the closest studies on this issue. Although their research was not focused on RD per se, they initiated a longitudinal study that investigated the cognitive skills and scholastic attainments of 8-year-olds who were selected on the basis of deficits in the phonological loop at 4 years of age. The phonological loop was assessed by short-term memory (STM) tasks that tapped the recall of verbal information. The authors investigated whether deficits in phonological STM performance during this developmental period had a direct impact on children’s attainment in the areas of language, arithmetic, and literacy. They reasoned that if phonological memory was a significant deterrent to learning, children with poor phonological memory should have low overall achievement. In fact, children who scored low on phonological STM measures assessed at 4 and 8 years of age performed at appropriate levels in all areas of vocabulary, language, number skills, and literacy. Gathercole and colleagues concluded that children did not experience learning difficulties in key domains over the early school years that could be attributed to poor phonological memory. However, they found that the relation with WM and literacy was significant (r = 56).

As an extension of this earlier work, our aim was to determine whether poor phonological memory and/or executive functioning may underlie subsequent growth in literacy in older children. It is possible that growth on academic tasks is more sensitive to phonological memory constraints in older children as language abilities approach adult levels. There is also some literature suggesting that tasks related to the phonological loop (i.e., STM) and the executive system are not clearly distinguishable in young children (i.e., high intercorrelation between WM and STM measures [Alloway et al., 2004, Gathercole et al., 2004]); therefore, the phonological system may play a more important role in later academic performance when older children with RD are included in the sample. Furthermore, Engle, Tuholski, Laughlin, and Conway (1999) suggested that STM and WM might not be distinguishable in young children whose rehearsal processes are unstable (i.e., used inconsistently). For example, several studies show weak to moderate correlations between memory span and articulation rates in children 8 years of age or younger (for a review, see Gathercole, 1998, pp. 6–7). This occurs because younger children’s time-based strategies (i.e., rehearsal) are unstable (Gathercole, 1998). Thus, we selected an age level where memory strategies are stable to better assess the processes that moderate the relation among WM, STM, and reading.

Assuming that WM does underlie reading growth, two models were tested in the current study. One is that poor phonological STM constrains growth on literacy measures.¹ Several studies that have compared skilled readers with children with RD assume that STM measures capture a subset of WM performance—the use and/or operation of the phonological loop (for a comprehensive review, see Swanson & Siegel, 2001). This is because successful performance on STM measures draws on two major components of the phonological loop: a speech-based phonological input store and a rehearsal process (Baddeley, 1986). Research to date on STM indicates that children with RD rehearse less and perform more poorly than do skilled readers on tasks requiring the short-term retention of order information (e.g., O’Shaughnessy & Swanson, 1998), suggesting inefficient use of the phonological rehearsal process (cf. Henry & Millar, 1993). Developmental differences in the phonological loop, therefore, might be expected to influence some aspects of growth in literacy (e.g., Leather & Henry, 1994).

Given that the phonological loop is partly controlled by the central executive system (i.e., the executive system shares some variance with the phonological loop), the development of reading may also be directly related to the controlling functions of the central executive system itself. Thus, in contrast to the previous model, the second model views age-related changes in the executive attention component of WM as underlying children’s reading performance. However, the model assumes that there is variance unique to particular subsystems of WM (executive processing and phonological loop) and that both subsystems contribute to children’s reading. Thus, in contrast to the aforementioned hypothesis suggesting that phonological processes play the dominant role in the mediating effects of WM on children’s reading, the second model assumes that executive processes also play a major role in mediating reading performance. There are clear expectations for this second model. The influence of WM on measures of reading follows automatically with improvements in controlled attention. Specifically, if growth in reading is explained by activities related to controlled attention, growth in WM should be correlated with growth in reading.

Because STM and WM tasks are strongly correlated, it may be difficult to determine whether reading deficits and poor WM performance are due primarily to a system related to storage (i.e., STM) or to a system that taps both storage and executive processing (i.e., WM). One means to untangle these possible sources of difficulties in reading is to compare subgroups of children with RD. Evidence for the contribution of different memory systems to reading deficits is suggested when different types of reading problems exhibit different profiles. Some studies have suggested that children who have combined deficits in reading and arithmetic (referred to as the comorbid group in the current study) reflect more generalized deficits related to the executive system than do children with RD only, that is, children who have deficits related only to reading (e.g., van der Sluis, van der Leij, & de Jong, 2005). Van der Sluis and colleagues (2005) suggested that children with RD do not exhibit problems in executive function, whereas children who show reading plus arithmetic difficulties exhibit problems in executive processing. Thus, of particular interest in the current study was to examine whether performance of children identified with only RD reflects deficits in the development of phonological STM, whereas performance of children with comorbid deficits in both reading and arithmetic reflects developmental deficits in the executive system.

The rationale behind these hypotheses is as follows. There is tremendous consensus in the literature that poor phonological coding is related to poor word recognition (e.g., Shankweiler et al., 1995; for reviews, see also Stanovich, 1990, Stanovich and Siegel, 1994); thus, a significant relation between STM and word recognition would be expected in the current study. Furthermore, the relation between WM and reading is well established (Daneman and Merikle, 1996, Engle et al., 1992). What is unclear from the literature, however, is whether specific deficits in the phonological storage and/or controlled attention underlie differences between low reading groups on WM measures. It seemed to us that if the WM system plays its major role in more generalized academic problems (i.e., below average word recognition and arithmetic), children with comorbid deficits would experience memory problems related to the executive system. In contrast, children with specific deficits in reading only would be impaired on phonological STM (storage) tasks.

The current study was designed to test whether growth in WM underlies growth in reading measures in subgroups of children with RD. The study addressed two questions:

• Do different growth patterns in WM emerge in subgroups of children with RD and skilled readers?

• Does an executive system (controlled attention) and/or a phonological storage system account for growth in reading performance?

Based on cross-sectional studies, we predicted that growth in WM underlies reading growth in children with RD only and children with more general academic deficits. We tested whether growth is stable (no change) or linear across age. We also tested whether more generalized WM deficits (controlled attention) and storage deficits underlie memory growth in the comorbid group when compared with children with RD only. Children with RD only and children with comorbid deficits were also compared with children who had comparable reading deficits but were low in verbal intelligence. Because more generalized deficits in achievement (problems in both reading and arithmetic) may be associated with verbal IQ, it was necessary to determine whether differences in WM growth are merely an artifact of verbal ability. Thus, also included in the comparison were children with low reading and low verbal IQ scores (referred to in this study as low verbal IQ readers). Nation, Adams, Bowyer-Crane, and Snowling (1999) suggested that the relation between WM and reading (i.e., comprehension) is explained by verbal intelligence and not by executive processing skills. They argued that low verbal IQ readers have a weakness in verbal skills that restricts their ability to store verbal information in STM. Likewise, Stothard and Hulme (1992) suggested that WM differences between good and poor readers (comprehenders) would be eliminated if differences in verbal IQ were controlled; however, no evidence was provided to support this prediction.