Agreement attraction is the well-documented, cross-linguistic phenomenon where a verb occasionally agrees not with its subject, as would be required by grammar, but instead with an unrelated noun that happens to match the verb’s relevant morphosyntactic features (’The key to the cabinets were ...’). Contenders for explaining agreement attraction fall into two broad classes: Purely morphosyntactic models specifically designed to explain agreement attraction, and more general sentence-processing models, such as the Lewis-Vasishth model, which explain attraction as a consequence of how linguistic structure is stored and accessed in memory. In the present research we disambiguate between these two classes by testing a surprising prediction made by the general sentence-processing models (but not the morphosyntactic models), namely that attraction should not be limited to morphosyntax but that semantic features of unrelated nouns can equally induce attraction. We conducted two single-trial forced-choice production experiments (N=1,072 and N=1,426) which both showed strong semantically induced attraction effects closely mirroring agreement attraction effects. While these results favor the general sentence processing models under consideration, we also present computational simulations showing clear shortcomings of one prominent member of this class. In sum, the results suggest that a hybrid model may be needed to explain the attraction phenomenon in its full breadth.

Gender stereotypes influence subjective beliefs about the world and this is reflected in our use of language. But do gender biases in language transparently reflect subjective beliefs? Or is the process translating thought to language itself biased? During the 2016 US (N=24,863) and 2017 UK (N=2,609) electoral campaigns, we compared participants’ beliefs about the gender of the next head of government with their use and interpretation of pronouns referring to the next head of government. In the US, even when the female candidate was expected to win, ’she’ references were rarely produced and induced substantial comprehension disruption. In the UK, where the incumbent female candidate was heavily favored, ’she’ was preferred in production but yielded no comprehension advantage. These and other findings suggest that the language system itself is as a source of implicit biases next to previously known biases such as those measured by the implicit association test.