Word learning is a complex phenomenon because it is tied to many different behaviors that are linked to multiple perceptual and cognitive systems. Further, recent research suggests that the course of word learning builds from effects at the level of individual referent selection or noun generalization decisions that accumulate on a moment-to-moment timescale and structure subsequent word learning behaviors. Thus, what is needed for any unified theory of word learning is 1) an account of how individual decisions are made across different contexts, including the details of how objects are encoded, represented, and selected in the course of a word learning behavior; and 2) a mechanism that builds on these individual, contextually specific decisions. Here, the authors present a Dynamic Neural Field (DNF) Model that captures processes at both the second-to-second and developmental timescales and provides a process-based account of how individual behaviors accumulate to create development. Simulations illustrate how the model captures multiple word learning behaviors such as comprehension, production, novel noun generalization (in yes/no or forced choice tasks), referent selection, and learning of hierarchical nominal categories. They also discuss how the model ties developments in these tasks to developments in object perception, working memory, and the representation and tracking of objects in space. Finally, the authors review empirical work testing novel predictions of the model regarding the roles of competition and selection in forced-choice and yes/no generalization tasks and the role of space in early name-object binding.
TopIntroduction
Word learning is a complex phenomenon because it is tied to many different behaviors. To learn even a single new word, children have to segment the target word from the ongoing speech stream, find the referent of the novel word in the current scene—which typically contains many possible referents—encode the novel word form, encode something about the referent such as where it was, its shape, color, material, what it was doing, etc., and store all this encoded information in such a way that the different pieces are linked and can be retrieved at a later point in time when the child needs to recognize the word or produce the name (Gupta, 2008; Capone & McGregor, 2006; Oviatt, 1982; Oviatt, 1980).
Word learning is also complex because these different behaviors are linked to multiple perceptual and cognitive systems—systems for orienting to sounds and distinguishing language from other noises; systems for finding regularities in the sound stream; systems for interacting in the social contexts in which language occurs; and systems for visually perceiving and categorizing objects. Each of these systems involves a host of sub-processes. For instance, visually perceiving and categorizing objects entails segmenting objects from a visual scene, integrating those objects across multiple feature dimensions (shape, color, material, size, etc.), and integrating this in-coming information with the learned organization of visual categories. Critically, these object processing details must somehow be integrated with the other systems—for instance systems for finding regularities in the sound stream and systems for interacting in the social contexts in which language occurs—required to learn a word.
Finally, word learning is complex because it is extended in time—children begin orienting to their name as early as the fourth month of life (Mandel, Jusczyk, & Pisoni, 1995), and word learning continues throughout the lifespan. Central to the model presented here is evidence of rapid changes in the speed of word learning in early development, including the vocabulary explosions in toddlers and young schoolchildren (Goldfield & Reznick, 1990; Mervis & Bertrand, 1995; Clark, 1993; although see McMurray, 2007; and Bloom, 2000, for debate regarding the nature of these explosions).
Given the complexity of word learning, a central challenge has been to establish empirical paradigms that effectively reveal the processes of word learning and to develop new theories that uncover the mechanisms that move word learning forward. One approach has been to focus on specific phenomena that seem to be particularly revealing of the processing that operates as children learn words. We focus on one such phenomenon here—research on the shape bias.
The shape bias refers to the tendency to generalize novel names for novel solid objects by shape. Children begin to demonstrate the shape bias after having acquired some nouns in their productive vocabulary (Landau, Smith, & Jones, 1988; Gershkoff-Stowe & Smith, 2004; Samuelson & Smith, 1999). Furthermore, there is evidence that children who are not acquiring vocabulary at the typical pace (e.g. late talkers) do not show a shape bias (Jones & Smith, 2005; Jones, 2003). Our own work has demonstrated that the processes that support the development of a shape bias are general and linked to the development of other word learning biases and to the acquisition of the vocabulary as a whole (e.g., Perry, Samuelson, Malloy, & Schiffer, 2010; Perry & Samuelson, 2011). Thus, a process-based, mechanistic account of the development of the shape bias should inform our understanding of how word learning changes in early development as well as how word learning is connected to developmental changes in other perceptual and cognitive systems both with typical and atypically developing individuals.