Bilingual categories as a window into semantic-conceptual mapping. Behavioural and fMRI evidence

Abstract

This presentation focuses on bilinguals' processing of categories that differ in their two languages, with the goal of gaining greater insight into the relationship between concepts and linguistically encoded classes. While concepts are fundamental to mental life (Murphy, 2002), debates still rage concerning the extent to which language influences cognition (see Lakoff, 1987; Bowerman, 1996; Gentner, 2003; Murphy, 2002; Elston-Güttler & Williams, 2008; Levinson, 1997). The present studies focus on cases in which one language lexicalizes a concept in a word, while another either lacks a word for it (e.g., English nut denotes a class including almonds, pecans, cashews, while there is no translation equivalent in Spanish (Gathercole et al., 2010) or divides up the conceptual space differently (e.g., English contrasts fingers and toes, while Catalan dits refers to both fingers and toes). Our two studies aim at exploring the behavioural and neural correlates of semantic and conceptual representations –often overlooked in language neuroscience with problematic results (Pylkännen 2011)-. The present studies use categories that differ across languages in their conceptual-semantic relation, ranging from categories that show a stronger ’conceptual’ component (classical/objectivistic categories), to ''categories'' that are purely ’linguistic/semantic’ (homonyms). (See appendix for stimuli examples). The first experiment was a category membership judgement study, where bilingual Spanish-Catalan and monolingual Spanish participants decided whether items presented visually belonged to a certain category. Participants’ performance was highest in classical (predominantly conceptual) categories followed by radial taxonomic (intermediate), then homonyms (linguistic) and finally radial thematic categories (less conceptual than radial taxonomic). Bilinguals performed overall better than monolinguals, but there were no differences between language groups in homonyms. Additionally, bilinguals tended to ’overextend’ in their judgements about category boundaries by choosing distractors as members of the categories. The second experiment used fMRI to explore L1 to L2 priming in a lexical decision task, specifically to examine whether the degree of priming and patterns of brain activation vary across category types. Spanish speakers who had English as their L2 and English monolinguals were given an explicit priming lexical decision task that used pairs of samecategory words as stimuli (see appendix). Results revealed an overall higher neural activation in bilinguals (performing in their L2) in areas associated with processing of visual word form (BA 37), classical language areas (BA 9, BA 47, BA 39) and memory (BA 36). Monolinguals showed lower neural activation except in memory areas (BA 6). In bilinguals, Classical categories were associated with higher neural activity in the visual cortex (BA 19) and areas hypothesized to integrate visual and motor information (BA 7). On the other hand, homonyms were associated with more activity in executive control areas (BA9, BA10, BA46), Broca's area (BA44), and the angular gyrus (involved in semantic processing BA39). Taken together, these results provide evidence for different mechanisms in category boundary processing in bilinguals, compared to monolinguals, and separate neural bases for conceptual and linguistic components. Finally, we argue for a more careful selection of stimuli in language neuroscience studies and a distinction between semantics and concepts.