Children’s Interpretation of Negative Determiners as a Window into Q-Spreading

It is well-known that children, unlike adults, would frequently give a no-answer to questions like those in (1) and (2) if they are presented without overt linguistic context, but become significantly more successful if the stimuli are presented in a context enriched (or otherwise manipulated) in various ways (Crain et al. 1996, Philip and Lynch 2000, Drozd and van Loosbroek 1998, Sugisaki and Isobe 2001, Rakhlin 2004). The main questions with regards to quantifier spreading, which have not yet been fully resolved, are whether it is a manifestation of children’s misinterpreting or misanalyzing the universal quantifier in some way (via a mechanism that has been placed at the level of syntax (Roeper et al. 2004), semantics (Philip 1995), or syntax-to-semantics mapping (Geurts 2004)), and if not, what causes such errors. An issue that had not been previously addressed with regards to “classic spreading” is the role played by indefinites in triggering it, the avenue that I propose to pursue. The hypothesis that the present experiment is aimed at testing was put forward in Rakhlin 2004. According to that hypothesis, interpreting universally quantified sentences with indefinites in conditions of contextual asymmetry, particularly visual asymmetry, causes “classic spreading”, the term I will continue to use for convenience. I believe this approach can account for a wide range of existing experimental data, which was difficult to unify previously. One question that is especially intriguing and that can be answered straightforwardly within this account is why the children’s rate of errors declines if the picture shown