ERP evidence for the activation of gender stereotypes: The case of Italian.

Language users have mental representations of words (e.g., occupation nouns and personal characteristics) that include information about the word’s stereotypical gender. This information is difficult to suppress during on-line language processing (e.g., Banaji & Hardin, 1996; Cacciari & Padovani, 2007; Oakhill, Garnham, & Reynolds, 2005). The few electrophysiological studies conducted on this topic showed that different neural processes are engaged in the processing of gender-stereotype information (Irmen, Holt, & Weisbrod, 2010: N400, P600; Osterhout, Bersick, & McLaughlin, 1997: P600; White, et al., 2009: N400). In this ERP study we investigated the activation of gender stereotypes in Italian using a priming paradigm adapted from Banaji and Hardin (1996). Our aim was, first, to establish how early this information becomes available to the reader, and, second, to uncover the ERP signature of the emergence of gender stereotypes in language. Participants were presented with a prime that could be: a masculine or feminine stereotypical gender noun (conducenteMASC “driver” vs. insegnanteFEM “teacher”); a masculine or feminine grammatically marked noun (pensionatoMASC “pensioner” vs. passeggeraFEM “passenger”). Each prime was followed by either a masculine or a feminine personal pronoun (Lui “he” vs. Lei “she”). Participants decided whether the pronoun was masculine or feminine, while their RTs and ERPs were recorded. Primes and targets were controlled for psycholinguistic variables (length, frequency); in addition, masculine and feminine stereotypes were matched in stereotype strength and valence. As in previous behavioural studies, participants were faster to judge the gender of the pronoun when preceded by a gender-congruent than gender-incongruent prime in both biological and stereotypical conditions. The ERP results suggest two different effects. First, when the pronouns were preceded by biological grammatically marked incongruent nouns (e.g., pensionato-lei; passeggera-lui), a larger negativity between 200 and 380 ms peaking around 300 ms (most prominent across frontal/central sites) emerged. Interestingly, when the pronouns were preceded by stereotypical primes, a negativity with similar latency and distribution emerged in the incongruent condition only for masculine pronouns. Second, an increased positivity between 380 and 500 ms peaking around 420 ms (most prominent across frontal/central sites) was observed when pronouns followed biological, but not stereotypical, gender-incongruent primes. The waveforms we obtained for biological gender violations are comparable to the N400 reported by Barber and Carreiras (2003). Our seemingly early and more frontal effect could be due to the use of function words (pronouns) rather than content ones as in Barber and Carreiras (2003). The positivity around 420 ms for biological gender violations appears to be in line with the P300 effect observed in Barber and Carreiras (2003) together with the N400. Crucially, our ERP results provide further support for online effects of stereotypical gender in language comprehension. When a role noun is read, the stereotypical gender associated with it, if any, is activated together with other lexical-semantic information and might prime gender-congruent nodes. Remarkably, the ERP confirmed a gender stereotype asymmetry (cfr. Cacciari & Padovani, 2007), in that male and female gender stereotypes affected the processing of pronouns differently. The results imply that participants seemed more accepting of female drivers than male teachers, suggesting that gender stereotypes (conveyed by occupation nouns or personal characteristics) might be less restrictive for females than males. According to social psychologists, one social group (e.g., males) can become more normative than another (e.g., females) (Hegarty & Pratto, 2001). Indeed, attitudes and stereotypes have been found to be influenced more by male exemplars than by female ones (Eagly & Kite, 1987). We can thus hypothesize that female gender stereotypes (e.g., insegnante “teacher”) recruited only female category members, while male gender stereotypes (e.g., conducente “driver”) recruited both male and female category members. This is because in our society, the male social group is more normative than the female one, being the “unmarked normative group” (Hegarty & Pratto, 2001). As a result, masculine pronouns that followed female stereotypes mismatched category norms, eliciting longer reading times and a more pronounced negativity, while feminine pronouns that followed male stereotypes did not.