Losing the thread: How three- and five-year-olds predict the outcome of a story when non-literal language is used to update events.

How do children learn to interpret nonliteral utterances nonliterally? A multidisciplinary report highlighting the role of reasoning about abstract meanings in nonliteral language comprehension is presented to shed light on why young children struggle to infer some, but not all, nonliteral meanings. An experimental paradigm using picture selection is then used to test differences in three- and five-year-old’s predictions when idioms are used to update stories. Norwegian-speaking children (N=162; N=86 females) are asked to predict story outcomes that are based on cognitive heuristics. The results show that five-year-olds are more likely than three-year-olds to choose literal interpretations of idioms as outcomes (δ = 0.12). Five-year-olds choose the correct outcome more when there is no literal outcome available (δ=0.15). The increase in literalism observed with age is explained through development in metalinguistic reflexivity. This suggests that children may increasingly hold speakers at their word, enabling access to abstract meanings over time.