Hubert Tardieu

Processing of anaphoric devices in young skilled and less skilled comprehenders: Differences in metacognitive monitoring.

This experiment investigated metacognitive monitoring in the processing of anaphors in 10–year-old skilled and less skilled comprehenders. Two tasks were used with expository texts. The direct self-evaluation task was carried out with consistent texts in which target anaphors were either repeated noun phrases or pronouns. Subjects had to read and to evaluate their own comprehension on a 6–point scale. After reading, subjects answered multiple-choice questions designed to test the processing of anaphors. In the inconsistency detection task, target anaphors were either repeated noun phrases or inconsistent noun phrases. Subjects had to read and detect inconsistencies. After reading, they answered multiple-choice questions. In both tasks, on-line measures (reading times for units containing target anaphors and for subsequent units, and look-backs) were collected in addition to off-line measures (ratings of comprehension, detection of inconsistencies and response to multiple-choice questions) in order to analyse indicators of implicit and explicit evaluation and revision activities. The results from the two tasks converged: less skilled comprehenders showed deficiencies in monitoring on measures of implicit and explicit evaluation and revision. Patterns of reading times revealed that less skilled comprehenders were sensitive to the difficulties in processing pronouns in the self-evaluation task and also sensitive to the lack of text cohesion in the inconsistency detection task. However, this sensitivity was weak and unable to trigger explicit activities. These results were interpreted in the framework of Karmiloff-Smiths (1986) model.

Levels of Representation and Domain-Specific Knowledge in Comprehension of Scientific Texts.

Abstract Two levels of representation in text comprehension were postulated: a propositional representation and a mental (or situation) model of the content of the text. An experiment tested the assumption that the role of domain-specific knowledge on comprehension takes place at the level of the mental model. Two groups of experts and novices in the domain of memory read two sets of texts on memory and physics. Reading was interrupted by paraphrase and inference questions presented 0, 2 or 4 sentences after the critical information. Response times analysis for memory texts showed that experts were faster than novices for inferences, whereas no difference was observed for paraphrases. Experts were also faster when the distance increased. No difference between groups was observed for the physics texts. These results tend to indicate that performance differences attributable to domain-specific knowledge reflect differences in the construction of the mental model.

Working-memory capacity and reading comprehension in young and older adults

SummaryThis study investigates the role of working-memory capacity in reading comprehension in young and older subjects. A task yielding separate measures for processing and storage components was used to assess working-memory capacity. A French version of the Nelson-Denny test was administered as a measure of abilities that underlie reading comprehension. In the working-memory task, recall performances were lower in older subjects. Nevertheless, the intercorrelations suggested that the age-related impairment was probably linked to the processing component. Mean scores on the reading-comprehension test did not differ between groups. However, scores were correlated with processing time on the working-memory task in younger subjects, but with storage capacity in older subjects.

Chapter 9 Illustrations, Mental Models, and Comprehension of Instructional Text

Abstract Considering that an illustration can be regarded as one possible external expression of a part of a mental model, we investigated the role of illustrations in the strengthening of a mental model built from an instructional text. In a first phase, two groups of subjects read a text dealing with the cellular division process. In a second phase, a group saw drawings alone which illustrated the process, whereas the other group read the text again. In each phase, two comprehension tests were used to evaluate the representation built by subjects. The course of reading was interrupted by questions assumed to reflect either a linguistic representation (paraphrases) or a mental model (inferences), and subjects had to verify inference statements at the end of the presentation. Results showed that gain in performance between the two phases was the same whether the subjects had seen the illustrations or the text again. This absence of difference was observed during as well as at the end of the presentation, and for paraphrases as well as for inferences. It therefore appears that the mental model has to reach a sufficient level of elaboration, in order that illustrations alone can reinforce it. Resume Considerant quune illustration peut ětre vue comme une expression possible ďune partie ďun modele mental, nous avons etudie le role ďillustrations dans ľenrichissement ďun modele mental construit a partir ďun texte a visee didactique. Dans une premiere phase, deux groupes de sujets lisaient un texte traitant du processus de la division cellulaire. Dans une deuxieme phase, un groupe voyait des dessins seuls qui illustraient le processus, alors que ľautre groupe relisait le texte. Dans chaque phase, deux tests de comprehension ont ete utilises pour evaluer les representations construites. Le cours de la lecture a ete interrompu par des questions supposees refleter soit une representation linguistique (paraphrases) soit un modele mental (inferences), et les sujets devaient verifier des enonces inferes en fin de presentation. Les resultats ont montre que le gain de performances entre les deux phases etait le měme que les sujets aient vu les illustrations ou relu le texte. Cette absence de difference a ete observee aussi bien en cours quen fin de presentation et pour les paraphrases comme pour les inferences. Il semble donc que le modele mental doive atteindre un niveau ďelaboration suffisant pour que des illustrations presentees seules puissent le renforcer.

Mémoire de travail à long terme: quelle est l'utilité de ce concept ? Emergence, concurrence et bilan de la théorie d'Ericsson et Kintsch (1995)

Resume nLa theorie de la memoire de travail a long terme d’Ericsson et Kintsch (1995) avance l’idee que, grâce a l’expertise dans un domaine, les individus sont capables d’utiliser une partie de leur memoire a long terme comme memoire de travail. Nous avons essaye d’evaluer l’utilite de cette notion en nous centrant sur trois pointsxa0: (1) en presentant la theorie des chunks et les limitations theoriques qui ont conduit a la proposition de memoire de travail a long termexa0; (2) en mettant en avant l’impact heuristique de la theorie d’Ericsson et Kintsch (1995) au travers des experiences realisees apres sa parutionxa0; et (3) en confrontant la theorie de la memoire de travail a long terme avec des theories parentes, afin d’evaluer l’impact des idees qui sont au cœur de cette theorie.

Faux souvenirs et vieillissement : les effets de l’âge sur les inférences prédictives

To study false memories in older adults, a lot of experiments used the DRM paradigm (Deese, Roediger et McDermott). Most of the time, the results showed that older adults make more false memories than young adults. To test this hypothesis with a more ecological material, we used a situation of text reading. When we read a text, we activate predictive inferences, which are anticipations of what will happen next. We constructed short texts inducing predictive inferences (represented by a target word not presented) to study false memories in young and older adults. For example, in the text <<Charlotte had her breakfast on the terrace when the bees beat about the bush of jam. She made a movement to move them away but one of them succeeds in being posed on its arm>>, the target word sting is not presented but represents the predictive inference. After the reading of the texts, we propose to the subjects a restitution task consisting in recalling texts with the first sentence as clue. Then, they made a recognition task composed of target words and lures; they had to say if they remembered having read these words in the texts. In these two tasks, the subjects tended to remember not presented target words, creating false memories. This effect was the same for the two age groups showing that, in an ecological situation like text reading, older persons make as many false memories as young adults.

Slow and fast adult readers in text comprehension

Reading speed is a component of reading ability tests designed to separate «good» and «poor» readers. The purpose of the two experiments reported in this paper were to study whether slow and fast adult readers differ in terms of the subprocesses involved in textual organization. For this, two variables were manipulated: title and type of text. Subjects were classified as slow and fast readers based on the median value of the distribution of reading times on a preliminary text.Data clearly show that reading speed is a reliable individual characteristic. In Experiment 1, the recall performance of slow and fast readers did not differ. No interaction between reading speed and the title and type of text variables were observed. In Experiment 2, these results were replicated and extended to a summary test. So, it appears that slow and fast readers do not process textual organization in a different way and show identical comprehension performances. Further research is necessary to identify those factors that characterize good comprehenders among slow and fast readers.