The influence of grammatical gender on conceptual representations of gender has proven to be a controversial topic in the linguistic relativity literature, with empirical evidence in support of the Linguistic Relativity Hypothesis being highly task and context-dependent, as well as being modulated by the type of items being investigated (animates/inanimates). In this paper, we take a megastudy approach in order to investigate differences in results based on explicit and implicit paradigms that modulate the role of language and gender in their design. We present analyses of three experiments focussing on participants (total N = 4,621) with a grammatically gendered L1 (Czech), a non-grammatically gendered L1 (English) and L1-Czech in L2-English, and on three distinct semantic categories – people, animals and inanimates (total Nitems = 1,208). Our results indicate that the most reliable effects of grammatical gender influencing conceptual gender (outside of the domain of people) are observed for items representing animals, with Czech participants showing congruency effects in both explicit and implicit paradigms, even in their L2. The evidence for effects on inanimates is substantially weaker and is highly restrained to explicit tasks. We discuss these results in relation to the Linguistic Relativity Hypothesis and highlight important methodological considerations for future research.

It has been argued that the incidental and arbitrary use of gender markings for inanimate concepts in language may affect the conceptualization or semantics of those inanimate concepts. The present article sought to replicate the findings of a classic paper that made this argument. Konishi used the potency dimension of the semantic differential method as an implicit measure of perceived gender. He reported that words for inanimate concepts of masculine grammatical gender were rated as higher in potency than words for the same concepts that had feminine grammatical gender. Two preregistered replication studies are reported here. The first was a conceptual replication of Konishi’s study that was conducted with 240 bilingual native speakers of either German or Spanish. Included in the study was a follow-up with 120 monolingual native English speakers. This data was used to test whether the grammatical gender in the native languages of German and Spanish speakers affected their sense of the potency of common inanimate categories when tested in a second language (English) in which they were fluent and the nouns had no grammatical gender. A second version of the study was conducted in the native languages of Spanish and German speakers, as a closer attempt at a replication of Konishi’s original study. The results of both studies provided evidence against the grammatical-gender hypothesis. Bayesian tests of both studies strongly favored the null hypothesis that there were no grammatical gender effects on implicit measures of perceived potency.

Previous studies demonstrate that lexical coding of colour influences categorical perception of colour, such that participants are more likely to rate two colours to be more similar if they belong to the same linguistic category (Roberson et al., 2000, 2005). Recent work shows changes in Greek–English bilinguals' perception of within and cross-category stimulus pairs as a function of the availability of the relevant colour terms in semantic memory, and the amount of time spent in the L2-speaking country (Athanasopoulos, 2009). The present paper extends Athanasopoulos' (2009) investigation by looking at cognitive processing of colour in Japanese–English bilinguals. Like Greek, Japanese contrasts with English in that it has an additional monolexemic term for ‘light blue’ (mizuiro). The aim of the paper is to examine to what degree linguistic and extralinguistic variables modulate Japanese–English bilinguals' sensitivity to the blue/light blue distinction. Results showed that those bilinguals who used English more frequently distinguished blue and light blue stimulus pairs less well than those who used Japanese more frequently. These results suggest that bilingual cognition may be dynamic and flexible, as the degree to which it resembles that of either monolingual norm is, in this case, fundamentally a matter of frequency of language use.