This talk investigates how we generalize properties of kinds based on the morphosyntax of their subject. While previous studies have looked at the ambiguity between generic and non-generic interpretations, these experiments focus on the distinct interpretations of various generic subjects of count nouns in English. By focusing on the morphosyntactic features of number and definiteness, these studies take into consideration the linguistic variety that is available for generic subjects of count nouns. Further, they separate properties that are typically either principally or statistically connected to an animal kind.

I will present two match-to-sample studies with adult participants. Both found a main effect of connection type, where property was chosen more for principled than statistical connections, p<.0001. In Experiment 1, each item was introduced with a bare-plural generic question (Do you know about vorzyds? Vorzyds…), andall generic subjects differed from the control (p<.001), but there was no difference among them. Experiment 2 only provided a label in lieu of the introductory sentence (A vorzyd. Vorzyds…) and obtained the predicted propertywordingblock orderinteraction. This demonstrates that morphosyntax can be used to distinguish between different types of generalizations.

This study is currently run with child participants. In addition to the visual stimuli, the experiment includes audio stimuli reading out the generic statements. Preliminary data show that there is no difference within the generic subjects. The difference between the generic wording conditions and the control is smaller than in the adult studies. This replicates previous findings that have been attributed to a higher tendency for children to generalize any property in instructional settings.