Spatial Cognition and Computation 1: 1–29, 1999.
© 1999 Kluwer Academic Publishers. Printed in the Netherlands.
Perception, object kind, and object naming
and MICHAEL LEYTON
University of Delaware;
Abstract. We investigated whether certain perceptual properties of objects could support
children’s and adults’ judgments of the range of shape changes permissible for a named object.
Three year-olds and adults saw a line drawing of a novel object and heard it named using a
count noun (e.g., “This is a dax.”). Then they judged whether shape or size changes of the
original could also be called by the same name (i.e., “Is this a dax?”). Children and adults
extended the object name to the size changes. In contrast, extension to shape changes strongly
depended on the particular characteristics of the objects. Objects with straight edges and sharp
corners elicited very low generalization to shape changes, consistent with a “shape bias”.
Objects with curved edges, curved and wrinkled edges, and curved and wrinkled edges plus
“eyes” elicited increasingly broad generalization to the same shape changes. In a compar-
able No-Word task, children’s and adults’ judgments were similar across all different object
types. The difference in generalization patterns over the two tasks suggests that only naming
systematically engaged representations of the objects that could support inferences about their
potential for shape change. The results are discussed in terms of the complex interactions of
perception, ontology, and labelling in the development of object naming.
Key words: count noun, lexical learning, naming, object, shape
The purpose of this paper is to investigate the role of certain perceptual prop-
erties in setting the boundaries for generalization of an object’s name. Our
broader question is whether speciﬁc perceptual properties can guide learners’
inferences about the range of transformations that an object can undergo and
still be called by the same name. We have in mind a range of different object
kinds that undergo very different transformations, and therefore can regularly
assume different shapes: Nutcrackers open and close, sponges deform when
wet, dogs and cats change conﬁguration as they move, etc. Although there is
considerable evidence that young children generalize novel object names on
the basis of object shape, little is known about the ﬂexibility with which these
generalizations are made, nor the richness of the underlying representations
of shape. Do children generalize only to static “snapshots” of an object’s
shape? Do they recognize that certain classes of objects (and not others)
might regularly undergo shape changes and still retain their identity and class
membership? If so, must they learn these item-by-item (observing cups that
never deform, sponges that shrink and expand, etc.), or are there more general