Current Research
Sign Up now to participate and join the fun!
Constraints on Infant Categorization
One of the most important accomplishments of infancy involves partitioning the world into meaningful categories (e.g., Mandler & McDonough, 1993; Quinn & Eimas, 1997). Infants' success at this task is particularly remarkable given the vast quantities of information that they must process in doing so. Like adults, infants are faced with an infinite array of objects, each characterized by an infinite number of attributes (e.g., size, shape, color). One of the principal goals of our research is to identify constraints that help to limit the information that enters into the categorization process, and to specify the mechanisms by which they do so. Thus far, our research has identified object functions and names as strong candidates.
1a. Object functions
We recently demonstrated that both 14- and 18-month-old infants were more likely to organize novel objects into categories after observing their functions (Booth, 2001; Booth & Waxman, 2002). Importantly, because the functions were not available at test, infants'
categorical responses were necessarily based on other object properties. This suggests not only that object functions are salient to infants, but also that these cues draw infants' attention to other similarities among novel objects. We are currently working towards answering the following questions regarding this effect.
- What are the specific consequences of object function on categorization? Clearly infants can form new categories without the support of object functions. We seek to determine the benefits of providing function over and above basic perceptual experiences with objects. Does function increase the efficiency with which categories are formed? Does it affect the memorability or generalizability of resulting representations?
- What is it about object function that supports its facilitative effect on categorization? Function is a complex and poorly defined construct. Our hope is to break it down into more specific components that can independently be tested for their influence on early categorization. Two of the components of particular interest are causal relations between objects and interesting outcomes and actions of a human agent.
- What are the mechanisms by which object functions exert their influence on categorization? We are particularly interested in whether functions enhance infants' attention in a general way by highlighting any and all commonalities among objects or whether function enhances attention in a more specific way by highlighting just those object features that are causally related to it.
1b. Object names
Our research (in collaboration with Dr. Sandra Waxman and the Project on Child Development) also demonstrates that words help infants as young as 11-months of age to isolate commonalities among objects (Waxman & Booth, 2003). Initially this effect is quite broad such that infants map both count nouns and adjectives onto both category-based (e.g., animals) and property-based (e.g., color) commonalities among objects. However, by 14 months of age, infants' are sensitive to subtle cues in adult speech that allow them to distinguish between these grammatical categories. They come to attend exclusively to category-based commonalities when hearing count nouns, but attend either to property-based commonalities exclusively or to a broader range of commonalities when hearing adjectives (Waxman & Booth, 2001, Booth & Waxman, 2003).
Interestingly, the influence of count noun naming on categorization is evident with completely novel objects in 18-, but not 14-month-olds (Booth & Waxman, 2002). The null results obtained for the younger age group appear to be due to infants' inability to detect a meaningful referent for the novel names presented to them. Because the stimuli were completely novel, and no conceptual properties were detectible, infants had no 'core meaning' to attach the novel words to. When infants were given a 'hint' regarding the deeper conceptual properties of the objects (i.e., their function), naming did facilitate categorization at 14 months. We are currently exploring other possible interactions between the effects of object names and object functions on categorization in infants and preschoolers. Specifically we are interested in whether these cues can work together to provide an optimal environment for learning new categories early in development.
Early word learning
Because we believe that early linguistic and conceptual development are inextricably linked, many of our projects focus on word learning in infants, toddlers, and preschoolers. We are particularly interested in the role that conceptual information plays in supporting the acquisition and extension of new words.
2a. Knowledge of ontological kind
In recent work, we demonstrated that young children's ontological knowledge about labeled objects influences their extension of novel words applied to those objects (Booth & Waxman, 2002, 2003). Three-year-olds extended novel words on the basis of similarities in shape alone when labeled objects were described as having conceptual properties typically associated with artifacts (e.g., has a special use, is kept in a particular location). In contrast, children extended labels on the basis of similarities in both shape and texture when the very same objects were described as having conceptual properties typically associated with animate kinds (e.g., has emotions, eats). Because the same objects were presented in both conditions, only the conceptual information provided could be responsible for the differences in extension patterns. We have since extended this important finding to 2-year-olds and have shown similar effects in infants as young as 18- to 22-months of age who have minimal productive vocabularies (Booth, Waxman, Huang & Hackenberg, 2005). We are currently attempting to better articulate the origins of these effects.
2b. Conceptual information
In recent studies, we began to explore the possibility that children can learn words more readily when conceptual information about their referents is provided. In teaching new words for pictured objects to 3-year-olds, we either highlighted obvious perceptual information (that was evident in the photographs), non-obvious perceptual information (e.g., it has purple bones or is spongy inside), or conceptually rich information (e.g., it can see in the dark or it is used to cut grass). Children appear to learn more words, and to learn words faster, when provided with conceptually-relevant information. We are currently conducting a similar study with toddlers to further explore the facilitative role of conceptual information (here, in the form of function) on learning new words for novel objects. We are also attempting to better articulate the mechanism underlying this effect, as well as its precise scope of impact on learning and retention.
3b. Gestural cues
It is well established that young children, even infants, utilize gestural cues to the referential intent of speakers (e.g., pointing) to disambiguate the meaning of new words. We are currently exploring a hierarchy of such cues in order to establish their relative impact on word learning in infants and toddlers. In our initial study, we taught 28- to 32-month-olds new words for novel objects by either looking at, pointing to, touching, or manipulating the intended referent. Learning improved with greater redundancy among cues, with the largest improvement evident when pointing was added to gazing. We are currently extending this investigation downward to younger infants. We are also currently assessing whether late-talking toddlers and young children diagnosed with autism are able to effectively capitalize on these cues, and whether they do so for the same reasons.
