In five picture-word interference experiments we explore the gender congrui
ty effect observed in Dutch in two languages, Spanish and Catalan. Particip
ants' performance was not affected by the relationship between the gender o
f the picture and the gender of the word. The results show that the gender
congruity effect is not a universal effect, but varies from language to lan
guage, depending on crucial characteristics of the gender/determiner select
ion system used to process a given language. Consistent with the crosslingu
istic hypothesis presented by Miozzo and Caramazza we argue that the retrie
val of the noun's gender is enough to specify the determiner's phonological
form in Dutch, but not in Catalan or Spanish, and this is the cause of the
failure to replicate the Dutch results in these two languages.