Word learning in preschoolers: are bilingual 3-year-olds less guided by mutual exclusivity than their monolingual counterparts?

Abstract

A fundamental question in developmental linguistics and developmental psychology is how young children learn new words. While some researchers suggest that words are primarily learned through experience, others argue that the acquisition process is guided by innate lexical biases. One of the most widely studied biases is the Mutual Exclusivity Bias (ME), which describes children’s preference for just one label per concept. The disambiguation effect in ME has been demonstrated extensively with ostensive paradigms requiring young monolingual children to choose between familiar and novel labels in identifying unfamiliar objects. However, evidence for ME within languages in bilingual children is mixed. In the present study, a productive naming paradigm was used to assess 3-year-olds’ tendency to adopt novel labels for familiar items (a variant on Merriman and Bowman’s (1989) rejection/correction effect). Five monolingual and 5 bilingual children aged 2;11-3;6 were tested in English. Following a training session when the experimenter applied novel labels to 3 of 12 pictures of familiar objects, the children played two successive naming games. The first game involved further reinforcement of the novel labels by the experimenter while the second game did not. In the first game, the bilingual children adopted novel labels more frequently (Mdn=.40) than the monolingual children (Mdn=.13) and Mann-Whitney’s (one-tailed exact) U=3.0, was significant, p<0.05 with a large effect size (r=-.63). In contrast, only one bilingual produced a novel label in the second game. Measures of receptiveness in the training session displayed asymmetries between production and comprehension. Overall the results suggest that experience of two languages plays an important role in learning novel labels. The findings are consistent with an account of ME as a heuristic learned from monolingual input, the application of which varies in bilingual preschoolers according to both ambient language and socio-pragmatic context. The results are discussed in the context of what insights can be gained from possible extensions to the experiment.