MODÈLE DE L'OPÉRATEUR ET MODÈLE DU PRESCRIPTEUR. LE CAS DES CONSIGNES DE RÉSOLUTION DE SITUATIONS INCIDENTELLES POUR LES CONDUCTEURS DE TRAINS

This paper reviews the difficulties involved in following and carrying out written instructions. Instructions, especially in professional contexts, can be considered to be prescribed tasks: the act of authoring instructions relates to the writer's task model whereas the act of using instructions relates to the operator's task model. The purpose of this study was to examine (1) any discrepancies between the two models and (2) the effects of frequency of execution and task complexity upon these discrepancies. Five instructions in flowchart form were selected from materials dealing with difficulties facing train drivers. Following a comparison between sixteen operators' verbal reports and the written instructions, we derived the operator's model and the instruction writer's model. Discrepancies between these two models can be classified into five groups: omission, addition, contradiction, inversion and modulation. The results are in part consistent with the view that the more the incident situation occurs, the greater the discrepancy between the task operator's model and the task writer's model. The results have yielded some useful data on the effect of task solving complexity on the number of deviations. For low task solving complexity, the operator's model seems to be larger than the writer's model ; while for high task solving complexity, the operator's model is smaller because of action and condition omissions. Taken together, these results suggest that a prototypical operator's model exists. Some suggestions for improving the use of written instructions are made. L'état actuel des connaissances sur les facteurs d'efficacité des consignes semble pouvoir être complété par une approche considérant d'une part un modèle de la tâche particulier à l'opérateur et d'autre part un modèle de la tâche du prescripteur reflété par la consigne. A partir d'une analyse de protocoles verbaux de seize conducteurs de trains et de cinq consignes pour situations d'incidents dans la conduite de train, l'existence d'écarts entre les deux modèles est mise en évidence. Un effet de la fréquence de la situation ainsi que de la complexité de la tâche apparaît pour certains types de ces écarts. Par ailleurs, on constate que la lecture des consignes par les opérateurs ne permet pas toujours à ceux-ci de prendre conscience de ces écarts.