Talk Title: Must be pragmatic: Child over-adherence to modal inferences (joint work with Ailís Cournane)
In this talk, I present the results of an experimental study done with 56 monolingual children aged between 3 and 6 y.o. and 9 dialect-matched adults (M = 35) from Sarajevo. The goal of the study was exploring whether preschool-aged children acquiring Bosnian/Croatian/Serbian (BCS) understand the syntactic cues that determine the flavor of morati ‘must’ as root (e.g., deontic obligation) or epistemic (e.g., inferential or evidential). Unlike the English must, which has been the focus of most previous studies on the acquisition of functional modals, morati has dedicated structures for root and epistemic uses. I will show syntactic arguments for a biclausal analysis of epistemic uses of modal verbs, and the results of a grammatical judgment questionnaire done on 25 adult native speakers of BCS that support it. Adults are categorical about (un)acceptability of sentences containing modal verbs in root or epistemic contexts.
In our experiment, participants engaged in a forced-choice picture selection modal flavor task done with English-learning children in Cournane (2015). Cournane has shown that children access epistemic interpretations from age 3 with structural cues for epistemicity, like progressive aspect (X must be Ying), but then gradually prefer epistemic interpretations of must regardless of structural cues and by age 5 treat must with a bare complement (X must Y) nearly uniformly as epistemic (Cournane & Perez-Leroux, subm.). To explain this we can appeal to the input, where must is primarily epistemic, and the availability of epistemic habitual readings for X must Y constructions. Cournane & Perez-Leroux find that even adults sometimes choose indirect evidence pictures to describe bare complement sentences (e.g. Michelle must swim) and ascribe the finding to these sentences having habitual/generic epistemic interpretations available (i.e., ‘…because Michelle has swimmer shoulders’).
Our results are similar to those found for English-speaking children and adults, which is surprising given the differences in the input, adults being categorical about grammaticality of different constructions in epistemic and root contexts, and the structural cues available in BCS compared to parallel English constructions. We argue that the account used for English cannot explain the same pattern in BCS. The pattern can, however, be accounted for in terms of pragmatics. We argue that by 5, both BCS and English children understand root and epistemic uses of must/morati, regardless of input and structural discrepancies; children differ significantly from adults by more frequently making inferences from obligations to actuality.