Development of object categorization in monolingual and bilingual children.

Most prominent models of categorization assume that categorization is similarity-based: Objects that have important properties in common tend to be given the same name, while objects that do not share important properties, tend to be named differently. However, Malt, Sloman, Gennari, Shi, and Wang (1999) showed that, even though similarity perception of common household objects is shared across cultures, speakers of different languages showed substantial differences in the way they segment the stimulus space by name. This finding shows that learning the words of a native language is not a straightforward one-to-one mapping process to entities they refer to. In a follow-up study using a similar set of household objects in a similar naming task with 5- to 14-year-olds, we showed that mastery of the appropriate names may take years for monolinguals (Ameel, Malt, & Storms, 2008). For bilingually raised children, the challenge to master two languages is even greater, since they have to maintain different mappings for their two languages. We also showed that bilingual adults are not fully native-like in either language: linguistic boundaries of corresponding categories in the two languages converge on a common naming pattern in both languages (Ameel, Storms, Malt, & Sloman, 2005; Ameel, Malt, Storms, & Van Assche, 2008). Taking the complexity and cross-linguistic variation of linguistic categories into account, the current research project focuses on the development of object naming, both in monolinguals and bilinguals. A first part of the research project is related to the development of object naming in monolinguals. In our monolingual developmental naming study (Ameel et al., 2008), we found that children gradually learned to attend to the adult feature sets and to assign the appropriate features the appropriate weights. The features, however, could not fully account for all the naming choices. As Malt et al. (1999) argued, other linguistic and cultural historic constraints that violate the similarity principle also play a role in naming patterns. This suggests that the learning process of object categorization during later lexical development comprises two aspects: similarity-based and language-specific learning. As of now, the relative contribution of both aspects is unknown. In the current research project, we want to disentangle both aspects and study how they develop over time. A straightforward approach to quantify both aspects is to identify outliers (i.e., objects that do not obey the similarity principle) in a multidimensional geometrical stimulus representation that reflects the similarity relations between the objects (in which similar objects are located close together and dissimilar objects are located further apart). Using different techniques to identify outliers, all objects can be classified as outliers or non-outliers in all age groups and thus the relative contributions of similarity-based and language-specific learning can be disentangled and the development of both learning aspects can be studied across later lexical development. Based on the artificial category learning literature and on findings from the developmental literature, we expect that object naming will initially be dominated too heavily by similarity-based learning, and that language-specific learning will gain in significance over time. Besides elaborating a method to quantify the two aspects of category learning, we also intend with the proposed research plan to develop a model for later lexical development that focuses on the evolution of similarity-based and language-specific categorization over time. Since most theorizing on lexical development has focused on the early stages of lexical development (during preschool years), and recent research has demonstrated that lexical development continues well past the early years of language acquisition (Ameel et al., 2008), a model for later lexical development would fill a crucial gap in the literature on language development. A second part of the research plan focuses on the development of object naming in bilingual children. The above mentioned finding of the merged naming pattern in bilinguals (Ameel et al., 2005) raises the question of how this pattern develops. On the one hand, it is possible that bilingual children acquire the same naming patterns as monolingual children because of similar language input, but that at a later age, an ergonomic reorganization takes place, induced by the cumulating cognitive requirements of the idiosyncrasies of both languages. On the other hand, it is also possible that the convergence takes place from the onset, because children may start to compromise (which leads to convergence) as soon as an idiosyncrasy in one language causes a naming difference between the two languages. We suggest to answer this question by comparing the naming data of compound bilingual children with the naming data of monolingual children. Compound bilingual children grow up with two native tongues. The two parents, each with a different native language, consistently speak their own language to the child. In follow-up research, we want to study coordinate bilingual children who acquire and use their two languages in distinct environments or separate contexts (e.g., home versus school) and subordinate bilingual children, who learn their second language at a later age (in school). We want to investigate how vulnerable the gradual shaping of the categories in their mother tongue is to influences of a second language, and how this is affected by the age at which they acquire the second language. Finally, we want to construct a computational model to explain the development and representation of the bilingual lexicon using a connectionist approach.

Keywords:Bilingual children, Monolingual children

Disciplines:Animal experimental and comparative psychology, Applied psychology, Human experimental psychology