Lexical development in bilinguals

The focus of this project is the representation of artifact categories, more specifically in the context of cross-linguistic comparison and bilingual lexical development. Across cultures similarity between artifact objects is perceived in much the same way. Despite this universal perception of inter-object similarity, there are substantial differences in naming patterns across languages. The dissociation between perceived similarity and naming poses a challenge for bilinguals. For bilinguals to be completely monolingual-like they have to acquire different naming patterns in each language. However, bilingual roughly equivalent categories resemble each other more closely than the corresponding monolingual categories, a finding that is called bilingual convergence. 

The first aim of this research project was to replicate the finding of convergence in the bilingual lexicon in adults. A successful replication offered confirmatory evidence and ensured that follow-up studies are not relying on coincidental characteristics of the participant sample. Second, this project evaluated the developmental trajectory for later lexical learning in children starting from age five to adults. Language specific violations of similarity driven naming cause challenges for monolingual as well as bilingual children. We examined three kinds of word-referent mappings in order to get a better understanding of these challenges using cross-linguistic comparison and comparing data of monolingual and bilingual children. Third, previous research concerning cross-linguistic comparison made use of data that were averaged over a group of participants of one language. To further explore cross-linguistic differences, I made use of Item Response Theory (IRT) models, which take into account individual differences. This allowed us to further uncover the nature of artifact category representations and sources of categorization differences, not only between groups of individuals but also between individuals.

A newly developed set of stimuli, consisting of 192 pictures of containers, takes a central role in answering the research questions above. The importance of the chosen set of stimuli is twofold. First of all, a replication with a larger set of stimuli validated previous findings and enabled us to investigate the generalizability of those findings. The new stimulus set contained more exemplars of the main lexical categories under investigation, leading to a more reliable, denser representation of the lexical domain. This increased density allowed for a more detailed comparison between the category boundaries of bilinguals versus monolinguals. Second, previous work using IRT modeling made use of a different type of stimuli, mainly consisting of superordinate categories with basic level words as instances. Whereas the previous work spans many types of superordinate categories, the stimulus set used in this project allowed us to examine the composition of basic level artifact categories in terms of their concrete instances in close detail.