Children's categorization of objects and scenes: a study in the internal structure of categories

Abstract

<p>This is a study of the internal structure of children's natural concepts. The application of this information is basic to our understanding of cognitive development, semantics and the development of memory. The study investigated four issues. The first issue dealt with whether children of different ages differ in the number of attributes they mention at the basic and superordinate levels for objects and scenes. The second issue considered was whether children of any age list significantly more attributes for basic level categories than they do for superordinate level categories. The third issue investigated was whether the most protoypical members of categories are those with the most attributes in common with other members of that category (i.e. family resemblance). The final issue studied was the effects of context on rated protoypicality. Will the item identified by children as most prototypical differ dependent upon the context provided by background scenes?;The design used was a 3 (age) x 2 (sex) x 2 (type of task: object or scene) x 2 (level of categorization: basic or superordinate) factorial design. Type of task and level of categorization are within-subject factors. Age and sex form the between-subject factors. Kindergarten (18 males and 18 females), second grade (19 males and 15 females) and fourth grade students (18 males and 17 females) from a rural midwestern elementary school served as subjects. All subjects participated in all of the tasks. The evidence showed first that fourth grade students list significantly more parts than the younger students. There was no significant difference between kindergarten and second grade performance. Secondly, the results indicate that all children listed significantly more attributes at the basic level than at the superordinate level. Finally the findings demonstrate that children of all three age groups do differ in the item they choose as most prototypical dependent upon the context of the background scene. With respect to the third issue the results were equivocal.</p>