Children and adults’ treatment of partial objects in bounded quantificational contexts

When asked to count under specific instructions, for example, to count “the forks,” or to identify the referent of “a fork,” children have been shown to treat whole forks and detached parts of forks on par, deviating systematically from adult behavior. A recent proposal links this behavior to the context-sensitive semantics of count nouns, which allows children and adults to allow partial objects to fall under the count noun extension depending on the demands of the context. Building on this, we test whether such contextual flexibility extends to quantificational linguistic contexts in which a modal imposes either an upper or a lower limit on a numerical expression in its structural scope (e.g. allowed to take three = “at most three,” have to have three = ’at least three’). We presented child and adult participants with a game scenario featuring these linguistic prompts and asked whether their treatment of partial objects systematically varies in relative to the boundedness conditions of the modal statement. Our results show that neither children (4;6–5;6) nor adults flexibly include or exclude partial objects to satisfy the bound conditions introduced by modals. Neither age group disregards the partial object for the purpose of satisfying an upper bound. However, in a context where the interaction of a modal and numerical expression induces a lower bound, adults assign partial objects a fractional value, while children simply treat them as wholes. These findings suggest that partial objects may be treated differently in a referential context versus in a quantificational context. In the latter case, adults—but not children—appear to access finer-grained measurement representations that support mapping quantities to fractional interpretations for enumeration.