Many studies that examined working memory impairments in children and adolescents with intellectual disabilities (ID) drew on Baddeley's multicomponent model (Baddeley, 1986 and Baddeley, 2007). In this model a central executive provides the attentional control of the working memory system and focuses, switches and divides the limited attentional resources. The phonological loop is concerned with storing and maintaining sound and speech-based information, while the visuo–spatial sketchpad is engaged in storing and maintaining visual and spatial material. The episodic buffer was added as a fourth component; it provides storage of information in a multimodal code and is capable of binding information from the other storage components and from long-term memory (Baddeley, 2000). Various studies comparing children with and without ID of the same chronological age have consistently found deficits in all four components that increased with the severity of the intellectual disability (Henry, 2001, Henry, 2010, Maehler and Schuchardt, 2009, Schuchardt et al., 2010 and van der Molen et al., 2009). In the tradition of the developmental–difference controversy (Zigler & Balla, 1982), research additionally focussed on the question whether these deficits in working memory do reflect a developmental lag and children with ID have the same cognitive structures as younger children matched for mental age (MA), or whether performance differences with MA matched children point to a structural difference or a qualitative deviation from typical development.
Because various studies reported that children with ID of nonspecific aetiology performed more poorly than MA children on some or on all phonological short-term memory (PSTM) tasks (Bayliss et al., 2005, Henry and MacLean, 2002, Henry and Winfield, 2010, Hulme and Mackenzie, 1992, Russell et al., 1996, Schuchardt et al., 2010, Schuchardt et al., 2011 and van der Molen et al., 2007), it is likely that there is a structural difference in the development of the phonological loop.
According to Baddeley, 1986 and Baddeley, 2007 model, the phonological loop comprises a phonological store and a subvocal articulatory rehearsal mechanism. To counteract the rapid decay of verbal memory traces in the phonological store, traces can be refreshed by rehearsing the items through covert or silent articulation and then be fed back into the phonological store. The most commonly suggested explanation for the aforementioned difficulties with PSTM tasks is that subvocal rehearsal is impaired in individuals with ID. Studies addressing this hypothesis used the word length paradigm to examine whether a word length effect (WLE) with superior recall of short compared to long words occurred in groups of individuals with ID. Some studies found a significant word length × group interaction with no WLE for the group with ID, but a significant WLE as rehearsal indicator for the MA control group (Hasselhorn and Mähler, 2007, Hulme and Mackenzie, 1992 and Rosenquist et al., 2003). Additionally, articulation speed correlated with word span in the samples of MA matched children but not in the samples of individuals with ID. Therefore, these studies suggested that subvocal rehearsal is impaired in individuals with ID and not developed according to their mental age. However, other studies (Schuchardt et al., 2011 and van der Molen et al., 2007) reported no word length × group interaction and found significant WLEs for children and adolescents with ID as well as for children of the MA control group indicating equal reliance on subvocal rehearsal.
The interpretation of these inconsistent results is further complicated by the fact that a WLE as well as a relationship between articulation speed and memory span not only could have been caused by subvocal rehearsal but also by output delays throughout responding. During full serial recall of lists of long words, more time elapses until the words in later serial positions can be said than during recall of lists of short words. Because of the longer output delay, there is more time for the memory traces to decay in lists of long words. This notion that verbal output delays produce WLEs is supported by several findings: the WLE grew larger across serial positions in verbal serial recall (Cowan et al., 1992; experiment 1); in mixed lists containing short words followed by long words or vice versa, recall for long words in the later serial position was superior to recall for short words which were preceded by long words allowing for more decay (Cowan et al., 1992; experiments 2 and 3). In verbal serial recall and to a lesser extent in probed recall (i.e. a memory task type in which not all words have to be reproduced, but one of the to-be-remembered words is probed per trial, e.g. it is required to name the first or third word when the first or third probe is presented) a WLE was also present under articulatory suppression when participants were required to rapidly say words like “the” to prevent subvocal rehearsal (Avons, Wright, & Pammer, 1994; experiments 1 and 2). Even though these experiments stress the role of output in producing a WLE, it is unlikely that output delays are the only source of the WLE because reduced but reliable WLEs were also found in experiments with serial recognition tasks that eliminated the output differences of verbal serial recall (Baddeley, Chincotta, Stafford, & Turk, 2002).
Most studies examining the WLE in children and adolescents with ID only tested word spans with verbal serial recall tasks so that it is not possible to decide whether the WLE found in the control group and sometimes in the experimental group indicated rehearsal or was only caused during output. Two studies were exceptional in this regard (Jarrold et al., 2000 and Russell et al., 1996), focusing on the PSTM of children with autism or children with Down syndrome and including control groups of children with ID of nonspecific aetiology. Russell et al. (1996) reported inconclusive results on the question whether children with ID differ in rehearsal from typically developing children with a mean mental age of 6 years. While the WLE did not differ significantly between groups, the correlation between articulation speed and memory span was only significant in the MA control group, but not in the group with ID. In their samples with mental ages of about four to five years, Jarrold et al. (2000) found WLEs only in a verbal serial recall task but not in a probed recall task. They therefore concluded that neither the individuals with ID nor the typically developing children engaged in spontaneous subvocal rehearsal. Likewise, experiments by Henry (1991) showed that in typically developing 5-year-olds, there was no WLE in probed recall tasks with no full verbal output requirements and suggested that the WLE in verbal serial recall tasks was caused by verbal output only. In 7-year-olds, however, a WLE could be found with probed recall tasks indicating that subvocal rehearsal starts to develop between the ages of 5 and 7 years.
Because there is a lack of studies that examine the WLE in children with ID focusing on a mental age of 6–8 years, and that take effects of verbal output on the WLE into consideration, we planned such a study. Since some previous studies reported a WLE for children with ID and a mental age of roughly seven years but others did not (whilst for the control groups the WLE was consistently found), we suppose that the WLE in verbal serial recall is smaller in children with ID and only caused by output delays, whereas output delays and subvocal rehearsal both contribute to the larger WLE in typically developing children matched for mental age.
If this suggestion is correct, we expect to find the following pattern of results when examining the WLE with sensitive verbal serial recall tasks: (a) In the group of children with ID a significant WLE should be present that is significantly smaller than the WLE in the MA control group aged 6–8 years. (b) In children with ID, the WLE should become evident only in later serial positions due to output delays but not in the first word of each trial. (c) In contrast, in children from the MA control group, the WLE should already be evident in the first serial position as rehearsal activities before output should contribute to the effect; additionally the WLE should increase during output from word position to word position. (d) In children with ID, experimentally minimizing the normally differing output delays between short and long words should eliminate the WLE. (e) In contrast, in the MA control group, such a technique should only reduce but not abolish the WLE.
