When Skunks are Similar to Giraffes and When They Are Not: Grammatical Gender Effects on Bilingual Cognition

When Skunks are Similar to Giraffes and when they are not: Grammatical Gender Effects on Bilingual Cognition Stavroula-Thaleia Kousta (s.kousta@ucl.ac.uk), Department of Psychology, 26 Bedford Way London WC1H 0AP United Kingdom David P. Vinson (d.vinson@ucl.ac.uk), Department of Psychology, 26 Bedford Way London WC1H 0AP United Kingdom Gabriella Vigliocco (g.vigliocco@ucl.ac.uk) Department of Psychology, 26 Bedford Way London WC1H 0AP United Kingdom Abstract We investigate how second language learning affects ‘think- ing for speaking’ in the first and second language by examin- ing the semantic effects of grammatical gender (present in Italian but absent in English). In an error-induction experi- ment we first establish a baseline, showing that gender affects the semantic substitution errors made by monolingual Italian speakers compared to monolingual English speakers. We then show that Italian—English bilinguals behave like monolin- gual English speakers when the task is in English, and like monolingual Italian speakers when the task is in Italian, hence exhibiting appropriate ‘thinking for speaking’ for each lan- guage. These results have implications for linguistic relativ- ity/determinism and models of bilingual semantic memory and processing. Keywords: bilingualism; linguistic relativity; thinking for speaking; grammatical gender; speech errors Introduction In order to speak any language, one has to pay attention to the distinctions obligatorily expressed in the language, and in this way every linguistic community differs from every other. In other words, there is linguistic relativity, since speakers of typologically different languages are required to verbalize different aspects of reality when constructing lin- guistic messages. What are the cognitive consequences of linguistic relativity? For Whorf (1956), to whom the linguis- tic relativity hypothesis is attributed, linguistic relativity is closely linked to linguistic determinism, that is, the proposal that language determines the way the external world is per- ceived, categorized and acted upon. Nowadays virtually nobody would like to claim that lan- guage has such a deterministic role on cognition as Whorf originally envisioned it. It is, however, still a matter of in- tense debate to what extent language affects (rather than determines) cognition. According to one hypothesis, the linguistic classifications imposed by language affect only those aspects of cognition that are actively engaged in the processes of speaking. This “thinking for speaking” pro- posal has been put forward by Slobin (1996), is currently more or less uncontroversial, and evidence in support of it abounds in the literature (see Vigliocco, Vinson, Paganelli, & Dworzynski, 2005). A stronger version of the linguistic determinism hypothe- sis argues that language also affects non-linguistic cogni- tion, above and beyond the processes involved in speaking and comprehending a particular language (e.g., speakers of different languages perceive the world in different ways due to linguistic differences). Data in support of this claim have been provided in numerous behavioral experiments (e.g. Imai & Gentner, 1997; Levinson, 1997; Lucy, 1992), but at the same time there is a substantial body of evidence that argues against this claim (e.g. Gennari et al., 2002; Li & Gleitman, 2002; Vigliocco et al., 2005). In other words, the main debate in the field focuses on the extent and perva- siveness of the effect of language on cognition. Up to the present, investigations of the link between lan- guage and thought have concentrated on monolingual popu- lations, with very little work carried out with bilingual speakers. Bilingualism, however, offers a unique test case for the potential role of language in shaping cognition, espe- cially when focusing on highly proficient bilingual speakers who acquired a second language after their first language was largely in place. If learning a first language affects cog- nition beyond ‘thinking for speaking’ for that same lan- guage, then bilingual speakers, even when they are highly fluent in their second language, should show evidence of transfer of linguistically motivated semantic/conceptual categories from their first language into their second, despite their high proficiency in their second language. In other words, their ‘thinking for speaking’ in their second language should be, at least to some extent, dependent on ‘thinking for speaking’ in the first. This prediction is based on the assumption that language, by means of its lexical and grammatical structure, habitually directs attention to a num- ber of conceptual distinctions. In this paper we investigate bilingual cognition in relation to a phenomenon, grammatical gender, which has attracted considerable interest in the literature on linguistic relativ- ity/determinism. The crucial questions are whether the bi-