Anouk Dieuleveut · [S-Lab] Figuring out modal force: evidence from children’s production and input (Amsterdam Colloquium practice talk)
This paper investigates how children figure out the force of modals: that possibility modals (e.g. can, might) express possibility, and necessity modals (e.g. must, have to), necessity. Given that necessary p entails possible p, any situation where a necessity modal is used is logically compatible with a possibility interpretation. What then prevents children from hypothesizing possibility meanings for necessity modals? Can children rely on evidence from negative environments—which reverse the direction of logical entailments (Gualmini & Schwarz 2009)? Do they need a bias for strong meanings (in the spirit of Berwick 1985)? Or are there sufficient conversational cues from the context in which the modals appear to give away their force? We first present results from a corpus analysis of children’s modal input, and a related Human Simulation Paradigm (HSP) experiment (Gillette et al. 1999) in which participants have to guess the underlying force of a modal blanked out from a short dialogue between a child and her mother. Our results show that the conversational context is highly informative about force for both possibility and necessity modals, thus that a necessity bias is in principle not necessary, but that negative contexts are unhelpful for necessity modals. We then present results from a corpus analysis of children’s modal productions, and a related HSP experiment testing the adult-likeness of child modal uses. Our results suggest that children master possibility modals early: by age two, they use them productively, and in an adult-like way. Their mastery of necessity modals is however less clear: children produce them less frequently, and in a non adult-like way. We argue that this asymmetry may be explained by the input: possibility modals are more frequent than necessity modals in speech to children. Together, our results suggest that learners can infer a modal’s force based on conversational cues alone, and thus need not rely solely on negative environments, nor on in-built learning biases.