The gender congruency effect across languages in bilinguals: A meta-analysis

In the study of gender representation and processing in bilinguals, two contrasting perspectives exist: integrated versus autonomous (Costa, Kovacic, Franck, & Caramazza, 2003, Bilingualism: Language and Cognition, 6, 181-200). In the former, cross-linguistic interactions during the selection of grammatical gender values are expected; in the latter, they are not. To address this issue, authors have typically explored the cross-linguistic gender congruency effect (GCE; a facilitation on the naming or translation of second language [L2] nouns when their first language [L1] translations are of the same gender, in comparison to those of a different gender). However, the literature suggests that this effect is sometimes difficult to observe and might vary as a function of variables, such as the syntactic structure produced to translate or name the target (bare nouns vs. noun phrases), the phonological gender transparency of both languages (whether or not they have phonological gender cues associated with the ending letter-e.g., "-a" for feminine words and "-o" for masculine words in Romance languages), the degree of L2 proficiency, and task requirements (naming vs. translation). The aim of the present quantitative meta-analysis is to examine the robustness of the cross-linguistic GCE obtained during language production. It involves 25 experiments from 11 studies. The results support a bilingual gender-integrated view, in that they show a small but significant GCE regardless of the variables mentioned above.