Our world knowledge is deeply embedded in language. The word “banana” immediately brings to mind the color “yellow”. Oftentimes, however, when we read about “banana”, we do not expect the color “yellow” to be explicitly mentioned as in “yellow banana”. Frequent omission of obvious information given our world knowledge is predicted by so-called informativity expectations, which capture our preferences for newsworthy and informative utterances. Here we present two studies investigating informativity expectations in a socially situated context of gender stereotypicality and testing whether expressions like “female nurses” are less preferred than expressions like “female directors”. While the results show a clear impact of informativity expectations, effects of gender stereotypes turn out to be difficult to overcome.

Natural languages often exhibit agreement, where two words must be matched for certain features. It's well known that people use knowledge about agreement to drive expectations during online processing. What is less well known is how the type of dependency mediates this expectation and thus the processing difficulty of a gender-mismatched word. To test this, we collect incremental processing data on three types of gender agreement mismatches in Russian: (i) past-tense verbs and subjects, (ii) attributive adjectives and nouns, (iii) predicate adjectives and nouns. We collect two types of incremental processing data: eye-tracking and Mouse-Tracking-for-Reading (MoTR), in which a participant reveals and reads text by moving their mouse, whose position is recorded. We find that while participants are surprised by ungrammatical conditions, this is mediated both by the type of agreement as well as the gender of the agreeing noun.