Kate Cain, Lancaster University, UK
Introduction to working memory and reading comprehension
Working memory tasks assess the ability to store and process information simultaneously and to retrieve for later recall. There is a strong relationship between children’s working
memory capacity and their ability to understand text: working memory deficits are reliably found in children who do poorly on assessments of reading comprehension and a child’s performance on a working memory task is a good indicator of their reading comprehension level, even after factors such as their vocabulary knowledge and word reading ability have been taken into account. In this talk, I will review the tasks used to assess working memory, and discuss how performance on these tasks is related to text comprehension in general, and key component skills of text comprehension.
Alix Seigneuric? and Marie-France Ehrlich??
? CESG, CNRS, Université de Bourgogne, Dijon, France
?? EPHE, University of Paris 5, France
Working memory capacity and reading comprehension in children:
Constraints on pronoun resolution.
The aim of our studies was to investigate the relationship between working memory capacity (WMC) and reading comprehension in children. WMC was described as a pool of limited resources that carry out processing and storage functions. It was assessed by a listening span test. The present talk is focused on the influence of WMC on the processing of anaphoric pronouns which ensure the cohesion of successive sentences. This processing plays a central role in the building of a coherent mental representation of the text meaning. It is assumed to be constrained by WMC.
The study was conducted on 48 fourth-grade children (9years 9months). In a first part, children were administered several tests assessing reading comprehension, WMC, and also vocabulary and decoding skills. In a second part, children were tested on the processing of pronouns by using a self-paced reading task followed by a comprehension question. Processing demands underlying the computation of the pronoun antecedent was varied by means of two factors: The availability of a gender cue on the pronoun and the distance between the pronoun and its antecedent. Three dependent variables were analyzed: Reading time of the sentence containing the pronoun, response time and response accuracy to the comprehension question.
The main results were the following:
1- WMC explained unique variance in reading comprehension when vocabulary and
decoding were controlled for. WMC also made a significant contribution to pronoun
processing performance (response time and response accuracy as criterion
2- In order to characterize how WMC affected pronoun resolution, two groups of 16
children classified as high span and low span were compared. Low span children
experienced more difficulties than high span children in computing the antecedent:
Response accuracy was lower and response time was longer. The two groups were
differently affected by the increasing of processing demands. High span children
spent longer reading the sentence containing the pronoun when pronoun could not
solved on the basis of the gender alone, suggesting that the pronoun was
processed as it was read. Low span children tended to delay processing until it was
required by the comprehension question. Concerning the distance factor, a negative
effect on response time was observed in low span, but not in high span children.
These interaction effects supported the hypothesis that WMC constrains pronoun
resolution in children.
Paola Palladino, University of Pavia, Italy
Poor comprehenders: are they overloaded by irrelevant information?
Italian is a transparent language that, compared to other languages such as English or French, has an almost perfect correspondence between sounds and symbols. Therefore learning to read Italian, after an initial phase, could result in partly independent skills, decoding and comprehension. In fact about 5-7% of children who can read at a normal speed and correctness for their age are not able to adequately understand text meaning (poor comprehenders-cattivi lettori). Their comprehension difficulty seems to be related to poor verbal working memory usually measured with complex double tasks in which only few, of many items processed, have to be recalled. Consistent results show that poor comprehenders’ failure in a working memory task is associated to a high number of intrusion errors, no longer relevant information processed in doing the memory task and wrongly recalled as target items (e.g. De Beni, Palladino, Pazzaglia & Cornoldi, 1998; Carretti, Cornoldi, De Beni & Palladino, 2004, Palladino, Cornoldi, De Beni & Pazzaglia, 2001). Poor comprehenders’ working memory performance seems to be damaged by a less efficient inhibition mechanism that fails in reducing activation of irrelevant information when entered in working memory. Recent data show that poor comprehenders keep active in working memory irrelevant information so that it would overload memory capacity reducing its efficiency.
Dr John Towse, Lancaster University, UK
Working memory capacity: Size isn’t everything!
Working memory involves the dynamic maintenance and transformation of labile information. Conventionally it is measured in terms of the amount of information that can successfully be recalled (the volume of working memory). Such convention measures of working memory are important predictors (among children and adults) of cognitive skill in many domains, including reading comprehension. Nonetheless, focusing only on the volume of information does not capture all of the richness of working memory, nor does it encourage a theoretical focus on all aspects of performance. Drawing on data from children and adults, I will argue that alternative measures can yield new insights into accounts of working memory and reading comprehension.
Dr Alice Spooner, University of Central Lancashire
Assessing comprehension with recognition
According to dual-process theories of memory recognition can be made on the basis of recollection or on the basis of familiarity alone. It appears that remember/familiar judgements in a sentence recognition test could be used to assess comprehension. Evidence suggests that people who tend to be more involved in rich, integrative text processing, resulting in superior comprehension, tend to make higher proportions of recognition judgements that are accompanied by an experience of recollection. In contrast,
people whose comprehension processes tend not to extend beyond text-base processing tend to make higher proportions of familiarity judgements. It therefore appears that sentence recognition tests that use a remember/familiar distinction could provide a reflection of reading comprehension ability. Such tests are easy to prepare, administer and score, making them potentially very useful in both research and educational assessment of comprehension.
Dr Nicola Botting, City University
Memory and comprehension in children with SLI
Recent research has highlighted two behavioural markers for specific language impairments (SLI): memory for non-words (Bishop and colleagues; Gathercole and colleagues); and ability repeat sentences (Conti-Ramsden et al, 2005). Studies have shown that even when language difficulties appear to have resolved, individuals with a history of SLI perform more poorly on these measures. However, few studies have examined the precise relationship of memory and (oral and written) comprehension in this population.
In a large longitudinal study following 242 children with SLI from 7 to 16 years, the relationships between non-word repetition performance, sentence recall and receptive language were examined. A large majority of children with poor oral comprehension at 16 also had poor reading scores (77/86; 90%) but a smaller majority (77/118; 65%) of those with reading comprehension difficulties had low oral comprehension scores. Early and concurrent oral comprehension skill (and change in comprehension over time) were found to relate to memory skills at 11 and 16. In addition, three main subgroups (those with expressive-SLI, expressive-receptive SLI, and complex SLI) showed significantly different non-word performance. However this result was not entirely reflected in written text comprehension where change in reading skill did not show a significant relationship with memory. Furthermore, overall children with low reading comprehension scores were significantly poorer at all memory tasks, but poor text comprehenders (with good decoding skills) were not different from competent SLI readers on CNrep at 11 or 14. Nevertheless, the poor comprehender group were similar to generally poor readers on recalling sentences.
Different aspects of memory may prove particularly useful in unpicking the subgroups and pathways of SLI, and possibly for predicting outcome and guiding remediation in the SLI population.
Prof. Susan E. Gathercole, University of York
Children with poor working memory: Cognitive and behavioural profiles, and classroom support
Low scores on tests of working memory are closely associated with poor academic progress in key areas of the curriculum including reading and maths. In a recent large-scale study, we have investigated the behavioural characteristics of children with poor working memory. The children typically have good social skills, but are reserved in the classroom, where they frequently fail on activities that involve following instructions, keeping track of progress in complex tasks, and combining storage with processing. The teachers typically judge the children to be inattentive. Two methods of supporting the
learning of these children will be considered. One involves classroom-based support
designed to prevent working memory overload, and the other is a computerised training
programme designed to enhance working memory skills.