Abstract
INTRODUCTION : Recent research has shown that learning difficulties are closely related to both metacognitive and motivational determinants. School failure often entails such perceptions as learned helplessness, negative self - image, low self - esteem and a tendency to attribute failure to incapacity and successful outcomes to luck. Besides, poor strategy attainment and generalisation during school tasks is ascribed to metacognitive functions like planning, anticipation and executive control. Combined attributional and metacognitive training is expected to raise feeble school performance. Therefore, the following research question has been formulated : How much executive control can be fostered by a decontextualised metacognitive training, and how important are motivational variables, such as causal attributions, for metacognitive intervention in pupils with learning difficulties ?
PARTICIPANTS : 36 adolescents with learning difficulties, 13 to 15 years old, were divided into 3 experimental groups of 12 participants each ; one Experimental Group (EG1) receiving a combined metacognitive and attributional training ; a second Experimental Group (EG2) receiving metacognitive training only ; and finally, a Control Group (GC). All pupils attended preparatory classes for Vocational Education in Geneva.
PROCEDURE : Educational intervention was divided into 3 main phases ; pre testing, metacognitive with / without attributional training, and post testing of near and far transfer. TRAINING METHOD : In order to foster dynamic interactions and social - cognitive confrontation, work in small groups implicating minimum adult intervention was implemented.
MATERIAL : School - based tasks and attributional beliefs questionnaire for pre and post testing, metacognitive oriented exercises (DELF program) and discussion concerning attributional beliefs for training phases, and " Home Delivery ", a computer - based program for near transfer.
RESULTS will be discussed in relation to attributional beliefs and metacognition in pupils with learning difficulties.
INTRODUCTION
The concept of metacognition refers to two different things : metacognitive knowledge on the one hand and executive functions on the other.
Flavell & Wellman (1977) defined metacognitive knowledge as the knowledge learners have about themselves, about task characteristics and about strategies.
The concept of executive control as introduced in metacognitive theories is not oriented towards actions but towards cognitive processes. Executive function is indicated when a child alters strategic behaviour over the course of a task in order to solve it. Authors suggested that there exist three general executive skills such as anticipation, strategy planning and self-control (Belmont, Butterfield & Ferretti, 1982).
According to Büchel (1995) the main question about metacognition could be formulated as follows : if I know the features of the task, the characteristics of my own cognitive functioning, and a certain number of strategies how can I anticipate, plan and control my learning activity ?
Metaknowledge is thus supposed to be the knowledge base of executive functions at a metacognitive level which in turn controls cognitive processes at a cognitive level.
From a process oriented view, pupils with learning difficulties lack metacognitive skills (Borkowski et Büchel, 1983 ; Büchel, 1990). Therefore authors proposed that if we want to secure transfer in learning we have to instruct executive skills. Executive skills are important for efficient and effective learning because they assist in the performance of cognitive strategies. Thus, there is a functional relation between transfer and learning.
Recent research has shown that in order to understand strategy use and transfer it is necessary to consider both the motivational and metacognitive systems. Accordingly, the adoption and transfer of strategies are directly related to the attribution system of a person.
With respect to academic achievement, attribution theory posits that students make causal attributions to explain their success or failure. Some of the perceived causes which have been identified are effort, ability, luck and task characteristics (Powers and Rossman, 1983). Weiner (1985) delineated two main dimensions of attribution patterns: locus (internal / external), and stability (stable / unstable).
Strategy use is found to be related to attributions of success and failure. Previous research has shown that attribution beliefs are directly related to achievement behaviour for pupils with mental retardation (Turner, 1998). Pupils who attribute failure to a lack of effort are likely to increase their efforts in future attempts to solve a task. These pupils perceive that they have control over their performance. Children with learning difficulties often develop motivational and personal problems as a consequence of their learning difficulties.
Borkowski (1984) reported that children who attributed success to effort were more strategic following strategy training than those who attributed success to non controllable factors such as luck.
Previous research has shown that groups who received combined attribution and strategy training performed better at transfer than did the strategy only and control groups. Those studies were conducted with hyperactive pupils, pupils with mental retardation or learning difficulties (Borkowski, Weyhing and Carr, 1988 ; Reid and Borkowski, 1987 ; Turner, Dofny and Dutka, 1994).
Therefore, the following research questions have been formulated :
1 does a decontextualised metacognitive training foster executive control ?
and
1 how important are motivational variables, such as causal attributions, for metacognitive intervention in pupils with learning difficulties ?
METHOD
Main research hypothesis
We expected that the group who received a double training would show higher transfer at cognitive tasks than the groups who received a single training or no training at all.
Participants of the study
Thirty six adolescents with learning difficulties were divided into three experimental groups. Each group consisted of twelve participants ; the first experimental group received a combined executive and attribution training ; the second experimental group received executive training only ; and finally, the control group received no training at all. All pupils attended preparatory classes for vocational education in Geneva.
Material
During testing sessions four experimental tasks were used in order to evaluate cognitive performance and transfer effects. These tasks were an attributions questionnaire, a computer-based task on planning, and two school-based tasks, the symmetries and the tangram.
Tangram and Symmetries
School-based tasks were the symmetries and the tangram. These tasks were paper-and-pencil exercises. Both tasks had not been taught in the school year during which research took place. The goal in symmetries is to reproduce the symmetric design of a model in a quadrille. The goal in tangram is to reproduce a model by designing and numbering in the right place seven geometric forms.
Attributions Questionnaire
Based on Weiner’s model we have also developed an attributions questionnaire. Attribution elements such as effort, ability, task and luck were proposed in order to explain four success and four failure outcomes in school achievement situations. Each hypothetical situation was read by the experimenter. The pupils rated the degree to which each attribution element was responsible for a certain outcome.
Planning task
Finally, the goal in the planning task was to prepare an itinerary in a simulated environment in order to execute a home delivery activity. A familiarisation phase preceded the main phase of the task. During this phase the participant learned to use the different task operations such as to trace an itinerary, to read a delivery command, to load and unload delivery packets. During the task, the participant had to trace an itinerary, load the delivery packets on a lorry, and finally complete the delivery procedure.
Strategic games
In order to examine transfer effects we worked on executive functions by the means of metacognitive oriented exercises of the programme DELF. In order to examine the role of causal attributions to these transfer effects, we also aimed at altering the attributional style.
For executive functions training, we applied the strategic games of the metacognitive program DELF. The strategic games are chess-like paper-and-pencil exercises. Pupils have to find the shortest way in order to arrive with their chess pawn at a certain position on a designed board. To do this, pupils have to read the rules and the instructions of the task, to anticipate, plan and control their procedure, and finally to justify for strategy selection.
For attribution training a relevant five-minute discussion followed the end of each strategic game. The pupils of the first experimental group were invited to think about why they obtained a certain outcome in the strategic exercise that preceded. Four attribution elements such as effort, ability, luck and task were proposed as possible explanations. Following this discussion, the pupils were invited to complete an attributions questionnaire relevant to the strategic game that preceded.
Design
Research took place in the school of the participants. Educational intervention was divided into 3 main phases ; pre testing, training, and post testing. Posttessting was carried out during three sessions after training.
The testing assessments were the same for all students. All three groups participated in testing sessions. Each experimental task was administered during 10 minutes.
The metacognitive training phase occurred during 4 sessions. The first experimental group received a double metacognitive training, which included attributions training and executive functions training. The second experimental group received a single metacognitive training which included executive functions training only. In order to foster dynamic interactions and social-cognitive confrontation, work in small groups implicating minimum adult intervention was implemented.
Attributions training was realised by discussing on the reasons participants obtained a certain outcome in the strategic game which preceded. In the end of the discussion participants completed an attributions questionnaire equivalent to the questionnaire used during testing sessions.
Research variables
For the tangram and the symmetries each time two lines or forms were attached a correct point was accorded.
For the questionnaire we have collected the points for each attribution item separately.
For the planning task, variables were classified according to two main categories : « task learning » and « planning performance ». Planning performance was further divided into 3 sub-categories « anticipation », « performance » and « distraction resistance ».
Results
Results for attributions questionnaire
First of all we wanted to examine the stability and validity properties of the attributions questionnaire.
In order to evaluate the test-retest stability, we calculated Spearman’s coefficients for each sub-scale separately. Results showed that the present questionnaire is relatively reliable for the present population.
In order to examine the attributional tendency of the participants we calculated the correlations between the two sub-scales for the same attribution element. Spearman’s correlation coefficients were calculated. Results revealed no significant correlation between success and failure outcomes for effort, ability, task and luck.
Prior to training, participants responses were greatly influenced by the simulated outcome. After training, we expected that the group who received the double training would understand better the simulated outcomes than the two other groups. Correlations between the two sub-scales for the same attribution element were calculated for each testing session. Results indicated that that the group who received the double training made better distinction between the attribution element and success or failure outcome than the two other groups. However, correlation coefficients for this group were not significant.
In addition, we wanted to examine the questionnaire’s validity. Therefore, we calculated the coefficients between the items that are supposed to measure the same causal dimension. Spearman’s correlations were calculated for each sub-scale separately. Results revealed that the questionnaire’s validity in relation to the causal dimensions previously presented was not significant.
Results for cognitive tasks
As far as cognitive performance is concerned we expected that the group who received a double training would perform better than the control group. To test this hypothesis we administered a one-tailed t-test of independent samples. We compared the Mean scores of the first experimental group with the Mean scores of the control group separately for the pre-test, the post-test 1 and the post-test 2. Group differences were considered at a zero point ten level of significance(1). ANOVA repeated measures was not carried out because for each testing session different exercises of symmetries and tangram were applied.
Results indicated that the first experimental group was significantly more performing than the control group in symmetries. In addition, the first experimental group has demonstrated higher anticipation during the planning task than the control group. In the post-tests, however, there were no significant differences in the performance of both groups in tangram and in the planning task.
We also wanted to examine whether executive training only could be as efficient as the combined training. Therefore, we tested the same hypothesis with the second experimental group who received a single executive training. In symmetries the second experimental group made it much better than the control group ; even though in the pre-test the control group was significantly better that the second experimental group. For tangram, results indicated that in the post-test 2 this group was significantly more performing than the control group. As far as the planning task is concerned there were no significant group differences.
Finally, we wanted to see whether the metacognitive training proved to be effective. For this reason we administrated a series of ANCOVA with the post-test scores as dependant variable and the pre-test scores as covariate. Analysis of covariance was carried out because pre and post-test materials were not directly comparable.
Results for the first experimental group yielded no significant main effects neither for tangram nor for symmetries. However, results for the second experimental group yielded significant main effects for symmetries but not for tangram. As far as symmetries are concerned, the metacognitive training was effective for this group.
Discussion
The results of the present study showed that the double training did not bring about expected transfer effects at cognitive tasks. In contrast, the group who received strategy training only performed better at transfer tasks than the group who received the combined training. Therefore, the hypothesis that additional attribution training could be implicated in transfer was not confirmed.
It is a fact that the population of the present research displays several behaviour problems. For this reason the pupils of the first experimental group were less collaborative than the pupils of the second experimental group. Even though both groups were comparable in terms of cognitive and metacognitive abilities the two groups proved to be unequal as far as co-operative work is concerned.
The results of the present study have also shown that the attributions questionnaire consists in a relatively reliable instrument. However, it seems that causal explanations are greatly influenced by success or failure outcomes.
Questions to be addressed
This study should be replicated in order to confirm present hypotheses. Future research should consider two different aspects :
First of all, the validity and reliability of attribution questionnaires should be considered. In addition, in the present quasi-experimental design, the research groups were unequal in terms of non metacognitive variables such as willingness to collaborate. There is a need to focus on metacognitive training with experimental rather than quasi-experimental designs. In fact in quasi-experimental designs the research groups are not always matched in a flawless manner.
 |
 |