Limousines and Corvettes: The development of subordinate-level categorization

• In Event: Categories and Concepts

Abstract

Despite extensive prior research on the structure of categories at multiple hierarchical levels (Markman, 1989; Murphy, 2003), important questions remain about the nature of subordinate-level categorization – particularly in the domain of human-made products (artifacts). Previous investigations have often conflated two distinct ways of categorizing artifacts beneath the basic level – one by kind and one by brand. For example, prior study of subordinate-level categories under the basic-level category, CAR, has included as examples both LIMOUSINE (a kind of car) and CORVETTE (a brand of car) (Collins, 2006). This conflation is problematic because membership in subordinate-level artifact kind categories is based on the maker’s intended function (Matan & Carey, 2001) whereas membership in subordinate-level brand categories is based on the maker’s identity. In two studies, we examined whether and when children distinguish between these two fundamentally different types of subordinate-level artifact categories.

In Study 1, participants (240 four- to seven-year-olds and adults) completed a forced-choice match-to-target task in which they extended a novel word from a target artifact (e.g., a fall glove) to one of two other artifacts from the same basic-level category – a kind match and a brand match. Participants learned that the kind match was made with the same intended function as the target but had a different maker (e.g., another fall glove made by a different company); and that the brand match was made with a different intended function but had the same maker (e.g., a winter glove made by the same company). Half the participants in each age group were assigned to the kind condition, where they heard the novel word modeled as a kind label (e.g., "This kind of glove is called a POGON"); the other half were assigned to the brand condition where they heard it modeled as a brand name ("This brand of glove is called POGON"). By five years, children selected the kind match more often in the kind condition than in the brand condition. By six years, children in the kind condition systematically chose the kind match. Only adults were systematic both in choosing the kind match in the kind condition and in choosing the brand match in the brand condition. The results leave open the question of when in development children in the brand condition systematically choose the brand match.

In Study 2, participants (144 four- to eight-year-olds and adults) completed a modified replication of the Study 1 brand condition, in which we simplified the task by describing the objects’ makers as “people” rather than “companies.” We replicated the key results from the Study 1 brand condition. Moreover, we found that eight-year-olds – like adults – systematically chose the brand match. Together, these results suggest that children distinguish between two distinct types of subordinate-level categories – kind and brand – by five years and show a systematic understanding of both types of subordinate-level categories by eight years. The findings highlight the need for the categorization literature to recognize and further explore the important distinction between – and the acquisition of – these two types of subordinate-level categories.

Authors

• Erica Dharmawan, University of British Columbia (UBC)
  Presenting Author

• D. Geoffrey Hall, University of British Columbia (UBC)
  Non-Presenting Author