Page 84 - Encyclopedia of Philosophy of Language

Language and Mind
Typically, these tasks involve objects characterized by variation in size, shape, color, or similar simple properties, usually denoted by adjectives—qualities, in a word. So, on the basis of the well-known studies mentioned above the following research findings may be identified:
Finding (1): Children in the age range 3-6 years can hold a quality in mind so as to organize free-sorting perform- ance, but cannot readily switch to a different principle of organization, nor coordinate two such principles in multiplicative fashion.
Such children then, according to the criterion of representational effectiveness, possess concepts of these qualities, although there are still some limi- tations to the flexibility with which they are employed. Younger children, although unable to form concepts of these properties, can execute simpler versions of the free-sorting task. Ricciuti (1965) showed that if the set of objects to be sorted consists of subsets of objects belonging to different simple kinds, such as dolls and boats, and if one requires only that the subjects should touch or handle these subsets successively (rather than form them into spatial groups), then even 1-year olds show some ability and 2-year olds can carry out the task, thus redefined. These findings have been con- firmed by Sugarman (1983), who also showed that 2- year olds will treat locally well-separated categories (for example, a set of green cylinders and a set of red
circles) as if they were kinds. So, taking Ricciuti's and Sugarman's results together one arrives at the second research finding:
Finding (2): Children in the age range 1-3 years can hold some sortal categories in mind so as to organize their free-sorting performance.
Such children can therefore form concepts of some kinds, but cannot form concepts of qualities. These sortals cover roughly the same ontological ground as Mervis and Rosch's notion of basic-level category. Also, the timing of this achievement, beginning around 18 months, coincides with Stage 6 of 'object permanence' (Piaget 1954). If this latter achievement is taken as marking the first construction of rep- resentationally effective individual concepts, then it would seem that the formation of concepts of simple individuals and concepts of basic-level categories are developmentally simultaneous. This is perhaps not surprising, in view of the next finding, after Mervis and Rosch, that individuals play an important role in the formation of such concepts:
Finding (3): Basic-level concepts are resemblance struc- tures. For any such concept and the population that employs it, some objects (stereotypes or prototypes) are better examples of the target category than others. Judged membership of such categories depends on similarity to the prototypes rather than on some (set of) common attribute(s).
Readers with philosophical backgrounds will be reminded of similar discussions in traditional meta- physics, notably in connection with the problem of universals (see Armstrong 1980). In that context (sometimes called 'first philosophy') the problem is to characterize the notions of object and property (by means of the metaphysical notions of particular and universal—or not, as the case may be)—-so as to give a satisfactory account of what things there are. The psychological context is different: it is to characterize a variety of mental representations—concepts—so as to give a satisfactory account of how wecome to know whatever things there are. However, as noted, there are many affinities between the two sorts of inves- tigation. In metaphysical discussions, whether realist or nominalist in tendency, such resemblance struc- tures have often been proposed as characterizations of properties. In realist analyses, beginning with Plato, there is a single external target against which resem- blance is measured, a pure or Ideal Form: in nominalist analyses, for example, the well-known discussion of games by Wittgenstein (1951), there is an endless chain of global resemblance. However, as has often been pointed out, one is left minimally with the universal
properties (relations) of resemblance and with the task of characterizing these. Similarly, in the account of concepts given in finding (3), one is led to wonder on what fundamental capacities the judgment of simi- larity to prototype depends. Substantial help is pro- vided here by study of the ranges of application of children's first names for basic-level categories. Stud- ies by Clark (1973) and especially Bowerman (1978) make it quite evident that judged similarity is by no means global but, rather, sharply structured by attri- butes, features, and qualities of shape, color, texture, etc. Although the theory of the development of word meaning proposed by Clark has now been discarded, it is sometimes forgotten that the data persist, and that these data show clearly that children's appre- hension of similarity is strongly structured by these qualities. Hence the finding:
Finding (4): Children younger than 2 years form concepts with pronounced resemblance structure. However, simi- larity to prototype clearly depends on formation of con- cepts of attributes, features, and qualities.
4. Paradoxes of Early Concept Formation
The findings (1) to (4) just described generate two formidable developmental paradoxes:
Paradox (1): Whereas studies of object sorting and hand- ling suggest that children younger than 3 years cannot yet form concepts of qualities, the studies of early word use suggest that they must have done so.
The distinction between minimal type-detectors (definition (a) above) and representationally effective type-conceptsmaybeeffectiveinresolvingthispuzzle. Or, conversely, this puzzle makes it evident that the
62